QGIS 2 Cookbook Become a QGIS power user and master QGIS data management, visualization, and spatial analysis techniques - Engenharia de Software (2025)

QGIS 2 Cookbook Become a QGIS power user and master QGIS data management, visualization, and spatial analysis techniques - Engenharia de Software (2)

QGIS 2 Cookbook Become a QGIS power user and master QGIS data management, visualization, and spatial analysis techniques - Engenharia de Software (3)

QGIS 2 Cookbook Become a QGIS power user and master QGIS data management, visualization, and spatial analysis techniques - Engenharia de Software (4)

QGIS 2 Cookbook Become a QGIS power user and master QGIS data management, visualization, and spatial analysis techniques - Engenharia de Software (5)

QGIS 2 Cookbook Become a QGIS power user and master QGIS data management, visualization, and spatial analysis techniques - Engenharia de Software (6)

QGIS 2 Cookbook Become a QGIS power user and master QGIS data management, visualization, and spatial analysis techniques - Engenharia de Software (7)

QGIS 2 Cookbook Become a QGIS power user and master QGIS data management, visualization, and spatial analysis techniques - Engenharia de Software (8)

QGIS 2 Cookbook Become a QGIS power user and master QGIS data management, visualization, and spatial analysis techniques - Engenharia de Software (9)

QGIS 2 Cookbook Become a QGIS power user and master QGIS data management, visualization, and spatial analysis techniques - Engenharia de Software (10)

QGIS 2 Cookbook Become a QGIS power user and master QGIS data management, visualization, and spatial analysis techniques - Engenharia de Software (11)

