Practical GIS: Learn novice to advanced topics such as QGIS, Spatial data analysis, and more

Installing the packages on Linux distributions is pretty straightforward. The dependencies are installed with the packages, when there are any. We only have to watch out for three things prior to installing the packages. First of all, the package name of the Apache web server can vary between different distributions. On distros using RPM packages (for example--Fedora, CentOS, and openSUSE), it is called httpd, while on the ones using DEB packages (for example--Debian and Ubuntu), it is called apache2. On Arch Linux, it is simply called apache.

The second consideration is related to distributions which do not update their packages frequently, like Debian. GeoServer has a hard dependency of a specific JRE (Java Runtime Environment). We must make sure we have it installed and configured as the default. We will walk through the Debian JRE installation process as it is the most popular Linux distribution with late official package updates. Debian Jessie, the latest stable release of the OS when writing these lines, is packed with OpenJDK 7, while GeoServer 2.11 requires JRE 8:

      update-alternatives --config java

The last consideration before installing the packages is related to the actual version of QGIS. Most of the distributions offer the latest version in a decent time after release; however, some of them like Debian do not. For those distros, we can use QGIS's repository following the official guide at http://www.qgis.org/en/site/forusers/alldownloads.html.

After all things are set, we can proceed and install the required packages. The order should not matter. If done, let's take a look at GeoServer, which doesn't offer Linux packages to install. It offers two methods for Linux: a WAR for already installed Java servlets (such as Apache Tomcat), and a self-containing platform independent binary. We will use the latter as it's easier to set up:

      cd <geoserver's folder>/bin
      ./startup.sh

Installing the required software on Windows only requires a few installers as most of the packages are bundled into the OSGeo4W installer.

Don't worry if you end up with no solutions, we will concentrate on GeoServer, which runs perfectly on Windows. Just make sure Apache is installed and working (i.e. http://localhost returns a blank page or the OSGeo4W default page), as we will need it later.

Installing the software on macOS could be the most complicated of all (because of GRASS). However, thanks to William Kyngesburye, the compiled version of QGIS already contains a copy of GRASS along with other GIS software used by QGIS. In order to install QGIS, we have to download the disk image from http://www.kyngchaos.com/software/qgis.

PostgreSQL and PostGIS are also available from the same site, you will see the link on the left sidebar. pgAdmin, on the other