Prévia do material em texto

<p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS 2 Cookbook</p><p>Become a QGIS power user and master QGIS</p><p>data management, visualization, and spatial</p><p>analysis techniques</p><p>Alex Mandel</p><p>Víctor Olaya Ferrero</p><p>Anita Graser</p><p>Alexander Bruy</p><p>BIRMINGHAM - MUMBAI</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS 2 Cookbook</p><p>Copyright © 2016 Packt Publishing</p><p>All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or</p><p>transmitted in any form or by any means, without the prior written permission of the publisher,</p><p>except in the case of brief quotations embedded in critical articles or reviews.</p><p>Every effort has been made in the preparation of this book to ensure the accuracy of the</p><p>information presented. However, the information contained in this book is sold without</p><p>warranty, either express or implied. Neither the authors, nor Packt Publishing, and its dealers</p><p>and distributors will be held liable for any damages caused or alleged to be caused directly</p><p>or indirectly by this book.</p><p>Packt Publishing has endeavored to provide trademark information about all of the companies</p><p>and products mentioned in this book by the appropriate use of capitals. However, Packt</p><p>Publishing cannot guarantee the accuracy of this information.</p><p>First published: April 2016</p><p>Production reference: 1250416</p><p>Published by Packt Publishing Ltd.</p><p>Livery Place</p><p>35 Livery Street</p><p>Birmingham B3 2PB, UK.</p><p>ISBN 978-1-78398-496-1</p><p>www.packtpub.com</p><p>www.it-ebooks.info</p><p>www.packtpub.com</p><p>http://www.it-ebooks.info/</p><p>Credits</p><p>Authors</p><p>Alex Mandel</p><p>Víctor Olaya Ferrero</p><p>Anita Graser</p><p>Alexander Bruy</p><p>Reviewers</p><p>Jorge Arévalo</p><p>Olivier Dalang</p><p>Ben Mearns</p><p>Commissioning Editor</p><p>Pramila Balan</p><p>Acquisition Editor</p><p>Vinay Argekar</p><p>Content Development Editor</p><p>Amey Varangaonkar</p><p>Merint Mathew</p><p>Technical Editor</p><p>Dhiraj Chandanshive</p><p>Copy Editor</p><p>Priyanka Ravi</p><p>Project Coordinator</p><p>Suzanne Coutinho</p><p>Proofreader</p><p>Safis Editing</p><p>Indexer</p><p>Rekha Nair</p><p>Production Coordinator</p><p>Aparna Bhagat</p><p>Cover Work</p><p>Aparna Bhagat</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>About the Authors</p><p>Alex Mandel is a geospatial scientist who has a PhD in geography and more than 12 years</p><p>of experience in applying GIS to a variety of projects. He has also taught courses on GIS,</p><p>geospatial programming, and Geoweb.</p><p>Víctor Olaya Ferrero is a GIS developer. He is the creator and main developer of the QGIS</p><p>Processing Framework. He is also the author of Sistemas de Información Geográfica, a free</p><p>book about the fundamentals of GIS.</p><p>I would like to thank my co-authors and everyone else in the QGIS</p><p>community.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Anita Graser studied geomatics at the University of Applied Sciences Wiener Neustadt,</p><p>Austria, from where she graduated with a master's degree in 2010. During her study, she</p><p>gained hands-on experience in the fields of geo-marketing and transportation research. Since</p><p>2007, she has been working as a geographic information system (GIS) expert with the</p><p>dynamic transportation systems group at the Austrian Institute of Technology (AIT), where</p><p>she focuses on analyzing and visualizing spatio-temporal data.</p><p>Anita serves on the OSGeo board of directors and the QGIS project steering committee. She</p><p>has been working with GIS since 2005, provides QGIS training courses, and writes a popular</p><p>blog on open source GIS at anitagraser.com.</p><p>I would like to thank my family, partner, and coworkers for their support</p><p>and encouragement. Of course, I also want to thank the whole QGIS</p><p>community for their continued efforts to provide the best open source</p><p>GIS experience possible.</p><p>Alexander Bruy is a GFOSS advocate and open source developer working on the QGIS</p><p>project. He has also maintained a collection of his own open source projects. He has been</p><p>working with QGIS since 2006, and now he is an OSGeo charter member and QGIS core</p><p>developer. He is also the author of QGIS By Example, Packt Publishing.</p><p>Alexander is currently a freelance GIS developer and works for various companies worldwide.</p><p>www.it-ebooks.info</p><p>anitagraser.com</p><p>http://www.it-ebooks.info/</p><p>About the Reviewers</p><p>Jorge Arévalo is a computer engineer from Universidad Autónoma de Madrid, UAM. He</p><p>started developing web applications with JS, PHP, and Python in 2007. In 2010, he began</p><p>collaborating with PostGIS and GDAL projects after participating in GSoC 2009, creating the</p><p>PostGIS Raster GDAL driver. He currently works as a technology trainer and Python/Django</p><p>developer. He also organizes hackathons with others at http://hackathonlovers.com/.</p><p>Jorge Arévalo has co-written the book Instant Zurb Foundation 4 for Packt Publishing. He has</p><p>also worked as reviewer for the books PostGIS Cookbook, OpenLayers 3 Beginner's Guide,</p><p>and Getting Started with Memcached, all of them for Packt Publishing.</p><p>I want to thank my wife Elena for her continuous love and support while</p><p>reviewing this book.</p><p>Olivier Dalang completed his master's degree in architecture and urban planning from</p><p>EPFL, Switzerland. He then worked as an urban planner at Team+ as a volunteer for the</p><p>NGO urbaMonde, which is active in Senegal, and as a researcher and lecturer at EPFL on</p><p>the Venice Time Machine project. He got more and more acquainted with QGIS through the</p><p>different positions he worked in. Being a programmer, he developed a few plugins, of which</p><p>a few are now in the core.</p><p>www.it-ebooks.info</p><p>http://hackathonlovers.com/</p><p>http://www.it-ebooks.info/</p><p>Ben Mearns lives in Philly, PA, where he consults, teaches, advises, speaks, and creates</p><p>geographical information. In private practice, he has previously been the lead geospatial</p><p>information consultant and instructor on GIS for natural resource management at the University</p><p>of Delaware. He has held other GIS and data positions at the Cartographic Modeling Lab at the</p><p>University of Pennsylvania, Princeton University, and Macalester College. He is currently writing</p><p>QGIS Blueprints with Packt Publishing, which will soon be available in the market.</p><p>I would like to thank my girlfriend, Catherine Moore, and mother, Raiana</p><p>Mearns, for their support during the review of this book.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>www.PacktPub.com</p><p>eBooks, discount offers, and more</p><p>Did you know that Packt offers eBook versions of every book published, with PDF and ePub</p><p>files available? You can upgrade to the eBook version at www.PacktPub.com and as a print</p><p>book customer, you are entitled to a discount on the eBook copy. Get in touch with us at</p><p>customercare@packtpub.com for more details.</p><p>At www.PacktPub.com, you can also read a collection of free technical articles, sign up</p><p>for a range of free newsletters and receive exclusive discounts and offers on Packt books</p><p>and eBooks.</p><p>TM</p><p>https://www2.packtpub.com/books/subscription/packtlib</p><p>Do you need instant solutions to your IT questions? PacktLib is Packt's online digital book</p><p>library. Here, you can search, access, and read Packt's entire library of books.</p><p>Why Subscribe?</p><p>f Fully searchable across every book published by Packt</p><p>f Copy and paste, print, and bookmark content</p><p>f On demand and accessible via a web browser</p><p>www.it-ebooks.info</p><p>www.PacktPub.com</p><p>www.PacktPub.com</p><p>https://www2.packtpub.com/books/subscription/packtlib</p><p>http://www.it-ebooks.info/</p><p>i</p><p>Table of Contents</p><p>Preface v</p><p>Chapter 1: Data Input and Output 1</p><p>Introduction 1</p><p>Finding geospatial data on your computer 2</p><p>Describing data sources 6</p><p>Importing data from text files 10</p><p>Importing KML/KMZ files 16</p><p>Importing DXF/DWG files 17</p><p>Opening a NetCDF file 19</p><p>Saving a vector layer 21</p><p>Saving a raster layer 22</p><p>Reprojecting a layer 24</p><p>Batch format conversion 25</p><p>Batch reprojection 29</p><p>Loading vector layers into SpatiaLite 31</p><p>Loading vector layers into PostGIS 34</p><p>Chapter 2: Data Management 39</p><p>Introduction 39</p><p>Joining layer data 40</p><p>Cleaning up the attribute table 42</p><p>Configuring relations 45</p><p>Joining tables in databases 47</p><p>Creating views in SpatiaLite 48</p><p>Creating views in PostGIS 51</p><p>Creating spatial indexes 54</p><p>Georeferencing rasters</p><p>that</p><p>the names of the joined columns are too long for the 10 character-name length limit of the</p><p>shapefile format. QGIS ensures that all columns in the exported shapefiles have unique</p><p>names even after the names have been shortened to only 10 characters. To do this, QGIS</p><p>adds incrementing numbers to the end of, otherwise, duplicate column names. If you save</p><p>the join from this example as a shapefile, you will see that the column names are altered to</p><p>census_w_1, census_w_2, and so on. Of course, these names are less than optimal to</p><p>continue working with the data. As described in How it works... in this recipe, the names for</p><p>the joined columns are a combination of joined layer name and column name. Therefore, we</p><p>can use the following trick if we want to create a shapefile from the join: we can shorten the</p><p>layer name. Just rename the layer in the layer list. You can even have a completely empty</p><p>layer name! If you change the joined layer name to an empty string, the joined column</p><p>names will be _STATE, _COUNTY, and so on instead of census_wake2000_pop_STATE</p><p>and census_wake2000_pop_COUNTY. In any case, it is good practice to document your</p><p>data and provide a description of the attribute table columns in the metadata.</p><p>In any case, it is very likely that you will want to clean up the attribute table of the new</p><p>dataset, and this is exactly what we are going to do in the next exercise.</p><p>Cleaning up the attribute table</p><p>There are many reasons why we need to clean up attribute tables every now and then.</p><p>These may be because we receive badly structured or named data from external sources, or</p><p>because data processing, such as the layer joins that we performed in the previous exercise,</p><p>require some post processing. This recipe shows us how to use attribute table and the Table</p><p>Manager plugin to rename, delete, and reorder columns, as well as how to convert between</p><p>different data types using Field Calculator.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>43</p><p>Getting ready</p><p>If you performed the previous recipe, just save the joined layer to a new shapefile; otherwise,</p><p>load census_wake2000_pop.shp. In any case, you will notice that the dataset contains a</p><p>lot of duplicate information, and the column names could use some love as well. To follow</p><p>this recipe, you should also install and enable the Table Manager plugin by navigating to</p><p>Plugins | Manage and Install Plugins.</p><p>How to do it…</p><p>1. Our first step to clean up this dataset is to delete duplicated information. From all</p><p>available columns, we only want to keep _STATE, _COUNTY, _TRACT, FIPSSTCO,</p><p>TRT2000, STFID, _POP2000, AREA, and PERIMETER.</p><p>2. To delete the other columns, enable editing using the Toggle editing mode button in</p><p>the upper-left corner of the attribute table or by pressing Ctrl + E. This activates the</p><p>Delete column button.</p><p>3. Alternatively, you can also press Ctrl + L to open the Delete attributes dialog. This</p><p>dialog allows us to delete multiple columns at once. Just select all the columns that</p><p>you want to be deleted, press OK, and QGIS will display the reduced attribute table.</p><p>It's worth noting that the changes will only be permanent</p><p>once you use the Save edits button or disable the editing</p><p>mode and confirm that you want to save the changes.</p><p>4. Next, we will rename columns to remove the leading underscores in some of the</p><p>column names. This can be done using the Table Manager plugin.</p><p>5. When you start the plugin (edit mode should be disabled), you will see a list of the</p><p>layer columns. The plugin allows you to change the order of columns, as well as</p><p>rename, insert, clone, and delete columns.</p><p>6. To rename a column, just select it in the list and press the Rename button. You'll then</p><p>be asked to provide a new name. Go ahead and remove the leading underscores</p><p>from _STATE, _COUNTY, _TRACT, and _POP2000.</p><p>7. Finally, using the Move up and Move down buttons, you can also rearrange the</p><p>column order to something more intuitive. We'd suggest moving STFID to the first</p><p>position and AREA and PERIMETER to the last.</p><p>8. If you press Save, the changes will be saved back to the layer source file. Alternatively,</p><p>you can also create a new file using Save as....</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>44</p><p>How it works…</p><p>The steps provided in this exercise are mostly limited to layers with shapefile sources. If you</p><p>use other input data formats, such as MIF, GML, or GeoJSON files, you will notice that the</p><p>Toggle editing button is grayed out because these files cannot be edited in QGIS. Whether</p><p>a certain format can be edited in QGIS or not depends on which functionality has been</p><p>implemented in the respective GDAL/OGR driver.</p><p>The GDAL/OGR version that is used by QGIS is either part of the QGIS</p><p>package (as in the case of the Windows installers) or QGIS uses the GDAL</p><p>library existing in your system (on Linux and Mac). To get access to specific</p><p>drivers that are not supported by the provided GDAL/OGR version, it is</p><p>possible to compile custom versions of GDAL/OGR, but the details of</p><p>doing this are out of the scope of this cookbook.</p><p>There's more…</p><p>Another common task while dealing with attribute table management is changing column</p><p>data types. Currently, it is not possible to simply change the data type directly. Instead, we</p><p>have to use Field Calculator (which is directly accessible through the corresponding button</p><p>in the Attributes toolbar or from the attribute table dialog) to perform conversions and create</p><p>a new column for the result.</p><p>In our census_wake2000_pop.shp file, for example, the tract ID, TRACT, is stored in a</p><p>REAL type column with a precision of 15 digits even though it may be preferable to simply</p><p>have it in a STRING column and formatted to two digits after the decimal separator. To</p><p>create such a column using Field Calculator, we can use the following expression:</p><p>format_number("TRACT",2)</p><p>Compared to a simple conversion (which would be simple, use tostring("TRACT"),</p><p>format_number("TRACT",2) offers the advantage that all values will be formatted to</p><p>display two digits after the decimal separator, while a simple conversion would drop these</p><p>digits if they are zeros.</p><p>Of course, it's also common to convert from text to numerical. In this case, you can chose</p><p>between toint() and toreal().</p><p>See also</p><p>f Have a look through the conversion functions in the Field Calculator Function list</p><p>to see the other available functions that can deal with date and time data types.</p><p>Usage of all these functions is explained in Selected function help directly in the</p><p>calculator dialog.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>45</p><p>Configuring relations</p><p>In the Joining layer data recipe, we discussed that joins only append additional columns</p><p>to existing features (1:1 or n:1 relationships). Using joins, it is, therefore, not possible to</p><p>model 1:n relationships, such as "one zip code area containing n schools". These kinds of</p><p>relationships can instead be modeled using relations. This recipe introduces the concept of</p><p>relations and shows how you can put them to use.</p><p>Getting ready</p><p>To follow this exercise, load zip code areas and schools from zipcodes_wake.shp and</p><p>schools_wake.shp.</p><p>How to do it…</p><p>Relations are configured in Project Properties. The dialog is very similar to the join dialog:</p><p>1. Define the two layers (Referencing/Child and Referenced/Parent), as well as the</p><p>fields containing the common values/IDs. As you want to model "one zip code area</p><p>contains n schools," the zip code dataset is the parent layer and the school dataset is</p><p>the child layer. The connection between both datasets is established based on the zip</p><p>code fields (ADDRZIPCOD and ZIPNUM), as shown in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>46</p><p>2. To verify that the relation is set up and working, you can either check the attribute</p><p>table in form view (button in the lower-right corner), as shown in the following</p><p>screenshot, or open an individual</p><p>feature form. You will find that the relation</p><p>information has been appended at the end of the form:</p><p>As the preceding screenshot shows, setting up this relation enables you to get access</p><p>to all schools within a certain zip code in a very convenient way. As the edit button</p><p>suggests, it is even possible to edit the school data from this view. You can simply edit</p><p>the values in the table view. You can add and delete schools from the dataset using</p><p>the + and X buttons. The next two buttons enable you to quickly add new entries to</p><p>the relation or to remove them.</p><p>How it works…</p><p>In this example, removing a school from the dataset works just fine, but adding a school via this</p><p>dialog makes less sense because you cannot create a point geometry through this process.</p><p>If you press the button to add to the relation, you will get a dialog that allows you to choose</p><p>which existing school you want to add. In the background, the school's ADDRZIPCOD value is</p><p>updated to match the zip code we just assigned it to.</p><p>Similarly, if you select a school and press the button to remove the relation, what actually</p><p>happens is that the school's ADDRZIPCOD value is set to NULL.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>47</p><p>Joining tables in databases</p><p>If you use a database (SpatiaLite or PostGIS) to store your data, vector and nonspatial, then</p><p>you also have the option of using the database and SQL to perform tables joins. The primary</p><p>advantages of this method include being able to filter data before loading in the map, perform</p><p>multitable joins (three or more), and have full control over the details of the join via queries.</p><p>Getting ready</p><p>You'll need at least two layers in either a SpatiaLite or PostGIS database. These two layers</p><p>need at least one column in common, and the column in common should contain unique</p><p>values in at least one table. In this case, our example uses the census_wake_2000 polygon</p><p>layer and census_wake_2000_pop.csv.</p><p>How to do it…</p><p>1. Open the DB Manager plugin that comes with QGIS. You can find this in the</p><p>Database menu.</p><p>2. Select your database from the tree on the left-hand side, use cookbook.db in</p><p>SpatiaLite (which was created in Chapter 1, Data Input and Output).</p><p>If you don't see this database listed, use Add SpatiaLite</p><p>Layer (the icon or the menu item), or right-click on SpatiaLite</p><p>in the Browser window to make a new connection and add it</p><p>to an existing database.</p><p>3. Now, open the SQL window (the second icon from the left in top toolbar of the</p><p>plugin window).</p><p>4. Put in the following SQL code to query and JOIN the tables:</p><p>SELECT *</p><p>FROM census_wake2000 Sas a</p><p>JOIN census_wake2000_pop AS b</p><p>ON a.stfid = b.stfid;</p><p>How it works…</p><p>SELECT lists all the columns that you want from the source tables; in this case, * means</p><p>everything. FROM is the first (left) table, as a is an alias, which is used so that there's less</p><p>typing later. JOIN is the second (right) table, and ON indicates which columns to should be</p><p>matched between the two tables. The rest of how this works in relational database theory is</p><p>best explained in other texts.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>48</p><p>There's more…</p><p>In databases, there's more than one type of join. You can perform a join where you retain</p><p>only the matches in both tables, or you can retain all content from the left (first table) and any</p><p>matches from the right. You can also control how a one-to-many relationship is summarized or</p><p>select specific records instead of aggregating.</p><p>If you want to save the results of a query you have two options. You can make a view or a new</p><p>table. A view is a saved copy of your query. Every time you open it, the query will be rerun. This</p><p>is great if your data changes because it will always be up-to-date, and this doesn't use any</p><p>additional disk space. On the other hand, a table is like saving a new file; it becomes a static</p><p>new copy of the results. This is good to repeatedly access the same answer, and it is usually</p><p>faster to use, especially for large tables.</p><p>See also</p><p>f Refer to the Creating views in SpatiaLite and Creating views in PostGIS sections in</p><p>this chapter to learn how to make views of the query results.</p><p>f For more general information on writing SQL queries refer to http://sqlzoo.net/</p><p>f Refer to Chapter 1, Data Input and Output, about using the cookbook.db database</p><p>Creating views in SpatiaLite</p><p>In a database, view is a stored query. Every time you open it, the query is run and fresh results</p><p>are generated. To use views as layers in QGIS takes a couple of steps.</p><p>Getting ready</p><p>For this recipe, you'll need a query that returns results containing a geometry. The example</p><p>that we'll use is the query from the Joining tables in databases recipe (the previous recipe)</p><p>where attributes were joined 1:1 between the census polygons and the population CSV. The</p><p>QSpatiaLite plugin is recommended for this recipe.</p><p>How to do it…</p><p>The GUI method is described as follows:</p><p>1. Using the QspatiaLite plugin (which is in the Database menu, if you've activated it)</p><p>place the following in the query:</p><p>SELECT *</p><p>FROM census_wake2000 as a</p><p>www.it-ebooks.info</p><p>http://sqlzoo.net/</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>49</p><p>JOIN census_wake2000_pop as b</p><p>ON a.stfid = b.stfid;</p><p>2. From the Option dropdown, select the last choice, Create Spatial View & Load in</p><p>QGIS, and set the Geometry field box value to the name of your geometry field from</p><p>your spatial layer. In this example, this is geom.</p><p>You can explore your data table fields in the left-hand side</p><p>to check the name of the fields that you need.</p><p>The SQL method is as described, as follows:</p><p>1. In Database | DB Manager, open SQL Window.</p><p>2. Write a query. In this example, this is the Join query from the previous recipe.</p><p>3. Convert this query to a view by adding CREATE VIEW as SELECT:</p><p>CREATE VIEW census_wake2000_pop_join AS</p><p>SELECT *</p><p>FROM census_wake2000 as a</p><p>JOIN census_wake2000_pop as b</p><p>ON a.stfid = b.stfid;</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>50</p><p>4. Register the view with the SpatiaLite metadata backend with a follow up query. This</p><p>function is case sensitive:</p><p>CREATE VIEW census_wake2000_pop_join AS</p><p>INSERT INTO views_geometry_columns</p><p>(view_name, view_geometry, view_rowid, f_table_name,</p><p>f_geometry_column,read_only)</p><p>VALUES ('census_wake2000_pop_join', 'geom', 'rowid',</p><p>'census_wake2000', 'geom',1);</p><p>This only works when the view geometry is based on the</p><p>geometry of a single table. If you need to generate new</p><p>geometries, you probably need a table.</p><p>5. The pattern is ('name of view','name of view geometry field','A Unique ID','name</p><p>of table the view gets its geometry from','name of geometry field in the original</p><p>table',read-only (1) or writable(0)).</p><p>6. After running the second query, you should be able to load the view in QGIS and see</p><p>the same fields as the join query.</p><p>How it works…</p><p>A view is actually stored in the database and is triggered when you load it. In this way, if you</p><p>change the original data tables, the view will always be up to date. By comparison, creating new</p><p>tables makes copies of the existing data, which is stored in a new place, or creates a snapshot</p><p>or freeze of the values at that time. It also increases the database's size by replicating data.</p><p>Whereas, a view is just the SQL text itself and doesn't store any additional data.</p><p>QGIS reads the metadata tables of SpatiaLite in order to figure out what layers contain</p><p>spatial data, what kind of spatial data they contain, and which column contains the geometry</p><p>definition. Without creating entries in the metadata, the tables appear as normal SQLite</p><p>tables, and you can only load attribute data without spatial representation.</p><p>As it's a view, it's really reading the geometries from the original tables. Therefore, any edits</p><p>to the original table will show up. New in SpatiaLite 4.x series, this makes it easier to create</p><p>writable views. If you use the spatialite-gui standalone application, it registers all the database</p><p>triggers needed</p><p>to make it work, and the changes made will affect the original tables.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>51</p><p>There's more…</p><p>You don't have to use ROWID as unique id, but this is a convenient handle that always exists</p><p>in SQLite, and unlike an ID from the original table, there's no chance of duplication in an</p><p>aggregating query.</p><p>See also</p><p>f Read more about writable-view at https://www.gaia-gis.it/fossil/</p><p>libspatialite/wiki?name=writable-view. This recipe is extremely similar to</p><p>the next one on PostGIS and demonstrates how interchangeable the two can be if you</p><p>are aware of the slight differences.</p><p>Creating views in PostGIS</p><p>In a database, a view is a stored query. Every time that you open it, the query is run and</p><p>fresh results are generated. To use views as layers in QGIS takes a couple of steps.</p><p>Getting ready</p><p>For this recipe, you'll need a query that returns results containing a geometry. The example</p><p>that we'll use here is the query from the Joining tables in databases recipe where attributes</p><p>were joined 1:1 between the census polygons and the population CSV.</p><p>How to do it…</p><p>The SQL method is described as follows:</p><p>1. In Database | DB Manager, open SQL Window.</p><p>2. Write a query; in this example, this is the join query that was written in the previous</p><p>exercise. If you want to see it right away but not necessarily retain it, check the Load</p><p>as new layer checkbox near the bottom:</p><p>SELECT *</p><p>FROM census_wake2000 as a</p><p>JOIN census_wake2000_pop as b</p><p>ON a.stfid = b."STFID";</p><p>www.it-ebooks.info</p><p>https://www.gaia-gis.it/fossil/libspatialite/wiki?name=writable-view</p><p>https://www.gaia-gis.it/fossil/libspatialite/wiki?name=writable-view</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>52</p><p>3. Now, execute the query by clicking on the Execute (F5) button:</p><p>4. After executing the query, to load it to the map check the Load as new layer box,</p><p>which will expand some additional options. Pick your unique integer (id_0) for</p><p>Column with unique integer values and geom for Geometry column. Name your</p><p>result in the Layer name (prefix) textbox and click on Load now!.</p><p>If you only needed to see this data in this particular QGIS</p><p>project, you can stop here. In order to make the database store</p><p>this query for other projects and users, continue this recipe.</p><p>5. Convert this query to a view by adding CREATE VIEW AS SELECT:</p><p>CREATE VIEW census_wake2000_pop_join AS SELECT *</p><p>FROM census_wake2000 as a</p><p>JOIN census_wake2000_pop as b</p><p>ON a.stfid = b."STFID";</p><p>6. Go back to DB Manager and hit the Refresh button (on the left). You should now see</p><p>your new view listed and be able to add it to the map.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>53</p><p>How it works…</p><p>QGIS reads the metadata tables or views of PostGIS in order to figure out what layers contain</p><p>spatial data, what kind of spatial data they contain, and which column contains the geometry</p><p>definition. Without creating entries in the metadata, the tables appear as normal PostgreSQL</p><p>tables, and you can only load attribute data without spatial representation.</p><p>As this is a view, it's really reading the geometries from the original tables. Therefore, any edits</p><p>to the original table will also show up.</p><p>There's more…</p><p>QGIS is really picky about having a unique ID for PostGIS tables and views. There are a few</p><p>tips to make this always work. Always include a numeric unique ID (as the first column is</p><p>recommended but not required, IDs must be integer columns (usually int4, but int8 should</p><p>work now too). Autoincrementing IDs are good idea. When you don't have such an ID field to</p><p>use from one of the underlying tables, you can add an ID on the fly with the following:</p><p>SELECT row_number() OVER() AS id_qgis, FROM table;</p><p>The downside of this is that you now have to list out all the fields that you want to use in the</p><p>view rather than using *. When creating tables, you'll want to turn this id_qgis field into an</p><p>auto-incrementing field if you plan to add records.</p><p>The other big catch is that if you make a new geometry by manipulating existing geometries,</p><p>QGIS isn't always aware of the results. In the previous example, the geometry is just</p><p>passed from the original table to the view unchanged, so it is properly registered in the</p><p>geometry_columns metadata of PostGIS. However, a new geometry doesn't exist in</p><p>the original table, so the trick is to cast the geometry result, as follows:</p><p>CREATE VIEW census_wake2000_4326 AS</p><p>SELECT id_0,</p><p>stfid,tractid,ST_Transform(geom,4326)::geometry(GeometryZ, 4326)</p><p>As geom</p><p>FROM census_wake2000;</p><p>QGIS doesn't always think that this is a valid spatial layer but adding to the Canvas should work.</p><p>The more specific you can be, the better. If you're not sure what</p><p>geometry type it is or if you have 3D (aka Z), check the entries in</p><p>the geometry_columns view.</p><p>Also, keep your eyes on Postgres's relatively new feature called Materialized Views. This is a</p><p>method of caching view results that don't update automatically, but they also don't require</p><p>whole new tables.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>54</p><p>See also</p><p>f Finer details from the PostGIS manual can be read at http://postgis.</p><p>refractions.net/docs/using_postgis_dbmanagement.html#Manual_</p><p>Register_Spatial_Column. This recipe is extremely similar to the previous one</p><p>and demonstrates how interchangeable these two can be if you are aware of the</p><p>slight differences.</p><p>f Read more about Materialized Views at http://www.postgresql.org/</p><p>docs/9.3/static/rules-materializedviews.html</p><p>Creating spatial indexes</p><p>Spatial indexes are methods to speed up queries of geometries. This includes speeding up</p><p>the display of database layers in QGIS when you zoom in close (it has no effect on viewing</p><p>entire layers).</p><p>This recipe applies to SpatiaLite and PostGIS databases. In the event that you've made a new</p><p>table or you have imported some data and didn't create a spatial index, it's usually a good</p><p>idea to add this.</p><p>You can also create a spatial index for shapefile layers. Take a look at</p><p>Layer Properties | General for the Create Spatial Index button. This will</p><p>create a .qix file that works with QGIS, Mapserver, GDAL/OGR, and other</p><p>open source applications. Refer to https://en.wikipedia.org/</p><p>wiki/Shapefile.</p><p>Getting ready</p><p>You'll need a SpatiaLite and a Postgis database. For ease, import a vector layer from the</p><p>provided sample data and do not select the Create spatial index option when importing. (Not</p><p>sure how to import data? Refer to Chapter 1, Data Input and Output, for how to do this.)</p><p>How to do it…</p><p>Using the DB Manager plugin (in the Database menu), perform the following steps:</p><p>1. Check whether the index does not exist. In DB Manager, open the database and then</p><p>open the table that you want to check. Looking at the properties on the right, you</p><p>should see a message just above Fields that looks like this:</p><p>www.it-ebooks.info</p><p>http://postgis.refractions.net/docs/using_postgis_dbmanagement.html#Manual_Register_Spatial_Column</p><p>http://postgis.refractions.net/docs/using_postgis_dbmanagement.html#Manual_Register_Spatial_Column</p><p>http://postgis.refractions.net/docs/using_postgis_dbmanagement.html#Manual_Register_Spatial_Column</p><p>http://www.postgresql.org/docs/9.3/static/rules-materializedviews.html</p><p>http://www.postgresql.org/docs/9.3/static/rules-materializedviews.html</p><p>https://en.wikipedia.org/wiki/Shapefile</p><p>https://en.wikipedia.org/wiki/Shapefile</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>55</p><p>2. However, what if no index was listed for the geom column? Then, we can make one</p><p>just by clicking the create it link. Or you can do this in a SQL window, as follows:</p><p>� For SpatiaLite, use the following:</p><p>SELECT CreateSpatialIndex('schools_wake', 'geom');</p><p>� For PostGIS, use the following:</p><p>CREATE INDEX sidx_census_wake2000_geom</p><p>ON public.census_wake2000 USING gist(geom);</p><p>3. Verify that the index exists, as follows:</p><p>� For PostGIS (the left-hand side of the following screenshot), on the right-hand</p><p>side, scroll to the</p><p>bottom looking for the Indexes section</p><p>� For SpatiaLite (the right-hand side of the following screenshot), you can see</p><p>the idx_nameoftable_geomcolumn listed as a table:</p><p>How it works…</p><p>When you create a spatial index, the database stores a bounding box rectangle for every</p><p>spatial object in the geometry column. These boxes are also sorted so that boxes near each</p><p>other in coordinate space are also near each other in the index.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>56</p><p>When queries are run involving a location, a comparison is made against the boxes, which is a</p><p>simple math comparison. Rows with boxes that match the area in question are then selected</p><p>to be tested in depth for a precise match, based on their real geometries. This method of</p><p>searching for intersection is faster than testing complex geometries one by one because it</p><p>quickly eliminates items that are clearly not near the area of interest.</p><p>There's more…</p><p>Spatial indexes are really important to speed up the loading time of database spatial layers</p><p>in QGIS. They also play a critical role in the speed of spatial queries (such as intersects).</p><p>Note that PostGIS will automatically use a spatial index if one is present. SpatiaLite requires</p><p>that you write queries that intentionally call a particular spatial index (Refer to Haute Cuisine</p><p>examples from the SpatiaLite Cookbook)</p><p>Also, keep in mind that only one spatial index per table can be used in a single query. This</p><p>really comes into play if you happen to have more than one spatial column or create a spatial</p><p>index in a different projection than the geometry (check out the PostGIS Cookbook by Packt</p><p>Publishing for more information).</p><p>If you plan to insert many records into a table with an existing spatial index,</p><p>you may want to disable or drop the index and recreate it after the import is</p><p>done. Otherwise, the index will be recalculated after each row is inserted. This</p><p>applies to nonspatial indexes too.</p><p>Do you want to check lots of tables at once? You can list all GIST indexes in PostGIS at once:</p><p>SELECT i.relname as indexname, idx.indrelid::regclass as</p><p>tablename,</p><p>am.amname as typename,</p><p>ARRAY(SELECT pg_get_indexdef(idx.indexrelid, k + 1, true)</p><p>FROM generate_subscripts(idx.indkey, 1) as k</p><p>ORDER BY k</p><p>) as indkey_names</p><p>FROM pg_index as idx</p><p>JOIN pg_class as i ON i.oid = idx.indexrelid</p><p>JOIN pg_am as am ON i.relam = am.oid</p><p>JOIN pg_namespace as ns ON ns.oid = i.relnamespace</p><p>AND ns.nspname = ANY(current_schemas(false))</p><p>Where am.amname Like 'gist';</p><p>To do something similar in SpatiaLite, use the following:</p><p>SELECT * FROM geometry_columns WHERE spatial_index_enabled = 1;</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>57</p><p>See also</p><p>f Information on SpatiaLite spatial index implementation can be found at https://</p><p>www.gaia-gis.it/fossil/libspatialite/wiki?name=SpatialIndex</p><p>f More details on using spatial indexes can be found at https://www.gaia-gis.</p><p>it/fossil/libspatialite/wiki?name=SpatialIndex</p><p>f Information about PostGIS implementation is at http://postgis.net/docs/</p><p>manual-2.0/using_postgis_dbmanagement.html#gist_indexes</p><p>f You can also check out Chapter 10, Maintenance, Optimization, and Performance</p><p>Tuning, of PostGIS Cookbook by Packt Publishing,</p><p>Georeferencing rasters</p><p>Sometimes, you have a paper map, an image of a map from the Internet, or even a raster</p><p>file with projection data included. When working with these types of data, the first thing you'll</p><p>need to do is reference them to existing spatial data so that they will work with your other data</p><p>and GIS tools. This recipe will walk you through the process to reference your raster (image)</p><p>data, called georeferencing.</p><p>Getting ready</p><p>You'll need a raster that lacks spatial reference information; that is, unknown projection</p><p>according to QGIS. You'll also need a second layer (reference map) that is known and you can</p><p>use for reference points. The exception to this is, if you have a paper map that has coordinates</p><p>marked on it or a spatial dataset that just didn't come with a reference file but you happen to</p><p>know its CRS/SRS definition. Load your reference map in QGIS.</p><p>This book's data includes a scanned USGS topographic map that's missing its o38121e7.tif</p><p>projection information. This map is from Davis, CA, so the example data has plenty of other</p><p>possible reference layers you could use, for example, the streets would be a good choice.</p><p>Actually, the world file was just renamed to o38121e7.tfw.orig</p><p>so that QGIS wouldn't detect it. You can use this later to compare your</p><p>georeference quality.</p><p>www.it-ebooks.info</p><p>https://www.gaia-gis.it/fossil/libspatialite/wiki?name=SpatialIndex</p><p>https://www.gaia-gis.it/fossil/libspatialite/wiki?name=SpatialIndex</p><p>https://www.gaia-gis.it/fossil/libspatialite/wiki?name=SpatialIndex</p><p>https://www.gaia-gis.it/fossil/libspatialite/wiki?name=SpatialIndex</p><p>http://postgis.net/docs/manual-2.0/using_postgis_dbmanagement.html#gist_indexes</p><p>http://postgis.net/docs/manual-2.0/using_postgis_dbmanagement.html#gist_indexes</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>58</p><p>How to do it…</p><p>On the Raster menu, open the Georeferencing tool and perform the following steps:</p><p>1. Use the file dialog to open your unknown map in the Georeferencing tool.</p><p>2. Create a Ground Control Point (GCP) of matches between your start coordinates and</p><p>end coordinates.</p><p>Building corners, street intersections, and things where line</p><p>features intersect or significant edge features can be found.</p><p>3. Add a point in your unknown map with GCP Add +. You can now enter the coordinates</p><p>(that is, if it's a paper map with known coordinates marked on it), or you can select a</p><p>match from the main QGIS window reference layer.</p><p>4. Repeat this process to find at least four matches. If you want to get a really good fit</p><p>do between 10-20 matches.</p><p>5. (Optional) Save your GCPs as a text file for future reference and troubleshooting:</p><p>Try to spread out your control points so that you have good</p><p>coverage of the whole map. It's all about averaging the</p><p>differences.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>59</p><p>6. Now, choose Transformation Settings, as follows:</p><p>7. You have a choice here. Generally, you'll want to use Polynomial. If you set 4+ points</p><p>for the first order, 6+ points for the second order and 10+ points for the third order,</p><p>The second order is the currently recommend one. This will be discussed in the</p><p>There's more… section of this recipe.</p><p>8. Set Target SRS to the same projection as the reference layer. (In this case, this is</p><p>EPSG:26910 UTM Zone 10n)</p><p>9. Output Raster should be a different name from the original so that you can easily</p><p>identify it.</p><p>Save your GCP list to the file. If you don't like the results,</p><p>come back and try a different algorithm or change</p><p>the number of GCPs used. If you want a reference for</p><p>comparison, look at the text o38121e6.tif.points</p><p>file in this book's data folder.</p><p>10. When you're happy with your list of GCPs click on Start Georeferencing in File or on</p><p>the green triangular button.</p><p>How it works…</p><p>A mathematical function is created based on the differences between your two sets of points.</p><p>This function is then applied to the whole image, stretching it in an attempt to fit. This is</p><p>basically a translation or projection from one coordinate system to another.</p><p>There's more…</p><p>Picking transformation types can be a little tricky, the list in QGIS is currently in alphabetical</p><p>order and not the recommended order. Polynomial 2 and Thin-plate-spline (TPS) are probably</p><p>the two most common choices. Polynomial 1 is great when you just have minor shift, zooming</p><p>(scale), and rotation. When you have old well-made maps in consistent projections, this will</p><p>apply the least amount of change. Polynomial 2 picks up from here and handles consistent</p><p>distortion. Both of these provide you with an error estimate as the Residual or RMSE (Root</p><p>Mean Square Error). TPS handles variable distortion, varying it's correction around each</p><p>control point. This will</p><p>almost always result in the best fit, at least through the GCPs that</p><p>you provide. However, because it varies at every GCP location, you can't calculate an error</p><p>estimate and it may actually overfit (create new distortion). TPS is best for hand-drawn maps,</p><p>nonflat scans of maps, or other variable distorted sources. Polynomial methods are good for</p><p>sources that had high accuracy and reference marks to begin with.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>60</p><p>If you really want a good match, once you have all your points, check the RMSE values in the</p><p>table at the bottom. Generally, you want this near or less than 1. If you have a point with a</p><p>huge value, consider deleting it or redoing it. You can move existing points, and a line will be</p><p>drawn in the direction of the estimated error. So, go back over the high values, zoom in extra</p><p>close, and use the GCP move option.</p><p>Sometimes, just changing your transformation type will help, as shown in the following</p><p>screenshot that compares Polynomial 1 versus Polynomial 2 for the same set of GCP:</p><p>Polynomial 1</p><p>Note the residual values difference when changing to Polynomial 2 (assuming that you have</p><p>the minimum number of points to use Polynomial 2):</p><p>Polynomial 2</p><p>Resampling methods can also have a big impact on how the output looks.</p><p>Some of the methods are more aggressive about trying to smooth out</p><p>distortions. If you're not sure, stick with the default nearest neighbor. This</p><p>will copy the value of the nearest pixel from the original to a new square</p><p>pixel in the output.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>61</p><p>See also</p><p>f When performing georeferencing in a setting where you need it to be very accurate</p><p>(science and surveying), you should read up on the different transformations and</p><p>what RMSE values are good for your type of data. Refer to the general GIS or Remote</p><p>Sensing textbooks for more information.</p><p>f For full details of all the features of the QGIS georeferencer, refer to the online</p><p>manual at http://docs.qgis.org/2.8/en/docs/user_manual/plugins/</p><p>plugins_georeferencer.html.</p><p>f The QGIS documentation has some basic information about how to pick</p><p>transformation type at http://docs.qgis.org/2.8/en/docs/user_manual/</p><p>plugins/plugins_georeferencer.html#available-transformation-</p><p>algorithms.</p><p>Georeferencing vector layers</p><p>For various reasons, sometimes you have a vector layer that lacks projection information.</p><p>This is often the case with CAD layers that were created only in local coordinates. When it is</p><p>possible, try to track down the original projection information. As a last resort, you can attempt</p><p>to warp the vector layer to match a known reference layer with the recipe described here.</p><p>Getting ready</p><p>You can open two instances of QGIS (or use one as you'll just be zooming back and forth a</p><p>lot). In one instance, load a reference layer, something in the projection that you want your</p><p>data to be in. Activate Coordinate Capture Plugin from the Manage Plugins menu.</p><p>In Windows, you need the osgeo4w shell for this recipe. If you don't have</p><p>a start menu item, look for the OSGeo4W.bat launcher in your QGIS or</p><p>OSGeo4w installation folder.</p><p>This example uses cad-lines-only.shp, which is the line layer extracted from the</p><p>CSS-SITE-CIV.dxf file. This file is a CAD rendering of design plans for Academy St.</p><p>in the town of Cary, Wake County, North Carolina.</p><p>www.it-ebooks.info</p><p>https://www.qgis.org/en/docs/user_manual/plugins/plugins_georeferencer.html</p><p>https://www.qgis.org/en/docs/user_manual/plugins/plugins_georeferencer.html</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/plugins/plugins_georeferencer.html#available-transformation-algorithms</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/plugins/plugins_georeferencer.html#available-transformation-algorithms</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/plugins/plugins_georeferencer.html#available-transformation-algorithms</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>62</p><p>How to do it…</p><p>1. Create a list of GCP matches between your unknown layer (cad-lines-only.shp)</p><p>and your reference layer (CarystreetsND83NC.shp).</p><p>2. Here are some specific adjustments to help with cad-lines-only.shp referenced</p><p>to CarystreetsND83NC.shp. These will make it easier to find matches between</p><p>the two layers:</p><p>1. Load cad-lines-only.shp, and adjust its style properties using a</p><p>rule-based style. Use the "Layer" = 'C-ROAD-CNTR' rule, which will only</p><p>show you street centerlines.</p><p>2. In your other QGIS session, load CarystreetsND83NC.shp in order to</p><p>find the matching area, open the attribute table, and apply the following</p><p>select expression: "Street" LIKE '%N ACADEMY%' OR "Street"</p><p>LIKE '%S ACADEMY%' OR "Street" LIKE '%CHATHAM%'. The filter</p><p>here highlights the three main streets of the original project, which is at the</p><p>intersection of Chatham and N/S Academy streets in the center of the town.</p><p>This may also be useful to change the color of the selected features to</p><p>make it easier to find. The traffic circles at either end of the project are</p><p>good landmarks:</p><p>3. Find an easy-to-identify feature that matches in both layers</p><p>(street intersections).</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>63</p><p>4. Use the coordinate capture plugin to copy the x,y value for the point in</p><p>both layers.</p><p>5. Save the coordinates in a text editor while you work.</p><p>6. Repeat this procedure until you have at least four pairs of points. Try to pick</p><p>points spread out across the whole layer:</p><p>There is currently no graphical interface in QGIS for the next</p><p>step, which uses the OGR library that comes with QGIS. Take</p><p>the list of points and using the ogr2ogr command-line, you're</p><p>going to apply the GCP to the unknown layer.</p><p>3. Each set of coordinate pairs will look as follows:</p><p>-gcp sourceX sourceY destinationX destinationY</p><p>4. Open a terminal (Mac or Linux) or an OSGeo4w shell (Windows).</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>64</p><p>5. Change to the directory where you have the data (Hint: cd /home/user/</p><p>Qgis2Cookbook/):</p><p>ogr2ogr -a_srs EPSG:3358 -gcp 2064886.09740 741552.90836</p><p>629378.595 226024.853 -gcp 2066610.97021 741674.39817</p><p>629903.420 226064.049 -gcp 2064904.46214 743055.63847</p><p>629384.784 226485.725 -gcp 2062863.85707 741337.65243</p><p>628762.587 225960.900 cad_lines_nd83nc.shp cad-lines-only.shp</p><p>-a_srs is the proj code for your reference layer.</p><p>The command pattern is ogr2ogr .</p><p>Other useful advanced options include -order to</p><p>indicate polynomial level (default is based on the number</p><p>of GCPs) or -tps to use Thin-plate-spline instead of</p><p>polynomial. For more options refer to http://www.</p><p>gdal.org/ogr2ogr.html.</p><p>6. Now, load your new cad_lines_nd83nc.shp file in the same project, as</p><p>CarystreetsND83NC.shp. They should line up without the need to enable</p><p>projection-on-the-fly:</p><p>www.it-ebooks.info</p><p>http://www.gdal.org/ogr2ogr.html</p><p>http://www.gdal.org/ogr2ogr.html</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>65</p><p>How it works…</p><p>Given the list of input coordinates and matching output coordinates, a math formula is derived</p><p>to translate between the two sets. This formula is then applied to all the points in the original</p><p>data. The result of this is a reprojected dataset from an unknown projection to a known</p><p>projection.</p><p>The original data is actually EPSG:102719, but we're pretending that we didn't</p><p>have this piece of information to demonstrate this example.</p><p>There's more…</p><p>When picking a reference layer, try to pick something in the projection that you want to use</p><p>for your maps and analysis. That way you only have to reproject once, as each additional</p><p>transformation can add an error. There's also more than one way to go about accomplishing</p><p>this task, including moving the data by hand.</p><p>In this particular, example the transformation is autoselected based on the number of GCP</p><p>point pairs. 4-5 is the first order polynomial, 6-9 is the second order polynomial, and 10+ is</p><p>the third order polynomial. Refer to the previous recipe</p><p>in this chapter for more information.</p><p>A related topic is Affine transformations when you simply want to shift or rotate a vector layer</p><p>by a known amount. The QgsAffine plugin is great if you already know the parameters, or</p><p>roughly know how far you want to rotate and shift the vector layer, as it then just needs</p><p>some math to get the parameters.</p><p>Maybe by the time you read this, all of the difficult things here will be worked</p><p>in a plugin. Keep an eye open, and try the experimental plugins Vector</p><p>Bender, vectorgeoref, and Affine Transformations.</p><p>See also</p><p>f This method is very similar to the Georeferencing Rasters recipe and many of the</p><p>same tips apply to both</p><p>f If you want to see how we got the CAD file into an SHP to begin with, look at Importing</p><p>DXF/DWG files in the Chapter 1, Data Input and Output</p><p>f See the Using Rule Based Rendering recipe in Chapter 10, Cartography Tips, for tips</p><p>on how to visualize the resulting CAD import better by applying attribute based rule</p><p>filtering</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>66</p><p>f Too lazy to do the math? You can also just use GvSig to do the math and make a</p><p>world file; refer to http://foss4gis.blogspot.com/2011/05/computing-</p><p>and-applying-affine.html</p><p>f If you want to do the math yourself see http://press.underdiverwaterman.</p><p>com/rotating-a-point-grid-in-qgis/</p><p>Creating raster overviews (pyramids)</p><p>Overviews, or pyramids, and resampling are all about making raster layers load faster when</p><p>zooming and panning in your map canvas, by reducing the amount of data loaded when not</p><p>zoomed in all the way.</p><p>Getting ready</p><p>You will need a large raster image.</p><p>Generally, you want to make a copy of the data as this method will likely</p><p>alter the original file if you choose to make 'internal' pyramids (easy to do</p><p>on accident).</p><p>How to do it…</p><p>1. Load your raster in QGIS. elev_lid792_1m.tif will work fine for this example.</p><p>2. Right-click on the layer name and open Properties.</p><p>3. Go to the Pyramids item on the left:</p><p>www.it-ebooks.info</p><p>ftp://199.72.17.76/Engineering/Academy%20Street%20Improvements%202013/DWGs/CSS-WATER%20&%20SEWER/CSS-SITE-CIV.dwg</p><p>ftp://199.72.17.76/Engineering/Academy%20Street%20Improvements%202013/DWGs/CSS-WATER%20&%20SEWER/CSS-SITE-CIV.dwg</p><p>ftp://199.72.17.76/Engineering/Academy%20Street%20Improvements%202013/DWGs/CSS-WATER%20&%20SEWER/CSS-SITE-CIV.dwg</p><p>http://foss4gis.blogspot.com/2011/05/computing-and-applying-affine.html</p><p>http://foss4gis.blogspot.com/2011/05/computing-and-applying-affine.html</p><p>http://press.underdiverwaterman.com/rotating-a-point-grid-in-qgis/</p><p>http://press.underdiverwaterman.com/rotating-a-point-grid-in-qgis/</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>67</p><p>4. Select the image sizes that you want to create pyramids for:</p><p>� Optionally, choose whether to store externally (safer) or internally (less files</p><p>to keep track of).</p><p>� Optionally, choose a resampling algorithm; Nearest Neighbor is the simplest,</p><p>but other methods may look smoother at the cost of more data manipulation</p><p>and compute time</p><p>5. Click on Build pyramids.</p><p>6. When this is completed, you'll notice the red X on the sizes that you picked will now</p><p>show a pyramid.</p><p>How it works…</p><p>Generating pyramids essentially makes copies of your original data resized for different</p><p>zoom levels. As you zoom out, the original data is resampled to fit the size of the screen. The</p><p>pyramids do the same thing, but they let you decide what resampling method to use and</p><p>generate this overview ahead of time. By generating them ahead of time, QGIS can load the</p><p>image faster when you change zoom levels.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>68</p><p>There's more…</p><p>Resampling is a fancy way of saying that at each zoom level that is now 1 pixel is more than 1</p><p>pixel from the original data, so they need to be averaged in some way and the result assigned</p><p>to the 1 pixel that is now available. Each of the different methods uses a different math</p><p>formula to decide the new value and how much to smooth that value with neighboring pixels</p><p>(so that it looks aesthetically pleasing). This is the same concept as when you shrink pictures</p><p>so that you can e-mail them to your friends.</p><p>If you chose to save them externally, your overviews are stored in elev_lid792_1m.tif.</p><p>ovr. Some other programs store the same thing in the .aux files; however, pyramid formats</p><p>are not universally compatible between GIS applications.</p><p>See also</p><p>f This is the same effect as using the GDAL gdaladdo command; refer to</p><p>http://gdal.org/gdaladdo.html</p><p>f More details from the QGIS documentation can be found at https://docs.</p><p>qgis.org/2.8/en/docs/user_manual/working_with_raster/raster_</p><p>properties.html</p><p>Building virtual rasters (catalogs)</p><p>When you have a lot of rasters (instead of one big raster) that are all part of the same dataset</p><p>(typically adjacent to each other), you don't want to load each file individually and then style</p><p>it. It's much easier to load one file and treat it as one layer. This recipe lets you do this without</p><p>actually creating a single monstrous raster, which can be difficult to work with.</p><p>Getting ready</p><p>You will need two or more raster files that have adjacent extents or only overlap partially</p><p>around the edges and are in the same projection. Ideally, the files should be of the same type,</p><p>such as all elevations, all air photos, and so on. For this recipe, the elevation rasters from the</p><p>OSGeo EDU (North Carolina) dataset will work.</p><p>How to do it…</p><p>1. (Optional) Load the elevation rasters to your current map.</p><p>2. Go to Raster Menu | Miscellanous | Build Virtual Raster (Catalog).</p><p>3. Check the Use visible raster layers checkbox or choose SELECT, browse to the</p><p>example data, and select all four.</p><p>www.it-ebooks.info</p><p>http://gdal.org/gdaladdo.html</p><p>https://docs.qgis.org/2.8/en/docs/user_manual/working_with_raster/raster_properties.html</p><p>https://docs.qgis.org/2.8/en/docs/user_manual/working_with_raster/raster_properties.html</p><p>https://docs.qgis.org/2.8/en/docs/user_manual/working_with_raster/raster_properties.html</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>69</p><p>4. SELECT and name an output file using the .vrt extension.</p><p>5. (Optional) Check the Load into canvas when finished checkbox if you want to see the</p><p>results immediately:</p><p>GDAL command line equivalent:</p><p>For example, gdalbuildvrt elevlid.vrt elevlid_</p><p>D782_6.tif elevlid_D783_6m.tif elevlid_</p><p>D792_6m.tif elevlid_D793_6m.tif.</p><p>How it works…</p><p>GDAL VRT format is an XML file that defines the location of each raster file relative to an</p><p>anchor file. It uses the existing spatial extent information of the rasters to figure out their</p><p>positions relative to each other and then anchors the set in the given coordinate system.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>70</p><p>There's more…</p><p>Using a VRT is all about saving time. When you have hundreds of raster files for one particular</p><p>dataset, you can combine them into a single file. However, this file could be gigantic in size and</p><p>somewhat impossible to work with. This is a quick way to be able use the files as a seamless</p><p>background layer. If you need to perform analysis, you'll likely need to either combine the layers</p><p>or loop over them individually.</p><p>You could also generate Tile Index (also in the Miscellaneous menu), which makes a shapefile</p><p>of the outlines of the rasters and puts the ID and path of the raster in the attribute table. This</p><p>would allow you to figure out which image you want to load for a given map without having to</p><p>load them all.</p><p>Finally, if you really want to make all of the files a single large file, use the context menu</p><p>(right-click on the loaded VRT layer and choose Save As). If you have overlaps, more</p><p>complicated situations, or want to merge without loading the files, first use the Merge</p><p>tool (also in the Miscellaneous menu). This can be tricky if your files overlap, you'll need</p><p>to decide how to handle the double data.</p><p>See also</p><p>f For another example,</p><p>please refer to http://manual.linfiniti.com/en/</p><p>rasters/data_manipulation.html#basic-fa-create-a-virtual-raster</p><p>f GDAL's gdalbuildvrt is the underlying tool; it's documentation can be found at</p><p>http://gdal.org/gdalbuildvrt.html</p><p>www.it-ebooks.info</p><p>http://manual.linfiniti.com/en/rasters/data_manipulation.html#basic-fa-create-a-virtual-raster</p><p>http://manual.linfiniti.com/en/rasters/data_manipulation.html#basic-fa-create-a-virtual-raster</p><p>http://gdal.org/gdalbuildvrt.html</p><p>http://www.it-ebooks.info/</p><p>71</p><p>3</p><p>Common Data</p><p>Preprocessing Steps</p><p>In this chapter, we will cover the following recipes:</p><p>f Converting points to lines to polygons and back – QGIS</p><p>f Converting points to lines to polygons and back – SpatiaLite</p><p>f Converting points to lines to polygons and back – PostGIS</p><p>f Cropping rasters</p><p>f Clipping vectors</p><p>f Extracting vectors</p><p>f Converting rasters to vectors</p><p>f Converting vectors to rasters</p><p>f Building DateTime strings</p><p>f Geotagging photos</p><p>Introduction</p><p>When working with other people's data, it is often not the exact format that you need for a</p><p>particular use. This chapter is all about taking the data that you do have and converting it to</p><p>what you actually need. It covers converting between different types of vectors (points, lines,</p><p>and polygons), between vectors and polygons, and cutting out only the parts that you need.</p><p>Taking data from how you get it and converting it to the format and layout that you need in</p><p>order to work with is often called 'data preprocessing'.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Common Data Preprocessing Steps</p><p>72</p><p>Converting points to lines to polygons and</p><p>back – QGIS</p><p>Sometimes your data is vector formatted (point, line, or polygon), but it is not the right kind</p><p>of vector for a particular type of analysis. Or perhaps you need to split a vector in a particular</p><p>way to facilitate some analysis or cartography. Thankfully, all vector formats are related, lines</p><p>are two or more connected points, polygons are lines whose first and last point are the same,</p><p>multipolygons are two or more polygons for the same record, and rings are nested polygons</p><p>where the inner polygon outlines an area to be excluded. This recipe covers how to convert</p><p>between the different vector types using built-in QGIS methods.</p><p>Getting ready</p><p>To convert points to lines or polygons, you will need a shapefile with an ID column that has a</p><p>single value shared between the points of the same line or polygon. In the following example,</p><p>we will use census_wake_2000_points.shp.</p><p>You will also need to install and activate the Points2One plugin. Refer to the following</p><p>website for how install plugins, http://docs.qgis.org/2.8/en/docs/user_manual/</p><p>plugins/plugins.html.</p><p>How to do it…</p><p>The following instructions show four different conversion methods, depending on the starting</p><p>data and the end data type. All of the tools are in the Vector menu:</p><p>Start by loading the census_wake_2000_points.shp layer.</p><p>www.it-ebooks.info</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/plugins/plugins.html</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/plugins/plugins.html</p><p>http://www.it-ebooks.info/</p><p>Chapter 3</p><p>73</p><p>Converting points to lines (or polygons)</p><p>1. Go to Vector | Points2One.</p><p>2. Choose to create either lines or polygons.</p><p>3. Pick the group ID; in this case, this is STFID.</p><p>4. Create the output filename: census_wake_2000_pt2lines.shp:</p><p>Converting lines to polygons</p><p>1. Go to Vector | Geometry Tools | Lines to Polygons.</p><p>2. Create the output filename: census_wake_2000_lines2poly.shp.</p><p>Converting polygons to lines</p><p>1. Go to Vector | Geometry Tools | Polygons to Lines.</p><p>2. Create the output filename: census_wake_2000_poly2lines.shp.</p><p>Converting polygons or lines to points</p><p>1. Go to Vector | Geometry Tools | Extract Nodes.</p><p>2. Create the output filename: census_wake_2000_poly2pts.shp.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Common Data Preprocessing Steps</p><p>74</p><p>How it works…</p><p>Converting to simpler types from more complex ones is fairly straightforward in simple cases.</p><p>Lines are just multiple points connected together and polygons are lines that start and end</p><p>with the same point. So, it's pretty easy to see how to deconstruct one geometry to simpler</p><p>geometries.</p><p>It's building up from points, which is a little trickier. In a line with three or more points, you</p><p>need to make sure that you have them in the correct order; otherwise, you'll end up with a</p><p>squiggle. When going to polygons, this can create bigger issues by leaving you with invalid</p><p>polygons that self-intersect. So, it's really important to order your points in your source table in</p><p>the same order that they will be combined. Reordering your data can be somewhat tricky. The</p><p>Points2One plugin now includes a sort order option; to use this, make sure that your attribute</p><p>table has a numeric column with the order of the points specified per group (you can restart</p><p>the numbering at 1 for each distinct grouping).</p><p>There's more…</p><p>You can also split or combine multipolygons with the Singleparts to Multiparts and</p><p>Multiparts to Singleparts commands.</p><p>When things get really tricky, you may need to switch to editing the shapes by hand or</p><p>custom scripts. A good example of this is when you want a polygon with a hole in the middle.</p><p>If you do go the route of editing by hand, make sure to turn on snapping so that your lines</p><p>are automatically snapped to existing points. The official documentation on snapping</p><p>can be found at http://docs.qgis.org/2.8/en/docs/user_manual/working_</p><p>with_vector/editing_geometry_attributes.html#setting-the-snapping-</p><p>tolerance-and-search-radius.</p><p>The Editing and Advanced Editing toolbars and additional editing related plugins offer the</p><p>ability to manipulate particularly tricky geometries, one at a time, if you need to.</p><p>Converting points to lines to polygons and</p><p>back – SpatiaLite</p><p>The goal of this recipe is identical to the previous recipe, but it covers how to perform the</p><p>process with data in a SpatiaLite database. You will to turn points into lines and lines into</p><p>polygons.</p><p>Not all methods are available; for those that are not available, you can use the previous</p><p>recipe. It will also work on a database layer; it just doesn't save the results to the database.</p><p>So, the results will need to be imported to the database after completion.</p><p>www.it-ebooks.info</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/working_with_vector/editing_geometry_attributes.html#setting-the-snapping-tolerance-and-search-radius</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/working_with_vector/editing_geometry_attributes.html#setting-the-snapping-tolerance-and-search-radius</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/working_with_vector/editing_geometry_attributes.html#setting-the-snapping-tolerance-and-search-radius</p><p>http://www.it-ebooks.info/</p><p>Chapter 3</p><p>75</p><p>Getting ready</p><p>You need to load a vector layer of points with a numeric ID indicating order, and an identifier</p><p>of unique lines or polygons that is shared between points of the same geometry. For example,</p><p>you can use census_wake_2000_points loaded into SpatiaLite with the geometry field</p><p>called geom.</p><p>How to do it…</p><p>Using DB Manager Plugin (comes with QGIS and is in the Database menu), the QspatiaLite</p><p>plugin, or an alternate SpatiaLite SQL application (command line or GUI), the following SQL</p><p>examples will perform the conversions between vector types.</p><p>Points to lines</p><p>1. Create a table with points grouped by common ID:</p><p>--Create table grouping points with shared stfid into lines</p><p>CREATE Table census_pts2lines AS</p><p>SELECT stfid,MakeLine(geom) as geom</p><p>FROM census_wake_2000_points</p><p>GROUP BY stfid;</p><p>The following screenshot shows what the screen will look like:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Common Data Preprocessing Steps</p><p>76</p><p>2. Register the new table as spatial:</p><p>--Register the new table's geometry so QGIS knows its a</p><p>spatial layer</p><p>SELECT</p><p>RecoverGeometryColumn('census_pts2lines','geom',3358,'LINES</p><p>TRING',2);</p><p>Some SQL interfaces can run multiple SQL statements in a row,</p><p>separated by a semicolon. However, there are also many interfaces</p><p>that can only perform one query at a time. Generally, run one query</p><p>at a time unless you know your software supports multiple queries;</p><p>otherwise, this may fail or silently only run the first query.</p><p>Lines to polygons</p><p>1. Create a table with lines grouped by common ID:</p><p>--Create table grouping lines with shared stfid into polygons</p><p>CREATE Table census_line2poly AS</p><p>SELECT stfid,ST_Polygonize(geom) as geom</p><p>FROM census_pts2lines</p><p>GROUP BY stfid;</p><p>2. Register the new table as spatial:</p><p>--Register the new table's geometry so QGIS knows its a</p><p>spatial layer</p><p>SELECT</p><p>RecoverGeometryColumn('census_line2poly','geom',3358,'POLYGON',2);</p><p>Double dashes (--) is the SQL character for a comment line.</p><p>It is used to include descriptive text that is ignored in a query.</p><p>How it works…</p><p>Based on the common identifier specified in GROUP BY, the SQL statement aggregates</p><p>multiple points into a new geometry of the type specified. After creating the new geometry and</p><p>saving the results to a table, registration of the spatial metadata allows Spatialite and QGIS to</p><p>know the table is a spatial layer.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 3</p><p>77</p><p>There's more…</p><p>In the second example, lines were converted to polygons. You could also go directly from</p><p>points to polygons with ST_Polygonize(ST_MakeLine(geom)).</p><p>Under the current versions of SpatiaLite, only aggregation to higher levels is fully supported.</p><p>If you wish to disaggregate geometries, you can use the QGIS vector tools from the previous</p><p>recipe.</p><p>See also</p><p>f SpatiaLite does have functions to dump specific points by first, last, or ID, one at a</p><p>time. Refer to the index of functions (Reference Guide) online for details at https://</p><p>www.gaia-gis.it/fossil/libspatialite/index</p><p>f The Converting points to lines to polygons and back – QGIS recipe in this chapter for</p><p>nondatabase methods</p><p>Converting points to lines to polygons and</p><p>back – PostGIS</p><p>The goal of this recipe is identical to the previous two recipes, but it covers how to perform</p><p>the process with data in a PostGIS database. You will use it to turn points into lines, and lines</p><p>into polygons.</p><p>Not all methods are available; for those not available, you can use the previous recipe. It will</p><p>also work on a database layer; it just doesn't save the results to the database. So, the results</p><p>will need to be imported to the database after completion.</p><p>Getting ready</p><p>You need to load a vector layer of points with a numeric ID indicating order, and an identifier</p><p>of unique lines or polygons that is shared between points of the same geometry. For example,</p><p>you can use census_wake_2000_points loaded into PostGIS with the geometry field called</p><p>geom. (Refer to Chapter 1, Data Input and Output, the Loading Vector Data into PostGIS recipe</p><p>to see how to load data into PostGIS.)</p><p>Import as single not multigeometries. Otherwise, you'll need to carry out</p><p>some extra steps in the queries to split the multigeometries before they</p><p>can be converted.</p><p>www.it-ebooks.info</p><p>https://www.gaia-gis.it/fossil/libspatialite/index</p><p>https://www.gaia-gis.it/fossil/libspatialite/index</p><p>http://www.it-ebooks.info/</p><p>Common Data Preprocessing Steps</p><p>78</p><p>How to do it…</p><p>Using DB Manager Plugin (this comes with QGIS and is in the Database menu) or</p><p>an alternate PostGIS SQL application (command line—pgsql or GUI—pgadmin III), the</p><p>following SQL examples will perform the conversions between vector types.</p><p>Converting points to lines</p><p>1. Run the following query:</p><p>CREATE VIEW pts2line AS</p><p>SELECT ROW_NUMBER() over (order by census_wake_2000_points</p><p>.stfid) as id, stfid, ST_MakeLine(geom) as geom</p><p>FROM census_wake_2000_points</p><p>GROUP BY stfid;</p><p>The following screenshot shows what the screen will look like:</p><p>To test the creation of new geometries, wrap the queries in CREATE VIEW,</p><p>as demonstrated in Chapter 2, Data Management. If the data is large or you</p><p>are happy with the results, you can swap in CREATE TABLE to make a new</p><p>table for more permanent storage.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 3</p><p>79</p><p>Converting lines to polygons</p><p>Run the following query:</p><p>CREATE VIEW line2poly AS</p><p>SELECT id,stfid,ST_MakePolygon(geom) as geom</p><p>FROM pts2line;</p><p>Want to go straight from point to polygons? Try ST_</p><p>MakePolygon(ST_MakeLine(geom)); the rest is</p><p>as shown in the first example: points to lines query.</p><p>Converting lines or polygons to points</p><p>Run the following query:</p><p>CREATE VIEW pts AS</p><p>SELECT ROW_NUMBER() over (order by a.id_0) as id,id_0 as</p><p>grpid,(a.a_geom).path[2] as path,</p><p>ST_GeometryType((a.a_geom).geom), ((a.a_geom).geom) as geom</p><p>FROM (SELECT id_0,(ST_DumpPoints(geom)) as a_geom FROM</p><p>"census_wake2000") as a;</p><p>What's ROW_NUMBER() about? This is a trick to ensure a</p><p>unique integer for each row. Some tools complain if you don't</p><p>have this; for example, DB Manager won't preview or load the</p><p>layer, even though direct loading in QGIS works fine.</p><p>How it works…</p><p>Based on the common identifier specified in GROUP BY, the SQL statement aggregates</p><p>multiple points into a new geometry of the specified type.</p><p>When dumping geometries to points, PostGIS actually dumps an array, including ID</p><p>information. This is why the example query is actually a nested set of queries. The first is to</p><p>dump the array of geometry information, and the second to extract the relevant parts of the</p><p>results in the format that we want them in.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Common Data Preprocessing Steps</p><p>80</p><p>There's more…</p><p>PostGIS has a few dump functions with different purposes in mind. Splitting geometries is</p><p>apparently a difficult concept for databases because aggregation is usually the only direction</p><p>functions can logically go. Disaggregation is claimed by some to be counter to how SQL</p><p>conceptually works and would require non-SQL logic.</p><p>See also</p><p>f For more details on the dump functions of PostGIS (ST_Dump, ST_DumpPoints, and</p><p>ST_DumpRings), refer to the PostGIS manual at http://postgis.net/docs/</p><p>manual-2.1/reference.html</p><p>f Refer to the Converting points to lines to polygons and back – QGIS recipe in this</p><p>chapter for the non-database methods</p><p>Cropping rasters</p><p>Sometimes, the raster data you have for a theme is just much larger than the actual extent of</p><p>your study area or map. Or, in the case of scanned maps, you have extra nonmap information</p><p>around the outside edge. In these cases, you want to cut out a portion of your raster.</p><p>Getting ready</p><p>You'll need a raster file that you want to cut a portion of. In this example, we will use the North</p><p>Carolina whole state elevation model (elev_state_500m.tif) and cut it with the outline of</p><p>Wake County (county_wake.shp). Load both of these files in a fresh QGIS project.</p><p>How to do it…</p><p>The easiest way to do this is to use a polygon mask layer. The vector mask can be a rectangle,</p><p>but it doesn't have to be. The outline of a single polygon works best, though.</p><p>An alternate method would be to determine the bounding box (bbox)</p><p>coordinates of the extent that you want with the Capture Coordinate</p><p>tool or to draw the rectangle directly on the map.</p><p>1. Go to Raster | Extraction | Clipper.</p><p>2. Set Input file (raster) as elev_state_500m.tif.</p><p>3. Set Output file using the Select button to pick a directory, and name the output</p><p>elev_wake_500m.tif.</p><p>www.it-ebooks.info</p><p>http://postgis.net/docs/manual-2.1/reference.html</p><p>http://postgis.net/docs/manual-2.1/reference.html</p><p>http://www.it-ebooks.info/</p><p>Chapter 3</p><p>81</p><p>4. Set No data value to -9999.</p><p>Why -9999? Setting No data value to something impossible</p><p>makes it more obvious later to other users. The value 0 is a</p><p>really bad choice as data can legitimately have a value of zero.</p><p>As some raster formats only support numbers and, in particular,</p><p>integers, a large negative number is a common choice.</p><p>5. Now change Clipping mode to Mask Layer and select county_wake.shp as</p><p>Mask Layer:</p><p>How it works…</p><p>The shape of your mask and the size of the raster cells in the source data will determine how</p><p>pixelated</p><p>the resulting raster will be. Zoom in to the results and compare the edge of the new</p><p>raster to the vector outline of the county. You'll notice that because of the 500 m wide pixels,</p><p>it's hard to exactly match the edge of the county exactly with whole pixels.</p><p>Note that this tool, as with all other tools in the GDAL Tools menu, is actually a graphical</p><p>interface to GDAL command-line tools.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Common Data Preprocessing Steps</p><p>82</p><p>There's more…</p><p>You'll notice that with this particular example, an issue that arises with converting rasters to</p><p>nonrectangular shapes; the edges are jagged as compared to the vector. If you need it to be</p><p>really smooth, there are a few options. You can decrease the pixel size, splitting current pixels</p><p>into multiple pixels using the -tr option. As with other Raster tools, you can use the pencil</p><p>icon to override the GDAL command-line options to add features not included in the interface.</p><p>In this case, the -tr option inline with the rest of the already formatted command:</p><p>gdalwarp -tr 100 100</p><p>This would make each pixel 100 units instead of the current 500 x 500.</p><p>Some important options to remember when saving TIFF files with GDAL are</p><p>number type (Integer versus Float) and compression. Both of these can</p><p>greatly impact the final file size. Refer to the Converting Vectors to Rasters</p><p>recipe later in this chapter for an example. Also, if you have a multicore CPU,</p><p>add -multi to take advantage of your CPU cores for faster processing of</p><p>most raster operations.</p><p>See also</p><p>f For a full list of the gdalwarp options refer to http://gdal.org/gdalwarp.html</p><p>Clipping vectors</p><p>Like rasters, occasionally you only need vector data to cover a certain area of study (area of</p><p>interest). Also, like rasters, you can use a layer defining the extent that you want to select only</p><p>for a portion of a vector layer to make a new layer. The tool that is used for this job is Clip; that</p><p>is, 'Cookie Cutter' because of how the results look afterwards.</p><p>Getting ready</p><p>For the example in this recipe, we will use geology.shp and clip it to the extent of Wake</p><p>County using census_wake2000.shp. Any vector layer with the aggregation of polygons</p><p>covering all of the county will work.</p><p>www.it-ebooks.info</p><p>http://gdal.org/gdalwarp.html</p><p>http://www.it-ebooks.info/</p><p>Chapter 3</p><p>83</p><p>How to do it…</p><p>1. Load the two, geology.shp and census_wake2000.shp, layers.</p><p>2. Open the clipping tool from Vector | Geoprocessing Tools | Clip:</p><p>3. Input layer is the layer that has to be cut; this is geology.shp.</p><p>4. Clip layer defines the boundaries that have to be cut.</p><p>There is no requirement that clip layer be contiguous. You</p><p>can cut any combination of shapes that you want, circles,</p><p>squares, triangles, and so on. They just need to be polygons.</p><p>5. (Optional) Check Add result to canvas so that you see the results immediately.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Common Data Preprocessing Steps</p><p>84</p><p>6. Select OK to run the tool:</p><p>How it works…</p><p>All clipping is based on the principle of intersection features. For each feature in Input, the</p><p>tool checks to see whether it intersects with the overall shape of clip layer. When it does</p><p>intersect, the algorithm then checks whether any part falls outside the intersection. When</p><p>a part lands outside, it is cut off.</p><p>There's more…</p><p>You have to be careful when using clip. If the original table contained columns that included</p><p>measurements such as area and perimeter, these values are copied from the original.</p><p>Therefore, they may not reflect the new size of and shapes that were cut.</p><p>Generally, all geometry operations and analysis should be done with layers in the same</p><p>projection in order to ensure consistent results. Also, many tools are not projection aware and</p><p>won't compensate for two source layers being in different projections.</p><p>Tools that create the intersection of objects (for example, in PostGIS's and SpatiaLite's</p><p>ST_Intersection) can provide you with similar results. However, you may need to perform</p><p>multiple steps: Intersect, then select by contains or intersection to eliminate unwanted data.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 3</p><p>85</p><p>See also</p><p>Refer to the next recipe in the chapter if you want a way to limit features without altering the</p><p>original spatial data</p><p>Extracting vectors</p><p>Clipping is great, except when you don't want to alter the original geometries, such as when</p><p>you want to select overlapping features. Or, in other cases, you just want filter the geometries</p><p>based on nonspatial attributes. To achieve both of these results, you can utilize the Selection</p><p>tools in combination with Save Layer As.. to extract just the features of interest. This recipe</p><p>uses spatial selection methods to extract a subset of original polygons without altering them.</p><p>Getting ready</p><p>We'll use the same data as the previous recipe, geology.shp and census_wake2000.shp.</p><p>How to do it…</p><p>1. Select polygons from geology.shp that overlap with Wake County</p><p>(census_wake2000.shp) by navigating to Vector | Research Tools |</p><p>Select by location.</p><p>2. Select the feature in geology.</p><p>3. Intersect the features in census_wake2000.</p><p>4. Modify the current selection by creating new selection.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Common Data Preprocessing Steps</p><p>86</p><p>5. Click on OK:</p><p>6. Now, you will see the matching features highlighted (by default in yellow):</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 3</p><p>87</p><p>7. If the selection looks good, use the Layer context menu, right-click, and click on</p><p>Save As... to create a file containing only the selection.</p><p>When in the Save As... (as described in Chapter 1, Data</p><p>Input and Output) dialog make sure to check the box next</p><p>to Save only selected features.</p><p>How it works…</p><p>This really goes back to the same fundamental concept of Intersection that most vector</p><p>analysis rely on. When you can test whether two features overlap, there are many different</p><p>operations possible based on the answer. You can select, deselect, or, as in the previous</p><p>recipe, select then cut to fit. In these cases, each polygon is tested for at least a partial</p><p>intersection, and the matches are then highlighted as the results.</p><p>There's more…</p><p>While this recipe demonstrates how to select a subset of data based on location, you can also</p><p>do the same thing based on attributes of the features with a query on the attribute table. Or,</p><p>you can combine attribute based selection, spatial selection, and hand selection graphically</p><p>on the map—any selection combination that you want can be saved as a new layer.</p><p>You may also notice in the Select by location tool that vectors can also be added or removed</p><p>from existing selections in case you want to perform more complicated operations involving</p><p>more than one type of criteria.</p><p>See also</p><p>f Refer to the documentation on PostGIS, SpatiaLite, or the PostGIS Cookbook,</p><p>by Packt Publishing, for how to perform similar operations using SQL in PostGIS</p><p>(SpatiaLite, queries are very similar)</p><p>Converting rasters to vectors</p><p>Sometimes, you need to convert data that is originally in raster format to a vector format</p><p>in order to perform vector-based analysis methods. Generally speaking, as rasters are</p><p>continuous datasets, converting them to polygons is more common than converting them to</p><p>lines or points.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Common Data Preprocessing Steps</p><p>88</p><p>Getting ready</p><p>You need a raster layer, preferably one with groups of the same valued pixels next to each</p><p>other. For this example, we'll use geology_30m.tif, as a 30 meter x 30 meter pixel should</p><p>give decent results.</p><p>The smaller the pixels, the smoother looking the resulting</p><p>vector will appear when zoomed out.</p><p>How to do it…</p><p>1. Load geology_30m.tif.</p><p>2. Go to Raster | Conversion | Polygonize:</p><p>3. Name the output geology_30m.shp.</p><p>4. (Optional) Name the output column geology, class or value.</p><p>5. Press OK to run the process.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter</p><p>3</p><p>89</p><p>6. Compare the results (colors are in a similar but different scale):</p><p>How it works…</p><p>For each pixel, the value is compared to its neighbors (there are different neighbor</p><p>algorithms). When two pixels next to each other have the same value, they are lumped into</p><p>a polygon. Additional neighbors of the same value get added to the polygon until pixels of</p><p>differing values are encountered. As it's pixel-based, the edge of the result will usually follow</p><p>the outline of the pixels, making for a jagged edge. This edge can be smoothed into straight</p><p>lines with additional options or other tools, such as the QGIS smoothing tool.</p><p>There's more…</p><p>The minimum number of pixels required to make a polygon or the maximum allowed value</p><p>difference to be counted as the same can be altered to drop out isolated pixels or to allow for</p><p>a range of values to be counted together.</p><p>Note that if each pixel is unique as compared to its adjacent neighbors, then you'll just end up</p><p>with a polygon for each pixel. Or if your raster is sufficiently large and varied, this process could</p><p>take days. You may want to reconsider whether you really need to convert or whether your</p><p>analysis can be done in raster. Another option would be to resample or reclassify the raster to</p><p>larger polygons first to decrease the data density. Or, you need to investigate remote sensing</p><p>type tools that perform classification to create related groupings of pixels based on similarity.</p><p>If you want to convert raster data to points, then you probably want to use the points sampling</p><p>tool. If you want to convert some portion of a raster to lines then you may need more</p><p>sophisticated feature extraction tools found; for example, in SAGA and GRASS, either through</p><p>the Processing toolbox or as standalone software. Or, you may even need to result to a mix of</p><p>pixel extraction and hand digitizing.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Common Data Preprocessing Steps</p><p>90</p><p>See also</p><p>f Chapter 7, Raster Analysis I, and Chapter 8, Raster Analysis II, on further raster</p><p>methods</p><p>Converting vectors to rasters</p><p>Occasionally, you want to convert vectors to rasters to facilitate using raster analysis tools</p><p>such as the raster calculator.</p><p>Getting ready</p><p>You'll need a vector layer; this can be a point, line or polygon layer. The best results generally</p><p>come from polygon layers. We will use geology_wake2000reclass.shp. This file is the</p><p>result of the earlier clipping vectors recipe with a new column added that codes the geology</p><p>types as integers. For reference, you'll also use elev_wake_500m.tif as a matching raster</p><p>for the area of interest.</p><p>How to do it…</p><p>In order to be useful in analysis, here's a checklist:</p><p>1. Is the vector data in the same projection as the rest of the raster analysis data?</p><p>If not, reproject it first. Check the following URL for help, https://docs.qgis.</p><p>org/2.8/en/docs/training_manual/vector_analysis/reproject_</p><p>transform.html.</p><p>2. Clip the vector data to the analysis extent. You may need to convert a raster into a</p><p>polygon mask to clip it (refer to the clipping vectors recipe earlier in this chapter).</p><p>You can only pick numeric fields as raster formats only store</p><p>a single number per cell. In order to keep the attribute that</p><p>you want, you may need to use the field calculator to create</p><p>a new field that reclassifies categories of text into a numeric</p><p>scheme (for example, 1 = water, 2 = land, and so on) before</p><p>performing the conversion. If you copy a unique ID as the</p><p>attribute, there are some tools later that let you rejoin the</p><p>original attribute table as a value attribute table (refer to the</p><p>GRASS functions in Processing Toolbox).</p><p>3. Load geology_wake2000reclass.shp and elev_wake_500m.tif.</p><p>www.it-ebooks.info</p><p>https://docs.qgis.org/2.8/en/docs/training_manual/vector_analysis/reproject_transform.html</p><p>https://docs.qgis.org/2.8/en/docs/training_manual/vector_analysis/reproject_transform.html</p><p>https://docs.qgis.org/2.8/en/docs/training_manual/vector_analysis/reproject_transform.html</p><p>http://www.it-ebooks.info/</p><p>Chapter 3</p><p>91</p><p>4. Open Properties of elev_wake_500m.tif:</p><p>1. In the Metadata section, scroll down to Dimensions. You will want to match</p><p>either the dimensions or resolution so that your new raster will match the</p><p>existing elevation data pixels.</p><p>2. Note that Dimensions are X: 134, Y: 124 and Bands: 1. The resolution is</p><p>499.637,-498.342 pixel size.</p><p>3. Close the dialog.</p><p>5. Now, open the conversion dialog by navigating to Raster | Conversion | Rasterize</p><p>and follow these steps:</p><p>1. Input the file as geology_wake2000reclass.</p><p>2. Name your output geology_wake.tif (you will get a warning to set the</p><p>size or resolution).</p><p>3. Set the raster size: Width to 134 and Height to 124.</p><p>4. Click on OK to run the process.</p><p>The following screenshot shows how the screen will look:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Common Data Preprocessing Steps</p><p>92</p><p>How it works…</p><p>A grid of pixels is created at the specified width, height, and extent. For each cell, an</p><p>intersection is performed with the underlying vector layer. If more than 50% of the cell</p><p>intersects with the vector, it's designated attribute is assigned to the cell.</p><p>There's more…</p><p>It's really important to match projection and extent before converting to raster. If you fail to do</p><p>so, then your pixels in different raster layers won't line up perfectly with each other, and either</p><p>tools won't work or they will introduce a resampling error. If this looks too pixelated (squares)</p><p>for your liking, consider creating the raster at a higher pixel density.</p><p>If you compare the vector version to the new raster, you'll notice that the area in the middle all</p><p>came out a similar color. This is due to the values used for classification, where the geology</p><p>that started with the same major component was given the same starting value (for example,</p><p>PZ all start with 40, and the last number changes based on the letters after PZ).</p><p>Looking at the new layer and want to get rid of the black surrounding the real data? This area</p><p>is no-data, refer to Chapter 8, Raster Analysis II.</p><p>See also</p><p>f There are other methods to calculate the new value of a pixel to make smoother</p><p>transitions or intermediate values when multiple polygons are with the same</p><p>pixel. Refer to the GRASS and SAGA methods in Processing Toolbox for more</p><p>sophisticated alternatives.</p><p>Building DateTime strings</p><p>Date and time data get stored in all sorts of ways. One of the more frustrating issues is that</p><p>some common GIS formats (Shapefiles) can't store date and time in the same field without</p><p>making it a string. This is fine for visual display but terrible for use with tools that use DateTime</p><p>for their functionality, such as the TimeManager Plugin (refer to Chapter 4, Data Exploration).</p><p>Getting ready</p><p>Use datetime-example.shp, which contains a variety of date and time representations to</p><p>play with.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 3</p><p>93</p><p>How to do it…</p><p>1. Load datetime-example.shp.</p><p>2. Open the attribute table of datetime-example.</p><p>3. Create a new field. As this is a shapefile, we'll need to use a String of length 30 (or</p><p>you can use the empty field called calculated).</p><p>4. Turn on layer editing (this is the pencil icon, which is the first icon to the left of the</p><p>window toolbar).</p><p>5. From the drop-down list, select Calculated.</p><p>6. Now press the Calculation button.</p><p>In older versions of QGIS, you'll need to open the Field</p><p>calculator, which also works in newer versions but has</p><p>slightly more steps.</p><p>7. In the calculator, we'll use the substr String operation (that is, Substring) in</p><p>combination with the || concatenation to rearrange the values from existing fields</p><p>into a valid DateTime:</p><p>1. The simplest variant is just to combine shpDate with Time and put a space</p><p>in between:</p><p>"shpdate" || ' '|| "Time"</p><p>2. For a bit more of a challenge, use Date:</p><p>substr("Date",7,4) ||'-'||substr("Date",1,2)||'-'||</p><p>substr("Date",4,2) || ' '|| "Time"</p><p>Note the use of single quotes (') to denote a string</p><p>as opposed</p><p>57</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>ii</p><p>Table of Contents</p><p>Georeferencing vector layers 61</p><p>Creating raster overviews (pyramids) 66</p><p>Building virtual rasters (catalogs) 68</p><p>Chapter 3: Common Data Preprocessing Steps 71</p><p>Introduction 71</p><p>Converting points to lines to polygons and back – QGIS 72</p><p>Converting points to lines to polygons and back – SpatiaLite 74</p><p>Converting points to lines to polygons and back – PostGIS 77</p><p>Cropping rasters 80</p><p>Clipping vectors 82</p><p>Extracting vectors 85</p><p>Converting rasters to vectors 87</p><p>Converting vectors to rasters 90</p><p>Building DateTime strings 92</p><p>Geotagging photos 95</p><p>Chapter 4: Data Exploration 99</p><p>Introduction 99</p><p>Listing unique values in a column 100</p><p>Exploring numeric value distribution in a column 102</p><p>Exploring spatiotemporal vector data using Time Manager 105</p><p>Creating animations using Time Manager 108</p><p>Designing time-dependent styles 110</p><p>Loading BaseMaps with the QuickMapServices plugin 112</p><p>Loading BaseMaps with the OpenLayers plugin 116</p><p>Viewing geotagged photos 119</p><p>Chapter 5: Classic Vector Analysis 125</p><p>Introduction 125</p><p>Selecting optimum sites 125</p><p>Dasymetric mapping 131</p><p>Calculating regional statistics 135</p><p>Estimating density heatmaps 137</p><p>Estimating values based on samples 139</p><p>Chapter 6: Network Analysis 143</p><p>Introduction 143</p><p>Creating a simple routing network 144</p><p>Calculating the shortest paths using the Road graph plugin 148</p><p>Routing with one-way streets in the Road graph plugin 150</p><p>Calculating the shortest paths with the QGIS network analysis library 152</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>iii</p><p>Table of Contents</p><p>Routing point sequences 156</p><p>Automating multiple route computation using batch processing 158</p><p>Matching points to the nearest line 159</p><p>Creating a routing network for pgRouting 160</p><p>Visualizing the pgRouting results in QGIS 163</p><p>Using the pgRoutingLayer plugin for convenience 166</p><p>Getting network data from the OSM 168</p><p>Chapter 7: Raster Analysis I 171</p><p>Introduction 171</p><p>Using the raster calculator 172</p><p>Preparing elevation data 175</p><p>Calculating a slope 177</p><p>Calculating a hillshade layer 180</p><p>Analyzing hydrology 183</p><p>Calculating a topographic index 190</p><p>Automating analysis tasks using the graphical modeler 192</p><p>Chapter 8: Raster Analysis II 199</p><p>Introduction 199</p><p>Calculating NDVI 199</p><p>Handling null values 203</p><p>Setting extents with masks 207</p><p>Sampling a raster layer 210</p><p>Visualizing multispectral layers 212</p><p>Modifying and reclassifying values in raster layers 216</p><p>Performing supervised classification of raster layers 219</p><p>Chapter 9: QGIS and the Web 223</p><p>Introduction 223</p><p>Using web services 224</p><p>Using WFS and WFS-T 226</p><p>Searching CSW 229</p><p>Using WMS and WMS Tiles 231</p><p>Using WCS 235</p><p>Using GDAL 238</p><p>Serving web maps with the QGIS server 242</p><p>Scale-dependent rendering 247</p><p>Hooking up web clients 251</p><p>Managing GeoServer from QGIS 256</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>iv</p><p>Table of Contents</p><p>Chapter 10: Cartography Tips 259</p><p>Introduction 259</p><p>Using Rule Based Rendering 260</p><p>Handling transparencies 266</p><p>Understanding the feature and layer blending modes 269</p><p>Saving and loading styles 273</p><p>Configuring data-defined labels 277</p><p>Creating custom SVG graphics 282</p><p>Making pretty graticules in any projection 286</p><p>Making useful graticules in printed maps 291</p><p>Creating a map series using Atlas 295</p><p>Chapter 11: Extending QGIS 299</p><p>Introduction 299</p><p>Defining custom projections 300</p><p>Working near the dateline 304</p><p>Working offline 309</p><p>Using the QspatiaLite plugin 311</p><p>Adding plugins with Python dependencies 313</p><p>Using the Python console 315</p><p>Writing Processing algorithms 318</p><p>Writing QGIS plugins 322</p><p>Using external tools 327</p><p>Chapter 12: Up and Coming 333</p><p>Introduction 333</p><p>Preparing LiDAR data 334</p><p>Opening File Geodatabases with the OpenFileGDB driver 337</p><p>Using Geopackages 339</p><p>The PostGIS Topology Editor plugin 342</p><p>The Topology Checker plugin 346</p><p>GRASS Topology tools 350</p><p>Hunting for bugs 354</p><p>Reporting bugs 358</p><p>Index 363</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>v</p><p>Preface</p><p>Location-based technology is the latest buzzword to explain tools related to spatial knowledge</p><p>and analysis. For those who work on map making, geospatial science, or any number of other</p><p>things with spatial data, Geographic Information Systems (GIS), which is the more traditional</p><p>name for such tools, is a field of study with decades of innovation.</p><p>QGIS (previously known as Quantum GIS), a cross-platform, free, and open source software,</p><p>provides a traditional desktop-based geographic information system. Unlike a traditional</p><p>system though, it is highly customizable, extendable, and, by design, works in tandem</p><p>with a ton of other GIS-related tools (more are added all the time).</p><p>QGIS is a crossing point of the free and open source geospatial world. While there are a</p><p>great many tools in QGIS, it is not one massive application that does everything, and it was</p><p>never really designed to be that from the beginning. It is rather a visual interface to much of</p><p>the open source geospatial world. You can load data from proprietary and open formats into</p><p>spatial databases of various flavors and then analyze the data with well-known analytical</p><p>backends before creating a printed or web-based map to display and interact with your</p><p>results. What's QGIS's role in all this? It's the place where you check your data along the way,</p><p>build and queue the analysis, visualize the results, and develop cartographic end products.</p><p>If you need to test modify one layer before doing a batch of 1000, use QGIS. Want to make</p><p>sure the results of that SQL query or script make sense, use QGIS. Need to tinker with</p><p>alternative methods of displaying your data to find the right colors, lines, and layers to convey</p><p>your message, use QGIS. Find something QGIS can't do, look at other tools it works with,</p><p>search the plugin list, write a plugin, or submit a new feature request. With such a mentality,</p><p>everything is possible—it just takes an understanding of how to get there.</p><p>This book is all about showing you how to do all these great things and all the new cool things</p><p>you didn't know you wanted to do. When QGIS doesn't do what you need or doesn't work with</p><p>the tools you want to use, we'll show you where to go and who to talk to about making your</p><p>dream of new functionality a reality.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Preface</p><p>vi</p><p>What this book covers</p><p>Over the course of 12 chapters, this book will take you from data input and output, through</p><p>data management and analysis, to creating print and web output, as well as extending QGIS.</p><p>Chapter 1, Data Input and Output, covers loading and saving data with special instructions</p><p>for trickier formats, batch conversions, and databases.</p><p>Chapter 2, Data Management, describes the basic manipulation of attributes, indexes, and</p><p>queries to make the use of your data more efficient.</p><p>Chapter 3, Common Data Preprocessing Steps, deals with converting data into the formats</p><p>you need for analysis, including vector to and from raster, transitioning through different</p><p>types of vectors, and cutting your data to just the important areas.</p><p>Chapter 4, Data Exploration, explores methods for visualizing and understanding the</p><p>information in your data.</p><p>Chapter 5, Classic Vector Analysis, shows the QGIS way of performing traditional analysis</p><p>methods of vector layers.</p><p>Chapter 6, Network Analysis, dives into the methods for analyzing routes and networks.</p><p>Chapter 7, Raster Analysis I, covers raster analysis that is primarily related to topography</p><p>and hydrology.</p><p>Chapter 8, Raster Analysis II, covers common raster analysis methods and introduces more</p><p>advanced multispectral and classification data handling.</p><p>Chapter 9, QGIS and the Web, explores the use of live data from the Web and how to put up</p><p>your own web map based on a QGIS project.</p><p>Chapter 10, Cartography Tips, reveals advanced tips and tricks to get the most out of the</p><p>cartographic tools in QGIS.</p><p>Chapter 11, Extending QGIS, shows you how to take QGIS beyond the out-of-the-box features</p><p>with plugins, customization, and add-on tools.</p><p>Chapter 12, Up</p><p>to double quotes ("), which indicate a</p><p>field name.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Common Data Preprocessing Steps</p><p>94</p><p>8. Experiment with the formulas, checking the results with the calculator preview at the</p><p>bottom. When satisfied, select OK, and then click on the Update All button:</p><p>9. If you're happy with the results, save the edits; if not, toggle editing off and choose</p><p>Close without saving.</p><p>How it works…</p><p>Substr takes three arguments: the field name, the starting position, and the number of</p><p>characters after this to include. The position index starts at 1.</p><p>There's more…</p><p>Using string manipulation, we've combined multiple fields into an ISO standard format for</p><p>DateTime, which other tools will recognize and be able to utilize.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 3</p><p>95</p><p>The fields included in this example data are as follows:</p><p>Field Type Explanation</p><p>Time String This is time in a 24 hour format: hours:minutes:seconds.</p><p>datetime String This is a DateTime String from a typical GPS GPX file.</p><p>calculated String This is a String that is long enough to hold date and time with</p><p>padding characters and a timezone UTC offset.</p><p>Date String This is a typical date format coming out of a spreadsheet.</p><p>shpdate Date This is a date format in the Date type within a shapefile, and</p><p>it can not hold time.</p><p>See also</p><p>f If using database layers, this can all be performed with SQL in SpatiaLite or PostGIS.</p><p>The ISO date time standard is 8601, and can be found at http://en.wikipedia.</p><p>org/wiki/ISO_8601.</p><p>Geotagging photos</p><p>Newer cameras and phones with built-in GPS can be wonderful tools for data collection, as</p><p>they help keep track of exactly where and when a picture was taken. However, not all cameras</p><p>have a built-in GPS. You can add geotags afterwards, either with a GPS log from a separate</p><p>GPS unit or just using a reference map and your memory or notes.</p><p>Getting ready</p><p>For this recipe, you'll need a some photos and either a GPS log (*.gpx), reference vector,</p><p>reference raster, or coordinates. We've provided centerofcalifornia.jpg in the</p><p>geotag folder, and the coordinates are in the image itself but also included as a point in</p><p>centerofcalifornia.shp.</p><p>You will also need the Geotag photos plugin, which requires the exiftool program to be</p><p>installed on your system. If exiftool didn't come with your install, you can easily get it from</p><p>the Web at http://www.sno.phy.queensu.ca/~phil/exiftool/ or at package</p><p>repositories (Linux).</p><p>www.it-ebooks.info</p><p>http://en.wikipedia.org/wiki/ISO_8601</p><p>http://en.wikipedia.org/wiki/ISO_8601</p><p>http://www.sno.phy.queensu.ca/~phil/exiftool/</p><p>http://www.it-ebooks.info/</p><p>Common Data Preprocessing Steps</p><p>96</p><p>How to do it…</p><p>This particular plugin assigns location per folder, so all photos in a folder will get the same</p><p>coordinates. This works well for batch assigning of general coordinates:</p><p>1. Start by loading your GPS log, or creating a new vector layer and digitizing the points</p><p>that you want to assign to photos. In this case, load centerofcalifornia.shp.</p><p>Don't have a lot of locations or just have coordinates written</p><p>down by hand? You can manually enter the information into</p><p>the plugin interface without an existing layer.</p><p>2. (Optional) Load a background reference layer.</p><p>3. Now that you have the layer that you want to associate with photos open the plugin in</p><p>Vector | Geotag and import photos | Geotag photos.</p><p>4. Select the layer and then the field (location) that you want to use as the label.</p><p>5. Now, click on Populate Table. One row should have been added to the interface.</p><p>6. Now, pick the photos by clicking under Path to Folder in the empty box for the row</p><p>that you want to assign:</p><p>1. You can type in the path or browse by clicking on ….</p><p>2. Pick the geotag folder.</p><p>3. Make sure to press Enter or click outside the box once you are back in the</p><p>main screen:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 3</p><p>97</p><p>7. (Optional) Check the box to rename geotagged files if you want the geotagged version</p><p>to be a new copy of the file instead of altering the original files (always keep an</p><p>original backup).</p><p>8. Click on OK.</p><p>How it works…</p><p>Exiftools writes to the built-in metadata of an image file to a section called EXIF. It's a standard</p><p>in photography to store extra data about photos that many software management tools can</p><p>easily read from. Latitude and Longitude in WGS 84 coordinates are the standard method</p><p>of encoding GPS data within the EXIF section.</p><p>There's more…</p><p>Now that you have a geotagged photo, you can upload it to sites such as Flickr, which will</p><p>display it on a map, or skip to the recipe Viewing Geotagged Photos in Chapter 4, Data</p><p>Exploration, for how to make a map in QGIS.</p><p>This plugin is very manual and assigns location per folder as it was created to work specifically</p><p>with camera traps. Instead, if you were travelling between each photo location and have a GPS</p><p>log, there are other non-QGIS tools to help you match GPS points with your photos. Digikam</p><p>(a photography management tool) has a function to geotag based on timestamp matches.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Common Data Preprocessing Steps</p><p>98</p><p>See also</p><p>f The OpenStreetMap wiki lists other free and paid options out there at</p><p>http://wiki.openstreetmap.org/wiki/Geotagging_Source_Photos</p><p>www.it-ebooks.info</p><p>http://wiki.openstreetmap.org/wiki/Geotagging_Source_Photos</p><p>http://www.it-ebooks.info/</p><p>99</p><p>4</p><p>Data Exploration</p><p>In this chapter, we will cover the following recipes:</p><p>f Listing unique values in a column</p><p>f Exploring numeric value distribution in a column</p><p>f Exploring spatiotemporal data using Time Manager</p><p>f Creating animations using Time Manager</p><p>f Designing time-dependent styles</p><p>f Loading BaseMaps with the QuickMapServices plugin</p><p>f Loading BaseMaps with the OpenLayers plugin</p><p>f Viewing geotagged photos</p><p>Introduction</p><p>This chapter focuses on recipes that will help you visually inspect and better comprehend</p><p>your data. The recipes in this chapter include methods of summarizing, inspecting, filtering,</p><p>and styling data, based on spatial and temporal attributes so that you can get a better feeling</p><p>for your data before you perform analysis. The primary goal is to create some visuals or</p><p>summaries of data that allow you, the human, to utilize your brain's ability to identify patterns</p><p>of interest. The better you understand your data, the easier it is to pick appropriate analysis</p><p>methods later.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Exploration</p><p>100</p><p>Listing unique values in a column</p><p>When investigating a new dataset, it is very helpful to have a way to quickly check which</p><p>values a column contains. In this recipe, we will use different approaches using both the GUI</p><p>and the Python console to list the unique values of POI classes in our sample POI dataset.</p><p>Getting ready</p><p>To follow this recipe, please load poi_names_wake.shp.</p><p>How to do it…</p><p>If you are simply looking for a solution based on the GUI, the List unique values tool is</p><p>available both in Vector | Analysis Tools as well as in the Processing Toolbox menu. You can</p><p>use either one of these to get a list of the unique values in a column. Having this tool available</p><p>in the Processing Toolbox menu makes it possible to include it in processing models and,</p><p>thus, automate the process. The following steps to list unique values use the Processing</p><p>Toolbox menu:</p><p>1. Start List unique values from the Processing Toolbox menu.</p><p>2. Select the poi_names_wake layer as Input layer and the class attribute as</p><p>Target field.</p><p>3. Click on Run and wait for the tool to finish. The results will be displayed in the</p><p>Results view, which will open automatically.</p><p>If you want to further customize this task, for example, by counting how often the values</p><p>appear in this dataset, it's time to fire up Python console:</p><p>1. Start Python Console, which you will find in the Plugins menu, and click on the Show</p><p>editor button (in the toolbar to the left of Python console) to open the editor window.</p><p>2. Paste the following short script into the editor, save it, and then click on the Run</p><p>script button. (Make sure that the POI layer is selected in the layer list.) It loops</p><p>through all features in the active layer and creates a dictionary object, which contains</p><p>all unique values and the corresponding counts:</p><p>import processing</p><p>layer = iface.activeLayer()</p><p>classes = {}</p><p>features = processing.features(layer)</p><p>for f in features:</p><p>attrs = f.attributes()</p><p>class_value = f['class']</p><p>if class_value in classes:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 4</p><p>101</p><p>classes[class_value] += 1</p><p>else:</p><p>classes[class_value] = 1</p><p>print classes</p><p>The following screenshot shows what the screen looks like:</p><p>How it works…</p><p>In the first line, we use import processing because it offers a very handy and convenient</p><p>function, processing.features(), to access layer features, which we use in line 7. It</p><p>is worth noting that, if there is a selection, processing.features() will only return an</p><p>iterator of the selected features.</p><p>In line 3, we get the currently active layer object. Line 5 creates the empty dictionary object,</p><p>which we will fill with the unique values and corresponding counts.</p><p>The for loop, starting on line 5, loops through all features of the layer. For each feature, we</p><p>get the value in the classification field (line 7). You can change the column name to analyze</p><p>other columns. Then, we only need to check whether this value is already present in the</p><p>classes dictionary (line 8) or whether we have to add it (line 11).</p><p>There's more…</p><p>If you are using an SQL database, such as Spatialite or PostGIS, you can achieve similar</p><p>results with a query using the COUNT and GROUP BY functions to count the number of</p><p>features or records per class:</p><p>SELECT class, COUNT(*)</p><p>FROM poi_name_wake</p><p>GROUP BY class</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Exploration</p><p>102</p><p>Exploring numeric value distribution in a</p><p>column</p><p>In this recipe, we will look at how to explore the properties of a column of numeric values. We</p><p>will look at the tools that QGIS offers and apply them to analyze the elevation values in our</p><p>sample POI dataset.</p><p>Getting ready</p><p>To follow this recipe, please load poi_names_wake.shp. If you followed the previous recipe,</p><p>Listing unique values in a column, you can continue directly from there.</p><p>How to do it…</p><p>A good way to get a first impression of the properties of a numeric column is using the</p><p>Basic Statistics tool from Vector. This allows you to calculate statistical values, such as the</p><p>minimum and maximum values, mean and median, standard deviation, and sum.</p><p>If you want to examine the distribution of elevation values, there is the handy Statist plugin.</p><p>Statist generates an interactive histogram representation of the value distribution:</p><p>1. Install Statist using Plugin Manager.</p><p>2. Start Statist from the Vector menu.</p><p>3. Specify Input vector layer and the attribute that you want to analyze (Target field),</p><p>then click on OK to compute the statistics.</p><p>4. Using the buttons below the diagram, you can zoom and pan the diagram, as well as</p><p>save the diagram image.</p><p>5. You can even customize the diagram by changing the title and axis labels and</p><p>ranges. Just use the right-most button with the green tick mark on it to open the</p><p>customization dialog:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 4</p><p>103</p><p>How it works...</p><p>Thanks to Python Console and the editor, we are not limited to the existing tools and plugins.</p><p>Instead, we can create or own specialized scripts such as the following one. This script creates</p><p>a short layer statistics report using HTML and the Google Charts Javascript API (for more</p><p>information and API docs refer to https://developers.google.com/chart/), which it</p><p>then displays in a QWebView window. Of course, you can use any other JavaScript charting API</p><p>as well. (Note that you need to be connected to the Internet for this script to work because it</p><p>has to download the Javascript.) We recommend using the editor that was introduced in the</p><p>previous recipe. Don't forget to select the layer in the legend:</p><p>import processing</p><p>from PyQt4.QtWebKit import QWebView</p><p>layer = iface.activeLayer()</p><p>values = []</p><p>features = processing.features(layer)</p><p>for f in features:</p><p>values.append( f['elev_m'])</p><p>myWV = wQWebView(None)</p><p>html=''</p><p>html+=''</p><p>html+='google.load("visualization","1",{packages:["corechart"]});'</p><p>html+='google.setOnLoadCallback(drawChart);'</p><p>www.it-ebooks.info</p><p>https://developers.google.com/chart/</p><p>http://www.it-ebooks.info/</p><p>Data Exploration</p><p>104</p><p>html+='function drawChart() { '</p><p>html+='var data = google.visualization.arrayToDataTable(['</p><p>html+='["%s"],' % (field_name)</p><p>for value in values:</p><p>html+='[%f],' % (value)</p><p>html+=']);'</p><p>html+='var chart = new google.visualization.Histogram('</p><p>html+='document.getElementById("chart_div"));'</p><p>html+='chart.draw(data, {title: "Histogram"});}'</p><p>html+='Layer: %s' % (layer.name())</p><p>html+='Values for %s range from: ' % (field_name)</p><p>html+='%d to %d' % (min(values),max(values))</p><p>html+=''</p><p>html+=''</p><p>myWV.setHtml(html)</p><p>myWV.show()</p><p>Of course, custom reports such as this one lend themselves to adding more details. For</p><p>example, we can create separate histograms for each POI class or add other types of charts,</p><p>such as scatter charts:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 4</p><p>105</p><p>The first part (lines 1 to 12) is very similar to the script explained in the previous recipe,</p><p>Listing unique values in a column: We get the active layer and collect all elevation values in</p><p>the values list.</p><p>The QWebView created on line 14 enables us to display the HTML content, which we then</p><p>generate in the following section (lines 16 to 34). First, we load the Google Charts Javascript.</p><p>The actual magic happens in the drawChart() function starting on line 21. Lines 22 to 26</p><p>create the data object, which is filled with the elevation values from our values list. The last</p><p>three lines of the function (lines 27 to 29) finally create and draw the histogram chart. Finally,</p><p>lines 30 to 34 contain the HTML body definition with the header stating the layer name and a</p><p>short introduction text that states the min and max elevation values.</p><p>See also</p><p>f For those who want to perform more advanced graphing and numerical analysis,</p><p>consider using the matplotlib python library or reading your data sources into R.</p><p>Aggregate functions in SatialLitetgis PostGIS can also provide you with min, max,</p><p>average, sum, and other summarization functions. For PostGIS, refer to http://</p><p>www.postgresql.org/docs/9.1/static/functions-aggregate.html.</p><p>Exploring spatiotemporal vector data using</p><p>Time Manager</p><p>In this recipe, we will look at exploring spatiotemporal vector data using the Time Manager</p><p>plugin. We'll use event data from the ACLED (Armed Conflict Location and Event Data</p><p>Project) at http://www.acleddata.com/about-acled/.</p><p>Getting ready</p><p>To follow this recipe, please load ACLED_africa_fatalities_dec2013.shp. The layer</p><p>style that you will see in the following screenshots consists of a simple circle marker at 50%</p><p>transparency with the data-defined size set to the number of fatalities of the incident. (You</p><p>can read more about styling in Chapter 10, Cartography Tips, and Learning QGIS book by</p><p>Packt Publishing.) If you want some additional geographic context, you can also load NE_</p><p>africa.shp, which contains the outline of Africa.</p><p>www.it-ebooks.info</p><p>http://www.postgresql.org/docs/9.1/static/functions-aggregate.html</p><p>http://www.postgresql.org/docs/9.1/static/functions-aggregate.html</p><p>http://www.acleddata.com/about-acled/</p><p>http://www.it-ebooks.info/</p><p>Data Exploration</p><p>106</p><p>How to do it…</p><p>Once the data is loaded, all event positions will be displayed. The default way to filter the</p><p>events, for example, to only see the</p><p>events from December 1, is to use Layer | Query and</p><p>enter a filter expression or query, such as the following:</p><p>"EVENT_DATE" >= '2013-12-01' AND "EVENT_DATE"</p><p>size.</p><p>For the time-dependent style, we will add a new definition to the Fill color property,</p><p>as shown in the following screenshot:</p><p>2. Confirm the changes and start the animation to watch the effect.</p><p>How it works…</p><p>Here is our color expression in more detail:</p><p>color_hsla(</p><p>0,</p><p>scale_linear(</p><p>day(age(todatetime(animation_datetime()),</p><p>todatetime("EVENT_DATE"))),</p><p>0,31,</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Exploration</p><p>112</p><p>100,0</p><p>),</p><p>50,</p><p>128</p><p>)</p><p>The expression consists of multiple parts, as follows:</p><p>f day(age(todatetime(animation_datetime()),todatetime("EVENT_</p><p>DATE"))): This calculates the number of days between the current animation time</p><p>given by animation_datetime() and calculates EVENT_DATE</p><p>f scale_linear: This transforms the age value (in days between 0 to 31 days) into</p><p>a value between 100 and 0 which is suitable for the saturation parameter of the</p><p>following color function</p><p>f color_hsla: This is one of many functions that are available in QGIS to create</p><p>colors. It returns a string representation of a color, based on its attributes for hue</p><p>(0 equals red), saturation (between 0 and 100 depending on our age function),</p><p>lightness (50 equals medium lightness) and alpha (128 equals 50% transparency)</p><p>You can speed up the fading effect by reducing the scale_linear parameter, domain_max,</p><p>from 31 days to a smaller value, such as 7, for a complete fade to gray within one week.</p><p>See also</p><p>f If you are interested in learning more about color models, such as the HSL color</p><p>model used in this recipe, we recommend the Wikibook on color models at</p><p>http://en.wikibooks.org/wiki/Color_Models:_RGB,_HSV,_HSL</p><p>Loading BaseMaps with the</p><p>QuickMapServices plugin</p><p>Often, when exploring your data, you may feel somewhat lost. Without the context of the</p><p>known world, a layer can seem like a blob of information floating in space. By adding an</p><p>atlas-style map, air photos, or another BaseMap, you can begin to see how your data fits in</p><p>the on-the-ground reality. However, adding such layers often takes considerable preprocessing</p><p>work; sometimes, you just don't want to go through this until you know you need it. What's the</p><p>solution? Use a premade layer, preferably fast-loading tiles, from a web service.</p><p>www.it-ebooks.info</p><p>http://en.wikibooks.org/wiki/Color_Models:_RGB,_HSV,_HSL</p><p>http://www.it-ebooks.info/</p><p>Chapter 4</p><p>113</p><p>Getting ready</p><p>The QuickMapServices plugin works best when you have another dataset that you want to</p><p>provide extra context for. Start by first loading such a layer and then zooming in to its extent.</p><p>You will need the following:</p><p>f A layer of interest to overlay (you can use Davis_DBO_Centerline-wgs84.shp)</p><p>f An active Internet connection (this may not work behind corporate proxies)</p><p>f You will need to install and activate the QuickMapServices plugin</p><p>How to do it…</p><p>Starting with a new QGIS project, follow these instructions to load BaseMap from the web with</p><p>the QuickMapServices plugin:</p><p>1. Start by first loading your local map layers:</p><p>2. Verify that the projection definition is correctly identified by QGIS.</p><p>The plugin will not turn on projection-on-the-fly for you</p><p>unless you change its settings. However, in order for most</p><p>tile services to work in QGIS, projection-on-the-fly must</p><p>be enabled and set to EPSG:3857 Psuedo/Web/Popular</p><p>Mercator. Other data will fail to line up if their projection is</p><p>not defined or read properly by QGIS.</p><p>3. Go to Web | QuickMapServices and select a layer to load from the Web. Wait a few</p><p>seconds for the tiles to be loaded:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Exploration</p><p>114</p><p>The default list of services is open and free. If you</p><p>want to use other services that have more limited</p><p>licensing restrictions, such as Google and Bing, you</p><p>need to change some of the plugin's settings. Refer</p><p>to the There's more… section of this recipe.</p><p>4. (Optional) Temporarily disable Rendering (the checkbox in the bottom panel) to avoid</p><p>constant redrawing while rearranging the layer order.</p><p>5. Rearrange your layers to move the new QuickMapService added layer to the bottom</p><p>of your layer list.</p><p>6. Zoom to your original layer's extent.</p><p>7. (Optional) If you turned Rendering off, reactivate Rendering now.</p><p>Are things not lining up? Try zooming in a little more or</p><p>panning slightly. Most of all, be patient! Depending on</p><p>your Internet connection, it can take a while to retrieve</p><p>the tiles.</p><p>The following screenshot shows how the screen will look:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 4</p><p>115</p><p>How it works…</p><p>The QuickMapServices plugin is a web-based tool. All of the BaseMaps come from the Internet</p><p>as you pan and zoom; none of the data comes from your computer or QGIS itself.</p><p>There are a few things to be cautious of when using the QuickMapServices plugin. It doesn't</p><p>always line up quite right, especially when zoomed out to big areas. First, check whether your</p><p>other layers' projections are defined correctly and then try to reset the map by slightly panning</p><p>to the side. The key idea to remember is that tiled services generally only exist for EPSG:3857</p><p>and at a very specific set of scales. QGIS will attempt to pick the closest matching scale and</p><p>resample the scale to make it fit. This also explains why loading such layers can sometimes</p><p>be slow.</p><p>There's more…</p><p>To add more restricted services, such as Google. Bing, and so on, perform the following steps:</p><p>1. Go to Web | QuickMapServices | Settings | Contributed Services.</p><p>2. Click on the Get contributed pack button:</p><p>While it may be legal to view the maps (most of the time), depending on layers that are</p><p>selected, it may not be legal to digitize maps based on them, print them, or, otherwise, save</p><p>them for offline use. The license varies by data source. So, make sure to check this for the</p><p>sources you want to use by going online and reading the Terms of Service on their websites.</p><p>If your use case is outside of generally viewing for quick reference, you will probably need to</p><p>spend some time obtaining a license or permission for your use.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Exploration</p><p>116</p><p>OpenStreetMap-based sources are often good choices as the licenses typically just require</p><p>attribution with no restrictions on use. The main layers that originally come with the plugin</p><p>are there because they have less restrictive licenses.</p><p>Finally, you may be wondering how QuickMapServices differs from the OpenLayers plugin</p><p>mentioned in the next recipe. For starters, this plugin is newer and currently supported. It</p><p>also solves some long-standing issues, especially in regards to printing. There is also the</p><p>contributed layers GitHub repository, which should make it easier for people to contribute</p><p>new layer definitions.</p><p>See also</p><p>f Additional tile services can be added by hacking the plugin code or using a GDAL</p><p>TMS layer (Refer to Chapter 9, QGIS and the Web). You can also substitute in WMS</p><p>services to serve a similar role without some of the same limitations of tiles. Refer to</p><p>Chapter 9, QGIS and the Web, for information about creating your own web services.</p><p>Loading BaseMaps with the OpenLayers</p><p>plugin</p><p>Often, when exploring your data, you may feel somewhat lost. Without the context of the</p><p>known world, a layer can seem like a blob of information floating in space. By adding an atlas-</p><p>style map, air photos, or another BaseMap, you can begin to see how your data fits in the</p><p>on-the-ground reality. However, adding such layers often takes considerable preprocessing</p><p>work; sometimes, you just don't want to go through this until you know you need it. What's the</p><p>solution? Use a premade layer from a web service.</p><p>This recipe is almost identical to the previous recipe. QuickMapServices</p><p>is a replacement for the OpenLayers plugin, which is being discontinued</p><p>(deprecated). We kept this recipe because it's still a commonly-mentioned</p><p>plugin and works slightly differently. However, please consider using</p><p>QuickMapServices.</p><p>Getting ready</p><p>The</p><p>Openlayers plugin works best when you have another dataset that you want to provide</p><p>extra context for. Start by first loading such a layer and then zooming in to its extent.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 4</p><p>117</p><p>You will need the following:</p><p>f A layer of interest to overlay (you can use Davis_DBO_Centerline-wgs84.shp)</p><p>f An active Internet connection (this may not work behind corporate proxies)</p><p>f You will need to install and activate the OpenLayers plugin</p><p>How to do it…</p><p>Starting with a new QGIS project, follow these instructions to load BaseMap from the Web with</p><p>the Openlayers plugin:</p><p>1. Start by first loading your local map layers.</p><p>2. Verify that the projection definition is correctly identified by QGIS.</p><p>The plugin will set and turn on projection-on-the-fly for you. In order</p><p>for most tile services to work in QGIS, projection-on-the-fly must be</p><p>enabled and set to EPSG:3857 Psuedo/Web/Popular Mercator.</p><p>Other data will fail to line up if their projection is not defined or read</p><p>properly by QGIS.</p><p>3. Go to Vector | OpenLayers Plugin and select a layer to load from the Web. Wait a few</p><p>seconds for the tiles to be loaded.</p><p>4. (Optional) Temporarily disable Rendering (the checkbox in the bottom panel) to avoid</p><p>constant redrawing while rearranging layer order.</p><p>5. Rearrange your layers to move the new OpenLayers added layer to the bottom of your</p><p>layer list.</p><p>6. Zoom to your original layer's extent.</p><p>7. (Optional) If you turned Rendering off, reactivate Rendering now.</p><p>Are things not lining up? Try zooming in a little more or</p><p>panning slightly. Most of all, be patient! Depending on</p><p>your Internet connection, it can take a while to retrieve</p><p>the tiles.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Exploration</p><p>118</p><p>The following screenshot shows how the screen will look:</p><p>How it works…</p><p>The OpenLayers plugin is a web-based tool, based on the similarly named OpenLayers library</p><p>to create web-based maps in an Internet browser. However, instead of displaying the maps in</p><p>the browser, this plugin renders them into an active QGIS canvas (that is, a map).</p><p>There are a few things to be cautious of when using the OpenLayers plugin. It doesn't always</p><p>line up quite right, especially when zoomed out to big areas. First, check whether your other</p><p>layers' projections are defined correctly and then try to reset the map by panning slightly to</p><p>the side. The key idea to remember is that tiled services generally only exist for EPSG:3857</p><p>and at a very specific set of scales. QGIS will attempt to pick the closest matching scale and</p><p>resample the scale to make it fit. This also explains why loading such layers can sometimes</p><p>be slow.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 4</p><p>119</p><p>There's more…</p><p>While it may be legal to view the maps, depending on layers selected, it may not be legal to</p><p>digitize maps based on them, print them, or, otherwise, save them for offline use. The license</p><p>varies by data source. So, make sure to check for the sources you want to use by going online</p><p>and reading the Terms of Service on their websites. If your use case is outside of generally</p><p>viewing for quick reference, you will probably need to spend some time obtaining a license</p><p>or permission for your use. OpenStreetMap-based sources are often good choices as the</p><p>licenses typically just require attribution.</p><p>See also</p><p>f Additional tile services can be added by hacking the plugin code or using a GDAL</p><p>TMS layer (Refer to Chapter 9, QGIS and the Web). You can also substitute in WMS</p><p>services to serve a similar role without some of the same limitations of tiles. Refer to</p><p>Chapter 9, QGIS and the Web, for information about creating your own web services.</p><p>Viewing geotagged photos</p><p>Keeping track of photographs by location can be an extremely useful tool, enabling you to</p><p>easily pull up relevant photos of a place and time. They provide local context about other data</p><p>collected in the same place, and they can provide office staff with a view of what people in the</p><p>field saw. You can think of this as your own personal Street View, which is just more focused</p><p>than Google's version.</p><p>Getting ready</p><p>For this recipe, you'll need a set of geotagged photos. We've included a set a photos in this</p><p>book's data for you to learn with. This is a collection of photos from downtown Davis that</p><p>highlights the density and variety of public art along several blocks.</p><p>This recipe also takes advantage of several plugins, as follows:</p><p>f Install and activate Photo2Shape</p><p>f Activate the core plugin, eVis (Event Visualization)</p><p>f (Optional) Install and activate OpenLayers Plugin</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Exploration</p><p>120</p><p>How to do it…</p><p>Follow these steps to view geotagged photo locations in QGIS:</p><p>1. In a QGIS project, enable the plugins listed in the Getting ready section.</p><p>2. (Optional) Load a reference layer to help you see the local context</p><p>(Davis_DBO_Centerline.shp and/or OpenStreetMap/Google Streets</p><p>via OpenLayers Plugin).</p><p>Keep in mind that GPS locations and geotagged photos are almost</p><p>always in Latitude and Longitude WGS84 coordinates (that is,</p><p>EPSG:4326). So, you'll need to turn on projection-on-the-fly to</p><p>make them line up with your reference layers.</p><p>3. Go to the Vector menu or locate the icon on the toolbar for Photo2Shape:</p><p>4. This will ask to you select the directory in which you have the geotagged photos and</p><p>set an output shapefile. (Use the davis-art folder as the input directory.)</p><p>5. You should now get a new shapefile of the point locations of your photos loaded in</p><p>the map. Use Zoom to Layer Extent to zoom in on the locations. You should see a</p><p>camera icon at the location of each photo.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 4</p><p>121</p><p>6. Looking at the attribute table, you can see all the information about the photos pulled</p><p>into the table, including the path to the photos on your computer:</p><p>Going a little further</p><p>If you want to be able to see the actual photos in QGIS and not just the locations, continue</p><p>with the next section of steps:</p><p>1. Enable the eVis plugin.</p><p>2. Once activated go to Database | eVis | eVis Event Browser.</p><p>3. In the new window that pops up, you can see the attributes in the bottom box</p><p>the photo:</p><p>1. If this is blank, go to the Options tab and check whether the correct field is</p><p>selected for the path to the photo, in this case, this is filepath.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Exploration</p><p>122</p><p>2. To make the tool remember this change, check the Remember This box and</p><p>click on the Save button at the bottom:</p><p>How it works…</p><p>In photography, there is a standard metadata format written by most cameras called Exif,</p><p>which is stored as part of the image file format. Normally, all images store the timestamp,</p><p>camera model, camera settings, and other general information about an image. When you</p><p>take a picture with a GPS enabled camera, it should write the latitude and longitude to the</p><p>photo's metadata. Other programs that are metadata-aware can then read this information at</p><p>any time. If you happen to touch up these photos, make sure to tell your software to keep or</p><p>copy the metadata from the original so that you retain the location information.</p><p>There's more…</p><p>Don't have a camera or phone with built-in geotagging? This is not a problem. There are many</p><p>ways to add location information by yourself. One such method is with the Geotag and import</p><p>photos plugin that lets you link photo data to known locations, and this can be found at</p><p>http://hub.qgis.org/projects/geotagphotos/wiki.</p><p>If you need something more sophisticated, there are many other tools out there. Digikam,</p><p>an open source photo management program, includes a geotagging tool that will attempt to</p><p>automatch a GPX file from a GPS to your photos, based on timestamps.</p><p>www.it-ebooks.info</p><p>http://hub.qgis.org/projects/geotagphotos/wiki</p><p>http://www.it-ebooks.info/</p><p>Chapter 4</p><p>123</p><p>Geotagged photos are also supported by many online photos</p><p>services, so you can easily</p><p>browse a map of the photos that you've uploaded. Flickr is probably the most well-known for</p><p>this, and it also includes a concept of geo-fences, where you can exclude certain locations</p><p>from being publicly known.</p><p>On the flip-side, you now have an idea about how to remove geotags from photos in case you</p><p>don't want their locations known if you share them online.</p><p>See also</p><p>f There are other methods of seeing photos in the map besides eVis, including HTML</p><p>map tips. Refer to Nathan's blog at http://nathanw.net/2012/08/05/html-</p><p>map-tips-in-qgis/.</p><p>f More information about geotagging with Digikam can be found at http://</p><p>docs.kde.org/development/en/extragear-graphics/kipi-plugins/</p><p>geolocation.html.</p><p>f You can also use Flickr to geotag and re-export your images, you can or create online</p><p>map mash-ups with their API.</p><p>www.it-ebooks.info</p><p>http://nathanw.net/2012/08/05/html-map-tips-in-qgis/</p><p>http://nathanw.net/2012/08/05/html-map-tips-in-qgis/</p><p>http://docs.kde.org/development/en/extragear-graphics/kipi-plugins/geolocation.html</p><p>http://docs.kde.org/development/en/extragear-graphics/kipi-plugins/geolocation.html</p><p>http://docs.kde.org/development/en/extragear-graphics/kipi-plugins/geolocation.html</p><p>http://www.it-ebooks.info/</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>125</p><p>5</p><p>Classic Vector Analysis</p><p>In this chapter, we will cover the following recipes:</p><p>f Selecting optimum sites</p><p>f Dasymetric mapping</p><p>f Calculating regional statistics</p><p>f Estimating density using heatmaps</p><p>f Estimating values based on samples</p><p>Introduction</p><p>This chapter will provide you with an introduction to some of the most-common GIS analysis</p><p>use cases. The recipes focus on step-by-step instructions, as well as a closer explanation of</p><p>the tools that are used to achieve the desired analysis results. This chapter includes recipes</p><p>on optimum site selection, using interpolation, and creating heat maps, as well as calculating</p><p>regional statistics.</p><p>Selecting optimum sites</p><p>Optimum site selection is a pretty common problem, for example, when planning shop or</p><p>warehouse locations or when looking for a new apartment. In this recipe, you will learn how to</p><p>perform optimum site selection manually using tools from the Processing Toolbox option, but</p><p>you will also see how to automate this workflow by creating a Processing model.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Classic Vector Analysis</p><p>126</p><p>In the optimum site selection in this recipe, we will combine different vector analysis tools to</p><p>find potential locations in Wake County that match the following criteria:</p><p>f Locations are near a big lake (up to 500 m)</p><p>f Locations are close to an elementary school (up to 500 m)</p><p>f Locations are within a reasonable distance (up to 2 km) from a high school</p><p>f Locations are at least 1 km from a main road</p><p>Getting ready</p><p>To follow this exercise, load the following datasets, lakes.shp, schools_wake.shp, and</p><p>roadsmajor.shp.</p><p>As all datasets in our test data already use the same CRS, we can get right to the analysis. If</p><p>you are using different data, you may have to get all your datasets into the same CRS first. In</p><p>this case, please refer to Chapter 1, Data Input and Output.</p><p>How to do it…</p><p>The following steps show you how to perform optimum site selection using the Processing</p><p>Toolbox option:</p><p>1. First, we have to filter the lakes layer for big lakes. To do this, we use the Select</p><p>by expression tool from the Processing toolbox, select the lakes layer, and enter</p><p>"AREA" > 1000000 AND "FTYPE" = 'LAKE/POND' in the Expression textbox,</p><p>as shown in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 5</p><p>127</p><p>2. Next, we create the buffers that will represent the proximity areas around lakes,</p><p>schools, and roads. Use Fixed distance buffer from the Processing Toolbox</p><p>option to create the following buffers:</p><p>1. For the lakes, select Distance of 500 meters and set Dissolve result by</p><p>checking the box as shown in the following screenshot. By dissolving the</p><p>result, we can make sure that the overlapping buffer areas will be combined</p><p>into one polygon. Otherwise, each buffer will remain as a separate feature</p><p>in the resulting layer:</p><p>It's your choice whether you want to save the</p><p>buffer results permanently by specifying an output</p><p>file, or you just want to work with temporary files</p><p>by leaving the Buffer output file field empty.</p><p>2. To create the elementary school buffers, first select only the schools with</p><p>"GLEVEL" = 'E' using the Select by Expression tool like we did for the</p><p>lakes buffer. Then, use the buffer tool like we just did for the lakes buffer.</p><p>3. Repeat the process for the high schools using "GLEVEL" = 'H' and a</p><p>buffer distance of 2,000 meters.</p><p>4. Finally, for the roads, create a buffer with a distance of 1,000 meters.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Classic Vector Analysis</p><p>128</p><p>3. With all these buffers ready, we can now combine them to fulfill these rules:</p><p>1. Use the Intersection tool from the Processing Toolbox option on the buffers</p><p>around elementary and high schools to get the areas that are within the</p><p>vicinity of both school types.</p><p>2. Use the Intersection tool on the buffers around the lakes and the result of</p><p>the previous step to limit the results to lakeside areas. Use the Difference</p><p>tool to remove areas around major roads (that is, the buffered road layer)</p><p>from the result of the previous (Intersection) steps.</p><p>4. Check the resulting layer to view the potential sites that fit all the criteria that we</p><p>previously specified. You'll find that there is only one area close to WAKEFIELD</p><p>ELEMENTARY and WAKEFIELD HIGH that fits the bill, as shown in the following</p><p>screenshot:</p><p>How it works…</p><p>In step 1, we used Intersection to model the requirement that our preferred site would be</p><p>near both an elementary and a high school. Later, in step 3, the Difference tool enabled us to</p><p>remove areas close to major roads. The following figure gives us an overview of the available</p><p>vector analysis tools that can be useful for similar analyses. For example, Union could be</p><p>used to model requirements, such as "close to at least an elementary or a high school".</p><p>Symmetrical Difference, on the other hand, would result in "close to an elementary or a high</p><p>school but not both", as illustrated in the following figure:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 5</p><p>129</p><p>Dissolve</p><p>Intersection Symmetrical Difference</p><p>DifferenceUnion</p><p>There's more…</p><p>We were lucky and found a potential site that matched all criteria. Of course, this is not always</p><p>the case, and you will have to try and adjust your criteria to find a matching site. As you can</p><p>imagine, it can be very tedious and time-consuming to repeat these steps again and again</p><p>with different settings. Therefore, it's a good idea to create a Processing model to automate</p><p>this task.</p><p>The model (as shown in the following screenshot) basically contains the same tools that we</p><p>used in the manual process, as follows:</p><p>f Use two select by expression instances to select elementary and high schools. As you</p><p>can see in the following screenshot, we used the descriptions Select "GLEVEL" = 'E'</p><p>and Select "GLEVEL" = 'H' to name these model steps.</p><p>f For elementary schools, compute fixed distance buffers of 500 meters. This step is</p><p>called Buffer "GLEVEL" = 'E'.</p><p>f For high schools, compute fixed distance buffers of 2,000 meters. This step is called</p><p>Buffer "GLEVEL" = 'H'.</p><p>f Select the big lakes using Select by expression (refer to the Select big lakes step)</p><p>and buffer them using fixed distance buffer of 500 meters (refer to the Buffer</p><p>lakes step).</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Classic Vector Analysis</p><p>130</p><p>f Buffer the roads using Fixed distance buffer (refer to the Buffer roads step). The</p><p>buffer size is controlled by the number model input called road_buffer_size. You can</p><p>extend this approach of controlling the model parameters using additional inputs to</p><p>all the other buffer steps</p><p>in this model. (We chose to show only one example in order</p><p>to keep the model screenshot readable.)</p><p>f Use Intersection to get areas near schools (refer to the Intersection: near schools</p><p>step).</p><p>f Use Intersection to get areas near schools and lakes (refer to the Intersection:</p><p>schools and lakes step).</p><p>f Use Difference to remove areas near roads (refer to the Difference: avoid roads</p><p>step).</p><p>This is how the final model looks like:</p><p>You can run this model from the Processing Toolbox option, or you can even use it as a</p><p>building block in other models. It is worth noting that this model produces intermediate results</p><p>in the form of buffer results (near_elementary, near highschool, and so on). While</p><p>these intermediate results are useful while developing and debugging the model, you may</p><p>eventually want to remove them. This can be done by editing the buffer steps and removing</p><p>the Buffer names.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 5</p><p>131</p><p>Dasymetric mapping</p><p>Dasymetric mapping is a technique that is commonly used to improve population distribution</p><p>maps. By default, population is displayed using census data, which is usually available for</p><p>geographic units, such as census tracts whose boundaries don't necessarily reflect the actual</p><p>distribution of the population. To be able to model population distribution better, Dasymetric</p><p>mapping enables us to map population density relative to land use. For example, population</p><p>counts that are organized by census tracts can be more accurately distributed by removing</p><p>unpopulated areas, such as water bodies or vacant land, from the census tract areas.</p><p>In this recipe, we will use data about populated urban areas, as well as data about water</p><p>bodies to refine our census tract population data.</p><p>Getting ready</p><p>To follow this exercise, please load the population data from census_wake2000_pop.shp</p><p>(the file that we created in Chapter 2, Data Management), as well as the urban areas from</p><p>urbanarea.shp, and the lakes from lakes.shp.</p><p>As all the datasets in our sample data already use the same CRS, we can get right into the</p><p>analysis. If you are using different data, you may have to first get all datasets into the same</p><p>CRS. In this case, please refer to Chapter 1, Data Input and Output, for details.</p><p>How to do it…</p><p>To create a new and improved population distribution map, we will first remove the</p><p>unpopulated areas from the census tracts. Then, we will recalculate the population density</p><p>values to reflect the changes to the area geometries by performing the following steps:</p><p>1. Use Clip from the Processing Toolbox option (or Clip by navigating to Vector |</p><p>Geoprocessing tools if you prefer this option—the results will be identical) on the</p><p>census tracts and urban area layers to create a new dataset, containing only those</p><p>parts of the census tracts that are within urban areas.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Classic Vector Analysis</p><p>132</p><p>2. Refine the results of the previous step further by removing the water bodies (the</p><p>lakes layer) using the Difference tool. The following screenshot shows the results</p><p>of this so far:</p><p>3. Now, we can calculate the population density of the resulting areas, as follows:</p><p>1. Enable editing.</p><p>2. Open Field calculator.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 5</p><p>133</p><p>3. Calculate a new population density (inhabitants per square km) using the</p><p>formula, "_POP2000" / ($area / 1000000):</p><p>4. Deactivate editing and save the changes.</p><p>It is worth noting that you don't necessarily have to make a new</p><p>column. If you only want to use the density values for styling</p><p>purposes, you can also enter the expression directly in the style</p><p>configuration. On the other hand, if you create a new column, you</p><p>can inspect the density values in the attribute table, export them,</p><p>or analyze them further.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Classic Vector Analysis</p><p>134</p><p>We are done, and you can now visualize the results using a Graduated renderer with, for</p><p>example, the Natural Breaks (Jenks) classification mode. The Jenks Natural Breaks classification</p><p>is designed to arrange values into "natural" classes by maximizing the variance between</p><p>different classes while reducing the variance within the generated classes. The following figure</p><p>shows the population density based on the original census data (on the left) and the results</p><p>after Dasymetric mapping (on the right):</p><p>How it works…</p><p>In the first step of this recipe, we used the Clip operation. As you most likely noticed, the</p><p>results of a Clip operation look very similar to the results of the Intersection tool, which</p><p>we used in the previous recipe of this chapter, Selecting optimum sites. Compare both the</p><p>results, and you will see the following differences:</p><p>f The layer resulting from an Intersection operation contains attributes from both</p><p>input layers, while the result of a Clip operation only contains attributes of the first</p><p>input layer.</p><p>f This also means that the layer order is important when using Clip, but this does not</p><p>change the output of Intersection (except for the attribute order in the attribute table).</p><p>f The Intersection result is also very likely to contain more features than the Clip result</p><p>(164 instead of 105 if you use our sample data census tracts and urban areas). This</p><p>is because the Intersect tool needs to create a new feature for every combination of</p><p>intersecting census tracts and urban areas, while the Clip tool only removes the parts</p><p>of the census tracts that are not within any urban area.</p><p>A popular way of thinking about the Clip operation is to imagine one layer as the cookie cutter</p><p>and the other layer as the cookie dough.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 5</p><p>135</p><p>Calculating regional statistics</p><p>Another classic spatial analysis task is calculating the areas of a certain type within regions,</p><p>for example, the area within a county that is covered by certain land use types, or the share of</p><p>different crops that is farmed in given municipalities.</p><p>In this recipe, we will calculate statistics of geological data for zip code areas. In particular, we</p><p>will calculate the total area of each type of rock per zip code area.</p><p>Getting ready</p><p>To follow this recipe, load zipcodes_wake.shp and geology.shp from our sample data.</p><p>Additionally, install and activate the Group Stats plugin using Plugin Manager.</p><p>How to do it…</p><p>Using the following steps, we can calculate the areas of certain rock types per zip code area:</p><p>1. Calculate the intersections between zip code areas and geological areas using the</p><p>Intersection tool in the Processing Toolbox option or from the Vector menu.</p><p>2. Using the Group Stats plugin, you can now calculate the total area per rock type</p><p>and zip code area, as follows:</p><p>1. Select the Intersection result layer as the input Layer.</p><p>2. Drag the ZIPCODE field to the Rows input area and the GEO_NAME field</p><p>to the Columns input area.</p><p>3. Drag the sum function and the Area value to the Value input area.</p><p>4. Click on Calculate to start the calculations.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Classic Vector Analysis</p><p>136</p><p>The following screenshot displays the complete configuration, as well as the results:</p><p>How it works…</p><p>The Group Stats plugin brings functionality, which is commonly known as pivot tables, to</p><p>QGIS. A pivot table is a data summarization tool, which is commonly found in applications,</p><p>such as spreadsheets or business intelligence software. As shown in this example, pivot</p><p>tables can aggregate data from an input table. Additionally, the Group Stats plugin offers</p><p>extended geometry functions, such as, Area and Perimeter for polygon input layers, or</p><p>Length for line layers. This makes using the plugin more convenient because it is not</p><p>necessary to first use Field calculator to add these geometric values to the attribute table.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 5</p><p>137</p><p>It is worth noting that you always need to put the</p><p>following two entries into the Value input area:</p><p>f An aggregation function, such as sum, average, or count</p><p>f A value field (from the input layer's attribute table) or geometry function, which should</p><p>be aggregated</p><p>Estimating density heatmaps</p><p>Whether they are animal sightings, accident locations, or general points of interest, many</p><p>point datasets can be interpreted more easily by visualizing the point density using a</p><p>heatmap. In this recipe, we will estimate the density of POIs in Wake county to find areas with</p><p>a high density.</p><p>Getting ready</p><p>Load the poi_names_wake.shp POI dataset from our sample data. Make sure that the</p><p>Heatmap plugin, which comes with QGIS by default, is enabled in Plugin Manager.</p><p>How to do it…</p><p>Using the following steps, we can calculate the POI heatmap:</p><p>1. Start the Heatmap plugin from the Raster menu.</p><p>2. Make sure that poi_names_wake is selected as Input point layer.</p><p>3. Select a location and filename for Output raster. You don't need to specify the file</p><p>extension because this will be added automatically, based on the selected Output</p><p>format. GeoTIFF is usually the first choice.</p><p>4. Select a search Radius of 1000 meters.</p><p>5. The Add generated file to map option should be activated by default. Click on OK to</p><p>create the default heatmap.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Classic Vector Analysis</p><p>138</p><p>6. By default, the heatmap layer will be rendered using the Singleband gray render</p><p>type. Change the render type to Singleband pseudocolor and apply a color ramp that</p><p>you like to improve the visualization, as show in the following screenshot:</p><p>If you want to control the size of the output raster, just enable the</p><p>Advanced section and adjust the number of Rows and Columns or Cell</p><p>size X and Cell size Y, accordingly. Note that changing rows and columns</p><p>will automatically recalculate the size of the cell and vice versa.</p><p>How it works…</p><p>The search radius, which is also known as the kernel bandwidth, determines how smooth the</p><p>heatmap will look because it sets the distance around each point at which the influence of the</p><p>point will be felt. Therefore, smaller radius values result in heatmaps that display finer details,</p><p>while larger values result in smoother heatmaps.</p><p>Besides the kernel bandwidth, there are also different kernel shapes to choose from. The</p><p>kernel shape controls the rate at which the influence of a point decreases with increasing</p><p>distance from the point. The kernel shapes that are available in the Heatmap plugin can be</p><p>seen in the following figures. For example, a Triweight kernel (the first on the bottom row)</p><p>creates smaller hotspots than the Epanechnikov kernel (the second on the bottom) because</p><p>the Triweight shape gives features a higher influence for distances that are closer to the point:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 5</p><p>139</p><p>The triangular kernel shape can be further adjusted using the Decay ratio setting. In the</p><p>preceding figure, you can see the shape for ratios of 0 (a solid red line), 0.5 (a dashed black</p><p>line), and 1 (a dotted black line), which is equal to the uniform kernel shape. You can even</p><p>specify values greater than 1. In this case, the influence of a feature will increase with the</p><p>distance from the point.</p><p>Estimating values based on samples</p><p>Interpolation is the idea that, with a set of known values, you can estimate the values of</p><p>additional points based on their proximity to these known values. This recipe shows you</p><p>how to use known values at point locations to create a continuous surface (raster) of value</p><p>estimates. Classic examples include weather data estimations that are based on weather</p><p>station data (think temperature or rainfall maps), crop yield estimates that are based on</p><p>sampling parts of a field, and like in this example in this recipe, elevation estimations that</p><p>are based on the elevation of sampled points.</p><p>Getting ready</p><p>Activate Interpolation Plugin via Plugin Manager.</p><p>Load a point layer with numeric columns, representing the feature of interest. For this recipe,</p><p>use the poi_names_wake.shp, and the elev_m column, which contains elevation in meters</p><p>for each point.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Classic Vector Analysis</p><p>140</p><p>How to do it…</p><p>1. Start by loading poi_name_wake.</p><p>2. Zoom to the layer extent.</p><p>3. Open the Interpolation tool by navigating to Raster | Interpolation | Interpolation.</p><p>Yes, it's on the Raster menu; the source data must be</p><p>a vector, but the results are a raster.</p><p>4. Select poi_names_wake for Input.</p><p>5. Select elev_m for Interpolation attribute.</p><p>6. Click on the Add button, your selection should appear in the box on the left-hand side.</p><p>7. Select Inverse Distance Weighted (IDW) for Interpolation Method.</p><p>8. Now, set the Extent and Cell Size properties. In Cellsize X and Cellsize Y, enter 100</p><p>and 100. This forces the output cells to be 100x100 units of the current projection.</p><p>Generally, if this was for analysis, you would attempt to</p><p>match the region of interest or other raster layers. In this</p><p>case, we just want to go for sensibly-sized cells. As the map</p><p>is in UTM, we will want cells to be integers that represent</p><p>metric units; 100 meters by 100 meters makes interpreting</p><p>the results easier.</p><p>9. Click on the Set to current extent button in the middle.</p><p>10. Next to Output file box, click on the button labeled ... to set the output path to save</p><p>the results:</p><p>11. Pick the folder and type in a name with no file extension, such as idw100m (the result</p><p>will be an ASCII raster .asc file), as shown in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 5</p><p>141</p><p>The wrench tool in the upper-right corner will let you change</p><p>the P value, which is the exponent in the denominator and</p><p>directly sets how much a point influences a nearby location,</p><p>as compared against more distanced points.</p><p>12. Check all your settings and then click on the OK button.</p><p>13. Now, wait patiently for your results, the smaller the size of the cell and the larger the</p><p>number of columns and data points, the longer the calculation will take, as shown in</p><p>the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Classic Vector Analysis</p><p>142</p><p>How it works…</p><p>The basic idea is that, at a given cell, you take the average of all the nearby points that are</p><p>weighted by their distance to the cell in order to estimate the value at your current location.</p><p>Inverse Distance Weighted (IDW) takes this one step further by giving more weight to values</p><p>that are closer to the given cell and less weight to values that are further. This function uses</p><p>an exponent factor P in order to greatly increase the role of closer points over distant points.</p><p>There's more…</p><p>Are the results not quite what you expected? There are a few parameters that can be adjusted;</p><p>these are primarily the P value and the size of the cell. Is this still not coming out the way that</p><p>you want? There are several other Interpolation tools that are accessible in Processing under</p><p>the SAGA, GRASS, and GDAL toolboxes, which allow you to manipulate more of the formula</p><p>parameters to refine the results.</p><p>Finally, depending on your data, IDW may not do a good job of interpolating. In the example</p><p>here, you can actually see how there are distinct circles around isolated points. This is generally</p><p>not a good result, and this needs a smoother transition to nearby points. If you have any control</p><p>over field sampling to begin with, keep in mind that regularly-spaced grids will usually provide</p><p>better results.</p><p>Do you not have control over the source data or you didn't get good results? Then, you may need</p><p>to look into other more complicated formulas that compensate for skew, strong directionality,</p><p>obstructions, and non-regular spacing of samples, such as Splines or Kriging, or Triangulated</p><p>Irregular Networks (TINs). There is lot of science and statistics behind the methods and</p><p>diagnostic tools to determine the best parameters. This is far too complicated a topic for this</p><p>recipe,</p><p>but it is well-covered in books on geostatistics.</p><p>See also</p><p>f http://docs.qgis.org/2.2/en/docs/user_manual/plugins/plugins_</p><p>interpolation.html</p><p>f http://en.wikipedia.org/wiki/Inverse_distance_weightinging</p><p>www.it-ebooks.info</p><p>http://docs.qgis.org/2.2/en/docs/user_manual/plugins/plugins_interpolation.html</p><p>http://docs.qgis.org/2.2/en/docs/user_manual/plugins/plugins_interpolation.html</p><p>http://en.wikipedia.org/wiki/Inverse_distance_weightinging</p><p>http://www.it-ebooks.info/</p><p>143</p><p>Network Analysis</p><p>In this chapter, we will cover the following recipes:</p><p>f Creating a simple routing network</p><p>f Calculating the shortest paths using the Road graph plugin</p><p>f Routing with one-way streets in the Road graph plugin</p><p>f Calculating the shortest paths with the QGIS network analysis library</p><p>f Routing point sequences</p><p>f Automating multiple route computation using batch processing</p><p>f Matching points to the nearest line</p><p>f Creating a network for pgRouting</p><p>f Visualizing the pgRouting results in QGIS</p><p>f Using the pgRoutingLayer plugin for convenience</p><p>f Getting network data from the OSM</p><p>Introduction</p><p>This chapter focuses on the common use cases that are related to routing within networks.</p><p>By far, the most common networks that are used to route are street networks. Other less</p><p>common cases include networks for indoor routing, that is, through rooms inside buildings,</p><p>or networks of shipping routes.</p><p>6</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Network Analysis</p><p>144</p><p>Networks and routing are in no way a GIS-only topic. You will find a lot of math literature</p><p>related to this, called Graph Theory. In this chapter, we will use the following terms to talk</p><p>about networks:</p><p>f A network (also known as graph) is a collection of connected objects</p><p>f These objects are called nodes (also known as vertices)</p><p>f The connections between nodes are called links (also known as edges)</p><p>The following figure explains these terms:</p><p>nodes (or vertices)</p><p>edges</p><p>(or links)</p><p>The two routing tools that are commonly used with QGIS are as follows:</p><p>f The Road graph plugin, which is one of the QGIS core plugins; that is, this plugin is</p><p>available in every QGIS installation, but you may have to activate it in Plugin Manager</p><p>f The PostGIS extension pgRouting, which can be used directly through the QGIS DB</p><p>Manager, or more comfortably through the pgRoutingLayer plugin, which can be</p><p>installed from the QGIS plugin repository using Plugin Manager</p><p>Creating a simple routing network</p><p>In this recipe, we will create a routing network from scratch using the QGIS editing tools. Even</p><p>though more and more open network data is available, there will still be numerous use cases</p><p>where necessary network data does not exist or is not available for use. Therefore, it is good to</p><p>know how to create a network and what to pay attention to in order to avoid common pitfalls.</p><p>For the task of network creation, the main difference between the Road graph plugin and</p><p>pgRouting is that pgRouting needs a network node (that is, link start or end node) at each</p><p>intersection while the Road graph plugin will also use intermediate link geometry nodes</p><p>to infer intersections if two links share a node. In this recipe, we will create a network,</p><p>which can be used in both tools.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 6</p><p>145</p><p>Getting ready</p><p>To follow this recipe, you only need a new empty QGIS project. Additionally, make sure you</p><p>have the Digitizing toolbar enabled (as shown in the following screenshot). We will create</p><p>an imaginary network, but if you want you can load a background map and digitize this:</p><p>How to do it…</p><p>Before we can start to create the network, there are a few things that need to be set up first:</p><p>1. Create a new shapefile line layer for the network. You don't need to add any extra</p><p>attributes besides the default ID attribute yet.</p><p>You can read more about creating new shapefiles in the Learning QGIS</p><p>book by Packt Publishing and the QGIS user guide at http://docs.</p><p>qgis.org/2.2/en/docs/user_manual/working_with_</p><p>vector/editing_geometry_attributes.html#creating-</p><p>a-new-shapefile-layer.</p><p>2. To ensure that we can digitize the network with valid topology, we'll activate snapping</p><p>next. Go to Settings | Snapping Options and activate snapping for your line layer</p><p>by enabling the checkbox to the left of it. Additionally, set the mode to to vertex and</p><p>choose a tolerance of at least 5.00000 pixels:</p><p>3. Now, we can enable editing for the line layer and then select the Add Feature tool</p><p>from the Digitizing toolbar to start digitizing.</p><p>www.it-ebooks.info</p><p>http://docs.qgis.org/2.2/en/docs/user_manual/working_with_vector/editing_geometry_attributes.html#creating-a-new-shapefile-layer</p><p>http://docs.qgis.org/2.2/en/docs/user_manual/working_with_vector/editing_geometry_attributes.html#creating-a-new-shapefile-layer</p><p>http://docs.qgis.org/2.2/en/docs/user_manual/working_with_vector/editing_geometry_attributes.html#creating-a-new-shapefile-layer</p><p>http://docs.qgis.org/2.2/en/docs/user_manual/working_with_vector/editing_geometry_attributes.html#creating-a-new-shapefile-layer</p><p>http://www.it-ebooks.info/</p><p>Network Analysis</p><p>146</p><p>4. Create the first line feature now, and give it the ID number 1. The line can have</p><p>as many nodes as you wish. We'll create a line with four nodes, as shown in the</p><p>following screenshot:</p><p>5. To draw the second line feature, start at the first or the last node of line 1. As we have</p><p>activated snapping, you will see that the node is being highlighted if you hover over it</p><p>with the mouse cursor. Draw a second line and give it the ID number 2.</p><p>The line in the preceding screenshot is drawn with a style that has circles</p><p>on the starting and ending points. You can reproduce this style by adding</p><p>the Marker line symbol levels to the line style or load network_links.</p><p>qml from our sample data. For more details about styling features, please</p><p>refer to Chapter 10, Cartography Tips.</p><p>6. Draw a few more lines (around 12 in total) forming a network. Make sure to pay</p><p>attention to the snapping and assign link IDs:</p><p>7. Disable editing, and confirm that you want to save the changes.</p><p>We will use this basic network as a starting point for the remaining recipes in this chapter.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 6</p><p>147</p><p>How it works…</p><p>By setting the snapping mode to to vertex, we made it possible to digitize the line network in</p><p>a way that ensures that lines, which should be connected, really contain a node at the exact</p><p>same position.</p><p>There's more…</p><p>You can validate the network topology by running the Topology Checker plugin, which is</p><p>installed with QGIS by default (you can read more about Topology Checker in Chapter 12,</p><p>Up and Coming):</p><p>1. Start Topology Checker from the Vector menu.</p><p>2. Click Configure to set up a topology rule, as shown in the following screenshot, and</p><p>click on Add Rule to add it to the list of rules to check. Then, close the settings by</p><p>clicking on OK:</p><p>3. Once this tool is configured, click on Validate All (the button with the checkmark) to</p><p>initiate the check. You will see the list of discovered errors displayed in the list above</p><p>the buttons, as shown in the following screenshot. Additionally, the dangling ends are</p><p>highlighted in red in the map:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Network Analysis</p><p>148</p><p>4. You can select the error entries in the list to jump to the line features that failed the</p><p>check. In our network, only lines with dead-ends should be listed. If you see an error</p><p>at an intersection, you should zoom closer and try to correct the node positions.</p><p>Calculating the shortest paths using the</p><p>Road graph plugin</p><p>This recipe shows you how to use the built-in Road graph plugin to calculate the shortest</p><p>paths in a network.</p><p>Getting ready</p><p>To follow this recipe, load network_pgr.shp from the sample data. Additionally, make sure</p><p>that the Road graph plugin is enabled in Plugin Manager.</p><p>How to do it…</p><p>The Road graph plugin enables us to route between two points that are</p><p>selected on the map.</p><p>Before we can use this, we have to configure it first, as follows:</p><p>1. Enable the Shortest path panel by navigating to View | Panels. This should add the</p><p>plugin panel to the user interface.</p><p>2. Go to Vector | Road graph | Settings to get to the configuration dialog. For now, the</p><p>default settings, as shown in the following screenshot, should be fine. Note that the</p><p>network layer is selected as Transportation layer. Click on OK to confirm the settings:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 6</p><p>149</p><p>3. Once the settings are configured, we can calculate our first route. Select the Start</p><p>and Stop locations using the buttons marked with crosshair icons. Activate the</p><p>crosshair button and then click in the map to select a location. This location will be</p><p>marked on the map, and the coordinates will be automatically inserted into the Start</p><p>or Stop input field.</p><p>4. Click on Calculate to initiate route computation. Depending on the size of the</p><p>network used, this step will either be very fast or it can take much more time. The</p><p>route will be highlighted in the map, and the route length and travel time will be</p><p>displayed, as shown in the following screenshot:</p><p>5. If you want to store the computed shortest path, click on Export and you will be able</p><p>to choose whether you want to create a new layer for the path or add the route to an</p><p>existing line layer.</p><p>6. To compute a new route, simply change the start and stop locations and click on</p><p>Calculate again.</p><p>7. Click on Clear to remove the route highlights when you are done.</p><p>How it works…</p><p>The Road graph plugin uses the QGIS network analysis library, which implements Dijkstra's</p><p>algorithm. For a given starting node, the algorithm finds the path with the lowest cost (that is,</p><p>the shortest path if the cost criterion is length or the fastest path if the cost criterion is time)</p><p>between this node and every other node in the network. This can also be used to find costs of</p><p>the shortest paths from a start node to a destination node by stopping the algorithm once the</p><p>shortest path to the destination node has been determined.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Network Analysis</p><p>150</p><p>In contrast to many simple routing tools, the Road graph plugin builds the network topology</p><p>automatically. As our network dataset is topologically sound (that is, there are no tiny</p><p>gaps where network edges meet), we can set up Road graph plugin settings with Topology</p><p>tolerance as 0. If you are using a network from a different source, it may not have been</p><p>created with the same attention to detail, and you may have to increase Topology tolerance</p><p>to get routing to work.</p><p>See also</p><p>f If you are interested in learning more about this algorithm, you can start at</p><p>http://wiki.gis.com/wiki/index.php/Dijkstra's_algorithm</p><p>Routing with one-way streets in the Road</p><p>graph plugin</p><p>When it comes to vehicle routing, it is often necessary to go into more detail and consider</p><p>driving restrictions, such as one-way streets. This recipe shows you how to use one-way street</p><p>information to route with the Road graph plugin.</p><p>Getting ready</p><p>To follow this recipe, load network_pgr.shp from the sample data. Additionally, make sure</p><p>that the Road graph plugin is enabled in Plugin Manager.</p><p>How to do it…</p><p>To demonstrate routing with one-way street information, we will first visualize the one-way</p><p>values, and then we will configure the Road graph plugin to use the one-way information,</p><p>as follows:</p><p>1. Before we start routing with one-way information, it is helpful to visualize the one-</p><p>way streets. It is worth noting that the one-way direction will depend on the direction</p><p>of the link geometry (that is, the direction the link was digitized in). The best way to</p><p>visualize the link direction is by assigning arrow symbols, as shown in the following</p><p>screenshot. You can load network_pgr.qml from our sample data to get the style:</p><p>www.it-ebooks.info</p><p>http://wiki.gis.com/wiki/index.php/Dijkstra's_algorithm</p><p>http://www.it-ebooks.info/</p><p>Chapter 6</p><p>151</p><p>There are many different ways to encode one-way information. In our dataset,</p><p>a forward direction is encoded as FT for "from-to", a backward direction as TF</p><p>for "to-from", and both ways as B for "both".</p><p>2. Then, we can configure the Road graph plugin to use the one-way information. To do</p><p>this, we have to choose the dir attribute as Direction field and enter the values for</p><p>forward (in link geometry) direction and reverse (against link geometry) direction:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Network Analysis</p><p>152</p><p>3. Once the plugin is configured, you can compute the shortest path as described in</p><p>the previous recipe, Calculating the shortest paths using the Road graph plugin. You</p><p>will see how the resulting routes differ from the normal (without one-way restrictions)</p><p>paths, as shown in the following screenshot where the algorithm avoids the one-way</p><p>links on the direct route and takes the longer route instead:</p><p>How it works…</p><p>When we use the default two-way setting, each network link is interpreted as a connection</p><p>from the start to end node, as well as a connection from the end to start node. By adding</p><p>one-way restrictions, this changes and the link is only interpreted as one connection now.</p><p>Besides FT, TF, and B, another common way to encode one-ways is 1 for in-link direction,</p><p>-1 for against-link direction, and 0 for both ways. In OpenStreetMap, you will find yes for the</p><p>in-link direction, no for both ways and -1 for the against-link direction (refer to http://wiki.</p><p>openstreetmap.org/wiki/Key:oneway for more details).</p><p>Calculating the shortest paths with the</p><p>QGIS network analysis library</p><p>As mentioned in the recipe, Calculating the shortest paths using the Road graph plugin, QGIS</p><p>comes with a network analysis library, which can be used from the Python console, inside</p><p>plugins, to process scripts, and basically anything else that you can think of. In this recipe,</p><p>we will introduce the usage of the network analysis to compute the shortest paths in the</p><p>Python console.</p><p>www.it-ebooks.info</p><p>http://wiki.openstreetmap.org/wiki/Key:oneway</p><p>http://wiki.openstreetmap.org/wiki/Key:oneway</p><p>http://www.it-ebooks.info/</p><p>Chapter 6</p><p>153</p><p>Getting ready</p><p>To follow this recipe, load network_pgr.shp from the sample data.</p><p>How to do it…</p><p>Instead of typing or copying the following script directly in the Python console, we recommend</p><p>opening the Python console editor using the Show editor button on the left-hand side of the</p><p>Python console:</p><p>1. Paste the following script into the editor:</p><p>import processing</p><p>from processing.tools.vector import VectorWriter</p><p>from PyQt4.QtCore import *</p><p>from qgis.core import *</p><p>from qgis.networkanalysis import *</p><p># create the graph</p><p>layer = processing.getObject('network_pgr')</p><p>director = QgsLineVectorLayerDirector(layer,-1,'','','',3)</p><p>director.addProperter(QgsDistanceArcProperter())</p><p>builder = QgsGraphBuilder(layer.crs())</p><p>from_point = QgsPoint(2.73343,3.00581)</p><p>to_point = QgsPoint(0.483584,2.01487)</p><p>tied_points =</p><p>director.makeGraph(builder,[from_point,to_point])</p><p>graph = builder.graph()</p><p># compute the route from from_id to to_id</p><p>from_id = graph.findVertex(tied_points[0])</p><p>to_id = graph.findVertex(tied_points[1])</p><p>(tree,cost) = QgsGraphAnalyzer.dijkstra(graph,from_id,0)</p><p># assemble the route</p><p>route_points = []</p><p>curPos = to_id</p><p>while (curPos != from_id):</p><p>in_vertex = graph.arc(tree[curPos]).inVertex()</p><p>route_points.append(graph.vertex(in_vertex).point())</p><p>curPos = graph.arc(tree[curPos]).outVertex()</p><p>route_points.append(from_point)</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Network Analysis</p><p>154</p><p># write the results to a Shapefile</p><p>result = 'C:\\temp\\route.shp'</p><p>writer = VectorWriter(result,None,[],2,layer.crs())</p><p>fet = QgsFeature()</p><p>fet.setGeometry(QgsGeometry.fromPolyline(route_points))</p><p>writer.addFeature(fet)</p><p>del writer</p><p>processing.load(result)</p><p>2. If you are using your own network dataset instead of network_pgr.shp, which is</p><p>provided with this book, adjust the coordinates of from_point</p><p>and Coming, hints at the future with cutting-edge plugins and how to</p><p>participate in the future development of QGIS.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Preface</p><p>vii</p><p>What you need for this book</p><p>We recommend installing QGIS 2.8 or later; you will need at least QGIS 2.4. During the writing</p><p>of this book, several new versions were released, approximately every 4 months, and most</p><p>recently, 2.14 was released. Most of the recipes will work on older versions, but some may</p><p>require 2.6 or newer. In general, if you can, upgrade to the latest stable release or Long Term</p><p>Support (LTS) version.</p><p>There are also a lot of side interactions with other software throughout many of these</p><p>recipes, including—but not limited to—Postgis 2+, GRASS 6.4+, SAGA 2.0.8+, and Spatialite</p><p>4+. On Windows, most of these can be installed using OSGeo4W; on Mac, you may need</p><p>some additional frameworks from Kyngchaos, or if you're familiar with Brew, you can use the</p><p>OSGeo4Mac Tap. For Linux users, in particular Ubuntu and Debian, refer to the UbuntuGIS</p><p>PPA and the DebianGIS blend.</p><p>Does all of this sound a little too complicated? If yes, then consider using a virtual machine</p><p>that runs OSGeo-Live (http://live.osgeo.org). All the software is preinstalled for you</p><p>and is known to work together.</p><p>Lastly, you will need data. For the most part, we've provided a lot of free and open data</p><p>from a variety of sources, including the OSGeo Educational dataset (North Carolina), Natural</p><p>Earth Data, OpenFlights, Wake County, City of Davis, and Armed Conflict Location & Event</p><p>Data Project (ACLED). A full list of our data sources is provided here if you would like</p><p>additional data.</p><p>We recommend that you try methods with the sample data first, only because we tested it.</p><p>Feel free to try using your own data to test many of the recipes; however, just remember that</p><p>you might need to alter the structure to make it work. After all, that's what you'll be working</p><p>with normally.</p><p>The following are the data sources for this book:</p><p>f OSGeo Educational Data: http://grass.osgeo.org/download/sample-data/</p><p>f Wake County, USA: http://www.wakegov.com/gis/services/pages/data.</p><p>aspx</p><p>f Natural Earth Data: http://www.naturalearthdata.com/</p><p>f City of Davis, USA: http://maps.cityofdavis.org/library</p><p>f Stamen Designs: http://stamen.com/</p><p>f Armed Conflict Location & Event Data Project: http://www.acleddata.com/</p><p>www.it-ebooks.info</p><p>http://live.osgeo.org</p><p>http://grass.osgeo.org/download/sample-data/</p><p>http://www.wakegov.com/gis/services/pages/data.aspx</p><p>http://www.wakegov.com/gis/services/pages/data.aspx</p><p>http://www.naturalearthdata.com/</p><p>http://maps.cityofdavis.org/library</p><p>http://stamen.com/</p><p>http://www.acleddata.com/</p><p>http://www.it-ebooks.info/</p><p>Preface</p><p>viii</p><p>Who this book is for</p><p>This book is for anyone who wants to do more with QGIS. It's aimed at an intermediate to</p><p>advanced audience that already has some experience using GIS (any). The goal is to master</p><p>the fundamentals of QGIS and launch you, the reader, to the next level of being a QGIS power</p><p>user and community member.</p><p>Whatever your current level of experience with QGIS, you will find a lot of new ways to put your</p><p>skills to good use. For those who are new to GIS, this book shows you many of the things that</p><p>are possible with GIS. For those of you who are new to QGIS, this book is a guide on how to do</p><p>the things you know from elsewhere with QGIS. For advanced users, this book is a reference</p><p>and cheat sheet to the common tools you use often but can't quite remember how they work.</p><p>And for all users, this book is filled with unknown and barely documented aspects of QGIS</p><p>that you didn't think possible but likely want to use.</p><p>Sections</p><p>In this book, you will find several headings that appear frequently (Getting ready, How to do it,</p><p>How it works, There's more, and See also).</p><p>To give clear instructions on how to complete a recipe, we use these sections as follows:</p><p>Getting ready</p><p>This section tells you what to expect in the recipe, and describes how to set up any software or</p><p>any preliminary settings required for the recipe.</p><p>How to do it…</p><p>This section contains the steps required to follow the recipe.</p><p>How it works…</p><p>This section usually consists of a detailed explanation of what happened in the previous</p><p>section.</p><p>There's more…</p><p>This section consists of additional information about the recipe in order to make the reader</p><p>more knowledgeable about the recipe.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Preface</p><p>ix</p><p>See also</p><p>This section provides helpful links to other useful information for the recipe.</p><p>Conventions</p><p>In this book, you will find a number of text styles that distinguish between different kinds of</p><p>information. Here are some examples of these styles and an explanation of their meaning.</p><p>Code words in text, database table names, folder names, filenames, file extensions,</p><p>pathnames, dummy URLs, user input, and Twitter handles are shown as follows: "Create a</p><p>Spatialite database if you don't already have one and name it cookbook.db."</p><p>A block of code is set as follows:</p><p>geom,id,elevation</p><p>LINESTRING(0 1, 0 2, 1 3),1,50</p><p>LINESTRING(0 -1, 0 -2, 1 -3),2,60</p><p>LINESTRING(0 1, 0 3, 5 4),3,70</p><p>When we wish to draw your attention to a particular part of a code block, the relevant lines or</p><p>items are set in bold:</p><p>geom,id,elevation</p><p>LINESTRING(0 1, 0 2, 1 3),1,50</p><p>LINESTRING(0 -1, 0 -2, 1 -3),2,60</p><p>LINESTRING(0 1, 0 3, 5 4),3,70</p><p>Any command-line input or output is written as follows:</p><p>-gcp sourceX sourceY destinationX destinationY</p><p>New terms and important words are shown in bold. Words that you see on the screen, for</p><p>example, in menus or dialog boxes, appear in the text like this: "Check the Use visible raster</p><p>layers checkbox or choose SELECT."</p><p>Warnings or important notes appear in a box like this.</p><p>Tips and tricks appear like this.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Preface</p><p>x</p><p>Reader feedback</p><p>Feedback from our readers is always welcome. Let us know what you think about this book—</p><p>what you liked or disliked. Reader feedback is important for us as it helps us develop titles</p><p>that you will really get the most out of.</p><p>To send us general feedback, simply e-mail feedback@packtpub.com, and mention the</p><p>book's title in the subject of your message.</p><p>If there is a topic that you have expertise in and you are interested in either writing or</p><p>contributing to a book, see our author guide at www.packtpub.com/authors.</p><p>Customer support</p><p>Now that you are the proud owner of a Packt book, we have a number of things to help you to</p><p>get the most from your purchase.</p><p>Downloading the example code</p><p>You can download the example code files for this book from your account at http://</p><p>www.packtpub.com. If you purchased this book elsewhere, you can visit http://www.</p><p>packtpub.com/support and register to have the files e-mailed directly to you.</p><p>You can download the code files by following these steps:</p><p>1. Log in or register to our website using your e-mail address and password.</p><p>2. Hover the mouse pointer on the SUPPORT tab at the top.</p><p>3. Click on Code Downloads & Errata.</p><p>4. Enter the name of the book in the Search box.</p><p>5. Select the book for which you're looking to download the code files.</p><p>6. Choose from the drop-down menu where you purchased this book from.</p><p>7. Click on Code Download.</p><p>Once the file is downloaded, please make sure that you unzip or extract the folder using the</p><p>latest version of:</p><p>f WinRAR / 7-Zip for Windows</p><p>f Zipeg / iZip / UnRarX for Mac</p><p>f 7-Zip / PeaZip for Linux</p><p>www.it-ebooks.info</p><p>www.packtpub.com/authors</p><p>http://www.packtpub.com</p><p>http://www.packtpub.com</p><p>http://www.packtpub.com/support</p><p>http://www.packtpub.com/support</p><p>http://www.it-ebooks.info/</p><p>Preface</p><p>xi</p><p>Downloading the color images of this book</p><p>We also provide you with a PDF file that has color images of the screenshots/diagrams used</p><p>in this book. The color images will help you better understand the changes in the output.</p><p>You can download this file from http://www.packtpub.com/sites/default/files/</p><p>downloads/QGIS2Cookbook_ColorImages.pdf.</p><p>and to_point for</p><p>the route's starting and ending points.</p><p>3. Change the file paths for the result layer depending on your operating system.</p><p>4. Make sure that the network layer is loaded and selected in the QGIS layer list.</p><p>5. Save the script and run it.</p><p>How it works…</p><p>On line 8, we created a QgsLineVectorLayerDirector object (http://qgis.org/api/</p><p>classQgsLineVectorLayerDirector.html), which contains the network configuration.</p><p>The constructor (QgsLineVectorLayerDirector(layer,-1,'','','',3)) parameters</p><p>are as follows:</p><p>f The network line layer</p><p>f The ID of the direction field: we set it to -1 because this script does not consider</p><p>one-ways</p><p>f The following three parameters are the values for the in link direction: reverse link</p><p>direction, and two-way</p><p>f The last parameter is the default direction: 1 for the in link direction, 2 for the reverse</p><p>direction, and 3 for the two-way</p><p>Line 10 creates the QgsGraphBuilder (http://qgis.org/api/</p><p>classQgsGraphBuilder.html) instance, which will be used to create the routing graph</p><p>on line 14.</p><p>On lines 11 and 12, we defined the starting and ending points of our route. To be able to route</p><p>between these two points, they have to be matched to the nearest network link. This happens</p><p>on line 13 in the makeGraph() function, which returns the so-called tied_points.</p><p>The actual route computation takes place on line 18 in the QgsGraphAnalyzer.</p><p>dijkstra() (http://qgis.org/api/classQgsGraphAnalyzer.html) function.</p><p>www.it-ebooks.info</p><p>http://qgis.org/api/classQgsLineVectorLayerDirector.html</p><p>http://qgis.org/api/classQgsLineVectorLayerDirector.html</p><p>http://qgis.org/api/classQgsGraphBuilder.html</p><p>http://qgis.org/api/classQgsGraphBuilder.html</p><p>http://qgis.org/api/classQgsGraphAnalyzer.html</p><p>http://www.it-ebooks.info/</p><p>Chapter 6</p><p>155</p><p>The while loop, starting on line 22, is where the script moves through the tree created by</p><p>Dijkstra's algorithm to collect all the vertices on the way and add them to the route_points</p><p>list, which becomes the resulting route geometry on line 31.</p><p>The writer for output route line layer is created on line 29, where we pass the file path, None</p><p>for default encoding, the [] for empty fields list, and the 2 for geometry type, which equals to</p><p>lines as well as the resulting layer CRS. The following lines, 30 to 32, create the route feature</p><p>and add it to the writer.</p><p>Finally, the last line loads the resulting shapefile, and this is displayed on the map, as</p><p>illustrated by the following screenshot:</p><p>See also</p><p>You can read more about QGIS's network analysis library online in the PyQGIS Developer</p><p>Cookbook at http://docs.qgis.org/testing/en/docs/pyqgis_developer_</p><p>cookbook/network_analysis.html.</p><p>www.it-ebooks.info</p><p>http://docs.qgis.org/testing/en/docs/pyqgis_developer_cookbook/network_analysis.html</p><p>http://docs.qgis.org/testing/en/docs/pyqgis_developer_cookbook/network_analysis.html</p><p>http://www.it-ebooks.info/</p><p>Network Analysis</p><p>156</p><p>Routing point sequences</p><p>In the recipes so far, we routed from one starting point to one destination point. Another use</p><p>case is when we want to compute routes that connect a sequence of points, such as the</p><p>points in a GPS track. In this recipe, we will use the point layer to route processing script to</p><p>compute a route for a point sequence. At its core, this script uses the same idea that was</p><p>introduced in the previous recipe, Calculating the shortest paths with the QGIS network</p><p>analysis library, but this computes several shortest paths one after the other.</p><p>Getting ready</p><p>To follow this recipe, load network_pgr.shp and sample_pts_for_routing.shp,</p><p>which contains a point layer that should be routed from the sample dataset.</p><p>Additionally, you need to get the point layer to route script from https://raw.</p><p>githubusercontent.com/anitagraser/QGIS-Processing-tools/master/2.6/</p><p>scripts/point_layer_to_route.py and save it in the Processing script folder, which</p><p>is set to C:\Users\youruser\qgis2\processing\scripts (on Windows), /home/</p><p>youruser/.qgis2/processing/scripts (on Linux), and /Users/youruser/.qgis2/</p><p>processing/scripts (on Mac) by default. Alternatively, save the point layer to route to the</p><p>folder configured in the Processing menu under Options | Scripts | Scripts folder.</p><p>How to do it…</p><p>To compute the route between the input points, you need to perform the following tasks:</p><p>1. Load the network and the point layer.</p><p>2. If you are using your own data, make sure that both layers are in the same CRS.</p><p>If they are in different CRS, you need to reproject them (for example, using the</p><p>Reproject layer tool from the Processing Toolbox option) before you continue.</p><p>3. Start the point layer to route tool from the Processing Toolbox option.</p><p>4. Pick the points and network input layers, make sure that Open output file</p><p>after running algorithm option is activated, and click on Run to start the route</p><p>computation. The resulting route layer will be loaded automatically:</p><p>www.it-ebooks.info</p><p>https://raw.githubusercontent.com/anitagraser/QGIS-Processing-tools/master/2.6/scripts/point_layer_to_route.py</p><p>https://raw.githubusercontent.com/anitagraser/QGIS-Processing-tools/master/2.6/scripts/point_layer_to_route.py</p><p>https://raw.githubusercontent.com/anitagraser/QGIS-Processing-tools/master/2.6/scripts/point_layer_to_route.py</p><p>http://www.it-ebooks.info/</p><p>Chapter 6</p><p>157</p><p>How it works…</p><p>The point layer to route tool uses the QGIS network analysis library. We already discussed the</p><p>basic use of this library in the previous recipe, Calculating the shortest paths with the QGIS</p><p>network analysis library. The main difference is that we now have to handle more than two</p><p>points. Therefore, the script fetches all points from the input point layer and ties or matches</p><p>them to the graph:</p><p>points = []</p><p>features = processing.features(point_layer)</p><p>for f in features:</p><p>points.append(f.geometry().asPoint())</p><p>tiedPoints = director.makeGraph(builder, points)</p><p>For each pair of consecutive points, the script then computes the route between the two</p><p>points just like we did in the Calculating the shortest paths with the QGIS network analysis</p><p>library recipe:</p><p>point_count = point_layer.featureCount()</p><p>for i in range(0,point_count-1):</p><p># compute the route between two consecutive points</p><p>The resulting route line layer contains one line feature for each consecutive point pair.</p><p>There's more…</p><p>Of course, you can also use the point layer to route tool to route between only two points</p><p>as well.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Network Analysis</p><p>158</p><p>See also</p><p>There is also a version of this script, which takes one-way information into account at</p><p>https://raw.githubusercontent.com/anitagraser/QGIS-Processing-tools/</p><p>master/2.2/scripts/point_layer_to_route_with_oneways.py.</p><p>Automating multiple route computation</p><p>using batch processing</p><p>If you have multiple input point layers, you can use Processing's batch processing capabilities</p><p>to speed up the process. In this recipe, we will compute routes for two input point layers at</p><p>once, but the same approach can be applied to many more layers.</p><p>Getting ready</p><p>To follow this recipe, load network_pgr.shp and the two point layers, sample_pts_for_</p><p>routing.shp and sample_pts_for_routing2.shp.</p><p>How to do it…</p><p>To get started, right-click on the point layer to route tool in the Processing Toolbox option</p><p>and select Execute as batch process. Then perform the following steps:</p><p>1. In the points column, click the … button and use the Select from open layers option</p><p>to select sample_pts_for_routing and sample_pts_for_routing2.</p><p>2. Select and remove the third line using the Delete row button at the bottom of</p><p>the dialog.</p><p>3. In the network column, click the … button and use the Select from open layers</p><p>option to select network_pgr. You can avoid having to pick the file twice by double-</p><p>clicking on the table header entry (where it reads network). This will autofill all rows</p><p>with the same network file path.</p><p>4. In the routes column, you need to pick a path for the resulting route files. In the Save</p><p>file dialog, which opens when you</p><p>click on the … button, you can specify one base</p><p>filename and click on Save. The following Autofill settings dialog lets you specify if</p><p>and how you want to have the rows filled. Use Autofill mode Fill with numbers and</p><p>Processing will automatically append a running number to the filename that you</p><p>specified. You can see an example in the following screenshot where we specified</p><p>route as the base filename.</p><p>www.it-ebooks.info</p><p>https://raw.githubusercontent.com/anitagraser/QGIS-Processing-tools/master/2.2/scripts/point_layer_to_route_with_oneways.py</p><p>https://raw.githubusercontent.com/anitagraser/QGIS-Processing-tools/master/2.2/scripts/point_layer_to_route_with_oneways.py</p><p>http://www.it-ebooks.info/</p><p>Chapter 6</p><p>159</p><p>5. Click on Run to execute the batch process. Both routes will be computed and loaded</p><p>automatically:</p><p>Matching points to the nearest line</p><p>In this recipe, we will use the QGIS network analysis library from Python console to match</p><p>points to the nearest line. This is the simplest form of what is also known as map matching.</p><p>Getting ready</p><p>To follow this recipe, load network_pgr.shp from the sample data.</p><p>How to do it…</p><p>The following script will match three points, QgsPoint(3.63715,3.60401),</p><p>QgsPoint(3.86250,1.58906), and QgsPoint(0.42913,2.26512), to the network:</p><p>1. Open Python console and its editor and then load or paste the following network_</p><p>analysis_match_points.py script:</p><p>import processing</p><p>from processing.tools.vector import VectorWriter</p><p>from PyQt4.QtCore import *</p><p>from qgis.core import *</p><p>from qgis.networkanalysis import *</p><p>layer = processing.getObject('network_pgr')</p><p>director = QgsLineVectorLayerDirector(layer,-1,'','','',3)</p><p>director.addProperter(QgsDistanceArcProperter())</p><p>builder = QgsGraphBuilder(layer.crs())</p><p>additional_points =</p><p>[QgsPoint(3.63715,3.60401),QgsPoint(3.86250,1.58906),QgsPoi</p><p>nt(0.42913,2.26512)]</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Network Analysis</p><p>160</p><p>tied_points = director.makeGraph(builder,additional_points)</p><p>result = 'C:\\temp\\matched_pts.shp'</p><p>writer = VectorWriter(result,None,[],1,layer.crs())</p><p>fet = QgsFeature()</p><p>for pt in tied_points:</p><p>fet.setGeometry(QgsGeometry.fromPoint(pt))</p><p>writer.addFeature(fet)</p><p>del writer</p><p>processing.load(result)</p><p>2. Make sure that the network layer is selected in the layer list.</p><p>3. Run the script. The results should be loaded automatically.</p><p>How it works…</p><p>This script uses the QGIS network analysis library's ability to match points to lines using the</p><p>makeGraph() function. The resulting tied_points list contains the coordinates of the</p><p>points on the network that are closest to the input points.</p><p>The 1 option on line 15 specifies that the output layer is of type point.</p><p>The for loop finally goes through all points in the tied_points list and creates point</p><p>features, which are then added to the result writer.</p><p>Creating a routing network for pgRouting</p><p>This recipe shows you how to import a line layer into PostGIS and create a routable network</p><p>out of it, which can be used by PostGIS's routing library, pgRouting. (For details about</p><p>pgRouting, please visit the project website at http://docs.pgrouting.org.)</p><p>The installation of PostGIS with pgRouting won't be covered in detail here because</p><p>instructions for the different operating systems can be found on the project's website at</p><p>http://docs.pgrouting.org/2.0/en/doc/src/installation/index.html.</p><p>If you are using Windows, both PostGIS and pgRouting can be installed directly from the Stack</p><p>Builder application, which is provided by the standard PostgreSQL installation, as described</p><p>at http://anitagraser.com/2013/07/06/pgrouting-2-0-for-windows-quick-</p><p>guide/.</p><p>www.it-ebooks.info</p><p>http://docs.pgrouting.org</p><p>http://docs.pgrouting.org/2.0/en/doc/src/installation/index.html</p><p>http://anitagraser.com/2013/07/06/pgrouting-2-0-for-windows-quick-guide/</p><p>http://anitagraser.com/2013/07/06/pgrouting-2-0-for-windows-quick-guide/</p><p>http://www.it-ebooks.info/</p><p>Chapter 6</p><p>161</p><p>Getting ready</p><p>To follow this exercise, you need a PostGIS database with pgRouting enabled. In QGIS, you</p><p>should set up the connection to the database using the New button in the Add PostGIS</p><p>Layers dialog. Additionally, you should load network_pgr.shp from the sample data.</p><p>How to do it…</p><p>These steps will create a routable network table in your PostGIS database:</p><p>1. Open DB Manager by navigating to Database | DB Manager.</p><p>2. In Tree on the left-hand side of the dialog, select the database that you want to load</p><p>the network to.</p><p>3. Go to Table | Import Layer/File to load the network_pgr layer into your database,</p><p>as shown in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Network Analysis</p><p>162</p><p>4. After network_pgr has been imported successfully, open the SQL window of DB</p><p>Manager by pressing F2, clicking on the corresponding toolbar button, or in the</p><p>Database menu.</p><p>5. pgRouting is a little picky when it comes to column data types. You will notice this when</p><p>you see Error, columns 'source', 'target' must be of type int4, 'cost' must be of type</p><p>float8. When we import network_pgr with QGIS's DB Manager, it creates the cost</p><p>column as numeric. As pgRouting won't accept numeric, we will use Table | Edit Table</p><p>in DB Manager to edit the cost column. Click on the Edit column button and change</p><p>Type from numeric to double precision (which equals the required float8).</p><p>6. Now that the data is loaded and ready, we can build the network topology. This will</p><p>create a new network_pgr_vertices_pgr table, which contains the computed</p><p>network nodes:</p><p>SELECT pgr_createTopology('network_pgr',0.001);</p><p>7. Once this topology is ready, we can test the network by calculating a simple shortest</p><p>path from the node number 16 to the node number 9:</p><p>SELECT pgr_dijkstra('SELECT id, source, target, cost</p><p>FROM network_pgr', 16, 9, false, false);</p><p>This will result in the following:</p><p>(0,16,6,1)</p><p>(1,17,7,1)</p><p>(2,5,8,1)</p><p>(3,6,9,1)</p><p>(4,11,15,1)</p><p>(5,9,-1,0)</p><p>How it works…</p><p>The preceding pgr_dijkstra query consists of the following parts:</p><p>f 'SELECT id, source, target, cost FROM network_pgr': This is a SQL</p><p>query, which returns a set of rows with the following columns:</p><p>f id: This is the unique edge ID (type int4)</p><p>f source: This is the ID of the edge source node (type int4)</p><p>f target: This is the ID of the edge target node (type int4)</p><p>f cost: This is the cost of the edge traversal (type float8)</p><p>f 16, 9: These are the IDs of the route source and target nodes (type int4)</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 6</p><p>163</p><p>f false: This is true if the graph is directed</p><p>f false: If true, the reverse_cost column of the SQL-generated set of rows will be</p><p>used for the cost of the traversal of the edge in the opposite direction</p><p>The results of pgr_dijkstra contain the list of network links that our route uses to get from</p><p>the start to the destination. The four values in reach result row are as follows:</p><p>f seq: This is the sequence number, which tells us the order of the links within the</p><p>route starting from 0</p><p>f id1: This is the node ID</p><p>f id2: This is the edge ID</p><p>f cost: This is the cost of the link (can be distance, travel time, a monetary value,</p><p>or any other measure that you chose)</p><p>See also</p><p>In the following recipe, Visualizing pgRouting results in QGIS, we will see how to use the</p><p>results of pgr_dijkstra to visualize the route on a map.</p><p>If you are interested in more pgRouting SQL recipes, you will find a whole chapter on this topic</p><p>in PostGIS Cookbook by Packt Publishing.</p><p>Visualizing the pgRouting results in QGIS</p><p>In the previous recipe, Creating a routing network for pgRouting, we imported a network layer,</p><p>built the topology, and finally tested the routing. Building on these results, this recipe will show</p><p>you how to visualize the routing results on a map in QGIS.</p><p>Getting ready</p><p>You should first go through the previous recipe, Creating a routing network for pgRouting, to</p><p>set up the necessary PostGIS tables. Alternatively, you can use your own network tables, but</p><p>be aware that you may have to alter</p><p>some of the SQL statements if your table uses different</p><p>column names.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Network Analysis</p><p>164</p><p>How to do it…</p><p>To visualize the results in QGIS, we can use the DB Manager SQL window, as shown in the</p><p>following screenshot. The extended query that we use here joins the routing results back to the</p><p>original network table to get the route link geometries, which we want to display on the map:</p><p>1. Open DB Manager by navigating to Database | DB Manager.</p><p>2. In Tree to the left of the dialog, select the database that you want to load the</p><p>network to.</p><p>3. Open the SQL window of DB Manager and configure it, as shown in the following</p><p>screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 6</p><p>165</p><p>Note that there must not be a semicolon at the end of the SQL statement.</p><p>Otherwise, loading the results as a new layer will fail.</p><p>4. Make sure that the Geometry column is selected correctly and click on the Load now!</p><p>button to load the query result as a new layer, as shown in the following screenshot:</p><p>How it works…</p><p>As pgr_dijkstra only returns a list with the IDs of the route edges, we need to get the edge</p><p>geometries from the original network table in order to display the route on the map. Therefore,</p><p>we join the routing results with the network table on id2 (which contains the edge ID) and the</p><p>network table's id column.</p><p>See also</p><p>To make using pgRouting from within QGIS more convenient, the pgRoutingLayer plugin</p><p>provides a GUI to access many of pgRouting's functions. You will find an introduction</p><p>to this plugin in the Using the pgRoutingLayer plugin for convenience recipe.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Network Analysis</p><p>166</p><p>Using the pgRoutingLayer plugin for</p><p>convenience</p><p>The previous recipe, Visualizing pgRouting results in QGIS, showed you how to manually</p><p>add pgRouting results to the map. In this chapter, we will use the pgRoutingLayer plugin</p><p>to get more convenient access to the functions that pgRouting offers, including the most</p><p>basic algorithms, such as Dijkstra's algorithm, which we have used so far, to more complex</p><p>algorithms, such as drivingDistance and alphashape, which can be used to visualize</p><p>catchment zones, also known as service areas.</p><p>Getting ready</p><p>You should first go through the previous recipe, Creating a routing network for pgRouting, to</p><p>set up the necessary PostGIS tables. Alternatively, you can use your own network tables, but</p><p>be aware that you may have to alter some of the SQL statements if your table uses different</p><p>column names.</p><p>Additionally, install the pgRoutingLayer plugin from Plugin Installer. You will need to enable</p><p>experimental plugins in Settings to view this.</p><p>How to do it…</p><p>The pgRoutingLayer plugin adds a new panel to the QGIS GUI, which allows convenient access</p><p>to the available routing functions. The following steps show you how to use this plugin:</p><p>1. First, you should select a database from the Database field that contains your routing</p><p>network table. The drop-down list contains all the configured PostGIS connections.</p><p>2. Next, you can select a function from the Function field that you want to use.</p><p>Let's try Dijkstra's algorithm first; select the dijkstra function. You will recognize</p><p>the parameters from the previous recipes where we wrote the pgRouting SQL</p><p>query manually.</p><p>3. Specify the parameters for the network table (edge_table) and the geometry, id,</p><p>source, target, and cost columns, as shown in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 6</p><p>167</p><p>4. Now, you can use the green + buttons beside the source_id and target_id input</p><p>fields to select the source and target nodes in the map.</p><p>5. When everything is configured, you can click on the Run button to compute and</p><p>display the route.</p><p>6. Next, you can switch functions and compute a service area. Select the alphashape</p><p>function. The rest of the input fields adapt automatically to the selected function.</p><p>7. Now, you can use the green + button right beside the source_id input field to select</p><p>the starting or center node of the service area.</p><p>8. Then, select the size of the service area by specifying the distance limit.</p><p>9. Finally, click on the Run button to compute and display the service area, as shown in</p><p>the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Network Analysis</p><p>168</p><p>How it works…</p><p>When we click on the Run button, the query results are visualized as a temporary overlay</p><p>on the map. If you want to save the output permanently, you can click on the Export button.</p><p>Currently, the Export button is only available for the routing functions but not for the service</p><p>area functions.</p><p>See also</p><p>For a detailed documentation on the pgRouting algorithms, refer to the project documentation</p><p>website at http://docs.pgrouting.org/2.0/en/doc/index.html.</p><p>Getting network data from the OSM</p><p>A popular data source for real-world routing applications is OpenStreetMap (OSM). This recipe</p><p>shows you how to prepare OSM data for usage with pgRouting using the osm2po command-</p><p>line tool to convert OSM data to an insert script for PostGIS. Finally, we will test the data</p><p>import using the pgRoutingLayer plugin.</p><p>Getting ready</p><p>Download osm2po from http://osm2po.de and unpack the download. Note that osm2po</p><p>requires Java to be installed on your machine.</p><p>You also need a pgRouting-enabled database to follow this recipe.</p><p>Additionally, you should have the pgRoutingLayer plugin installed and enabled because we will</p><p>use this to test the OSM data import.</p><p>You can use the wake.pbf OSM file from our sample data, or download your own data from</p><p>services such as http://download.geofabrik.de.</p><p>How to do it…</p><p>Open the command line to perform the following steps. If you are working on Windows, we</p><p>recommend using the osgeo4W Shell:</p><p>1. Go to the osm2po folder and open osm2po.config in a text editor. Look for the</p><p>following configuration line and remove the # at the beginning of the line to activate</p><p>the pgRouting export:</p><p>postp.0.class = de.cm.osm2po.plugins.postp.PgRoutingWriter</p><p>www.it-ebooks.info</p><p>http://docs.pgrouting.org/2.0/en/doc/index.html</p><p>http://osm2po.de</p><p>http://download.geofabrik.de</p><p>http://www.it-ebooks.info/</p><p>Chapter 6</p><p>169</p><p>2. Now use osm2po to convert the OSM .pbf file to SQL. Adjust the file paths for your</p><p>system, as follows:</p><p>D:\osm2po-5.1.0>java -jar osm2po-core-5.1.0-signed.jar prefix=wake</p><p>"C:\tmp\OSM_NorthCarolina\wake.pbf"</p><p>3. When osm2po is finished, you should see the following:</p><p>INFO Services started. Waiting for requests at</p><p>http://localhost:8888/Osm2poService</p><p>4. You should now find a folder with the name of the prefix (that is, wake) inside the</p><p>osm2po folder. This contains a log file, which in turn provides a command-line</p><p>template to import the OSM network to PostGIS:</p><p>INFO commandline template:</p><p>psql -U [username] -d [dbname] -q -f</p><p>"D:\osm2po-5.1.0\wake\wake_2po_4pgr.sql"</p><p>5. Using this template, we can easily import the .sql file into an existing database,</p><p>as follows:</p><p>D:\osm2po-5.1.0\wake>psql -U [username] -d cookbook -q -f D:\</p><p>osm2po-5.1.0\wake\wake_2po_4pgr.sql</p><p>6. Now, the data is ready for use in QGIS. When we connect to the cookbook database,</p><p>we can see the wake_2po_4pgr table:</p><p>7. Finally, we can use the pgRouting Layer plugin to test the OSM data import by</p><p>calculating a service area of 0.1 hours (the distance value) around the 43679</p><p>(source_id) source node number:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Network Analysis</p><p>170</p><p>How it works…</p><p>The network table created by osm2po contains, among others, the following useful columns:</p><p>f km: This is the length of the network edge</p><p>f kmh: The is the speed on the edge, depending on the street class and the values</p><p>specified in the osm2po configuration</p><p>f cost: This is the travel time computed using km/kmh</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>171</p><p>7</p><p>Raster Analysis I</p><p>In this chapter, we will cover the following recipes:</p><p>f Using the raster calculator</p><p>f Preparing elevation data</p><p>f Calculating a slope</p><p>f Calculating a hillshade layer</p><p>f Analyzing hydrology</p><p>f Calculating a topographic index</p><p>f Automate analysis tasks using the graphical modeler</p><p>Introduction</p><p>Raster analysis is a classic area in GIS analysis. This chapter shows you some of the most</p><p>important and common tasks of raster analysis. Elevation data is commonly stored as raster</p><p>layers, and in this format, it is particularly suitable to run a large variety of analysis. For this</p><p>reason, terrain analysis has traditionally been one of the main areas of raster analysis, and</p><p>we will show you some of the most common operations that are related to Digital Elevation</p><p>Models (DEM), from simple analysis, such as slope calculation, to more complex ones, such</p><p>as drainage network delineation or watershed extraction.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis I</p><p>172</p><p>Using the raster calculator</p><p>The raster calculator is one of the most flexible and versatile tools in QGIS. This allows you</p><p>to perform algebraic operations based on raster layers, and compute new layers. This recipe</p><p>shows you how to use it.</p><p>Getting ready</p><p>Open the catchment_area.tif file. The file should look like the following screenshot:</p><p>How to do it…</p><p>1. Open the Processing Toolbox option and find the algorithm called Raster calculator</p><p>by searching for it using the search box. Double-click on the algorithm item to execute</p><p>it, as shown in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 7</p><p>173</p><p>2. Click on the button in the Input layers field to open the layer selector. There is only</p><p>one layer available: the catchment_area layer. Select this layer.</p><p>3. In the Formula field, enter ln(a).</p><p>4. Click on Run to run the algorithm. The resulting layer will be added to the QGIS</p><p>project, as follows:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis I</p><p>174</p><p>How it works…</p><p>The layers selected in the layer selector are referred to using a single letter in alphabetical</p><p>order (a for the first one, b for the second one, and so on). In this case, we selected just one</p><p>layer, so we can refer to it as a in the formula.</p><p>The formula calculates a natural logarithm of the values in the catchment area layer. The</p><p>distribution of values in this layer is not homogeneous because it contains a large number</p><p>of cells with low values and just a few of them with very large values. This causes the</p><p>rendering of the layer to be not very informative with most of the colors in the color ramp</p><p>not even being used.</p><p>The resulting layer is much more informative because applying the logarithm alters the</p><p>distribution of values, resulting in a more explicit rendering.</p><p>There's more…</p><p>QGIS contains a raster calculator module outside of Processing. You can find this by</p><p>navigating to Raster | Raster calculator...:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 7</p><p>175</p><p>This interface resembles an actual calculator, and it is more intuitive and user friendly. On</p><p>the other hand, this lacks many of the functions that are available in the Processing raster</p><p>calculator (the logarithm that we have computed, for instance, is not available). This also</p><p>cannot be used in automated processes, such as scripts or graphical models, which are</p><p>only available for the Processing algorithms.</p><p>On the other hand, the QGIS built-in calculator supports multiband layers, while the</p><p>Processing one is limited to single-band ones.</p><p>See also</p><p>f The QGIS raster calculator is described in more detail in the QGIS manual at</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/working_with_</p><p>raster/raster_calculator.html</p><p>Preparing elevation data</p><p>In this recipe, we will show you how to perform terrain analysis in QGIS. Terrain analysis</p><p>algorithms assume certain characteristics in the DEMs that are used as inputs, so it is</p><p>important to know them and prepare these DEMs if they are needed. This recipe shows</p><p>you how to do this.</p><p>Getting ready</p><p>Open the dem_to_prepare.tif layer. This layer contains a DEM in the EPSG:4326 CRS</p><p>and elevation data in feet. These characteristics are unsuitable to run most terrain analysis</p><p>algorithms, so we will modify this layer to get a suitable one.</p><p>How to do it…</p><p>1. Reproject the layer to the EPSG:3857 CRS, using the Save as... entry in the context</p><p>menu that appears by right-clicking on the layer name.</p><p>2. Open the resulting reprojected layer.</p><p>3. Open the Processing raster calculator and select the reprojected layer as the only</p><p>raster input in the Input layers field. Enter a * 0.3048 in the Formula field. Run</p><p>the algorithm.</p><p>www.it-ebooks.info</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/working_with_raster/raster_calculator.html</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/working_with_raster/raster_calculator.html</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis I</p><p>176</p><p>How it works…</p><p>Most of the algorithms that we are going to use assume that the horizontal units (the unit</p><p>used to measure the size of the cell) are the same as the units used in the elevation values</p><p>that are contained in the layer. If the layer does not meet this requirement, the result of the</p><p>analysis will be wrong.</p><p>Our input layer uses a CRS with geographic coordinates (degrees). As elevation cannot</p><p>be measured in degrees, the layer cannot have the same units for horizontal and vertical</p><p>distances, and it is not ready to be used for terrain analysis.</p><p>By reprojecting the layer to the EPSG:3857 CRS, we get a new layer in which coordinates are</p><p>expressed in meters. This is a unit that is more suitable for the type of analysis that we plan to</p><p>run. Actually, after the reprojection, the units are meters only near the equator, but this gives</p><p>us enough precision for this case. If more precise calculations are needed, a local projection</p><p>system should be used.</p><p>The next step is converting the elevation values in feet to elevation values in meters. Knowing</p><p>that 1 foot = 0.3048 meter, we just have to use the calculator to apply this formula and</p><p>convert the values in the reprojected layer.</p><p>There's more…</p><p>There are other things that must be taken into account when running a terrain analysis</p><p>algorithm to ensure that results are correct.</p><p>One common problem is dealing with different cell sizes. An assumption that is made by most</p><p>terrain analysis algorithms (and also most of the ones not related to terrain analysis) is that</p><p>cells are square. That is, their horizontal and vertical values are the same. This is the case in</p><p>our input layer (you can verify this by checking the layer properties), but it may not be true for</p><p>other layers.</p><p>In this case, you should export the layer and define the sizes of the cells of the exported layer</p><p>to have the same value. Right-click on the layer name and select Save as.... In the save dialog</p><p>that will appear, enter the new sizes of the cells in the lower part of the dialog:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 7</p><p>177</p><p>Calculating a slope</p><p>Slope is one of the most basic parameters that can be derived from a DEM. It corresponds</p><p>to the first derivative of the DEM, and it represents the rate of change of the elevation. It is</p><p>computed by analyzing the elevation of each cell and comparing this with the elevation of</p><p>the surrounding ones. This recipe shows you how to compute slope in QGIS.</p><p>Getting ready</p><p>Open the DEM that we prepared in the previous recipe.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis I</p><p>178</p><p>How to do it…</p><p>1. In the Processing Toolbox option, find the Slope algorithm and double-click on it</p><p>to open it:</p><p>2. Select the DEM in the Input layer field.</p><p>3. Click on Run to run the algorithm.</p><p>The slope layer will be added to the QGIS project.</p><p>How it works…</p><p>Slope is calculated from a DEM elevation model by analyzing the cells around a given one.</p><p>This analysis is performed by the slope algorithm from the GDAL library.</p><p>There's more…</p><p>There are several ways of using the slope algorithms in QGIS. Here are some comments and</p><p>ideas about this.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 7</p><p>179</p><p>Using a ratio for elevation values</p><p>If the units of elevation are not the same as the horizontal units, you can convert them,</p><p>as we did in the previous recipe, using the raster calculator. However, the slope module</p><p>contains an option to convert them on-the-fly by entering the conversion factor in the Scale</p><p>field. Note that this option is not available in other terrain analysis modules that we will use,</p><p>so it's still good practice to create a layer with the correct units, which can be used without</p><p>any further processing.</p><p>Other slope algorithms</p><p>The Processing framework contains algorithms that rely on several external applications and</p><p>libraries. These libraries sometimes contain similar algorithms, so there is more than one</p><p>option for a given analysis.</p><p>If you switch the presentation mode of the toolbox from simplified to advanced using</p><p>the lower part of the drop-down list and then type slope in the search box, you will see</p><p>something like the following screenshot:</p><p>Calculating the slope</p><p>Try using the GRASS or SAGA algorithm to calculate the slope. Each of them has different</p><p>parameters and options, but all of them perform similar calculations and create slope layers.</p><p>Apart from Processing, you can also perform analysis using the Raster Terrain Analysis plugin.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis I</p><p>180</p><p>See also</p><p>f The Using the raster calculator recipe in the beginning of this chapter</p><p>Calculating a hillshade layer</p><p>A hillshade layer is commonly used to enhance the appearance of a map and display</p><p>topography in an intuitive way, by simulating a light source and the shadows it casts.</p><p>This can be computed from a DEM by using this recipe.</p><p>Getting ready</p><p>Open the DEM that we prepared in the Preparing elevation data recipe.</p><p>How to do it…</p><p>1. In the Processing Toolbox option, find the Hillshade algorithm and double-click on it</p><p>to open it:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 7</p><p>181</p><p>2. Select the DEM in the Input layer field. Leave the rest of the parameters with their</p><p>default values.</p><p>3. Click on Run to run the algorithm.</p><p>The hillshade layer will be added to the QGIS project, as follows:</p><p>How it works…</p><p>As in the case of the slope, the algorithm is part of the GDAL library. You will see that the</p><p>parameters are very similar to the slope case. This is because slope is used to compute the</p><p>hillshade layer. Based on the slope and the aspect of the terrain in each cell and using the</p><p>position of the sun that is defined by the Azimuth and Altitude fields, the algorithm computes</p><p>the illumination that the cell will receive. This is based on a focal analysis, so shadows are not</p><p>considered and are not a real illumination value, but they can be used to render and to display</p><p>the topography of the terrain.</p><p>You can try changing the values of these parameters to alter the appearance of the layer.</p><p>There's more…</p><p>As in the case of slope, there are alternative options to compute the hillshade. The SAGA one</p><p>in the Processing Toolbox option has a feature that is worth mentioning.</p><p>The SAGA hillshade algorithm contains a field named method. This field is used to select the</p><p>method that is used to compute the hillshade value, and the last method that is available.</p><p>Raytracing differs from the other ones as it models the real behavior of light, making an</p><p>analysis that is not local but that uses the full information of the DEM instead because it</p><p>takes into account the shadows that are cast by the surrounding relief. This renders more</p><p>precise hillshade layers, but the processing time can be notably larger.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis I</p><p>182</p><p>Enhancing your map view with a hillshade layer</p><p>You can combine the hillshade layer with your other layers to enhance their appearance.</p><p>As you used a DEM to compute the hillshade layer, it should be already in your QGIS project</p><p>along with the hillshade itself. However, this will be covered by the hillshade because of the</p><p>new layers produced by Processing are added on top of the existing ones in the layers list.</p><p>Move it to the top of the layer list so that you can see the DEM (and not the hillshade layer)</p><p>and style it to something like the following screenshot:</p><p>Lets see the steps to enhance the map view with a hillshade layer:</p><p>1. In the Properties dialog of the layer, move to the Transparency section, and set the</p><p>Global transparency value to 50%, as shown in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 7</p><p>183</p><p>2. Now, you should see the hillshade layer through the DEM, and the combination of</p><p>both of them will look like the following screenshot:</p><p>Another way of doing this is using the blending modes in QGIS. You can find more information</p><p>about this in the recipe, Understanding the feature and layer blending modes of Chapter 10,</p><p>Cartography Tips, or in the QGIS manual at http://docs.qgis.org/2.8/en/docs/</p><p>user_manual/working_with_vector/vector_properties.html#style-menu.</p><p>Analyzing hydrology</p><p>A common analysis from a DEM is to compute hydrological elements, such as the channel</p><p>network or the set of watersheds. This recipe shows you the steps to do these analysis.</p><p>Getting ready</p><p>Open the DEM that we prepared in the Preparing elevation data recipe.</p><p>www.it-ebooks.info</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/working_with_vector/vector_properties.html#style-menu</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/working_with_vector/vector_properties.html#style-menu</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis I</p><p>184</p><p>How to do it…</p><p>1. In the Processing Toolbox option, find the Fill Sinks algorithm and double-click on it</p><p>to open it:</p><p>2. Select the DEM in the DEM field and run the algorithm. This will generate a new</p><p>filtered DEM layer. From now on, we will just use this DEM in the recipe and not</p><p>the original one.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 7</p><p>185</p><p>3. Open Catchment Area and select the filtered DEM in the Elevation field:</p><p>4. Run the algorithm. This will generate a catchment area layer:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis I</p><p>186</p><p>5. Open the Channel network algorithm and fill it in, as shown in the following</p><p>screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 7</p><p>187</p><p>6. Run the algorithm. This will extract the channel network from the DEM, based on the</p><p>catchment area, and it will then generate it as both a raster and vector layer:</p><p>7. Open the Watershed basins algorithm and fill it in, as shown in the following</p><p>screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis I</p><p>188</p><p>8. Run the algorithm. This will generate a raster layer with the watersheds calculated</p><p>from the DEM and the channel network. Each watershed is a hydrological unit that</p><p>represents the area that flows into a junction, which is defined by the channel network:</p><p>How it works…</p><p>Starting from the DEM, the preceding steps follow a typical workflow for hydrological analysis:</p><p>f First, the sinks are filled. This is a required preparation whenever you plan to perform</p><p>a hydrological analysis. The DEM may contain sinks where a flow direction cannot be</p><p>computed, which represents a problem to model the movement of water across these</p><p>cells. Removing these sinks solves this problem.</p><p>f The catchment area is computed from the DEM. The values in the catchment area</p><p>layer represent the area that is upstream of each cell. That is, the total area in which</p><p>if water is dropped, it will eventually pass through the cell.</p><p>f Cells with high values of the catchment area will likely contain a river, while cells</p><p>with lower values will have overland flow. By setting a threshold on the catchment</p><p>area values, we can separate the river cells (the ones above the threshold) from</p><p>the remaining ones and extract the channel network.</p><p>f Finally, we compute the watersheds associated with each junction in the channel</p><p>network that was extracted</p><p>in the last step.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 7</p><p>189</p><p>There's more…</p><p>The key parameter in the preceding workflow is the catchment area threshold. If a larger</p><p>threshold is used, fewer cells will be considered as river cells, and the resulting channel</p><p>network will be sparser. As the watershed is computed based on the channel network,</p><p>this will result in a lower number of watersheds.</p><p>You can try this yourself with different values of the catchment area threshold. Here, you can</p><p>see the result for threshold is equal to 1,000,000 in the following screenshot:</p><p>The channel network has been added to help you understand the structure of the resulting set</p><p>of watersheds.</p><p>Here, you can see the result for a threshold of 50,000,000 in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis I</p><p>190</p><p>Note that in this last case, with a higher threshold value, there is only one single watershed in</p><p>the resulting layer.</p><p>The threshold values are expressed in the units of the catchment area which, as the size of</p><p>the cell is assumed to be in meters, are in square meters.</p><p>Calculating a topographic index</p><p>As the topography defines and influences most of the processes that take place in a given</p><p>terrain, the DEM can be used to extract many different parameters, which give us information</p><p>about these processes. This recipe shows you how to calculate a popular one, which is called</p><p>the Topographic wetness index, which estimates the soil wetness based on the topography.</p><p>Getting ready</p><p>Open the DEM that we prepared in the Preparing elevation data recipe.</p><p>How to do it…</p><p>1. Calculate a slope layer using the Slope, aspect, curvature algorithm from SAGA in the</p><p>Processing Toolbox option. Calculate a catchment area layer using the Catchment</p><p>area algorithm from the Processing Toolbox option. Note that you must use a</p><p>sink-less DEM, such as the one that we generated in the previous recipe with</p><p>the Fill sinks algorithm.</p><p>Open the Topographic wetness index algorithm from the Processing Toolbox option</p><p>and fill it in, as shown in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 7</p><p>191</p><p>2. Run the algorithm. This will create a layer with the topographic wetness index,</p><p>indicating the soil wetness in each cell:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis I</p><p>192</p><p>How it works…</p><p>The index combines slope and catchment area, two parameters that influence the soil</p><p>wetness. If the catchment area value is high, this means that more water will flow into the cell,</p><p>thus, increasing its soil wetness. A low value of slope will have a similar effect because the</p><p>water that flows into the cell will not flow out of it quickly.</p><p>This algorithm expects the slope to be expressed in radians. This is the reason why the Slope,</p><p>aspect, curvature algorithm has to be used because it produces its slope output in radians.</p><p>The other Slope algorithm that you will find, which is based on the GDAL library, creates a</p><p>slope layer with values expressed in degrees. You can use this layer if you convert its units</p><p>using the raster calculator.</p><p>There's more…</p><p>Other indices that are based on the same input layers can be found in different algorithms</p><p>in the Processing Toolbox option. The Stream Power Index and the LS factor both use the</p><p>slope and catchment area as inputs as well, and they can be related to potential erosion.</p><p>Automating analysis tasks using the</p><p>graphical modeler</p><p>Most analysis tasks involve using several algorithms. Repeating the same analysis with a</p><p>different dataset or different input parameters requires using them one by one, making this</p><p>task tedious and error-prone. You can automate analysis workflows using the Processing</p><p>graphical modeler, which allows you to define a workflow graphically and wrap it in a single</p><p>algorithm. This recipe introduces the main ideas about the modeler and creates a simple</p><p>model as an example.</p><p>Getting ready</p><p>No special preparation is needed in QGIS for this recipe, but make sure that you have read the</p><p>previous recipe about computing a topographic index. This recipe will create a model based</p><p>on the workflow in that recipe, so it is important that you understand it.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 7</p><p>193</p><p>How to do it…</p><p>1. Open the graphical modeler by navigating to Processing | Graphical modeler:</p><p>2. Double-click on the Raster Layer item to add a raster input. In the dialog that will</p><p>appear to define the input, name it DEM and set it as mandatory:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis I</p><p>194</p><p>3. Click on OK to add the input to the canvas:</p><p>4. Move to the Algorithms tab. Double-click on the Slope, aspect, curvature algorithm</p><p>and set the algorithm definition, as shown in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 7</p><p>195</p><p>5. Close the dialog by clicking on the OK button. This will be added to the modeler</p><p>canvas, as follows:</p><p>6. Add the Catchment area algorithm to the model by double-clicking on it in the</p><p>algorithm list and filling in the dialog, as shown in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis I</p><p>196</p><p>7. Finally, add the Topographic wetness index algorithm, defining it as shown in the</p><p>following screenshot:</p><p>8. The final model should look like the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 7</p><p>197</p><p>9. Enter a name and a group to identify the model and save it by clicking on the Save</p><p>button. Do not change the save location folder, because Processing will only look</p><p>for it in the default location, you can however change the name of the model. Close</p><p>the modeler dialog. If you now go to the Processing Toolbox option, you will find a</p><p>new algorithm in the Models section, which corresponds to the workflow that you</p><p>have just defined:</p><p>How it works…</p><p>The model automates the workflow and wraps all the steps into a single one.</p><p>By saving the model in the models folder, Processing will see this when updating the toolbox</p><p>and will include it along with the rest of algorithms so that it can be executed normally.</p><p>See also</p><p>f More information about the graphical modeler can be found in the Processing</p><p>chapter of the QGIS manual at http://docs.qgis.org/2.8/en/docs/user_</p><p>manual/processing/modeler.html</p><p>www.it-ebooks.info</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/processing/modeler.html</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/processing/modeler.html</p><p>http://www.it-ebooks.info/</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>199</p><p>8</p><p>Raster Analysis II</p><p>In this chapter, we will cover the following recipes:</p><p>f Calculating NDVI</p><p>f Handling null values</p><p>f Setting extents with masks</p><p>f Sampling a raster layer</p><p>f Visualizing multispectral layers</p><p>f Modifying and reclassifying values in raster layers</p><p>f Performing supervised classification of raster layers</p><p>Introduction</p><p>Following the previous chapter, this chapter introduces some additional techniques for raster</p><p>analysis. This chapter will show you how to work with images, how to modify raster values and</p><p>classify them, and how raster layers can be used along with vector layers, thus extending the</p><p>set of tools that were introduced in the recipes in the previous chapter.</p><p>Calculating NDVI</p><p>The Normalized Differential Vegetation Index is a very popular vegetation index that gives us</p><p>useful information about the presence or absence of live green vegetation.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis II</p><p>200</p><p>Getting ready</p><p>NDVI is calculated using a band with red spectral reflectance values, and another one with</p><p>near-infrared reflectance values. In the sample dataset, you will find two image files named</p><p>red.tif and nir.tif that can be used to compute NDVI. A project named ndvi.qgs is</p><p>available, which contains these two layers and a landsat image corresponding to this same</p><p>area. Open this project.</p><p>How to do it…</p><p>1. Open the Processing</p><p>Toolbox menu and find the algorithm called Vegetation</p><p>index[slope based]. Double-click on the algorithm item to execute it:</p><p>2. Select the red.tif layer in the Red Band field and the nir.tif layer in the Near</p><p>Infrared Band field. Click on Run to run the algorithm.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 8</p><p>201</p><p>3. The algorithm will produce a set of layers with different vegetation indices, NDVI is</p><p>among them:</p><p>How it works…</p><p>All vegetation indices that are computed by the algorithm are based on the relation between</p><p>red and near-infrared reflectances. Leaf cells scatter solar radiation in the near-infrared</p><p>reflectance and absorb radiation in the red reflectance, which can be used to predict the</p><p>location of healthy green vegetation based on these two values.</p><p>NDVI is computed with the formula given in the following section.</p><p>There's more…</p><p>As the formula of the NDVI is rather simple, you can calculate it without using a specific</p><p>algorithm, just by going to the raster calculator. You can use the one integrated in the</p><p>Processing Framework or the QGIS built-in on. You can see how you should fill the parameters</p><p>in the QGIS Raster calculator to compute the NDVI, based on the two proposed sample layers</p><p>in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis II</p><p>202</p><p>Extracting bands</p><p>The vegetation indices algorithm requires the red and infrared values to be in two separate</p><p>layers, each of them with a single band. However, it's common to have both of them in a</p><p>multiband image. To be able to use these bands, you must separate them, extracting them</p><p>into two separate files.</p><p>This can be done using the GDAL translate algorithm. The project contains a multiband</p><p>image named landsat.tif with the red band in band number 3 and infrared band in</p><p>band number 4:</p><p>1. Open the Translate algorithm in the Processing Toolbox menu.</p><p>2. Fill its parameters, as shown in the following screenshot, to export the infrared band:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 8</p><p>203</p><p>3. Run this again, as shown in the following screenshot, to export the red band:</p><p>The Translate algorithm uses the GDAL library underneath. You can also use this library as an</p><p>independent tool from the console. At the lower part of the algorithm dialog, you will find a text</p><p>field where you will see the equivalent console call to your current algorithm configuration.</p><p>Handling null values</p><p>Null values are a particular type of values that are used to indicate cells where the value for a</p><p>given layer is not defined. Understanding how to use them is important to avoid wrong results</p><p>when performing analyses but also to use them as a tool to get better and more correct</p><p>results. This recipe explains some of the fundamental ideas about null values in raster layers.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis II</p><p>204</p><p>How to do it…</p><p>The watershed.tif layer contains the area of a watershed. Cells inside the watershed</p><p>are cells from which water will eventually flow into the outlet point of the watershed. The</p><p>remaining cells belong to a different watershed. To mask the DEM with the watershed mask,</p><p>follow these steps:</p><p>1. Open the watershed.tif layer.</p><p>2. Open the identify tool and check whether the cells that belong to the watershed have</p><p>a value of 1, and the ones outside, have a value of no data.</p><p>3. Try clicking inside and outside the watershed; in your Identify Results dialog, you will</p><p>see the results, as shown in the following screenshot:</p><p>4. Now, let's calculate some statistics of the raster layer. Open the Raster layer</p><p>statistics algorithm in the Processing Toolbox menu.</p><p>5. Select the watershed layer in the input layer field and click on OK to run the</p><p>algorithm. The result is a short text output that looks like the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 8</p><p>205</p><p>Only the cells with a value of 1 have been considered, and the average value in the layer is</p><p>equal to 1.</p><p>The layer has 610 columns and 401 rows, but the total number of valid cells is much lower</p><p>than 610 x 410. These are the cells that have been used to compute the statistics.</p><p>How it works…</p><p>Raster layers always cover a rectangular region. However, in some circumstances, the land</p><p>object that the layer represents might not be rectangular. This might be due to a purely</p><p>geophysical reason (imagine a layer with water temperature that contains non-water cells),</p><p>political ones (a layer with a DEM of a given country with no data available for a neighboring</p><p>country), or many others. In any case, a value is needed for these cells to indicate that no data</p><p>is available. An arbitrary value is selected and used. As such, this is usually a value that is not</p><p>a logical and/or feasible value for the variable that is stored in the layer.</p><p>In the case of the example layer, the value used is -99999, which is the default value set for</p><p>no-data values. This means that, when the identify tool shows no data, it has actually selected</p><p>a value of -99999 in this case.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis II</p><p>206</p><p>Algorithms in the Processing framework systematically ignore no-data cells, and do not use</p><p>their values. You can clearly see this in the preceding example. A large part of the cells in</p><p>the layer have a value of 1 (the ones that belong to the watershed), but many of them have</p><p>a value of -99999. The average value of the cells should then be different from 1, but as</p><p>-99999 is defined as the no-data value, all cells with this value are ignored. The average</p><p>of the layer is, therefore, equal to 1.</p><p>There's more…</p><p>Null values should be considered not only when performing an analysis, but also when we just</p><p>want to render a layer that contains them.</p><p>Controlling the rendering of null values</p><p>Null values are also considered separately when rendering a raster layer. You can choose to</p><p>select them using a given color (as set by the current color palette), or to not render them</p><p>at all. To make all cells with null values transparent, open the layer properties and go to the</p><p>Transparency section. Make sure that the No data value checkbox is checked, as shown in</p><p>the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 8</p><p>207</p><p>Setting extents with masks</p><p>The extent of a layer can be set using a second layer, which acts as a mask. This recipe shows</p><p>you how to do this.</p><p>How to do it…</p><p>To mask the DEM with the watershed mask, follow these steps:</p><p>1. Open the watershed.tif layer and the dem.tif layer.</p><p>2. Open the Raster Calculator algorithm present in the Processing Toolbox menu.</p><p>3. In Main input layer, select the DEM, and in the Additional layers field, select the</p><p>watershed layer.</p><p>4. In the Formula field, enter the formula, a*b.</p><p>5. Click on Run to run the algorithm. You will get a masked DEM, as follows:</p><p>How it works…</p><p>When using the raster calculator, all operations involving a no-data value will result in another</p><p>no-data value. This means that, when multiplying the DEM layer and the mask layer, in the</p><p>cells that contain no-data values in the mask layer, the value in the resulting layer will be</p><p>a no-data value, no matter which elevation value is found in the DEM layer for this cell.</p><p>As cells inside the watershed in the mask layer have a value of 1, the result is a layer with</p><p>elevation values for watershed cells and no-data values for the remaining ones.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis II</p><p>208</p><p>There's more…</p><p>Here are some additional ideas about masks.</p><p>Restricting analysis to a given area</p><p>Once we have masked the area of interest (in this case, the watershed), all analysis that</p><p>we perform will be restricted to this. For instance, let's calculate the average elevation of</p><p>the watershed:</p><p>1. Open the Basic statistics for raster layers algorithm.</p><p>2. Select the masked DEM in the Input layer field and click on Run to run the algorithm.</p><p>You will get the statistics on the</p><p>Results window:</p><p>These values have been computed using only valid cell values and ignoring the no-data ones,</p><p>which means that they refer to the watershed and not the the full extent of the raster layer.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 8</p><p>209</p><p>Removing superfluous no-data values</p><p>Sometimes, you might have more no-data values that are needed in a raster layer, as in the</p><p>case of the proposed watershed layer. To reduce the extent of the layer and just have the</p><p>minimum extent that covers the valid data, you can use the Crop to data algorithm:</p><p>1. Open the Crop to data algorithm in the Processing Toolbox menu.</p><p>2. Enter the masked DEM in the Input layer field and click on Run to run the algorithm.</p><p>The resulting layer should look like this when you disable transparency for no-data</p><p>values:</p><p>Note that, if you have opted to render no-data cells as transparent pixels, you</p><p>will see no visual difference between the original and the cropped layer.</p><p>Masking using a vector mask</p><p>Masking a raster layer can also be done using a polygon vector layer. The watershed.shp</p><p>file contains a single polygon with the area of the watershed that we have already used to</p><p>mask the DEM. Here is how to use this to mask that DEM without using the raster mask:</p><p>1. Open the Clip grid with the polygon algorithm.</p><p>2. Select the DEM in the Input field.</p><p>3. Select the vector layer in the Polygons field.</p><p>4. Click on Run to run the algorithm. The clipped layer will be added to the QGIS project.</p><p>In this case, the clip algorithm automatically reduces the extent of the output layer to the</p><p>minimum extent defined by the polygon layer, so there is no need to run the Crop to data</p><p>algorithm afterwards.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis II</p><p>210</p><p>Sampling a raster layer</p><p>Data from a raster layer can be added to a points layer by querying the value of the layer in</p><p>the coordinates of the points. This process is known as sampling, and this recipe explains</p><p>how to perform it.</p><p>Getting ready</p><p>Open the dem.tif raster layer and the dem_points.shp vector layer:</p><p>How to do it…</p><p>1. In the Processing Toolbox menu, find the Add grid values to points algorithm and</p><p>double-click on it to open it:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 8</p><p>211</p><p>2. Select the DEM in the Grids field.</p><p>3. Select the point layer in the Points field.</p><p>4. Click on Run to run the algorithm.</p><p>A new vector layer will be created. This contains the same points as the input layer, but the</p><p>attribute table will have an additional field with the name of the selected raster layer and</p><p>the values corresponding to this layer in each point:</p><p>How it works…</p><p>The coordinates of the points are taken, and the value of the pixel in which the layer falls is</p><p>added to the resulting points layer.</p><p>This method assumes that the value of a cell is constant in all the area covered by this cell. A</p><p>different approach is to consider that the value of the cell represents its value only in the center</p><p>of the cell and perform additional calculations to compute the value at the exact sampling</p><p>point using the values of the surrounding cells as well. This can be done using several different</p><p>interpolation methods, which can be selected in the Interpolation method selector, changing</p><p>the default value, which only uses the value of the cell where the sampling point falls.</p><p>Layers are assumed to be in the same CRS and no reprojection is performed. If this is not the</p><p>case, the value added to the vector layer might not be correct.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis II</p><p>212</p><p>There's more…</p><p>Here, you can find some ideas about how to combine a raster and vector layer in different</p><p>situations.</p><p>Other raster-vector data transfer operations</p><p>Data coming from a raster layer can also be added to other types of vector layers. In the</p><p>case of a vector layer with polygons, the Grid statistics for polygons algorithm can be used,</p><p>as follows:</p><p>1. Open the watershed.shp file that we used in the previous recipe.</p><p>2. Open the Grid statistics in the Polygons field.</p><p>3. Select the raster layer to clip in the Grids field.</p><p>4. Select the polygon layer with the mask in the Polygons field.</p><p>5. Select the statistics to be calculated from the remaining parameters. For instance,</p><p>to calculate just the mean elevation, leave the Mean field selected and unselect</p><p>the others.</p><p>6. Click on Run to run the algorithm.</p><p>The resulting layer is a new polygon layer with the watershed and an additional field in the</p><p>attributes table, containing the mean elevation value for each polygon.</p><p>If more statistics are selected, the result will have a larger number of additional fields added,</p><p>one for each new parameter computed and each selected grid.</p><p>Visualizing multispectral layers</p><p>Multispectral layers can be rendered in different ways depending on how bands are used. This</p><p>recipe shows you how to do this and discusses the theory behind it.</p><p>Getting ready</p><p>Open the landsat.qgs project.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 8</p><p>213</p><p>How to do it…</p><p>1. The Landsat image, when opened with the default configuration, looks something like</p><p>the following screenshot:</p><p>2. Double-click on the layer to open its properties and move to the Style section:</p><p>1. Select the band number 4 in the Red band field.</p><p>2. Select the band number 3 in the Green band field.</p><p>3. Select the band number 2 in the Blue band field.</p><p>Your style configuration should be like the following:</p><p>3. Click on OK.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis II</p><p>214</p><p>The image should now look like the following:</p><p>How it works…</p><p>Colors representing a given pixel are defined using the RGB color space, which requires three</p><p>different components. A normal image (such as the one you will get from a digital camera)</p><p>has three bands containing the intensity for each one of these three components: red, green,</p><p>and blue.</p><p>Multispectral bands, such as the one used in this recipe, have more than three bands and</p><p>provide more detail in different regions of the electromagnetic spectrum. To visualize these,</p><p>three bands from the total number of available bands have to be chosen and their intensities</p><p>have to be used as intensities of the basic red, green, and blue components (although they</p><p>might correspond to a different region of the spectrum, even outside the visible range). This is</p><p>known as a false color image.</p><p>Depending on the combination of the bands that are used, the resulting image will convey</p><p>a different type of information. The combination chosen is frequently used for vegetation</p><p>studies, as it allows you to separate coniferous from hardwood vegetation as well as</p><p>providing information about vegetation health.</p><p>The combination is applied, in this case, to a Landsat 7 image, which is taken with the</p><p>ETM+ sensor. The wavelengths covered by each band are as follows (in micrometers):</p><p>f Band 1: 0.45 - 0.515</p><p>f Band 2: 0.525 - 0.605</p><p>f Band 3: 0.63 - 0.69</p><p>f Band 4: 0.75 - 0.90</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 8</p><p>215</p><p>f Band 5: 1.55 - 1.75</p><p>f Band 6: 10.40 - 12.5</p><p>f Band 7: 2.09 - 2.35</p><p>There's more…</p><p>Different combinations are frequently used for Landsat layers. One of them is the following:</p><p>f Select the band number 3 in the Red band field</p><p>f Select the band number 2 in the Green band field</p><p>f Select band number 1 in the Blue band field</p><p>This is a natural color combination, as the bands used for the R, G, and B components</p><p>actually have the wavelengths corresponding to the colors red, green, and blue:</p><p>If you are using an image that is not a Landsat 7 one, each band will have a different</p><p>meaning, and using the same combination of band numbers will yield different results. The</p><p>meaning of each band must be checked in order to understand the information displayed by</p><p>the rendered image.</p><p>See also</p><p>f Landsat data is freely available. If you want to download Landsat data corresponding</p><p>to a given region, visit http://landsat.gsfc.nasa.gov/.</p><p>Here, you can find</p><p>more information about where and how you can download it.</p><p>www.it-ebooks.info</p><p>http://landsat.gsfc.nasa.gov/</p><p>http://www.it-ebooks.info/</p><p>Raster Analysis II</p><p>216</p><p>Modifying and reclassifying values in raster</p><p>layers</p><p>A very useful technique to work with raster data is changing their values or grouping them into</p><p>categories. In this recipe, we will see how to do this.</p><p>Getting ready</p><p>Open the DEM file that we used in previous recipes.</p><p>How to do it…</p><p>We will classify the elevation in three groups:</p><p>f Lower than 1,000m</p><p>f Between 1,000 and 2,000m</p><p>f Higher than 2,000m</p><p>To do this, follow these steps:</p><p>1. Open the Change grid values algorithm from the Processing Toolbox menu. Set the</p><p>Replace condition parameter to Low Value</p><p>for geographic data and a few methods to put your maps online.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>224</p><p>Using web services</p><p>There are quite a few different types of web-based map service that can be loaded in QGIS.</p><p>Each type of web service provides data; often, this is the same data in different ways. This</p><p>recipe is about helping you figure out what type of web service you want to consume, and</p><p>conversely what type of web service you may want to create for others to use.</p><p>Getting ready</p><p>This recipe is all about thinking, so you don't need anything in particular to start. It does help</p><p>if you have a project in mind and some type of data you are interested in using or creating.</p><p>Most, if not all, of these methods require an Internet connection or a local server, providing</p><p>these services. Of course, if you have the data locally, you should probably just load it directly.</p><p>To help with the following section, here's a list of acronyms:</p><p>f CSW: This is Catalog Service for the Web</p><p>f WFS: This is Web Feature Service</p><p>f WFS-T: This is Web Feature Service, Transactional</p><p>f WCS: This is Web Coverage Service</p><p>f WMS: This is Web Map Service</p><p>f WMTS: This is Web Map Tile Service</p><p>f TMS: This is Tile Map Service</p><p>f XYZ: This is an "X Y Z" Service (there really isn't a formal name for this one because</p><p>it's not an official standard)</p><p>How to do it…</p><p>1. Start by answering the following questions with regards to using data over</p><p>web services:</p><p>� Do you already know where to find the data that you want?</p><p>� Do you need to edit or apply custom styling to this data?</p><p>� Do you care if the data is vector or raster?</p><p>� Do you need the data at its original resolution or quality?</p><p>� Do you need data at specific resolutions or in a specific projection?</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>225</p><p>2. Use the following decision matrix to pick out which services are appropriate to your</p><p>use case:</p><p>Criteria CSW WFS WFS-T WCS WMS Tiles (WMTS,</p><p>XYZ, TMS)</p><p>Do you already know where to</p><p>find the data you want?</p><p>No Yes Yes Yes Yes Yes</p><p>Do you need to edit or apply</p><p>custom styling to the data?</p><p>Yes Yes Yes No No</p><p>Do you care if the data is</p><p>vector or raster?</p><p>Vector Vector Raster Raster Raster</p><p>Do you need the data at its</p><p>original resolution or quality?</p><p>Yes Yes Yes No No</p><p>Do you need data at specific</p><p>resolutions or in a specific</p><p>projection?</p><p>Yes Yes Yes Yes No</p><p>Recipe number in this chapter. 3 2 2 5 4 4, 6</p><p>Now that you've found the appropriate recipe for the web service that you want to use, jump</p><p>to this recipe later in this chapter. If you're still not sure, read on for more hints on how to pick</p><p>the correct service. This recipe applies both to how you use web services and how to decide</p><p>what web services to offer (if you put up a web service for other people to use).</p><p>Generally speaking, from left (WFS) to right (Tiles) the speed of the service increases. Tiles</p><p>serve data the fastest to end users but with the most limitations.</p><p>How it works…</p><p>For each of the services, there is a QGIS tool (built-in or plugin). This tool stores your list of web</p><p>servers and connection settings for each service. When you go to load a layer from a chosen</p><p>server for a particular protocol an up-to-date list of layers is requested from the server (that is,</p><p>the GetCapabilities XML). You then get to pick off this list the layers that you would like to add</p><p>to the map canvas (and depending on service type, the projection, and the file type).</p><p>There's more…</p><p>Vector is generally slower, as more data needs to be transmitted as the data grows. Raster</p><p>formats are a fixed number of pixels onscreen, so it's always approximately the same amount</p><p>of data per screen load.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>226</p><p>WMS, WMTS, WFS, and WCS are sometimes referred to as W*S as a</p><p>collective group of related services that behave similarly.</p><p>Each situation will have additional considerations. For example, if you need a specific</p><p>projection, you probably can't use a Tile service because these are usually only in very</p><p>specific projections (Web/Spherical Mercator). Or perhaps, you want to print large paper</p><p>maps. Then, you probably want WFS or WCS in order to get the full resolution possible over</p><p>your entire region.</p><p>One of the most common mistakes is to think that you need vector data when you actually just</p><p>need a background tile that incorporates vector data. A great example of this is road data. If</p><p>you don't actually need to style, select, or individually manipulate road data, and then a Tile</p><p>or WMS type layer will be much faster.</p><p>See also</p><p>f For more information, read the standards at the OGC website, http://</p><p>opengeospatial.org</p><p>Using WFS and WFS-T</p><p>Web Feature Services (WFS) is an OGC standard method to access and, in some cases, edit</p><p>(WFS-T) vector data over the Internet. When you need full attribute tables, local style control,</p><p>or editing, WFS is the way to go. Like most other web services, the biggest advantage over a</p><p>local layer is that you don't have to copy or load the whole layer at once.</p><p>If you just need to view the layer, often WMS or a Tile service</p><p>(described in other recipes within this chapter) are more efficient.</p><p>Getting ready</p><p>You need the URL of a WFS service to use and a working Internet connection. We will use the</p><p>public Mapserver demo website (http://demo.mapserver.org/).</p><p>To try WFS-T, which involves editing, you will need to get access to a</p><p>service (typically, password protected) or make one yourself. Do you</p><p>need a WFS-T test server? This is a great case where OSGeo-Live</p><p>comes in handy, as you can run your own WFS-T server in a virtual</p><p>machine at http://live.osgeo.org.</p><p>www.it-ebooks.info</p><p>http://opengeospatial.org</p><p>http://opengeospatial.org</p><p>http://demo.mapserver.org/</p><p>http://live.osgeo.org</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>227</p><p>How to do it…</p><p>1. Find a WFS service that you want to use and copy the GetCapabilities URL. For this</p><p>example, we will use the Mapserver demo website, http://demo.mapserver.org/.</p><p>As with other web services, it's more efficient if you load your local</p><p>layers and zoom to their extent first. This enables you to not waste time</p><p>loading data from web services for extents outside your area of interest.</p><p>2. Open the add WFS dialog by clicking on the following icon:</p><p>3. Create a New connection.</p><p>4. Assign Name of your choosing and paste in the URL field for the WFS service,</p><p>(http://demo.mapserver.org/cgi-bin/wfs?SERVICE=WFS&VERSION=1.0.</p><p>0&REQUEST=GetCapabilities):</p><p>On future usage of the same service, this will already be in your list of</p><p>services, so you only have to add it once.</p><p>5. Save your edits by clicking on the OK button.</p><p>6. Now select the service from the drop-down menu.</p><p>7. Query for a list of layers by clicking on the Connect button.</p><p>www.it-ebooks.info</p><p>http://demo.mapserver.org/</p><p>http://demo.mapserver.org/cgi-bin/wfs?SERVICE=WFS&VERSION=1.0.0&REQUEST=GetCapabilities</p><p>http://demo.mapserver.org/cgi-bin/wfs?SERVICE=WFS&VERSION=1.0.0&REQUEST=GetCapabilities</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>228</p><p>8. Select the layer or layers that you want to add to the map; either continents or</p><p>cities works for this example:</p><p>9. When your selection is complete, use the Add button to place the layer in the map.</p><p>10. Rearrange the render order of the map by dragging layers up and down in the list.</p><p>11. Pan and zoom to make fresh requests for WFS data to be loaded to the view.</p><p>WFS layers can be restyled with standard layer properties. Also,</p><p>the information tool and the attribute table will appear as other</p><p>vector layers.</p><p>How it works…</p><p>For each web service, there is a main URL. When you browse to this URL and add the</p><p>GetCapabilities parameter (QGIS does this for you), the returned result is an XML file,</p><p>which describes the services that are offered by the server. The client, QGIS, parses the list of</p><p>layers for you to choose from, and once you pick the layer(s), uses the additional information</p><p>in the</p><p>Errata</p><p>Although we have taken every care to ensure the accuracy of our content, mistakes do</p><p>happen. If you find a mistake in one of our books—maybe a mistake in the text or the</p><p>code—we would be grateful if you could report this to us. By doing so, you can save other</p><p>readers from frustration and help us improve subsequent versions of this book. If you find</p><p>any errata, please report them by visiting http://www.packtpub.com/submit-errata,</p><p>selecting your book, clicking on the Errata Submission Form link, and entering the details of</p><p>your errata. Once your errata are verified, your submission will be accepted and the errata will</p><p>be uploaded to our website or added to any list of existing errata under the Errata section of</p><p>that title.</p><p>To view the previously submitted errata, go to https://www.packtpub.com/books/</p><p>content/support and enter the name of the book in the search field. The required</p><p>information will appear under the Errata section.</p><p>Piracy</p><p>Piracy of copyrighted material on the Internet is an ongoing problem across all media. At</p><p>Packt, we take the protection of our copyright and licenses very seriously. If you come across</p><p>any illegal copies of our works in any form on the Internet, please provide us with the location</p><p>address or website name immediately so that we can pursue a remedy.</p><p>Please contact us at copyright@packtpub.com with a link to the suspected pirated material.</p><p>We appreciate your help in protecting our authors and our ability to bring you valuable content.</p><p>Questions</p><p>If you have a problem with any aspect of this book, you can contact us at questions@</p><p>packtpub.com, and we will do our best to address the problem.</p><p>www.it-ebooks.info</p><p>http://www.packtpub.com/sites/default/files/downloads/QGIS2Cookbook_ColorImages.pdf</p><p>http://www.packtpub.com/sites/default/files/downloads/QGIS2Cookbook_ColorImages.pdf</p><p>http://www.packtpub.com/submit-errata</p><p>https://www.packtpub.com/books/content/support</p><p>https://www.packtpub.com/books/content/support</p><p>http://www.it-ebooks.info/</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>1</p><p>1</p><p>Data Input and Output</p><p>In this chapter, we will cover the following recipes:</p><p>f Finding geospatial data on your computer</p><p>f Describing data sources</p><p>f Importing data from text files</p><p>f Importing KML/KMZ files</p><p>f Importing DXF/DWG files</p><p>f Opening a NetCDF file</p><p>f Saving a vector layer</p><p>f Saving a raster layer</p><p>f Reprojecting a layer</p><p>f Batch format conversion</p><p>f Batch reprojection</p><p>f Loading vector layers into SpatiaLite</p><p>f Loading vector layers into PostGIS</p><p>Introduction</p><p>If you want to work with QGIS, the first thing you need is spatial data. Whether you want to</p><p>prepare a nice-looking map layout or perform spatial analysis, you need to open some data to</p><p>work with. This chapter deals with the basic input and output commands, which will allow you</p><p>to use data in several different formats and also export to the most convenient format</p><p>in case you want to use it in different applications or share with others.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>2</p><p>Automation is possible for many of the operations that you will see in this cookbook. This</p><p>chapter contains some recipes that use automation to process a set of input files.</p><p>Finding geospatial data on your computer</p><p>This recipe shows you how to use the QGIS browser to locate and open spatial data.</p><p>Getting ready</p><p>Before you start working, make sure that you have copied the sample dataset to your</p><p>filesystem and you have it located.</p><p>How to do it…</p><p>There are several ways of locating and opening a data file to open it in QGIS, but the most</p><p>convenient of these is the QGIS browser:</p><p>1. To enable this, go to the View | Panels menu and enable the Browser checkbox in</p><p>it. The browser will be shown by default in the left-hand side of the QGIS window, as</p><p>shown in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>3</p><p>Browser contains a tree with all the available sources of spatial data. This includes</p><p>data files in your filesystem, databases, and remote services.</p><p>2. Navigate to the folder where you copied the sample dataset, and you will see a list of</p><p>available data files, as shown in the following screenshot:</p><p>Not all files are shown but just the ones that are identified as valid</p><p>data sources.</p><p>3. To add a file to your project, just right-click on it and select Add Layer:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>4</p><p>4. Multiple selections are allowed. In that case, select the Add selected layer menu.</p><p>Another way of opening a file is by just dragging it and dropping it into the QGIS</p><p>canvas. Dragging multiple files is allowed, as well, as shown in the following</p><p>screenshot:</p><p>How it works…</p><p>The browser acts as a file explorer that is directly linked to QGIS, which only shows valid data</p><p>files and can be used to easily add them to a QGIS project.</p><p>There's more…</p><p>There are a few more things that you need to know that are related to this recipe. They are</p><p>explained in the following sections.</p><p>Adding layers with the Layer menu</p><p>As an alternative to the browser, the Layer menu contains a set of entries. Each of them deals</p><p>with a different type of data. They give you some additional options, and they might allow you</p><p>to work with formats that are not directly supported by the browser.</p><p>Adding a folder to Favorites</p><p>Navigating to the folder where your data is located can be tedious. If you use a given folder</p><p>regularly, you can right-click on it and select Add as favorite. The folder will appear on the</p><p>Favorites section at the top of the browser tree.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>5</p><p>Nonfile data sources</p><p>The browser also shows non-file data, such as remote services. Services have to be defined</p><p>before they appear on the corresponding section in the browser. To add a service, right-click</p><p>on the service name and select New connection.... A dialog will appear to define the service</p><p>connection parameters.</p><p>As an example, try adding the following WMS service, using the WMS entry in the browser,</p><p>as shown in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>6</p><p>A new entry will appear, containing the layers offered by the service, as shown in the</p><p>following screenshot:</p><p>Describing data sources</p><p>You can get additional information about a data file before opening it. This recipe shows you</p><p>how to explore the properties of a data origin.</p><p>Getting ready</p><p>Before you start working, make sure that you have copied the sample dataset to your</p><p>filesystem and that you have it located.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>7</p><p>How to do it…</p><p>1. In the QGIS browser, navigate to the folder with your sample dataset. Select the</p><p>elev_lid792_1m file and right-click on it. In the context menu, select Properties.</p><p>A dialog like the one in the following screenshot will appear:</p><p>This dialog displays the properties of a raster layer.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>8</p><p>2. Now, let's select a vector layer instead. Select the elev_lid792_randpts.shp file,</p><p>right-click on it, and select Properties. The information dialog will look like</p><p>the following:</p><p>How it works…</p><p>In the upper part of the description window, you will see a field named Provider. Provider</p><p>defines the type or data origin and who takes care of reading the data and passing it to QGIS.</p><p>For raster layers, you will see gdal as Provider. For most file-based vector layers, ogr will</p><p>be the provider that will appear. They refer to the GDAL and OGR libraries, two open-source</p><p>libraries that are used by many GIS programs to access both raster and vector data.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>9</p><p>There's more…</p><p>If the data is already loaded in QGIS, you can access the information about it in the Properties</p><p>section of the layer (right-click on the layer name to select the Properties entry in the context</p><p>menu). In</p><p>XML to look up the data at the specific URL.</p><p>Data requests are limited to the visible bounding box of the map canvas. This limits the</p><p>amount of data that is requested. At least this is how it should work. However, features that go</p><p>off screen will likely be included in their entirety to maintain geometry integrity. So, expect that</p><p>loading large vector layers over WFS has the potential to be extremely slow.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>229</p><p>There's more…</p><p>WFS-T services typically require passwords and are designed to work over the Internet. If you</p><p>are working within a local network, you may consider just using PostGIS layers. Either way,</p><p>it should also be noted that versioning and conflict resolution are not automatic, requiring</p><p>the service backend to be configured to support such features.</p><p>Searching CSW</p><p>CSW is a catalog web service. Its main function is to provide discoverability of geographic data</p><p>and link you to usable data either by download or by any other of the web services that are</p><p>mentioned in this chapter.</p><p>Getting ready</p><p>This recipe uses the MetaSearch plugin. It requires the pycsw and owslib libraries installed</p><p>in your system's Python. Refer to the Adding plugins with Python dependencies recipe of</p><p>Chapter 11, Extending QGIS, for help on installing pycsw and owslib if you don't know how</p><p>to do this.</p><p>How to do it…</p><p>1. Open the MetaSearch plugin by navigating to Web | MetaSearch | MetaSearch.</p><p>If you don't see the Web menu, check the plugin manager and</p><p>ensure that MetaSearch is enabled (checkmark).</p><p>2. Pick a service from the dropdown on the right: UK Location Catalogue</p><p>Publishing Service.</p><p>If you don't see any services in the drop-down list, click on the</p><p>Services tab and use the Add default services button.</p><p>3. Type a search term in box on the left: Park.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>230</p><p>4. (Optional) Set an extent to limit the search. In this case, use Map Extent.</p><p>Global searches often return too many results, or they cause the</p><p>connection to time out while waiting for all the results. As with other</p><p>web services, it is advisable to load a reference layer and zoom to</p><p>the area of interest first before trying to search them. The third tab,</p><p>Settings, allows you to adjust the timeout. Increase this if you're</p><p>getting too many timeout errors.</p><p>5. Click on the Search button and wait for the results:</p><p>1. Double-click on any of the results to see additional details.</p><p>2. If a selected result is available as a loadable layer, one or more of the service</p><p>buttons at the bottom of the screen will be enabled. To understand more</p><p>about how to use each of these choices, refer to the other recipes in this</p><p>chapter on WMS, WFS, and WCS:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>231</p><p>How it works…</p><p>MetaSearch queries websites that provide catalogs in the CSW standard, which is defined</p><p>by the OGC. Once your request parameters are sent, the receiving website queries its online</p><p>database for matches. If matches are found, metadata about the results is sent back to the</p><p>client, in this case, QGIS.</p><p>CSW currently includes options to search by keyword and spatial extent. Future versions may</p><p>enable setting time frames.</p><p>There's more…</p><p>If you pick opening an additional service that is based on the results, Metasearch will create a</p><p>temporary service registration and open the correct service dialog. Unfortunately, at this time,</p><p>you need to then scroll through the available layers to find the one that you want and actually</p><p>add it to the map.</p><p>Additional future CSW searches will ask if you want to override the</p><p>existing connection. You must say yes. If you find yourself using the</p><p>same W*S service, consider copying the GetCapabilities URL and</p><p>making a new permanent entry in the correct service dialog.</p><p>You can add more catalogs to search on the Services tab within the plugin. You will need</p><p>to find the CSW GetCapabilities URL on the website that you want to query. Most of the</p><p>common geoportal-type websites now support CSW, including (but not limited to) Geonode,</p><p>Geonetwork, and the ESRI Geoportal.</p><p>CSW is a relatively new standard when compared to some of the others, and it seems to be</p><p>hard to find services that consistently work and actually offer WMS, WCS, or WFS of the layers</p><p>in their catalog.</p><p>See also</p><p>f The Adding plugins with Python dependencies recipe of Chapter 11, Extending QGIS,</p><p>for help on installing pycsw and owslib if you don't know how</p><p>Using WMS and WMS Tiles</p><p>Web Map Services (WMS), one of the first OGC web services created, provides a method for</p><p>dynamic raster generation served over the Web. They are a compromise between the flexibility</p><p>of WFS and the speed of Tile services.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>232</p><p>Getting ready</p><p>There are several iterations of WMS, and QGIS supports most of them. To use a WMS, you</p><p>need to give QGIS the GetCapabilities URL of the service that you want to view data from.</p><p>How to do it…</p><p>1. Find a WMS service that you want to use and copy the GetCapabilities URL. In this</p><p>recipe, we can use the Geoserver demo website (http://demo.opengeo.org/</p><p>geoserver/web/).</p><p>As with other web services, it's more efficient if you load your</p><p>local layers and zoom to their extent first. This enables you to not</p><p>waste time loading data from web services for extents outside</p><p>your area of interest.</p><p>2. Open the Add WMS dialog.</p><p>3. Create a New connection.</p><p>4. Assign a Name of your choosing and paste in the URL (http://demo.opengeo.</p><p>org/geoserver/ows?service=wms&version=1.3.0&reque</p><p>st=GetCapabilities):</p><p>www.it-ebooks.info</p><p>http://demo.opengeo.org/geoserver/web/</p><p>http://demo.opengeo.org/geoserver/web/</p><p>http://demo.opengeo.org/geoserver/ows?service=wms&version=1.3.0&reque st=GetCapabilities</p><p>http://demo.opengeo.org/geoserver/ows?service=wms&version=1.3.0&reque st=GetCapabilities</p><p>http://demo.opengeo.org/geoserver/ows?service=wms&version=1.3.0&reque st=GetCapabilities</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>233</p><p>On future usage of the same service, this will already be in</p><p>your list of services, so you only have to add it once.</p><p>5. Save your edits by clicking on OK.</p><p>6. Now select the service from the drop-down list.</p><p>7. Query for a list of layers using the Connect button.</p><p>8. Select the layer or layers that you want to add to the map:</p><p>You can select one or more layers. If you select multiple layers, they</p><p>will be merged and only appear as a single layer in the QGIS Layers</p><p>list. The Layer Order tab lets you arrange the WMS layers within the</p><p>combined layer. This is important when one of the layers is opaque</p><p>and has 100% continuous data, allowing you to put other data on</p><p>top of it visually.</p><p>9. There are several other options, including image type and projection:</p><p>� For an image type, PNG is a good default as it supports lossless compression</p><p>and transparency. If you don't need transparency and are okay with a little</p><p>data loss, JPG can be used for smaller files, so they are faster to load.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>234</p><p>� When picking projection, if you can use the original projection of the</p><p>data (if you know it), you will get the least resampling. Otherwise, pick</p><p>something that matches the other data that you plan to use in conjunction</p><p>with the WMS.</p><p>Not all image types and projections are available; this depends</p><p>on what the server offers. If one image type doesn't seem to</p><p>work, try a different one before reporting a bad server.</p><p>10. When your selection is complete, use the Add button to place the layer in the map.</p><p>11. Rearrange the render order of the map by dragging layers up and down in the list.</p><p>12. Pan and zoom to make fresh requests for WMS data to be loaded to the view.</p><p>How it works…</p><p>When you pan and zoom the map, a request with the bounding box of the viewable extent and</p><p>scale is sent to the service. The server then renders an image that matches the request and</p><p>passes it back</p><p>to the client (in this case, QGIS).</p><p>There's more…</p><p>Some WMS services now also support tiling under the Web Map Tiling Service (WMTS)</p><p>protocol. From the client's perspective, this not really different from WMS in usage. On the</p><p>server side, after each request the results are cached so that if the same extent and scale is</p><p>requested, the cached version can be delivered instead of creating the results from scratch.</p><p>For you, the end user, this should result in faster loading if a service provides WMTS.</p><p>When configuring a WMTS, use the WMTS URL instead of the WMS URL. One example would</p><p>be the Geoserver demo site's WMTS:</p><p>http://demo.opengeo.org/geoserver/gwc/service/</p><p>wmts?REQUEST=GetCapabilities</p><p>Once successful, this will take you to the Tilesets tab, where you can pick which layer and</p><p>projection of the available options you want to load. As the Tiles are premade or cached, you</p><p>will usually not have the option to combine multiple layers at once and will need to load them</p><p>one at a time:</p><p>www.it-ebooks.info</p><p>http://demo.opengeo.org/geoserver/gwc/service/wmts?REQUEST=GetCapabilities</p><p>http://demo.opengeo.org/geoserver/gwc/service/wmts?REQUEST=GetCapabilities</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>235</p><p>WMS-C is an earlier version of the WMTS standard. In usage, it's pretty</p><p>much the same, though the URL pattern may look more similar to the WMS.</p><p>See also</p><p>f See the QGIS documentation for more information about the WMS capabilities of</p><p>QGIS at http://docs.qgis.org/2.8/en/docs/user_manual/working_</p><p>with_ogc/ogc_client_support.html#ogc-wms</p><p>Using WCS</p><p>A Web Coverage Service (WCS) differs greatly in use case from the other services, but it</p><p>behaves very similarly. The goal of WCS is to allow users to extract a region of interest from</p><p>a large raster data that is hosted remotely. Unlike a WMS or Tiled set, WCS is a clip of the</p><p>original data in full resolution and usually in the original projection. This format is ideal if</p><p>you need the raster data for analysis purposes and not just visualization.</p><p>www.it-ebooks.info</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/working_with_ogc/ogc_client_support.html#ogc-wms</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/working_with_ogc/ogc_client_support.html#ogc-wms</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>236</p><p>Getting ready</p><p>For this recipe, you need a WCS to connect to. Check with your data providers to see</p><p>whether they offer WCS. For this recipe, we can use the OpenGeo Geoserver Demo site</p><p>at http://demo.opengeo.org/geoserver/web/.</p><p>How to do it…</p><p>1. Open a web browser and go to http://demo.opengeo.org/geoserver/web/.</p><p>2. On the right-hand side, you'll see a list of web services that are available; right-click</p><p>on WCS 1.1.1 and copy the link.</p><p>3. In QGIS, open the WCS dialog.</p><p>4. Select New to create a new server entry.</p><p>5. In the boxes, perform the following:</p><p>1. Provide a name so that you remember which service this is.</p><p>2. Paste the URL that you copied earlier in the URL box (http://demo.</p><p>opengeo.org/geoserver/ows?service=wcs&version=1.1.1&requ</p><p>est=GetCapabilities):</p><p>6. Click on the OK button.</p><p>www.it-ebooks.info</p><p>http://demo.opengeo.org/geoserver/web/</p><p>http://demo.opengeo.org/geoserver/web/</p><p>http://demo.opengeo.org/geoserver/ows?service=wcs&version=1.1.1&request=GetCapabilities</p><p>http://demo.opengeo.org/geoserver/ows?service=wcs&version=1.1.1&request=GetCapabilities</p><p>http://demo.opengeo.org/geoserver/ows?service=wcs&version=1.1.1&request=GetCapabilities</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>237</p><p>7. Now, you'll be back on the Add Layer(s) from a WCS Server dialog:</p><p>1. Click on the Connect button.</p><p>2. After the list is populated, select a layer to add to the map. Click on the</p><p>Add button. Try the Blue Marble layer.</p><p>3. Now, click on the Close button to return to your map:</p><p>8. You should now see the Blue Marble layer loaded.</p><p>If you zoom in to the level of a US State or a European</p><p>country, you will see the image start to pixelate. Blue Marble</p><p>is a low resolution image put together by NASA that roughly</p><p>shows what the whole world looks like from space, cloud</p><p>free. It is meant as a general view of the whole world and</p><p>does not contain fine details.</p><p>9. (Optional) Use the Save As option to download a portion or the entire WCS layer at its</p><p>full original resolution.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>238</p><p>How it works…</p><p>WCS, like other web services, sends a bounding box request to the server, which in turn</p><p>delivers the raster data to QGIS. Unlike WMS, no rendering is done on the server side,</p><p>the raw original raster data is sent. This could mean the following:</p><p>f There was no resampling of the image before it was sent</p><p>f You can apply your own styling to the data that is delivered</p><p>Keep in mind that requesting the full extent of a high resolution raster will result in a large</p><p>amount of data transfer. This is unlike Tiles or WMS, which at most return the exact number</p><p>of pixels in the viewable area at the resolution that is requested.</p><p>As with other web services, it is recommended that you zoom in to your area</p><p>of interest before loading a WCS.</p><p>There's more…</p><p>Another bonus of WCS over WMS is that because WCS delivers the original data, it is not</p><p>limited to a 3-band RGB image. You can use WCS to view and download Multi or Hyperspectral</p><p>data (4+ bands common in remote-sensing applications).</p><p>Currently, QGIS only supports 1.0.x and 1.1.x, not WCS 2.x; at least not yet!</p><p>Using GDAL</p><p>The QuickMapsServices and OpenLayers plugins, as described in the Loading BaseMaps with</p><p>the QuickMapServices plugin and Loading BaseMaps with the OpenLayers plugin recipes in</p><p>Chapter 4, Data Exploration, are awesome as they put a reference layer in your map session.</p><p>The one downside, however, is that it is a hassle to add new layers. So, if you come across or</p><p>build your own Tile service and want to use it in QGIS, this recipe will let you use almost any</p><p>Tile service.</p><p>Getting ready</p><p>You will need a web browser, text editor, and the URL of a web-based XYZ (sometimes called</p><p>TMS) service—one that allows you to make requests without an API key. We're going to use</p><p>the maps at http://www.opencyclemap.org/.</p><p>Viewing the JavaScript source (a good tool for this is Firebug, or other web-developer tools for</p><p>the browser), we can view the source URLs for the tiles.</p><p>www.it-ebooks.info</p><p>http://www.opencyclemap.org/</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>239</p><p>How to do it…</p><p>1. Open http://www.opencyclemap.org/ in a web browser.</p><p>2. Now, figure out the URL for the tiles by looking at the source code:</p><p>1. Look in map.js and you'll see the layer definition:</p><p>var cycle = new OpenLayers.Layer.OSM("OpenCycleMap",</p><p>["https://a.tile.thunderforest.com/cycle/${z}/${x}/${y}.</p><p>png",</p><p>"https://b.tile.thunderforest.com/cycle/${z}/${x}/${y}.</p><p>png",</p><p>"https://c.tile.thunderforest.com/cycle/${z}/${x}/${y}.</p><p>png"],</p><p>{ displayOutsideMaxExtent: true,</p><p>attribution: cycleattrib, transitionEffect: 'resize'}</p><p>);</p><p>2. Or, you can look at the image files your browser downloads. If you put</p><p>https://a.tile.thunderforest.com/cycle/13/1325/3143.png</p><p>into a browser, it will load that one tile.</p><p>3. The pattern is pretty straight forward:</p><p>////.</p><p>In this particular case, the Tile Index pattern is the TMS style;</p><p>refer to http://www.maptiler.org/google-maps-</p><p>coordinates-tile-bounds-projection/.</p><p>4. To turn this into a layer in QGIS, open up a text editor and paste in the following</p><p>definition. This definition tells GDAL which driver to use and the server URL pattern</p><p>with z, x, and y as variables. Save the file as opencyclemap.xml:</p><p>http://c.tile.thunderforest.com/cycle/${z}/${x}/</p><p>${y}.png</p><p>-20037508.34</p><p>20037508.34</p><p>20037508.34</p><p>-20037508.34</p><p>18</p><p>1</p><p>www.it-ebooks.info</p><p>http://www.opencyclemap.org/</p><p>https://a.tile.thunderforest.com/cycle/13/1325/3143.png</p><p>http://www.maptiler.org/google-maps-coordinates-tile-bounds-projection/</p><p>http://www.maptiler.org/google-maps-coordinates-tile-bounds-projection/</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>240</p><p>1</p><p>top</p><p>EPSG:3785</p><p>256</p><p>256</p><p>3</p><p>5. You can now load the layer using the Raster dialog or the browser:</p><p>Note that there are two listings for opencyclemap.xml; only</p><p>the one with the square-shaped icon will work (that is, a raster),</p><p>as tiles are a raster format.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>241</p><p>How it works…</p><p>The XML file defines the parameters of the service; however, because XYZ-style servers don't</p><p>follow a standard, the URL pattern varies slightly for each server and the servers do not have</p><p>a GetCapabilities function that describes available layers. By telling GDAL how to handle the</p><p>URL, you are wrapping a nonstandard format into a typical GDAL layer, which QGIS can easily</p><p>be loaded as a raster.</p><p>There's more…</p><p>One additional tip when using Spherical Mercator (EPSG:3785) is that you can set a custom</p><p>list of scales (Zoom Levels) in QGIS. The following set of scales can be loaded per QGIS</p><p>project, and will change the dropdown at the bottom right. These scales match the scales</p><p>that most servers will provide, so you get the best viewing experience:</p><p>1. Go to File | Project Properties.</p><p>2. Select the General tab.</p><p>3. Check the Project Scales checkbox.</p><p>4. Load the scales.xml file that is provided:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>242</p><p>This technique is not limited to just tile services. Many other formats that GDAL works</p><p>with can be wrapped for easier usage in QGIS. This is a similar method to Virtual Raster</p><p>Tables (VRT) layers mentioned in the Creating raster overviews (pyramids) recipe in</p><p>Chapter 2, Data Management.</p><p>Lastly, you may ask why a new plugin using this method doesn't replace the OpenLayers</p><p>plugin. Such an idea has been under discussion for a while; the key sticking point is that</p><p>accessing some layers, such as Google, Bing, and so on, with this method may violate the</p><p>Terms of Service as they do not keep the Copyright, Trademark, and Logo in the correct place.</p><p>Also, caching and printing such layers may not be legal. In general, avoid using proprietary</p><p>data when possible to reduce licensing issues.</p><p>See also</p><p>f This recipe and method has actually been known and discussed in many QGIS</p><p>venues. The most frequently cited example is available at http://www.3liz.com/</p><p>blog/rldhont/index.php?post/2012/07/17/OpenStreetMap-Tiles-in-</p><p>QGIS.</p><p>f The full explanation of options for GDAL can be found at http://www.gdal.org/</p><p>frmt_wms.html.</p><p>Serving web maps with the QGIS server</p><p>QGIS and the Web is not all about consuming data, it can also be used to serve data over the</p><p>Web for others to view online or consume in other web clients (such as QGIS). Keep in mind</p><p>that setting up your own web service is not the easiest way to make a web map (refer to the</p><p>Hooking up web clients recipe in this chapter). This is, however, a great way to transition all</p><p>the hard work that you've put into a QGIS project file into something other people can see</p><p>and use.</p><p>Getting ready</p><p>For this recipe, you need a working installation of the QGIS server. This involves running a</p><p>standard web server (such as Apache or Nginx). There are many ways to set up the server, so</p><p>please see the official documentation at http://hub.qgis.org/projects/quantum-</p><p>gis/wiki/QGIS_Server_Tutorial.</p><p>Once you have the QGIS server running, then you just need a QGIS project with the configuration</p><p>outlined in this recipe.</p><p>www.it-ebooks.info</p><p>http://www.3liz.com/blog/rldhont/index.php?post/2012/07/17/OpenStreetMap-Tiles-in-QGIS</p><p>http://www.3liz.com/blog/rldhont/index.php?post/2012/07/17/OpenStreetMap-Tiles-in-QGIS</p><p>http://www.3liz.com/blog/rldhont/index.php?post/2012/07/17/OpenStreetMap-Tiles-in-QGIS</p><p>http://www.gdal.org/frmt_wms.html</p><p>http://www.gdal.org/frmt_wms.html</p><p>http://hub.qgis.org/projects/quantum-gis/wiki/QGIS_Server_Tutorial</p><p>http://hub.qgis.org/projects/quantum-gis/wiki/QGIS_Server_Tutorial</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>243</p><p>How to do it…</p><p>1. Open QGIS.</p><p>2. Load up and style some layers:</p><p>� You need at least one vector layer to offer a WFS.</p><p>� You need at least one raster layer to offer a WCS.</p><p>� WMS can be any combination of layers, you can choose to server each as an</p><p>independent layer or as a combined layer.</p><p>3. Edit the Project properties in File | Project Properties:</p><p>1. Open the OWS server tab.</p><p>2. Check the Service Capabilities box to enable GetCapabilities.</p><p>3. Fill out some of the boxes so that end users know what your server is about,</p><p>who runs it, and how to contact you:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>244</p><p>4. Now examine the WMS capabilities section:</p><p>Most of these features are optional optimizations. Pick and choose</p><p>what suits your needs.</p><p>5. Here, you can set the maximum extent that clients should expect.</p><p>6. The CRS restrictions option lets you limit what projections are allowed.</p><p>7. Exclude Layers allows you to have layers in your project that don't show up on</p><p>the web.</p><p>8. Add geometry to feature response is an optional enhancement if you are building</p><p>a web map and you want to be able to work with the actual vectors (if it is a vector</p><p>to begin with).</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>245</p><p>9. GetFeatureInfo precision is about how close a user has to click to query a location. If</p><p>you have a lot of data, you probably want this number to be small; but if you have only</p><p>a few features, making this bigger makes it easier for end users.</p><p>10. Set Maximums for GetMap request if you want to reduce the load on your server</p><p>by limiting how much data a user can request at once. This is a good idea for a</p><p>public server. 2560 x 2048, as shown in the screenshot, is enough pixels for an</p><p>HD-resolution screen to be filled in a single request.</p><p>11. Next, take a look at the WFS capabilities section:</p><p>Only enable WFS if you want users to be able to request vector</p><p>data as vectors. This can be more intensive than WMS on your</p><p>bandwidth. Also, do not enable WFS-T features unless you secure</p><p>your server to only permitted users.</p><p>1. Check the Published box next to any vector layers that you want to be usable</p><p>over WFS.</p><p>2. To enable WFS-T, check the Update, Insert, and Delete checkboxes. As they</p><p>are separate, you can choose to only allow new data (Insert), only allow</p><p>edits to existing data (Update), or only allow removal of data (Delete). Insert</p><p>would be the safest option as it prevents editing or deletion of existing data.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>246</p><p>12. Finally, take a look at WCS capabilities:</p><p>This an all or none feature. Don't enable this unless you want users</p><p>to be able to download the original raster data.</p><p>13. When you are done setting options, click on the OK button.</p><p>14. Now, save the project in a place where the QGIS server has access to it.</p><p>In Apache, this is usually a folder such as /var/www/.</p><p>15. Once saved, you can test access from any OGC-compliant web client:</p><p>1. For a simple test, use a fresh QGIS project and the Add WMS dialog.</p><p>2. The GetCapabilties URL will look something like http://localhost/cgi-</p><p>bin/qgis_mapserv.fcgi?map=/usr/local/share/qgis/QGIS-</p><p>NaturalEarth-Example.qgs.</p><p>The key part of this URL that is somewhat unique to QGIS server is the</p><p>map parameter, which is followed by the full system path to the QGIS</p><p>project file.</p><p>This may seem odd, but adding your QGIS server as a WMS in QGIS is</p><p>a great way to test whether it's working.</p><p>How it works…</p><p>The QGIS server is a middleman that takes in web service requests and translates</p><p>them into</p><p>QGIS internal calls, returning the requested data or rendered images, which are delivered to</p><p>the end user via the web server.</p><p>www.it-ebooks.info</p><p>http://localhost/cgi-bin/qgis_mapserv.fcgi?map=/usr/local/share/qgis/QGIS-NaturalEarth-Example.qgs</p><p>http://localhost/cgi-bin/qgis_mapserv.fcgi?map=/usr/local/share/qgis/QGIS-NaturalEarth-Example.qgs</p><p>http://localhost/cgi-bin/qgis_mapserv.fcgi?map=/usr/local/share/qgis/QGIS-NaturalEarth-Example.qgs</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>247</p><p>There's more…</p><p>The QGIS server contains many options that allow you to control which types of service to</p><p>offer, which layers to offer over each service, and how to style these services. Alternatives to</p><p>the QGIS server include MapServer and GeoServer (refer to the Managing GeoServer from</p><p>QGIS section in this chapter).</p><p>See also</p><p>f For more details, refer to the main documentation for the QGIS server at http://</p><p>hub.qgis.org/projects/quantum-gis/wiki/QGIS_Server_Tutorial.</p><p>f Once you create a service, test it by adding your service to a QGIS project. Refer to</p><p>the previous recipes in this chapter for how to add WMS, WFS, or WCS services.</p><p>Scale-dependent rendering</p><p>While they are not specifically for web services, being able to change the styling and presence</p><p>of data based on the scale of the map can have a huge impact on the speed and readability</p><p>of web services. Unlike printed maps, web maps are viewed at multiple scales. This variation</p><p>in scales often requires different cartography to keep the map legible and usable.</p><p>Getting ready</p><p>You'll need a QGIS project, preferably one with a high data density or differing levels of</p><p>information. A good example is road data, where you have major, minor, local, and other</p><p>variants of road classification. caryStreets.shp converted from CAD in a previous</p><p>chapter is a good example.</p><p>How to do it…</p><p>1. Open QGIS and load caryStreets.shp.</p><p>2. Now, open the attribute table and look for an attribute to filter in. In</p><p>caryStreets.shp, there are several potential columns to use, such as</p><p>StreetType, Major_Road, and Main_Road.</p><p>StreetType appears to be classes, whereas the other two</p><p>columns appear to be True or False flags. Any of these are</p><p>decent candidates for filtering rules.</p><p>www.it-ebooks.info</p><p>http://hub.qgis.org/projects/quantum-gis/wiki/QGIS_Server_Tutorial</p><p>http://hub.qgis.org/projects/quantum-gis/wiki/QGIS_Server_Tutorial</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>248</p><p>3. Now, open the Properties section for the layer:</p><p>1. Switch to the Style tab to edit the symbology.</p><p>2. Change the top-right dropdown to Rule Based Rendering.</p><p>3. Create a new rule (green plus sign).</p><p>4. In the pop up dialog set Label to Major Roads and Filter to "Major_Road" = 't'.</p><p>5. (Optional) You can use the expression builder to build the filter statement and test it.</p><p>Click on the … button to open the dialog.</p><p>You could create two copies of Major with different scale ranges</p><p>so that as you zoom in, the major roads become thicker at the</p><p>same time that minor roads are enabled.</p><p>This is what your layer looks like before and after you create the first rule:</p><p>6. Now, add another rule for minor roads by filtering for "Major_Road" = 'f'.</p><p>7. This time, you're going to enable the Scale range option.</p><p>8. Set Minimum (exclusive) to 1:100,000. For any scale bigger than this, the features</p><p>will be hidden. For Maximum (inclusive), type in 1:0, which will disable the</p><p>Max filter.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>249</p><p>9. Pick a different line type and/or color for the minor roads:</p><p>10. You should now have two rules, one for major roads and one for minor roads:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>250</p><p>You don't have to open the edit rule dialog; you can directly modify</p><p>parts of the rules in the Rule Based Rendering page.</p><p>11. Go back to the map and zoom in to 1:50,000, then zoom out to 1:250,000. The</p><p>minor roads should appear and disappear as you change past the 1:100,000 scale:</p><p>How it works…</p><p>The goal with scale-dependent rendering is generally to make your map readable at many</p><p>different zoom levels. By setting the Min and Max scales for each layer or subfeatures within</p><p>a layer, you can declutter a map for readability. The rendering engine just checks the scale</p><p>against each rule before deciding what to render.</p><p>There's more…</p><p>Scale-dependent rendering can be used in several ways. This can be used to change the</p><p>styling based on zoom or hide or reveal data based on the zoom level. However, it's also</p><p>not limited to just changing styles or layers. You can also perform scale-dependent labeling,</p><p>which is part of data-driven labeling described in the Configuring data-defined labels recipe</p><p>in Chapter 10, Cartography Tips, of this book.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>251</p><p>Scale rules also work on raster layers; however, this only allows you to turn a raster on and off.</p><p>It doesn't allow you to change its appearance.</p><p>If you have a QGIS server set up from earlier in this chapter, the scaling rules should apply</p><p>to your web services (WMS and WFS).</p><p>You probably don't want to use something as complex as a street</p><p>layer via WFS in a web browser because it's almost guaranteed to</p><p>crash. Stick to pushing such complex layers as Tiles or WMS.</p><p>See also</p><p>The Rule Based Rendering has a lot of features crammed into it. However, this is not yet</p><p>a comprehensive guide to everything that it can do, so you'll need to explore and perform</p><p>Internet searches for now.</p><p>Hooking up web clients</p><p>Sometimes, the best way to share a map is to build a website with a map embedded in it.</p><p>There are many methods to accomplish this goal, ranging from a simple dump of a few layers</p><p>to a highly-interactive map, which is based on web services. There are many web clients that</p><p>will work with standard OGC services. This recipe will show you how to build a simple web map</p><p>using Leaflet—a popular JavaScript library that is used to create web maps.</p><p>Getting ready</p><p>You will need the qgis2leaf plugin and some sample data. The schools_wake.shp (Points)</p><p>and census_wake_2000.shp files make for a good example.</p><p>How to do it…</p><p>1. Install and enable the qgis2leaf plugin.</p><p>Make sure to check out the qgis2web plugin, which is a newer</p><p>variant that works similarly but has some different options.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>252</p><p>2. Load up some layers to make a map composition.</p><p>Make a copy of your layer and eliminate unnecessary columns that</p><p>you don't need to show on the web map. Reducing the size of the</p><p>attribute table will make it easier to read popups with information</p><p>and speed up web page loading.</p><p>3. Style the map as you want it to appear online.</p><p>Styling can be really tricky. Leaflet and other web map libraries don't</p><p>support 100% of the same options as QGIS. Try making a few maps,</p><p>changing settings, and re-exporting these maps a few times to</p><p>figure out how to get it the way that you want. It may not look good</p><p>in QGIS but look good in the export.</p><p>4. (Optional) Configure labels. In this example, label the School names.</p><p>Only black labels are currently supported. Though you can probably</p><p>customize the CSS and JavaScript (js) after the export if you need</p><p>labels in a different style.</p><p>5. Open the qgis2leaf plugin from its icon on the toolbar or from the Web menu:</p><p>1. Click on the GetLayers button to add the layers from your map to the</p><p>export list.</p><p>2. There are lots of options here, and they are optional. Go ahead and check</p><p>Create Legend. If you made labels, also check Export Labels and labels</p><p>on hover.</p><p>Create Cluster is a fantastic option if you have a lot of points on the</p><p>map. This will group points into a circle with a number indicating</p><p>how many points are near there. As you zoom in, they will split apart</p><p>into smaller groups, until at some zoom, all the points are in their</p><p>original spot.</p><p>3. For the frame size, you can pick a size of the page that you want the map to</p><p>take up (in pixels). However,</p><p>fullscreen works well if the map is the only thing</p><p>that you care about.</p><p>4. Go ahead and add a tile-based base layer; Stamen Terrain is an interesting</p><p>choice. Keep in mind that you can only have one of these on at a time, but</p><p>you can toggle between them.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>253</p><p>5. Pick an output folder location and fill in the remaining map information</p><p>describing how you want it to show up in the results.</p><p>6. Export the project:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>254</p><p>6. After exporting, the map should open in your web browser. If it doesn't, open your</p><p>operating system file explorer (or web browser) and navigate to the output folder.</p><p>You should see a new folder called export_year_month_day_hour_minute_</p><p>seconds (for example, export_2015_02_19_11_34_05). Inside this folder is</p><p>index.html. Open this file with a web browser to see your map:</p><p>7. Note that all the vectors are clickable, and the popup will display the attribute table</p><p>information. If you turned on labels and hover, then hovering over a point will display</p><p>the name.</p><p>How it works…</p><p>The qgis2leaf plugin converts your map into something that is compatible with the web.</p><p>Generally, this means converting vector data to the GeoJSON format and generating an</p><p>HTML page (web page) with some JavaScript to create and populate the map.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>255</p><p>Raster layers are trickier, and if you can, try to stick to using Tile or WMS services to serve</p><p>them. Refer to the next section to see how to use Tiles or WMS.</p><p>If you need host tiles locally, try using the QTiles plugin to generate them.</p><p>There's more…</p><p>The next logical step is to make the map dynamic based on a web service. To do this, you can</p><p>swap static files for web services:</p><p>1. Add a WMS layer to the map (you can use the previous recipe in this chapter on QGIS</p><p>server if you have it running). Add an external source WMS, such as the USGS NAIP</p><p>Airphoto. (Here's the GetCapabilities URL, http://isse.cr.usgs.gov/arcgis/</p><p>services/Orthoimagery/USGS_EROS_Ortho_NAIP/ImageServer/WMSServe</p><p>r?request=GetCapabilities&service=WMS).</p><p>2. Re-export with the same settings:</p><p>www.it-ebooks.info</p><p>http://isse.cr.usgs.gov/arcgis/services/Orthoimagery/USGS_EROS_Ortho_NAIP/ImageServer/WMSServer?request=GetCapabilities&service=WMS</p><p>http://isse.cr.usgs.gov/arcgis/services/Orthoimagery/USGS_EROS_Ortho_NAIP/ImageServer/WMSServer?request=GetCapabilities&service=WMS</p><p>http://isse.cr.usgs.gov/arcgis/services/Orthoimagery/USGS_EROS_Ortho_NAIP/ImageServer/WMSServer?request=GetCapabilities&service=WMS</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>256</p><p>Now that you've created the Leaflet map, if you wanted to get into JavaScript</p><p>programming, all of the code that you need is in the directory produced, either</p><p>directly in index.html or in the js folder. In particular, you can see exactly</p><p>how layers are styled and added to the map.</p><p>You don't have to use Tiles or WMS for raster layers but this is recommended. If you do want to</p><p>use a local file, be warned there is a bug currently where some exports fail unless your raster</p><p>is converted to a .jpg format image in EPSG:4326 projection.</p><p>See also</p><p>f Don't forget to look at the documentation for the Leaflet JavaScript library on how to</p><p>customize your results after the export at http://leafletjs.com/.</p><p>f qgis2web plugin aims to combine qgis2leaflet and qgis2ol3 plugins (https://</p><p>plugins.qgis.org/plugins/qgis2web/), which means it also includes export</p><p>to OpenLayers 3 that is very similar to Leaflet but uses the OpenLayers JavaScript</p><p>library. Lizmap (http://www.3liz.com/en/lizmap.html) and QGIS Web Client</p><p>(https://github.com/qgis/QGIS-Web-Client) are two more popular options</p><p>that add more elaborate prebuilt interfaces but require a little more setup.</p><p>Managing GeoServer from QGIS</p><p>QGIS does not only serve as a frontend for the QGIS server, but it can also serve as a frontend</p><p>for other similar servers. GeoServer is one of the most popular ones, and you can configure</p><p>it from QGIS, upload layers, or even edit the style of a GeoServer layer using the QGIS</p><p>symbology tools.</p><p>Getting ready</p><p>For this recipe, you will need the GeoServer Explorer plugin. This can be installed using</p><p>Plugin Manager.</p><p>You will also need a running instance of GeoServer. We will assume that you have a local</p><p>one running on port 8080, but you can replace the corresponding URL with the one of the</p><p>GeoServer instance that you have available, whether local or remote.</p><p>How to do it…</p><p>1. Open the GeoServer Explorer by navigating to Web | GeoServer | Geoserver</p><p>Explorer. The explorer will appear on the right-hand side of the QGIS window.</p><p>www.it-ebooks.info</p><p>http://leafletjs.com/</p><p>https://plugins.qgis.org/plugins/ol3/</p><p>https://plugins.qgis.org/plugins/ol3/</p><p>http://www.3liz.com/en/lizmap.html</p><p>http://www.3liz.com/en/lizmap.html</p><p>https://github.com/qgis/QGIS-Web-Client</p><p>https://github.com/qgis/QGIS-Web-Client</p><p>http://www.it-ebooks.info/</p><p>Chapter 9</p><p>257</p><p>2. Click on the GeoServer Catalogs item in the explorer tree, and then select the New</p><p>catalog button.</p><p>3. Complete the fields in the dialog that will appear to define a new catalog and</p><p>click on OK:</p><p>4. The new catalog will be added to the explorer tree, and you can now browse</p><p>its content.</p><p>5. Open the zipcodes_wake.shp layer and style it.</p><p>6. In the QGIS Layer List, drag the entry corresponding to the zipcode_wake.shp layer</p><p>and drop it on the catalog item in the GeoServer Explorer. The layer will be uploaded</p><p>and added to the default workspace of the catalog.</p><p>7. You can check whether the layer is now in the catalog by opening a web browser</p><p>and going to the GeoServer web interface at http://localhost:8080/</p><p>geoserver/web/.</p><p>How it works…</p><p>The GeoServer Suite plugin communicates with GeoServer using its REST API. By linking QGIS</p><p>with the GeoServer REST API, it allows you to easily configure many elements that, otherwise,</p><p>should be configured manually, such as the styling of layers.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>QGIS and the Web</p><p>258</p><p>There's more…</p><p>The Geoserver Explorer plugin has a lot of features to work with GeoServer. Here are some</p><p>additional ideas so that you can explore them. For more information, check out the plugin</p><p>help at http://boundlessgeo.github.io/qgis-geoserver-plugin/.</p><p>Editing a remote style</p><p>Once the layer is in the GeoServer catalog, you can edit its style without having to upload the</p><p>layer again. Just open the Styles branch in the explorer tree under the corresponding catalog</p><p>and double-click on the style to edit, or select the Edit... item in the context menu that is</p><p>shown when right-clicking on the element:</p><p>The QGIS symbology dialog will be opened, and you can edit the style in there. Once you close</p><p>the dialog, the style will be uploaded and updated in the catalog.</p><p>Support for multiple formats</p><p>The GeoServer REST API only supports shapefiles for vector layers, but you can drag and drop</p><p>any layer in any format that is supported by QGIS. This plugin will take care of converting it</p><p>before uploading, in case this is needed.</p><p>See also</p><p>Don't have a Geoserver instance, it's pretty easy to setup for testing. See http://geoserver.</p><p>org/ for details.</p><p>www.it-ebooks.info</p><p>http://boundlessgeo.github.io/qgis-geoserver-plugin/</p><p>http://geoserver.org/</p><p>http://geoserver.org/</p><p>http://www.it-ebooks.info/</p><p>259</p><p>10</p><p>Cartography Tips</p><p>In this chapter, we will cover the following recipes:</p><p>f Using Rule Based Rendering</p><p>f Handling transparencies</p><p>f Understanding the feature and layer blending modes</p><p>f Saving and loading styles</p><p>f Configuring data-defined labels</p><p>f Creating custom SVG graphics</p><p>f Making pretty graticules in any projection</p><p>f Making useful graticules in printed maps</p><p>f Creating a map series using Atlas</p><p>Introduction</p><p>Cartography has changed quite a bit in the past decade as more people transition to purely</p><p>electronic map products on a device or on the Web. While some types of visualizations are</p><p>better suited to different media, many of the underlying tools and techniques can actually be</p><p>applied across the board. This chapter covers a variety of tools that enable you, the QGIS user,</p><p>to maximize the readability and beauty of your maps.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>260</p><p>Using Rule Based Rendering</p><p>In the past, if you wanted to apply a wildly different style to more than one type of data in</p><p>the same source, the only way to do this was to duplicate or subset a layer. With Rule Based</p><p>Rendering, you now just have to create rules that are applied on-the-fly. This opens a huge</p><p>door on cartographic possibilities with different features in the same layer not only having</p><p>different colors but also different fill types, transparency, line type, and all manner of other</p><p>customizations. Extending from categorized symbology, rules also allow for mixing and</p><p>inheritance, allowing for intermediate categories or some shared properties and reducing</p><p>the amount of work to create elegant symbology.</p><p>Getting ready</p><p>Rule Based Rendering is built-in to vector symbology. So, you'll need a good complicated</p><p>vector layer to fully utilize its potential. A road layer is often a good use case, but for this</p><p>example we'll go slightly simpler with busroutesall.shp.</p><p>How to do it…</p><p>1. Load the busroutesall.shp layer.</p><p>2. Right-click on the layer name in the Layers window, select Properties, then pick</p><p>Style on the left-hand side of the new window.</p><p>3. Change the symbology drop-down type to Rule-Based.</p><p>4. Pick the attributes that you want to use to differentiate between groups of features:</p><p>1. In this case, let's edit the initial rule (double-click on the rule or the Edit icon</p><p>between + (add) and - (remove).</p><p>2. Rules can be based on attribute table values or geometry properties,</p><p>including on-the-fly calculated values. First let's style routes shorter than</p><p>2,000 map units apply here. In the Filter box type $length 2000 (don't forget to test this).</p><p>3. Pick some other symbology that differs quite a bit so that it's easy to tell</p><p>them apart (such as a different line type). Click on OK and then click on</p><p>Apply to see to the two rules in action.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>262</p><p>6. Now, things get really interesting. Let's add a subrule by either right-clicking on a rule</p><p>or by highlighting a rule and clicking on the Refine current rules dropdown:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>263</p><p>7. Select Add categories to rule:</p><p>1. In the subdialog, select Route.</p><p>2. Pick a color ramp and/or line style, click on Classify, and then</p><p>click on OK:</p><p>3. Before you click on Apply, edit the main rule and uncheck the Symbol box</p><p>(otherwise, the Rule and Sub Rules list will be additive, which can be useful</p><p>in some cases).</p><p>8. Now, when you look at the Rule list, you will see subrules under their parents.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>264</p><p>9. Finally, let's add a third top-level rule that is not based on the length:</p><p>1. Make a rule filter on the ROUTE name that contains a. The rule will look like:</p><p>"Route" LIKE '%a'.</p><p>2. Pick a line symbol that will make these routes stick out even with their</p><p>current coloring and click on Apply:</p><p>10. Play around some more; there are all sorts of things you can do, from partial</p><p>string matching to splitting by even or odd numbers ("ROUTE" % 2 = 0 is</p><p>even-numbered).</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>265</p><p>11. Finally, the map looks like the following:</p><p>How it works…</p><p>Each rule is processed in the rendering order specified from top to bottom, the last rule being</p><p>drawn last and, therefore, on top. The rules are added to any existing style that is already</p><p>applied to feature. You can change the rendering order by changing the rule order or by</p><p>applying a render order. The filters work just like attribute filters in the field calculator or</p><p>the table search. All of the symbology options are available to vectors and can be applied</p><p>to one or many rules. You can group rules by scale-rendering rules too.</p><p>There's more…</p><p>There are way too many possible ways to use Rule Based Rendering than can be described</p><p>here. You can create rendering groups that inherit rules from their parent and apply their own.</p><p>Each feature given a unique ID could have a completely different look. The big improvement</p><p>over using traditional single symbol, categorized, or graduated symbology is that you don't</p><p>have to edit every possible group, and you can more easily stack rules, mixing and matching</p><p>all the original methods.</p><p>There are some catches. Not everything you do with Rule Based Rendering is possible with</p><p>web services; so, before you go too crazy, consider your output format and test your ideas</p><p>before spending too much time on this.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>266</p><p>Handling transparencies</p><p>Transparency is a lack of pigment or color, such that you can see through one feature to the</p><p>feature beneath. You can think of this as being similar to tinted or stained glass; some light is</p><p>allowed to pass through and reflect off what's inside. When used right, transparency can help</p><p>emphasize or de-emphasize features in a map composition. It can also be used to blend two</p><p>layers to look as if they are one layer.</p><p>Getting ready</p><p>This recipe demonstrates transparency for both vectors and rasters, so we'll need an example</p><p>of each. The lakes.shp and elevlid_D782_6.tif layers will work well for demonstration</p><p>purposes. Load both of these layers in a fresh project, putting lake.shp on top.</p><p>How to do it…</p><p>1. With a vector layer loaded, open Properties and the Style tab.</p><p>2. On the right-hand side of the dialog, you will see a Transparency slider at 0%</p><p>(this means 100% solid or opaque).</p><p>3. Adjust the slider to the right and apply the changes to see them in the map:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>267</p><p>Using a bold or dark color will make it easier to notice the change. You will</p><p>also notice that the Simple fill option shows the original color.</p><p>4. Now to demonstrate this on a raster, first reset the lakes back to 0%.</p><p>5. Swap the order in the Layers list so that elevlid is on top.</p><p>6. Now, open Properties of elevid and the Transparency tab.</p><p>7. The Global transparency option will change the value evenly for the whole raster.</p><p>Set it to 50% and apply it. You should now be able to see the lakes layer, which was</p><p>hidden below:</p><p>No data value is always 100% transparent no matter what Global</p><p>transparency is set to. Use this to easily eliminate values that you</p><p>don't want to show up at all.</p><p>8. (Optional) You may have noticed that below the Global transparency slider is Custom</p><p>transparency options. This will allow you to make particular values more transparent</p><p>than others. You can either assign specific values to specific transparencies, or you</p><p>can add a band to the raster (or use a multiband raster), which specifies the amount</p><p>of transparency to apply to the rest of the raster (some data formats, such as GeoTiff,</p><p>call this an Alpha Transparency band):</p><p>� Reset the global back to 0% (otherwise, this is applied in too)</p><p>� Use the green + sign to add some values</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>268</p><p>� From 100 To 125, 25% transparent</p><p>� From</p><p>125 To 150, 75% transparent</p><p>� Click Apply and notice the lower elevation lakes are harder to see:</p><p>How it works…</p><p>This is really a computer graphics thing, but the simplest explanation is that you're telling the</p><p>computer to combine a percentage of two different layers in the same location instead of the</p><p>top layer's value covering. Based on the math of the original colors and their transparency,</p><p>a blended color is calculated for each pixel on the screen.</p><p>This doesn't begin to explain all the possible variations of appearance that can be achieved</p><p>by mixing multiple layers and multiple transparencies, only tinkering can show you this.</p><p>There's more…</p><p>One classic example of transparencies is to mix hillshades and airphotos. You can place either</p><p>layer on top and then adjust the transparency to let the other show through. Generally, you</p><p>would place the hillshade underneath in this case (but either can work). The end result is a</p><p>landscape that appears to have 3D relief, but it looks like an airphoto.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>269</p><p>Another classic example is to create a mask layer with a hole cut out around the region that</p><p>you want to emphasize. You now place the mask layer on top. Before adding transparency,</p><p>it blocks everything but the hole. Then, you slowly add transparency so that you can see</p><p>surrounding regions, but they are muted and stand out less. For this technique, try a black,</p><p>gray, or white fill for the mask layer. Each will have a slightly different look.</p><p>When styling vectors, you can apply different transparencies to different features in the same</p><p>layer if you use Rule Based Rendering. Each rule can have a different transparency value and</p><p>the entire layer can have yet another transparency modifier in the Layer Rendering section.</p><p>Lastly, keep in mind that not all output formats handle transparency well. In particular, be</p><p>careful using color gradients with transparencies when exporting to PDF. Generally, PNG</p><p>handles transparency, SVG may work or at least allow to you to edit the transparency after</p><p>export, unlike image formats.</p><p>Understanding the feature and layer</p><p>blending modes</p><p>In this recipe, we will look at the different layer and feature blending modes. Using these</p><p>tools, we can achieve special rendering effects, which you may already know from other</p><p>graphics programs.</p><p>Getting ready</p><p>To follow this recipe, you just need to load stamen.png and effect.png from our sample</p><p>data. Make sure that stamen (left-hand side in the following screenshot) is the lower layer</p><p>and effect (right-hand side in the following screenshot) is the upper layer. To test the feature</p><p>blending modes, load blending.shp:</p><p>(Background maps "Watercolor" and "Toner" by Stamen Design, under CC BY 3.0. Data</p><p>by OpenStreetMap, under CC BY SA).</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>270</p><p>How to do it…</p><p>Using the two raster layers, we can try the different blending modes. Of course, this works for</p><p>vector layers, as well:</p><p>1. Double-click on the effects layer to open Layer Properties.</p><p>2. You can find the blending settings by going to Layer Properties | Style | Color</p><p>Rending together with other helpful controls for Brightness, Contrast, Saturation,</p><p>and more, as shown in the next screenshot:</p><p>3. Change the Blending mode and click on Apply to see the results.</p><p>4. Similarly, for vector layers, such as our blending layer, we find the blending mode</p><p>settings by going to Layer Properties | Style | Layer rendering, as shown in</p><p>the following screenshot:</p><p>The main difference is that, for vector layers, we can control how features are blended</p><p>together, and how the result is then blended to the underlying layers using the Feature</p><p>blending and Layer blending modes, respectively. The feature blending mode is applied</p><p>on a per-feature-basis.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>271</p><p>The following screenshot shows the differences between feature and layer blending:</p><p>Feature and/or layer blending in action (from left to right): feature blending only, layer blending only,</p><p>feature and layer blending combined (background maps "Watercolor" and "Toner" by Stamen Design,</p><p>under CC BY 3.0. Data by OpenStreetMap, under CC BY SA).</p><p>The following is an explanation of the preceding screenshots:</p><p>f The leftmost image shows that Feature blending mode is set to Multiply, while Layer</p><p>blending mode is set to default, Normal. This results in a map where the vector</p><p>features are rendered on top of each other using the Multiply mode before the</p><p>whole layer is overlaid on top of the lower layer(s).</p><p>f The center image instead shows Normal Feature blending mode combined with</p><p>Multiply Layer blending mode. You can see how the features can block each other</p><p>out because they are drawn on top of each other.</p><p>f Finally, the rightmost image shows both Layer blending mode and Feature blending</p><p>modes being set to Multiply. In this combination, the Multiply rule is applied on both</p><p>the feature and the layer level and, therefore, we can see features and the underlying</p><p>background layer(s) shining through the features in the upper layer.</p><p>How it works…</p><p>Based on the selected blending mode, the pixel colors (in the RGB mode) of the lower and</p><p>upper layers are mixed as described next. For illustration and quick reference, the following</p><p>figure shows the results of all 12 blending modes (from left to right and top to bottom), except</p><p>for the Normal setting, which does not mix the colors but only uses the alpha channel of the</p><p>upper layer to blend with the layer below it:</p><p>f Lighten: The Lighten mode selects the maximum of each RGB component from</p><p>the foreground and background pixels. Be aware that the results tend to be jagged</p><p>and harsh.</p><p>f Screen: The Screen mode paints light pixels from the upper layer over the lower layer,</p><p>but it skips the dark pixels.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>272</p><p>f Dodge: The Dodge mode will brighten and saturate the lower layer based on the</p><p>lightness of the upper layer. This means that brighter colors in the upper layer cause</p><p>the saturation and brightness of the lower layer to increase. This works best if the top</p><p>pixels aren't too bright; otherwise, the effect is quite extreme.</p><p>f Addition: The Addition mode adds the pixel values of both layers. If the result</p><p>exceeds 1 (in the case of RGB), the respective areas are displayed in white.</p><p>f Darken: The Darken mode creates a result that retains the smallest RGB</p><p>components of both layers. Therefore, this is the opposite of the Lighten mode</p><p>and, just as with Lighten, the results tend to be jagged and harsh.</p><p>f Multiply: The Multiply mode multiplies the values of both layers, thus resulting</p><p>in a darker picture.</p><p>f Burn: The Burn mode causes darker colors in the upper layer to darken the lower</p><p>layer. Burn can be used to tweak and colorize underlying layers.</p><p>f Overlay: The Overlay mode combines the Multiply and Screen blending modes. As a</p><p>result, light parts become lighter and dark parts become darker.</p><p>f Soft light: The Soft light mode is very similar to Overlay, but it uses a combination</p><p>of Burn and Dodge. This is supposed to emulate shining a soft light on an image.</p><p>f Hard light: The Hard light mode is also very similar to Overlay. It is supposed to</p><p>emulate projecting a very intense light on an image.</p><p>f Difference: The Difference mode subtracts the values of the upper layer from the</p><p>lower layer (or the other way around) to always get a positive value. Blending with</p><p>black (which has an RGB value of 0,0,0) produces no change.</p><p>f Subtract: The Subtract mode subtracts the values of one layer from the other.</p><p>In the case of negative values, black is displayed:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>273</p><p>Overview of the 12 blending modes (background maps "Watercolor" and "Toner" by Stamen Design, under CC BY 3.0.</p><p>Data by OpenStreetMap, under CC BY SA): first row: Lighten, Screen, and Dodge;</p><p>second row: Addition, Darken,</p><p>and Multiply; third row: Burn, Overlay, and Soft light; fourth row: Hard light, Difference, and Subtract.</p><p>Saving and loading styles</p><p>What's better than making an awesome style for your feature layers? Being able to easily</p><p>share and reuse them. Both vector and raster styles can be saved and reused—however,</p><p>in slightly different ways.</p><p>Getting ready</p><p>For this recipe, you need two similar vector layers and a set of two similar raster layers. In the</p><p>example data that is provided, use two of the bus route shapefiles and two of the elevation</p><p>rasters (for example, elevlid_D782_6.tif).</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>274</p><p>How to do it…</p><p>First we'll start by copying and pasting styles for vector layers:</p><p>1. Load up two bus route shapefiles and two elevlid rasters.</p><p>2. The simplest method is to copy styles for vectors or rasters. Just right-click on the layer</p><p>name in the list and select Copy Style from the Style menu. Then, right-click on the</p><p>layer that you want to apply this to and select Paste Style from the Style menu. You</p><p>can only copy styles between layers of the same type (for example, Point to Point):</p><p>Try to copy and paste the style of one bus route to the second bus route using the</p><p>right-click menus:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>275</p><p>3. Now to export styles for later use, right-click on a layer and navigate to Properties |</p><p>Symbology.</p><p>4. In the bottom-right, there is a Save Style button in the Style menu. The output</p><p>choices are QGIS Layer Style File… (aka .qml) or SLD File.... Both formats are</p><p>XML text files, QGIS Layer Style... is recommended for maximum compatibility:</p><p>5. SLD is compatible with some other web map systems, but it will not capture your</p><p>QGIS style 100% except in the simplest cases (not all the same options exist in SLD).</p><p>QML is the native QGIS style file. Note the Load style option for later usage.</p><p>6. Go ahead and apply a new style to one of the bus route layers.</p><p>7. Then, save the symbology to QGIS Layer Style File (qml).</p><p>8. In the property dialog of the second bus route file, click Load Style... and pick the file</p><p>that you just saved.</p><p>You can open this file in a text editor and make customizations or,</p><p>for example, batch-find and replace values.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>276</p><p>Rasters are slightly trickier in that you can save a symbology file, but you can also save a color</p><p>table. The color table is a text file that lists raster value ranges and associated color codes.</p><p>It's a much simpler format to hand-edit than XML (QGIS Layer Style File), but does not retain</p><p>things like transparency or classification settings:</p><p>1. Go ahead and apply a new color gradient to one of the elevlid layers.</p><p>2. Save just the color table to a .txt file with the disk button (above the color table on</p><p>the right end of the button row; refer to the screenshot).</p><p>3. In the property of the second elevid file, load the color table and pick the file that you</p><p>just saved.</p><p>4. Apply the changes to your layer style:</p><p>Note that the same Style menu is available as was in the vector</p><p>properties. You can use this to save and load QGIS Layer Style File (.qml)</p><p>just as we did earlier.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>277</p><p>How it works…</p><p>Normally the style information for a layer is saved in the .qgs (if you save your project)</p><p>project file. The various export methods just package up the style information for a layer into</p><p>a separate file in a generic manner (not associated with the original data). This lets you apply</p><p>similar styles to similar data sources.</p><p>Vector symbology is stored in a special XML file that ends in the .qml extension. You can read</p><p>or edit the file if you want, and it can be produced via scripts or copied and pasted to create</p><p>mashups of multiple styles.</p><p>Raster symbology can also be stored in a .qml file. However, there's an additional option to</p><p>export the classification ranges and colors to a simple text file, one value or range of values</p><p>and one color code per line.</p><p>There's more…</p><p>The second format SLD (Style Layer Descriptors) is very common in web services. While not</p><p>all features of QGIS styling have equivalents in SLD, it's still a good starting point to share your</p><p>style across software platforms such as Mapserver or Geoserver.</p><p>Configuring data-defined labels</p><p>If there was a list of top features of QGIS, data-defined labels would be high on that list. They</p><p>offer the ease of automatic labeling with the customization of freehand labeling. You can mix</p><p>and match automatic and custom edits, storing the values in a table for later reference.</p><p>Getting ready</p><p>There are a couple of useful plugins for data-defined labeling which will add the extra attribute</p><p>fields that you need to either an existing layer or make a new layer just for labels. Download</p><p>and install Layer to labeled layer and Create labeled layer.</p><p>How to do it…</p><p>1. Open QGIS and load census_wake2000.shp.</p><p>2. Create a copy of the layer using the Save As dialog, and save the layer as census_</p><p>wake2000_label.shp. (You don't always have to do this but this process does</p><p>modify the table, so it's a good idea to make a backup.)</p><p>3. Highlight census_wake2000_label.shp in the layer list.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>278</p><p>4. Run the Layer to labeled layer plugin (Plugins | Layer to Labeled layer plugin):</p><p>1. Set Label Field to STFID.</p><p>2. Click on OK:</p><p>5. If you look at the attribute table now, you will see a whole bunch of new fields,</p><p>starting with the Lbl prefix, which are NULL:</p><p>6. Now, ensure that you have the Label toolbar open (View | Toolbars | Label):</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>279</p><p>7. Either in the layer (by navigating to Properties | Labels) or using the first button on</p><p>Label Toolbar, Layer Labeling Options, open the label management dialog.</p><p>8. Throughout the dialogs, you will see markers next to each field. A yellow one indicates</p><p>a data-defined attribute, a white marker is the same setting for all:</p><p>If you want to control additional attributes at this point, add a new</p><p>field to the layer. Then, return to this dialog and select the white</p><p>icon to pick the name of the field to use.</p><p>9. Now, you are ready to make custom edits to various labels and have the table store</p><p>the settings. Depending on the setting, there are a couple of ways to make the edits.</p><p>Note that you must toggle editing on the layer before you can change the labels:</p><p>1. You can edit the field directly in the table either by hand, or you can use the</p><p>field calculator to automate repetitive patterns (for example, give all major</p><p>roads the same Font and Color label).</p><p>2. For some attributes, such as X,Y and rotation, you can also edit by hand in</p><p>the map using the Label Toolbar option.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>280</p><p>Example: moving and rotating a label</p><p>1. Toggle editing by clicking on the following icon:</p><p>2. On the Label Toolbar menu, select the Move Label button. Now, click on a label and</p><p>drag it to a new location, releasing the mouse button when you are done moving the</p><p>label. Note that you must ensure that the X and Y fields in step 38 are set for this tool</p><p>to be usable:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>281</p><p>If you check the attribute table you will see that in the LblX and</p><p>LblY fields, the values have now been saved for the labels that</p><p>you moved.</p><p>3. Now, try the Rotate Label button. See if you can make some of the labels fit inside</p><p>their polygons using the move and rotate:</p><p>You can also use the Change Label button to edit the other properties of a specific</p><p>label that you select. This is really nice when you just need some fine-tuning.</p><p>4. Save your edits and toggle editing off to keep your changes.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>282</p><p>How it works…</p><p>The basic premise is that you keep an extra</p><p>set of attributes in a table often as additional</p><p>fields to your existing table.</p><p>You could add fields to your attribute table by hand, and assign them</p><p>to label properties. Using the Layer to Labeled Layer plugin does this</p><p>for you.</p><p>These fields if you set them are used in determining the location, size, font, color, angle, and</p><p>so on, of the label for the given row. If you don't set them, then the automatic settings from</p><p>the labeling engine are kept.</p><p>There's more…</p><p>Data-defined labels are powerful in that you can combine automated, calculated, and custom-</p><p>edited values. They are automated from the built-in labeling engine and calculated using</p><p>the field calculator to populate the data-defined fields (for example, with if statements</p><p>or calculations that are based on other attributes). Lastly, by just making these little hand</p><p>tweaks, you can fix a few not-quite labels that misbehave.</p><p>Note that you don't have to use data-defined labeling on an existing layer. You can create just</p><p>a label layer with the Create labeled layer plugin. In other software, user-defined labeling</p><p>is often called Annotation layers. QGIS also has annotation layers. These are layers where</p><p>you click to add a label to the map and then write and style it however you want. The biggest</p><p>problem is that these layers are not associated with the data that they label. You can't easily</p><p>give them to someone else, and if a label name or style changes, you have to chase down</p><p>and hand-edit every fix. In QGIS, data-defined labeling solves almost all the shortcomings of</p><p>annotation layers because it actually saves to a shapefile with all its properties as fields.</p><p>Creating custom SVG graphics</p><p>This recipe is all about making your map unique by creating custom icons, north arrows, or</p><p>even fill patterns.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>283</p><p>Getting ready</p><p>You will need a vector illustration program (for example, Inkscape or Adobe Illustrator).</p><p>Don't have one? There are several free and open source options available on</p><p>all platforms. Many people in the QGIS community use Inkscape (http://</p><p>inkscape.org), but you can also use LibreOffice Draw or OpenOffice Draw.</p><p>The most common proprietary software equivalent is Adobe Illustrator.</p><p>You will also need a text editor, such as TextEdit (Mac), Notepad, Notepad++ (Windows).</p><p>How to do it…</p><p>1. Open up your vector illustration program.</p><p>2. Set the canvas to a reasonable size to work with. Square ratios tend to work well</p><p>because the icon will eventually be used to mark points in QGIS; 100x100 pixels</p><p>is fine.</p><p>3. Draw a simple shape such as a square, circle, or star. Make sure you go most of the</p><p>way towards the edges and fill the whole page.</p><p>Remember that you will be using this drawing at sizes closer to 8-32 pixels;</p><p>it's just really annoying to work at these scales. when creating and editing</p><p>illustrations</p><p>4. Save the drawing as an .svg file.</p><p>5. Now, open the .svg file in a text editor, search and find the style line of your object,</p><p>and replace it with the following lines:</p><p>stroke-width="param(outline-width) 1"</p><p>stroke="param(outline) #000"</p><p>fill="param(fill) #FFF"</p><p>If working with a complex icon, set your line to a specific color code that is</p><p>different from all other colors in the drawing. Make a note of the color code so</p><p>that you can use it to search the .svg file in your text editor.</p><p>www.it-ebooks.info</p><p>http://inkscape.org</p><p>http://inkscape.org</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>284</p><p>The before-after scenario when this code has been incorporated is shown in the</p><p>following table:</p><p>Before After</p><p>6. Save your changes.</p><p>7. Now, start up QGIS and load a point layer.</p><p>8. Go to Properties | Style.</p><p>9. In the symbology options, there are two levels of objects that make a symbol: the</p><p>marker and then a sublevel of actual symbols that combine to make it.</p><p>10. Select the subobject, which is usually labeled Simple Marker by default.</p><p>11. Now, change the dropdown in the upper-right to SVG Marker.</p><p>12. Below the box displaying the symbol options look for the … button and select to load</p><p>an SVG from file. Use this to select the .svg file that you previously created.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>285</p><p>13. Once imported, you should be able to change the fill color of the symbol (if you</p><p>performed Step 5):</p><p>You may need to adjust the size and widths in large amounts for changes</p><p>to be apparent. Make use of the Apply button to see the changes in the</p><p>map but keep the dialog open for easy adjustment.</p><p>How it works…</p><p>The special text that you add to the .svg file is a marker or placeholder. QGIS looks for these</p><p>particular words and then utilizes them to insert symbol changes on-the-fly as the SVG is read</p><p>into the program.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>286</p><p>There's more…</p><p>While this recipe demonstrated a very simple SVG, this method applies to more complicated</p><p>symbols.</p><p>Also note that in Settings | Options | System, you can set paths to folders of SVGs so that</p><p>all of them will be available in the symbology dialogs all the time.</p><p>See also</p><p>QGIS also lets you customize fill patterns using SVG symbols. The QGIS Training Manual has</p><p>a good example of this at http://docs.qgis.org/2.8/en/docs/training_manual/</p><p>basic_map/symbology.html#hard-fa-creating-a-custom-svg-fill.</p><p>Making pretty graticules in any projection</p><p>A graticule is a set of reference lines on a map that help orient a map reader. They are often</p><p>set at, and labeled, with the coordinates. The tricky part about using graticules, however,</p><p>is projections. If you don't make them correctly, instead of smooth curves between the line</p><p>intersections, you get awkward unusual shapes (mostly straight lines). The default QGIS</p><p>graticule creator is not projection-friendly, so in this recipe, you'll see an add-on processing</p><p>algorithm that does this. This recipe is about ensuring you get nice, smooth, and properly-</p><p>labeled graticules.</p><p>Getting ready</p><p>You don't really need much for this recipe other than a bounding box and a coordinate interval</p><p>that you want to space the lines at. Usually, these will be in Latitude, Longitude WGS 84</p><p>(EPSG:4326), and decimal degrees, respectively, since the whole point of a graticule is</p><p>to add reference lines that help orient a user.</p><p>How to do it…</p><p>1. Start by downloading a Processing Toolbox algorithm specifically for this task called</p><p>Lines Graticule:</p><p>1. Open the Processing Toolbox.</p><p>2. Go to Scripts | Tools | Get scripts from on-line scripts collection:</p><p>www.it-ebooks.info</p><p>http://docs.qgis.org/2.8/en/docs/training_manual/basic_map/symbology.html#hard-fa-creating-a-custom-svg-fill</p><p>http://docs.qgis.org/2.8/en/docs/training_manual/basic_map/symbology.html#hard-fa-creating-a-custom-svg-fill</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>287</p><p>3. In the Not Installed section, check the box for the Lines Graticule algorithm.</p><p>4. Click on the OK button to install the algorithm.</p><p>Every time that you use a tool, it's good to check for updates.</p><p>You will see something like the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>288</p><p>2. Now that you've downloaded the algorithm, open it by navigating to Scripts | Vector</p><p>(it's called Lines graticule though the code is actually pygraticule.py):</p><p>3. You can fill in the parameters by hand if</p><p>you know them or use the … button to get</p><p>values from your existing project.</p><p>4. For now, you can use the defaults that will make a graticule for the whole world. The</p><p>outputs are determined by outfile and graticule. These parameters are optional, you</p><p>can choose to pick one, both, or neither. If you want a GeoJSON file, set the outfile. If</p><p>you want a shapefile, set the graticule (if you want the results to autoload afterwards,</p><p>make sure that the second output is set to temporary or a real file, just not blank).</p><p>Refer to the Help tab for details about each parameter. There are two really important</p><p>values to control the graticule:</p><p>1. The spacing value denotes how often to draw a line (when doing world-scale</p><p>maps, 20 or 30 degrees works well).</p><p>2. The density value denotes how often to put nodes:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>289</p><p>The more nodes, the smoother the curves; however, you get a bigger file</p><p>that takes longer to make. Picking the right density may require trial and</p><p>error to find the largest density before you notice the lines stop curving</p><p>smoothly for a given map scale.</p><p>5. Once you've chosen your settings, click on Run.</p><p>6. After it runs, a vector layer should get loaded with the results. This won't look all that</p><p>exciting, just straight lines making a grid.</p><p>7. The real magic is to now enable projection on-the-fly with one of the many decent</p><p>world-wide projections such as "World Robinson (EPSG:54030):</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>290</p><p>8. (Optional) If it doesn't look like the image, but instead still has straight lines that are</p><p>oddly spaced, you need to disable the QGIS rendering simplification:</p><p>1. Pick the layer from Properties | Rendering.</p><p>2. Make sure that Simplify geometry is disabled:</p><p>9. (Bonus) Generate a vector grid from Vector | Research tools. The difference looks</p><p>like the following:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>291</p><p>How it works…</p><p>Graticules are basically line layers (though sometimes they are also polygons). If you draw a</p><p>grid with nodes only at the points where two lines intersect, you can easily see how distorting</p><p>the grid will lead to blocky shapes. The key to smooth graticules is adding additional line</p><p>nodes in between the intersections (that is, increase the node density).</p><p>It's important to note that, when using projections that don't cover the whole world</p><p>(for example, polar or stereographic projections), pick bounding box values that fall</p><p>within the projection limits; otherwise, you may get errors when trying to reproject.</p><p>There's more…</p><p>The primary advantages of graticules in the main map canvas are that you can use them</p><p>as references while working in QGIS, include them in web and digital maps, and have full</p><p>control of the labels and symbology. The method used here differs from other graticule (grid)</p><p>tools in QGIS because it focuses on putting Latitude/Longitude lines with smooth curves as</p><p>references into any projection. Other grid tools focus more on making regular squares across</p><p>a map to subdivide a region.</p><p>The main advantages of the print composer method (next recipe) are its ability to make</p><p>multiple coordinate systems easily and to add tick marks around the outside edge of a map.</p><p>Tick marks are what you commonly see on navigation-oriented maps, such as USGS Topo</p><p>quads, and other printed maps.</p><p>See also</p><p>Lines graticule (aka Pygraticule) can also be used as a pure Python script; for updates and</p><p>more information, refer to https://github.com/wildintellect/pyGraticule.</p><p>To learn how to write your own processing toolbox algorithms, refer to the Writing processing</p><p>algorithms recipe in Chapter 11, Extending QGIS.</p><p>Making useful graticules in printed maps</p><p>A graticule is a set of reference lines on a map that help orient a map reader. They are</p><p>often set at and labeled with the coordinates. For traditional printed maps that are intended</p><p>for navigation and surveying tasks where you want to mark the geographic coordinates,</p><p>sometimes in multiple coordinate systems. This recipe is about adding such reference</p><p>lines to a Print Composer map.</p><p>www.it-ebooks.info</p><p>https://github.com/wildintellect/pyGraticule</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>292</p><p>Getting ready</p><p>You will need a map, typically of a small area (several miles or km across). For this recipe,</p><p>elevlid_D782_6.tif works well.</p><p>How to do it…</p><p>1. Load elevlid_D782_6.tif.</p><p>2. Turn on Projection on-the-fly by selecting UTM Zone 17 N, WGS 84 (EPSG:32617).</p><p>3. Now create New Print Composer.</p><p>4. In Print Composer, select Add New Map, and then draw a rectangle on the canvas.</p><p>5. Now that you have the map, in the dialogs on the right-hand side of the screen select</p><p>the Item Properties tab.</p><p>6. Scroll down or collapse sections until you see the Grids section.</p><p>7. Use the green plus (+) symbol to add a new grid.</p><p>8. Now, edit Interval X and Interval Y to 1,000 map units. (Make sure to tab to the next</p><p>field or click on Enter for the values to stick.):</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>293</p><p>The current map units are UTM-based, meters, which means the</p><p>lines will be 1,000 meters or 1 km apart.</p><p>9. Just below the Interval section, change Line Style and make the lines red so that</p><p>they are easier to see.</p><p>10. Now, scroll down even further to the Draw coordinates section and check the box to</p><p>enable labels for the grid lines:</p><p>11. Once this is enabled, change the top and bottom orientation to vertical, and change</p><p>the font color to red.</p><p>12. Now, to create a second grid, scroll back up to the Grids section and click the green</p><p>plus sign (+) again to add a second grid.</p><p>13. This time, change the CRS (Coordinate Reference System) to WGS 84 (EPSG:4326),</p><p>which is the most common Latitude and Longitude system that people use.</p><p>14. Make this grid be spaced 0.01 map units (that is, degrees in this case) and change</p><p>the style to blue to contrast with the other grid.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>294</p><p>15. Now, scroll down and add Draw coordinates. Also, make the top and bottom</p><p>vertical-oriented so that they avoid the first grid. You can also change the font</p><p>color to match the lines:</p><p>How it works…</p><p>Reference graticules are evenly spaced lines with marked coordinates. Based on your settings</p><p>the composer calculates the positioning of the lines from the map data coordinates. The key</p><p>to making useful graticules in the print composer is to select intervals that are often enough</p><p>to provide reference but not so often that they cover a large portion of the map. It's also</p><p>important to pick intervals that have nice rounded numbers, so that it's easy to calculate the</p><p>value half way between two lines.</p><p>There's more…</p><p>There are two ways to make grids/graticules in QGIS: the print composer for printed maps or</p><p>as a layer in QGIS for printed web maps as an internal usage.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>295</p><p>The main advantages of the print composer method are the ability to do multiple coordinate</p><p>systems easily and to add tick marks around the outside edge of a map. Tick marks are what</p><p>you commonly see on navigation oriented maps, such as USGS Topo quads.</p><p>The primary advantages of graticules in the main map canvas are that you can use them as</p><p>references while working with QGIS and have full control of the labels and symbology. Refer to</p><p>the Making pretty graticules in any projection recipe in this chapter for how to make graticules</p><p>in the main map interface.</p><p>Creating a map series using Atlas</p><p>In this recipe, we will use the Print Composer Atlas functionality to automatically create a PDF</p><p>map book with a series of maps.</p><p>Getting ready</p><p>To follow this recipe, load zipcodes_wake.shp and geology.shp from our sample data. In</p><p>the following screenshots, the zipcodes_wake layer was styled with a simple white border,</p><p>while the geology layer is styled with random colors.</p><p>How to do it…</p><p>The Print Composer Atlas feature</p><p>the sections displayed in the left-hand side, select the Metadata section. You will</p><p>see a box containing all the information corresponding to the layer data origin:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>10</p><p>Functionality provided by the GDAL library, which (mentioned earlier) acts as a provider for</p><p>raster layers, is also available in the Raster menu. This includes processing and data analysis</p><p>methods, but it also includes the information tool that is used to describe a raster data</p><p>source. You will find it by navigating to Raster | Miscellaneous | Info:</p><p>See also</p><p>f This is a more complex way to retrieve properties as you can call the tool by adjusting</p><p>the parameters with more details to get additional information. To know more, check</p><p>the gdalinfo help page at http://www.gdal.org/gdalinfo.htm .</p><p>Importing data from text files</p><p>Data can be imported from text files, providing some additional about how the geometry</p><p>information is stored in the text. This recipe shows you how to create a new points layer,</p><p>based on a text file.</p><p>www.it-ebooks.info</p><p>http://www.gdal.org/gdalinfo.html</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>11</p><p>How to do it…</p><p>1. Select the Add delimited text layer menu entry from the Layer menu. You will see a</p><p>dialog like the following one:</p><p>2. In the upper field, enter the path to the elev_lid792_randpts.csv file in the</p><p>sample dataset. That file contains a points layer as text.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>12</p><p>3. Once you enter the file path or select it in the file browser that can be opened by</p><p>clicking on the Browse button, the fields in the lower part of the dialog will be filled,</p><p>as shown in the following screenshot:</p><p>We are using a CSV file that has values separated by commas, so you</p><p>must select the CSV option in the Format field.</p><p>The X field and Y field drop-down lists will be populated with the fields that are</p><p>available, which are described in the first line of the text file. Select X for X field</p><p>and Y for Y field. Now, QGIS knows how to create the geometries and has enough</p><p>information to create a new layer from the text file.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>13</p><p>4. Enter a name for the layer in the Layer name field and click on OK. The layer will be</p><p>added to the QGIS project, as shown in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>14</p><p>5. No information about the CRS is contained in the text file or entered in the</p><p>parameters dialog, so it must be added manually. In this case, the CRS used is</p><p>EPSG:3358. To set this as the CRS of the layer, right-click on the layer name and</p><p>select Set layer CRS:</p><p>6. In the CRS selection dialog, select the EPSG:3358 CRS and click on OK. The layer</p><p>now has the correct CRS.</p><p>How it works…</p><p>Data is read from the text file and processed to create geometries. All the fields in the table</p><p>(all data in a row in the text file) are also added, including the ones used to create the</p><p>geometries, as you will see by right-clicking on the layer and selecting Open attribute table,</p><p>as shown in the following screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>15</p><p>Along with the CSV file, this file may contains a CSVT file, which describes the types of the</p><p>fields. This is used by QGIS to set the appropriate type for the attributes table of the layer. If</p><p>the CSVT file is missing, as in our example's case, QGIS will try to figure out the type based on</p><p>the values for each field.</p><p>There's more…</p><p>Layers created from text files are not restricted to point files. Any geometry can be created</p><p>from the text data. However, if it is not a point, instead of selecting two columns, you must</p><p>place all the geometry information in a single one and enter a text representation of the</p><p>geometry. QGIS uses the Well-Known Text (WKT) format, which is a text markup language for</p><p>vector geometries, to describe geometries as strings. Here is an example of a very simple CSV</p><p>file with line features and two attributes:</p><p>geom,id,elevation</p><p>LINESTRING(0 1, 0 2, 1 3),1,50</p><p>LINESTRING(0 -1, 0 -2, 1 -3),2,60</p><p>LINESTRING(0 1, 0 3, 5 4),3,70</p><p>See also</p><p>f To know more about the WKT format, you can go to http://en.wikipedia.org/</p><p>wiki/Well-known_text</p><p>www.it-ebooks.info</p><p>http://en.wikipedia.org/wiki/Well-known_text</p><p>http://en.wikipedia.org/wiki/Well-known_text</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>16</p><p>Importing KML/KMZ files</p><p>KML and KMZ files are used and produced by Google Earth and are a popular format. This</p><p>recipe shows you how to open them with QGIS.</p><p>How to do it…</p><p>1. To open a KML layer, select Layer/Add vector layer.... In the dialog that opens, click</p><p>on the Browse button to open the file selector dialog. Select the Keyhole Markup</p><p>Language (KML) format and then select the file that you want to load. In the example</p><p>dataset, you can find several KML files. Select the elcontour1m.kml file. Click on</p><p>OK in the vector layer selector dialog, and the layer will be added to your project, as</p><p>shown in the following screenshot:</p><p>KMZ files can also be opened in QGIS.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>17</p><p>2. Go to Layer | Add vector layer.... In the dialog that opens, click on the Browse button</p><p>to open the file selector dialog. Select the All files option to view all the files and then</p><p>select the elcontour1m.kmz file. There is not a KMZ file type defined in QGIS, but</p><p>QGIS supports it because the underlying OGR library can read KMZ files as well.</p><p>3. Click on OK on the open layer dialog to open the selected layer.</p><p>From the layers contained in the KMZ file, you must select one of them. In this case, only a</p><p>layer is contained in the elcontour1m.kmz file, so it is loaded automatically. The layer will</p><p>be added to your QGIS project.</p><p>How it works…</p><p>KMZ files are compressed files that contain a set of layers. When you select it, the OGR library</p><p>will unzip the content of this file and then open the layers that it contains.</p><p>If just a single layer is contained, you will not see the layer selection dialog. QGIS will</p><p>automatically open the only layer in the KMZ file.</p><p>There's more…</p><p>As KMZ is not recognized as a supported format, the KMZ file will not appear in the QGIS</p><p>browser. However, the browser supports zipped files, and a KMZ file is actually a zipped file</p><p>with KML files inside it. Unzip it in a folder and then you will be able to use the QGIS Browser</p><p>to open the layers it contains.</p><p>Importing DXF/DWG files</p><p>CAD files, such as DXF and DWG files, can be opened with QGIS. This recipe shows you how to</p><p>do this.</p><p>How to do it…</p><p>1. To open a DXF layer, select Add vector layer... in the Layer menu. In the dialog that</p><p>opens, click on the Browse button to open the file selector dialog. Select the Autocad</p><p>DXF format and then the file that you want to load.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>18</p><p>2. In the example dataset, you can find several DXF files. Select the Wake_</p><p>ApproxContour_100.dxf file. Click on OK in the vector layer selector dialog and</p><p>the layer will be added to your project, as shown in the following screenshot:</p><p>How it works…</p><p>DXF files are read as normal vector layers although they do not have the same structure</p><p>as a regular vector layer as they do not allow adding arbitrary attributes to each geometry.</p><p>There's more…</p><p>The example DXF file that you opened contained just one type of geometry. DXF files can,</p><p>however, contain several of them: in this case, they cannot be added to QGIS in one layer.</p><p>When this happens, QGIS will ask you to select the type of geometry that you want to open.</p><p>In the sample dataset, you will find a file named CSS-SITE-CIV.dxf. Open it and you will</p><p>see the following dialog:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>19</p><p>Select one of the available geometries, and a layer will be added to your QGIS project.</p><p>Opening DWG files</p><p>DWG is a closed format of Autodesk. This</p><p>will create one map for each feature in the so-called</p><p>Coverage layer. In this recipe, the zipcodes layer will serve as a Coverage layer, and we will</p><p>create one map for each zipcode feature:</p><p>1. Click on the New Print Composer button or press Ctrl + P to get started. You will be</p><p>prompted to set a title for the new composer. This can be left empty if you want QGIS</p><p>to generate a title automatically.</p><p>2. Click on the Add new map button and drag open a rectangle on the composer page</p><p>to create a map item for the main map.</p><p>3. To activate the Atlas functionality, we enable the map item's Controlled by atlas</p><p>checkbox. The following screenshot shows the fully configured map's item properties.</p><p>In the Controlled by atlas section, we can select which zoom mode Atlas should use:</p><p>1. Margin around feature: This is the most flexible option, which tells Atlas to</p><p>zoom to the feature while keeping the specified margin percentage around</p><p>the feature.</p><p>2. Predefined scale (best fit): This tells Atlas to use the one predefined project</p><p>scale (configurable in Project Properties | General | Project scales) where</p><p>the feature best fits in.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>296</p><p>3. Fixed scale: This keeps the same scale for all maps of the series; the scale</p><p>is configured in the map's Main properties, that is, 100,000 in the following</p><p>screenshot:</p><p>4. Next, we add a label for the title using the Add new label button. This title label will</p><p>display the zip code polygon feature's NAME value which will be automatically updated</p><p>by Atlas for each map in the series. To achieve this, we insert the following expression</p><p>in the input field of the label item's Main properties:</p><p>[%attribute( @atlas_feature, 'NAME' ) %]</p><p>5. To finalize the Atlas configuration, we need to go to the Atlas generation tab.</p><p>There, we first have to enable the Generate an atlas checkbox. This activates</p><p>the Configuration section, where we can pick the Coverage layer and set it to</p><p>the zipcodes_wake layer, as shown in the following screenshot.</p><p>6. To preview the Atlas output, we can now click on the Preview Atlas button. This</p><p>button is only active if the Generate an atlas checkbox in the Atlas generation tab is</p><p>enabled. Once the preview mode is active, you can step through the map series using</p><p>the arrow buttons right besides the Preview Atlas button.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 10</p><p>297</p><p>7. When we are happy with the preview, we can export the map series. The output</p><p>behavior is controlled by the configuration in the Atlas generation tab's Output</p><p>section, which you can also see in the following screenshot. Atlas supports exporting</p><p>to separate image, SVG, or PDF files. Activate the Single file export when possible</p><p>option to combine all maps into one PDF and click on the Export Atlas as PDF</p><p>button, as shown in the following screenshot:</p><p>How it works…</p><p>The Atlas feature provides access to a series of variables related to the current feature. We</p><p>already used this to display the NAME value of the current feature in the title label using the</p><p>[%attribute( @atlas_feature, 'NAME' ) %] expression. Besides @atlas_feature,</p><p>you have access to the following variables:</p><p>f @atlas_feature: This is the feature ID of the current Atlas feature. This makes it</p><p>possible to use this information in rules to, for example, hide or highlight features</p><p>based on their ID.</p><p>f @atlas_geometry: This is the geometry of the current Atlas feature and can be</p><p>used in rules to, for example, only show features of other layers if their geometry</p><p>intersects the Atlas feature geometry.</p><p>f @atlas_featurenumber: This is the number of the current Atlas feature.</p><p>f @atlas_totalfeatures: This is the total number of features in the Atlas</p><p>coverage layer.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Cartography Tips</p><p>298</p><p>There's more…</p><p>Overview maps are a great way to provide context to more detailed main maps. To add</p><p>an overview map (as shown in the upper-right corner of the composition in the following</p><p>screenshot), you need to add a second map item to the composition. To turn this map item</p><p>into an overview map, go to Item properties | Overviews and click on the button with the</p><p>green plus sign. This will add an Overview 1 entry and enable the Draw "Overview 1"</p><p>overview configuration GUI:</p><p>f Map frame: The Map frame drop-down list enables us to define the main map that</p><p>should be referenced by the overview map. By default, the map items are named</p><p>Map 0, Map 1, Map 2, and so on, depending on the order they were added to the</p><p>composition. Therefore, we will select the Map 0 entry if the main map was the first</p><p>item that was added to the composition.</p><p>f Frame style: The Change … button can be used to choose a style for the overview</p><p>frame. Usually, this will be a simple fill with transparency.</p><p>f Blending mode: These are supported by overview frames, as explained in detail in</p><p>the Understanding the feature and layer blending modes recipe.</p><p>f Invert overview: Enable the Invert overview checkbox if you want to apply the</p><p>overview frame style to the areas outside the extent of the main map.</p><p>f Center on overview: Enable the Center on overview checkbox if you want the</p><p>overview map to automatically pan to center on the extent of the main map.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>299</p><p>11</p><p>Extending QGIS</p><p>In this chapter, we will cover the following topics:</p><p>f Defining custom projections</p><p>f Working near the dateline</p><p>f Working offline</p><p>f Using the QSpatiaLite plugin</p><p>f Adding plugins with Python dependencies</p><p>f Using the Python console</p><p>f Writing Processing algorithms</p><p>f Writing QGIS plugins</p><p>f Using external tools</p><p>Introduction</p><p>QGIS can do many things on its own. However, as with all software, there are limits to its</p><p>default abilities. The great news is there are many ways to extend QGIS to do even more</p><p>through built-in customization options, existing add-on plugins, creating new analysis</p><p>algorithms, creating your own plugins, and using external software that compliments QGIS.</p><p>This chapter covers just a few of the common customizations and plugins that haven't been</p><p>mentioned in other chapters, and how you can get started with making your own add-ons to</p><p>share with others.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Extending QGIS</p><p>300</p><p>Defining custom projections</p><p>Map projections stump just about everybody at some point in their GIS career, if not more</p><p>often. If you're lucky, you just stick to the common ones that are known by everyone and your</p><p>life is simple. Sometimes though, for a particular location or a custom map, you just need</p><p>something a little different that isn't in the already vast QGIS projections database. (Often,</p><p>these are also referred to as Coordinate Reference System (CRS) or Spatial Reference</p><p>System (SRS).)</p><p>I'm not going to cover what the difference is between a Projection, Projected Coordinate</p><p>System, and a Coordinate system. From a practical perspective in QGIS, you can pick the one</p><p>that matches your data or your intended output. There's lots of little caveats that come with</p><p>this, but a book or class is a much better place to get a handle on it.</p><p>Getting ready</p><p>For this recipe, we'll be using a custom graticule, a grid of lines every 10 degrees</p><p>(10d_graticule.json.geojson), and the Natural Earth 1:10 million coastline</p><p>(ne_10m_coastline.shp).</p><p>How to do it…</p><p>1. Determine what projection your data is currently in. In this case, we're starting with</p><p>EPSG:4236, which is also known as Lat/Lon WGS84.</p><p>2. Determine what projection you want to make a map in. In this example, we'll be</p><p>making an Oblique Stereographic projection centered on Ireland.</p><p>3. Search the existing QGIS projection list for a match or similar projection. If you open</p><p>the Projection dialog and type Stereographic, this is a good start.</p><p>4. If you find a similar projection and just want to customize it, highlight the proj4</p><p>string and copy the information. NAD83(CSRS) / Prince Edward</p><p>means that the specification of the format is not</p><p>available. For this reason, QGIS, like other open source applications, does not support DWG</p><p>files. To open a DWG file in QGIS, you need to convert it. Converting it to a DXF file is a good</p><p>option as this will let you open your file in QGIS without any problem. There are many tools to</p><p>do this. The Teigha converter can be found at http://opendesign.com/guestfiles/</p><p>TeighaFileConverter and is a popular and reliable option.</p><p>Another option is using the free service offered by Autodesk, called Autocad 360, which can</p><p>be found at https://www.autocad360.com/.</p><p>Opening a NetCDF file</p><p>The NetCDF data is a data format, which is designed to be used with array-oriented scientific</p><p>data, and it is frequently used for climate or ocean data, among others. This recipe shows you</p><p>how to open a NetCDF file in QGIS.</p><p>How to do it…</p><p>NetCDF files are raster files, and they can be opened using the Add raster layer menu. Select</p><p>NGMT NetCDF Grid for CDF as the file format in the file selection dialog that you will see,</p><p>and select the rx5dayETCCDI_yr_MIROC5_rcp45_r2i1p1_2006-2100.nc file from</p><p>the example dataset. Click on OK.</p><p>www.it-ebooks.info</p><p>http://opendesign.com/guestfiles/TeighaFileConverter</p><p>http://opendesign.com/guestfiles/TeighaFileConverter</p><p>https://www.autocad360.com/</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>20</p><p>How it works…</p><p>The proposed NetCDF file contains a single variable, which is opened as a regular raster layer.</p><p>There's more…</p><p>A NetCDF file can contain contain multiple layers. In this case, QGIS will prompt you to select</p><p>the one that you want to add from the ones contained in the specified file.</p><p>When only one layer is available, it is opened directly, as in the previously described example.</p><p>The NetCDF Browser plugin</p><p>Another way of opening NetCDF files is using the NetCDF Browser plugin. Select the Manage</p><p>and install plugins... menu to open the plugin manager. Go to the Not installed section</p><p>and type netcdf in the search field to filter the list of available plugins. Select the NetCDF</p><p>Browser plugin and click on Install plugin to install it. Close the plugin manager.</p><p>The plugin is now installed, and you can open it by selecting NetCDF Browser in the</p><p>Plugins menu:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>21</p><p>Select the NetCDF file in the upper field. The other fields will be updated with the content of</p><p>the selected file. Select a layer from the available ones and click on Add to add the layer to</p><p>your QGIS project.</p><p>Saving a vector layer</p><p>QGIS supports multiple formats, not just to read vector layers but to also save them.</p><p>This recipe shows you how to export a vector layer, converting it to a different format.</p><p>Getting ready</p><p>You will use the layer named poi_names_wake.shp in this recipe. Make sure that it is</p><p>loaded in your QGIS project.</p><p>How to do it…</p><p>1. Right-click on the name of the points layer in the QGIS table of contents and select</p><p>the Save as... menu. You will see the following window:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>22</p><p>2. Let's suppose that you want to use this layer to create a web map. A popular format</p><p>supported by libraries, such as Leaflet of OpenLayers 3, is the GeoJSON format.</p><p>Select GeoJSON in the format field and enter a path and filename in the Save as field.</p><p>3. In the Save as dialog, click on OK. The GeoJSON file will be created.</p><p>How it works…</p><p>The OGR library, which is used by QGIS to read and open files, is also used to write them.</p><p>Not all of the formats that are supported for reading purposes are also supported for</p><p>writing purposes.</p><p>You can export even the layers that are not originally file-based to a file, such as a layer</p><p>coming from a PostGIS database or a WFS connection. Just select the layer in the table</p><p>of contents and proceed as just explained.</p><p>There's more…</p><p>The Save as dialog allows additional configuration beyond what you have seen in the example</p><p>in this recipe.</p><p>Fine-tuning the export operation</p><p>Depending on the format that you select to export your layer, different options are available</p><p>to configure how the layer is exported.</p><p>The options are shown by clicking on the More options button. Select GeoJSON as the export</p><p>format and then display the options for that particular format. The COORDINATE PRECISION</p><p>option controls the number of decimal places to write in the output GeoJSON file. The default</p><p>precision is too high for almost all cases, and most of the time, having three or four decimal</p><p>places is more than enough. Set the precision to 4, enter a valid path and filename, and export</p><p>the layer by clicking on OK. Your points layer will now be saved in a smaller GeoJSON file. You</p><p>can open this with a text editor to verify that the coordinates are expressed with the selected</p><p>precision or compare its size with the one created without specifying a precision value.</p><p>Opening the layer after creating it</p><p>If you want to work with the layer after it is created, check the Add saved file to map box. The</p><p>output layer will be opened and added to your current QGIS project.</p><p>Saving a raster layer</p><p>Raster layers can be exported to a different file. The export process can be used to crop the</p><p>layer or perform resampling, creating a modified layer. This recipe shows you how to do this.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>23</p><p>Getting ready</p><p>Open the elev_lid792_1m layer in your QGIS project.</p><p>How to do it…</p><p>1. Right-click on the name of the raster layer in the QGIS table of contents and select</p><p>the Save as... menu. You will see the following window:</p><p>2. In the Resolution fields, replace both of them with a value of 2. The original</p><p>resolution (the size of the cell) is 1, as you saw in a previous recipe.</p><p>3. Enter an output file path in the Save as field.</p><p>4. Click on OK. The layer will be saved with a coarser resolution than the original one.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>24</p><p>How it works…</p><p>The GDAL library is used to save the file. Not all formats supported for input are also</p><p>supported for output, but the most common ones are supported for both operations.</p><p>There's more…</p><p>The layer can be exported with a reduced extent. In the QGIS canvas, zoom to a small part of</p><p>the raster layer. Then open the Save as dialog. In the Extent section, click on the Map view</p><p>extent button. The bounding coordinates of the current map view will be placed in the four</p><p>coordinate fields.</p><p>Enter a file path to save the file to and click on OK. A layer with a reduced extent covering only</p><p>the region shown in the map view will be exported.</p><p>Reprojecting a layer</p><p>Layers may be in a CRS other than the one that is best for a given task. Although QGIS</p><p>supports on-the-fly reprojection when rendering, other tasks, such as performing spatial</p><p>analysis, may require using a given CRS or having all input layers in the same one.</p><p>This recipe shows you how to reproject a vector layer.</p><p>Getting ready</p><p>Open the layer named Davis_DBO_centerline.shp from the sample dataset.</p><p>How to do it…</p><p>The Davis_DBO_centerline.shp layer uses a CRS with feet as the unit, which makes</p><p>this unsuitable for certain operations. We plan to use this layer in future recipes to calculate</p><p>routes and work in metric units, so including this in a CRS that uses them is then a much</p><p>better option:</p><p>1. Right-click on the layer name in the table of contents and select Save as....</p><p>2. Select Selected CRS in the drop-down list to specify a different output CRS.</p><p>Click on the Browse button to select a CRS. You will see the CRS selector dialog.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>25</p><p>3. You will be converting the point to the EPSG:26911 CRS. Use the filter box to find it</p><p>among the list of available CRSs and select it. Then click on OK.</p><p>4. Click on OK in the Save as dialog to create the layer. A new shapefile will be created</p><p>with the projected lines.</p><p>How it works…</p><p>Reprojecting is done by the OGR library when it saves</p><p>the file because this is one of the</p><p>options that it supports.</p><p>There's more…</p><p>Raster layers can be reprojected in a similar way:</p><p>1. In the Save as dialog, for raster layers, you can find a CRS field with a Browse button.</p><p>2. Click on it to open the CRS selector, and select the destination CRS.</p><p>3. When you click on OK, the raster layer will be exported using the selected CRS</p><p>instead of its original one.</p><p>Batch format conversion</p><p>The Save as dialog can be used to convert the format of a single layer. When several layers</p><p>have to be converted, it is a better idea to use some automation. This recipe shows you how</p><p>to easily convert an arbitrary number of layers.</p><p>Getting ready</p><p>No previous preparation is needed. Batch conversion is not performed based on open</p><p>layers but performed directly on files, so there is no need to open layers in QGIS before</p><p>converting them.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>26</p><p>How to do it…</p><p>1. Open the Processing Toolbox menu by selecting Toolbox in the Processing</p><p>menu. The Processing Toolbox menu is the main element of the QGIS Processing</p><p>framework, and it is used to call its algorithms:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>27</p><p>2. In the filter box of the Processing Toolbox menu, type save to filter the list of</p><p>available algorithms. Locate the Save selected features algorithm, right-click on</p><p>it, and select Execute as batch process. The batch processing interface will be</p><p>displayed, as shown in the following screenshot:</p><p>3. In the upper cell in the Input layer column, click on the ... button and select</p><p>Select from filesystem. A file selector dialog will appear. Select the content of the</p><p>batch_conversion folder in the dataset. It should have a total of three files. Click</p><p>on OK on the file selection dialog. The batch processing interface should now have</p><p>all these selected files, one in each row in the parameters table.</p><p>4. In the Output layer column, click on the button in the first row. A dialog for saving the</p><p>file will be opened. Select a file path in your filesystem where you want to save the</p><p>output files and type converted.geojson as the output filename. Click on OK</p><p>and a new dialog like the one shown in the following screenshot will appear:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>28</p><p>5. Select Fill with parameter values in the first field and Input layer in the second</p><p>one. Click on OK. All the rows in the table will now have an output value, which</p><p>was created using the entered filename as a prefix, followed by the name of the</p><p>input layer.</p><p>6. To avoid layers being loaded after they are created, set the first cell in the Load into</p><p>QGIS column to No. Then, double-click on the column header to automatically copy</p><p>this value to all the rows below.</p><p>7. With the table already complete, you can launch the batch conversion process by</p><p>clicking on Run. The GeoJSON files will be created in the specified paths.</p><p>How it works…</p><p>The conversion is performed by an algorithm from the QGIS Processing framework. Processing</p><p>algorithms can be run either as individual algorithms or, in this case, in a batch process.</p><p>Outputs of Processing algorithms can be created in all formats supported by QGIS. The format</p><p>is selected using the corresponding extension in the filename and, unlike in the case of saving</p><p>a single layer, does not have to be selected in a field or list. Using geojson as the extension</p><p>for your output files, you tell processing that you want to generate a file in this format.</p><p>Although the algorithm saves only the selected features of the layer, if there is no selection,</p><p>it will use all the layer features. This is the default behavior of all algorithms in processing.</p><p>As there is no selection in the layers that you have converted, all of their features will have</p><p>been used.</p><p>When converting files this way, the additional options from the Save as dialog are not</p><p>available, and the default configuration values are used.</p><p>There's more…</p><p>You can also convert vector layers with another more complex algorithm from the Processing</p><p>Toolbox menu, which allows you to enter the configuration parameters used by the underlying</p><p>OGR library that takes care of the process. It's called Export vector. Find it in the toolbox,</p><p>right-click on it, and select Execute as batch process:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>29</p><p>In this case, the output format is not controlled by the extension of the output filename as it</p><p>happens with other processing algorithms according to what has been already explained.</p><p>Batch reprojection</p><p>Layers can be reprojected in a batch operation without having to enter parameters individually</p><p>on the Save as dialog. This recipe shows you how to reproject a set of layers to a different</p><p>CRS using an algorithm from the Processing Toolbox menu. You will see how to reproject all</p><p>the files accompanying the Davis_DBO_centerline.shp file that you reprojected in the</p><p>Reprojecting a layer recipe.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>30</p><p>How to do it…</p><p>1. In the filter box of the Processing Toolbox menu, type Reproject to filter the list of</p><p>available algorithms. Locate the Reproject layer algorithm, right-click on it, and select</p><p>Execute as batch process. The batch processing interface will be shown, as follows:</p><p>2. In the upper cell of the Input layer column, click on the ... button and select Select</p><p>from filesystem. A file selector dialog will appear. Select the content of the davis</p><p>folder in the dataset and add the files to the table.</p><p>3. In the first cell in the Target CRS column, click on the ... button. A CRS selector will</p><p>appear. Select the EPSG:26911 CRS, as you did in a previous recipe when converting</p><p>a single layer. Copy the value to the rest of rows in the column by double-clicking on</p><p>the column header.</p><p>4. Set all the values in the Reprojected layer column. Select a file in the first cell, and</p><p>then use the Fill with parameter value option to automatically fill the rest of rows.</p><p>5. Once the table is complete, click on Run to reproject the layers.</p><p>How it works…</p><p>The reprojection algorithm is a part of the Processing framework, so you can select the output</p><p>format by changing the output file extension. You can use this to not only reproject a set of</p><p>input layers but to also convert their format, all in a single step.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>31</p><p>There's more…</p><p>Raster layers can also be reprojected with another algorithm from the Processing Toolbox</p><p>menu named Warp (reproject). These inputs are rather similar to the ones in the reprojection</p><p>tool for vector layers with some additional parameters that are specific to raster layers. Select</p><p>the algorithm, right-click on it, and select Execute as batch process to run it and convert a</p><p>set of raster layers.</p><p>Loading vector layers into SpatiaLite</p><p>SpatiaLite is a single file relational database that is built on top of the well-known SQLite</p><p>database. It can store many layers of various types, including nonspatial tables. Interfaces to</p><p>the format also allow the ability to run spatial queries of various kinds. It's a highly-flexible and</p><p>portable format that is great for everyday use, especially when working on standalone projects</p><p>or with only one user at a time. SpatiaLite works in a similar manner to PostGIS without the</p><p>need to configure or run a database server.</p><p>Getting ready</p><p>Pick a vector layer and load it up in QGIS. This step is optional, as you can pick the source</p><p>layer from the filesystem in a later dialog.</p><p>How to do it…</p><p>1. Create a SpatiaLite database if you don't already have one and name it cookbook.db.</p><p>The easiest way to do this is with the Browser tab, as shown in the following</p><p>screenshot:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>32</p><p>2. Then, pick one of the following methods to import your data. The first option is faster,</p><p>but the second option gives</p><p>you more control over the import settings:</p><p>� Import method 1—the fast method</p><p>1. In the QGIS Browser tab, find the layer that you want to copy to</p><p>the database.</p><p>2. Drag and drop this layer on the Spatialite DB entry.</p><p>If you have a lot of files listed, this will be quite difficult as the browser</p><p>doesn't scroll during the drag operation. You can optionally open a</p><p>second browser window and drag the layer across. Also, note that this</p><p>defaults to multi-type geometry. If you need to control the options, use</p><p>the next method.</p><p>� Import method 2—the standard method</p><p>1. Open DB Manager from the Database menu.</p><p>2. Expand the Spatialite item to list your databases. Expand the database</p><p>that you want to connect to.</p><p>3. Click on the following import layer icon:</p><p>4. A dialog will pop up, providing you with import options.</p><p>SQL databases are usually case insensitive, so you can use</p><p>all lower case characters. Also, never use spaces or special</p><p>characters in table names; this can just lead to headaches later.</p><p>An occasional underscore is okay.</p><p>5. Select the layer to import from the drop-down list.</p><p>6. Fill in a name for the new table.</p><p>7. In most cases, the only thing left to do is check the Create spatial</p><p>index checkbox.</p><p>8. If this works, great. Now, you can load the layer to the map and verify</p><p>that it's identical to the input.</p><p>This method is more similar to traditional database import and very</p><p>similar to the PostGIS recipe next in this chapter.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>33</p><p>How it works…</p><p>QGIS converts your geometry to a format that is compatible with SpatiaLite and inserts it,</p><p>along with the attribute table. Afterwards, it updates the metadata tables in SpatiaLite to</p><p>register the geometry column and build the spatial index on it. These two postprocesses make</p><p>the database table appear as a spatial layer to QGIS and speed up the loading of data from</p><p>the table when panning and zooming.</p><p>There's more…</p><p>The import dialog contained a few other features that are often useful. You can reproject</p><p>data as part of the import process if you want, or you can specify the projection if QGIS didn't</p><p>detect it properly. You can also name the geometry column something different than the</p><p>default, geom; for example, utmz10n83 (this is normally not recommended). You can specify</p><p>the character encoding of the text in the event that it's not handled correctly.</p><p>You can even use the dialog to append data to an existing table; for example, you have</p><p>multiple counties with the same data structure that come as two separate files, but you</p><p>want them all in one layer.</p><p>If, for some reason, the layer didn't import the way that you want, delete it and redo the</p><p>import. If you delete layers, make sure to learn how to vacuum the database to recover</p><p>the now empty space in the file and shrink its total size (this is not automatic).</p><p>Look for the Vacuum option as a button in many graphical tools. If you</p><p>don't see it, no worries, just run the SQL, VACUUM;.</p><p>What happens if this fails? Databases can be really picky sometimes. Here are some common</p><p>issues and solutions:</p><p>f It could be character encoding (accents, non-Latin languages), which requires that</p><p>you specify the encoding.</p><p>f It could be picky about mixing multilayers with regular layers. Multilayers is when you</p><p>have several separate geometries that are part of one record. For example, Hawaii is</p><p>actually many islands. So, if you only have one row representing Hawaii, you need to</p><p>cram all the island polygons into one geometry field. However, if you mix this with North</p><p>Dakota, which is just a polygon, the import will fail. If you have this problem, you'll need</p><p>to perform the import on the command-line using ogr2ogr and its newish feature, -nlt</p><p>PROMOTE_TO_MULTI, which converts all single items to multi-items to fix this.</p><p>f Depending on your original source, you may have a mix of points, lines, and polygons.</p><p>You'll either need to convert this to a Geometry Collection, or you need to split each</p><p>type of geometry into a separate layer. Geometry Collections are currently poorly-</p><p>supported in many GIS viewers, so this is only recommended for advanced users.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>34</p><p>See also</p><p>If you need more advanced settings or can't get the QGIS tool to work, you may need to use</p><p>the QspatiaLite Plugin (install this with Manage Python Plugins under the Plugins menu), the</p><p>spatialite-gui (download this from https://www.gaia-gis.it/fossil/spatialite_</p><p>gui/index) application, or the ogr2ogr command line (this comes with QGIS, which is part</p><p>of OSGeo4w shell on Windows, or the terminal on Mac or Linux).</p><p>Loading vector layers into PostGIS</p><p>PostGIS is the spatial add-on to the popular PostgreSQL database. It's a server-style database</p><p>with authentication, permissions, schemas, and handling of simultaneous users. When</p><p>you want to store large amounts of vector data and query them efficiently, especially in a</p><p>multicomputer networked environment, consider PostGIS. This works fine for small data too,</p><p>but many users find its configuration too much work when SpatiaLite may be better suited.</p><p>Getting ready</p><p>Pick a vector layer and load it in QGIS. You will also need to have a working copy of Postgres/</p><p>PostGIS running, a PostGIS database created, and an account that allows table creation.</p><p>BostonGIS maintains a decent tutorial on installation for Windows, and</p><p>getting a PostGIS set up for everyone. You can find this at http://</p><p>www.bostongis.com/?content_name=postgis_tut01#316.</p><p>You should configure QGIS to be aware of your database and its connection parameters by</p><p>creating a new database item in the PostGIS load dialog or by right-clicking on PostGIS in the</p><p>Browser tab and selecting New Connection:</p><p>You can find more information about PostGIS at http://docs.qgis.org/2.8/en/docs/</p><p>user_manual/working_with_vector/supported_data.html#postgis-layers.</p><p>How to do it…</p><p>Now that you can connect to a PostGIS database, you are ready to try importing data:</p><p>1. Open DB Manager from the Database menu.</p><p>2. Expand the PostGIS item to list your databases. Expand the database that you want</p><p>to connect to, and you should be prompted to authenticate (if you haven't saved your</p><p>password in the settings).</p><p>3. Expand the list and select the Public schema.</p><p>www.it-ebooks.info</p><p>https://www.gaia-gis.it/fossil/spatialite_gui/index</p><p>https://www.gaia-gis.it/fossil/spatialite_gui/index</p><p>http://www.bostongis.com/?content_name=postgis_tut01#316</p><p>http://www.bostongis.com/?content_name=postgis_tut01#316</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/working_with_vector/supported_data.html#postgis-layers</p><p>http://docs.qgis.org/2.8/en/docs/user_manual/working_with_vector/supported_data.html#postgis-layers</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>35</p><p>In general, unless you are performing advanced work and understand how</p><p>Postgres schemas work, place your layers in the Public schema. This is the</p><p>default that everyone expects.</p><p>4. Click on the following import layer icon:</p><p>5. A dialog will pop up, providing you with import options.</p><p>SQL databases are usually case insensitive, so you can use all lowercase.</p><p>Also, never use spaces or special characters in table names; this can just</p><p>lead to headaches later. An occasional underscore is okay.</p><p>6. Select the layer to import from the drop-down list.</p><p>7. Fill in a name for the new table.</p><p>8. Check whether schema is set to public.</p><p>9. In most cases, the only thing left to do is check the Create spatial index checkbox:</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Input and Output</p><p>36</p><p>How it works…</p><p>QGIS converts your geometries to a format that is compatible with PostGIS, and inserts it,</p><p>along with importing the attributes. Afterwards, it updates the metadata views in PostGIS</p><p>to register the geometry column and build the spatial index on it. These two post-processes</p><p>make the database table appear as a spatial layer to QGIS and speed up the loading of data</p><p>from the table when panning</p><p>and zooming.</p><p>There's more…</p><p>The options presented in the dialog are not all the options that are available. If you need</p><p>more control or advanced options present, you'll likely be looking at the command-line tools:</p><p>shp2pgsql (a graphical plugin for pgadmin3 is available on some platforms) and ogr2ogr.</p><p>The shp2pgsql tool generally only handles shapefiles. If you have other formats, ogr2ogr can</p><p>handle everything that QGIS is capable of loading. You can also use these tools to develop</p><p>batch import scripts.</p><p>To import large or complicated CSV or text files, you sometimes will need to use the pgadmin3</p><p>or psql command-line interface to Postgres.</p><p>Need even more control? Then, consider scripting. OGR and Postgres both have very capable</p><p>Python libraries.</p><p>Another option is using the OpenGeo Suite plugin, which has some additional options, such as</p><p>allowing importing multiple layers into a single table or into one table per layer. To learn more</p><p>about this, including how to install it, refer to http://qgis.boundlessgeo.com/static/</p><p>docs/intro.html.</p><p>What happens if this fails? Databases can be really picky sometimes:</p><p>f It could be character encoding (accents, non-Latin languages), which requires</p><p>specifying the encoding.</p><p>f It could be picky about mixing multilayers with regular layers. Multilayers is when you</p><p>have several separate geometries that are part of one record. For example, Hawaii</p><p>is actually many islands. So, if you only have one row representing Hawaii, you need</p><p>to cram all the island polygons into one geometry field. However, if you mix this with</p><p>North Dakota that is just a polygon, the import will fail. If you have this problem, you'll</p><p>need to perform the import on the command-line using ogr2ogr and its new feature,</p><p>-nlt PROMOTE_TO_MULTI, which converts all single items to multi-items, to fix this.</p><p>f Depending on your original source, you may have a mix of points, lines, and polygons.</p><p>You'll either need to convert this to a Geometry Collection, or you need to split each</p><p>type of geometry into a separate layer. Geometry Collections are currently poorly</p><p>supported in many GIS viewers, so this is only recommended for advanced users.</p><p>www.it-ebooks.info</p><p>http://qgis.boundlessgeo.com/static/docs/intro.html</p><p>http://qgis.boundlessgeo.com/static/docs/intro.html</p><p>http://www.it-ebooks.info/</p><p>Chapter 1</p><p>37</p><p>See also</p><p>For more information on PostGIS installation and setup, refer to http://postgis.net/</p><p>install.</p><p>For a more in-depth text on using PostGIS, there are many books available, including Packt</p><p>Publishing's PostGIS Cookbook.</p><p>www.it-ebooks.info</p><p>http://postgis.net/install</p><p>http://postgis.net/install</p><p>http://www.it-ebooks.info/</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>39</p><p>2</p><p>Data Management</p><p>In this chapter, we will cover the following recipes:</p><p>f Joining layer data</p><p>f Cleaning up the attribute table</p><p>f Configuring relations</p><p>f Joining tables in databases</p><p>f Creating views in SpatiaLite</p><p>f Creating views in PostGIS</p><p>f Creating spatial indexes</p><p>f Georeferencing rasters</p><p>f Georeferencing vector layers</p><p>f Creating raster overviews (pyramids)</p><p>f Building virtual rasters (catalogs)</p><p>Introduction</p><p>One of the reasons to use QGIS is its many features that enable management and analysis</p><p>preparation of spatial data in a visual manner. This chapter focuses on common operations</p><p>that users need to perform to get data ready for other uses, such as analysis, cartography, or</p><p>input into other programs.</p><p>In this chapter, you will find recipes to manage vector as well as raster data. These recipes</p><p>cover the handling of data from both file and database sources.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>40</p><p>Joining layer data</p><p>We often get data in different formats and information spread over multiple files. Therefore, one</p><p>important skill to know is how to join attribute data from different layers. Joining data is a way to</p><p>combine data from multiple tables based on common values, such as IDs or categories.</p><p>This exercise shows you how to use the join functionality in Layer Properties to join geographic</p><p>census tract data to tabular population data and how to save the results to a new file.</p><p>Getting ready</p><p>To follow this exercise, load the census tracts in census_wake2000.shp using Add Vector</p><p>Layer (you can also drag and drop the shapefile from the file browser to QGIS) and population</p><p>data in census_wake2000_pop.csv using Add Delimited Text Layer.</p><p>You can also load the .csv text file using Add Vector Layer, but this will load</p><p>all data as text columns because the .csv file does not come with a .csvt</p><p>file to specify data types. Instead, the Add Delimited Text Layer tool will scan</p><p>the data and determine the most suitable data type for each column.</p><p>How to do it…</p><p>To join two layers, there has to be a column with values/IDs that both layers have in common.</p><p>If we check the attribute tables of the two layers that we just loaded, we will see that both</p><p>have the STFID field in common. So, to join the population data to the census tracts, use the</p><p>following steps:</p><p>1. Open the Layer Properties option of the census_wake2000 layer (for example,</p><p>by double-clicking on the layer name in the Layers list) and go to Joins.</p><p>2. To set up a new join action, press the green + button in the lower-left corner of</p><p>the dialog.</p><p>3. The following screenshot shows the Add vector join dialog, which allows you to</p><p>configure the join by selecting Join layer, which you want to use to join the census</p><p>tracts and the columns containing the common values/IDs (Join field and Target field):</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Chapter 2</p><p>41</p><p>If you want to change a join, you just need to select the join</p><p>definition from the list and then press the edit button with the</p><p>pencil icon, which you find below the list. This will reopen the</p><p>join definition dialog, and you can make your changes.</p><p>4. When you press OK, the join definition will be added to the list of joins, as shown in</p><p>the following screenshot.</p><p>5. To verify that you set up the join correctly, close Layer Properties and open attribute</p><p>table to see whether the population columns have been added and are filled with data.</p><p>How it works…</p><p>Joins can be used to join vector layers and tabular layers from many different file and database</p><p>sources, including (but not limited to) Shapefiles, PostGIS, CSV, Excel sheets, and more.</p><p>When two layers are joined, the attributes of Join layer are appended to the original layer's</p><p>attribute table. If you want, you can use the Choose which fields are joined option to select</p><p>which of the fields from the population layer should be joined to the census tracts. Otherwise,</p><p>by default, all fields will be added. The number of features in the original layer is not changed.</p><p>Whenever there is a match between the values in the join and the target field, the new</p><p>attribute values will be filled; otherwise, there will be NULL values in the new columns.</p><p>www.it-ebooks.info</p><p>http://www.it-ebooks.info/</p><p>Data Management</p><p>42</p><p>By default, the names of the new columns are constructed from join layer name with</p><p>underscore followed by join layer column name. For example, the STATE column of</p><p>census_wake2000_pop becomes census_wake2000_pop_STATE. You can change this</p><p>default behavior by enabling the Custom field name prefix option, as shown in the previous</p><p>screenshot. With these settings, the STATE column becomes pop_STATE, which is</p><p>considerably shorter and, thus, easier to handle.</p><p>There's more…</p><p>The join that you've created now only exists in memory. None of the original files have been</p><p>altered. However, it's possible to create a new file from the joined layers. To do this, just use</p><p>Save as … from the Layer menu or Context menu. You can choose between a variety of data</p><p>formats, including the ESRI shapefile, Mapinfo MIF, or GML.</p><p>Shapefiles are a very common choice as they are still the de facto standard GIS data</p><p>exchange format, but if you are familiar with GIS data formats, you will have noticed</p>
  • E-Book Neuroarquitetura no combate ao estresse e a ansiedade
  • Prova - Benefícios Econômicos e Socioambientais_ FICOINV84_Ética e Risco da Reputação no Mercado Financeiro
  • resposta simulado pratica de ensino em gestao educacional
  • E-book-Anki-Para-Concursos-2 0
  • Gerenciando_projetos_e_produtos_com_scrum
  • slidesgo-fundamentos-da-engenharia-de-requisitos-da-teoria-a-pratica-20241012155601N1zN
  • slidesgo-fundamentos-da-arquitetura-de-software-estruturas-e-praticas-eficazes-20241012160035ubuF
  • slidesgo-explorando-metodologias-de-desenvolvimento-abordagens-e-praticas-eficazes-20241012155036Ztql
  • Aquisição e Desenvolvimento de Linguagem - Unifatecie
  • slidesgo-explorando-metodologias-de-desenvolvimento-abordagens-e-praticas-eficazes-20241012155036Ztql
  • PROJETO DE EXTENSÃO II - ENGENHARIA DE SOFTWARE - PROGRAMA DE INOVAÇÃO E EMPREENDEDORISMO
  • Cadastro e Configuração na AWS
  • Aula_Redes de Petri-1
  • Se você tem um problema e está em busca de uma solução inovadora, os pensamentos divergente e convergente são conceitos chave e ferramentas necessá...
  • Inovação aberta é o uso de entradas e saídas ............................. de conhecimento para acelerar a inovação interna e expandir os mercados ...
  • A criação de mensagens publicitárias para mídias impressas perpassa a compreensão e o desenvolvimento do que é uma peça gráfica. São materiais impr...
  • Sistemas dinâmicos são sistemas fora deequilíbrio, caracterizados por estados que mudamcom o tempo. São usados para modelar e fazerprevisões de sis...
  • Qual é a principal função do framework NumPy conforme usado no código exemplo?
  • Qual o valor de 5 3 54 − 3 3 16?3 3 26 3 621 3 29 3 29 3 3
  • Considere o seguinte fragmento de texto: “Técnicas e critérios de teste fornecem ao projetista de software uma abordagem sistemática e teoric...
  • A avaliação laboratorial da hemostasia envolve testes que avaliam o sistema de coagulação e o sistema fibrinolítico. Assinale a alternativa que ind...
  • Qual das alternativas abaixo, apresenta de maneira mais completa, a definição posta na política educacional brasileira a respeito do público com AH...
  • Considere as assertivas abaixo: I. O objetivo da cadeia de custodia é garantir a preservação do local do crime. II. As impressoes digitais são poss...
  • A Cultura da qualidade\: Pergunta 3Selecione uma opção: a. diz respeito à melhoria contínua, à prevenção de erros, à atenção e à responsabilidade ...
  • (VUNESP/2015 - adaptada) No cenário brasileiro, identifica-se um processo que rompe com o modelo hospitalocêntrico, para promover um modelo de base...
  • Na arquitetura de software, a escolha adequada dos padrões arquiteturais visa garantir que o sistema atenda aos requisitos funcionais e não funcion...
  • Colaborar - Av2 - Sistemas Operacionais - B
  • Questao 55 - Filosofias de Supervisao
QGIS 2 Cookbook Become a QGIS power user and master QGIS data management, visualization, and spatial analysis techniques - Engenharia de Software (2025)
Top Articles
Kembalikan Mimpi Amerika - PDF Free Download
Bedtime Stories for Adults: Soothing Stories and Relaxing
Sensitive information: Types, definitions, and examples
Latest Posts
12+ Fran Fine Hair Tips
31 Shampoos for Curly Hair, According to the Pros
Recommended Articles
Article information

Author: Carmelo Roob

Last Updated:

Views: 5700

Rating: 4.4 / 5 (45 voted)

Reviews: 84% of readers found this page helpful

Author information

Name: Carmelo Roob

Birthday: 1995-01-09

Address: Apt. 915 481 Sipes Cliff, New Gonzalobury, CO 80176

Phone: +6773780339780

Job: Sales Executive

Hobby: Gaming, Jogging, Rugby, Video gaming, Handball, Ice skating, Web surfing

Introduction: My name is Carmelo Roob, I am a modern, handsome, delightful, comfortable, attractive, vast, good person who loves writing and wants to share my knowledge and understanding with you.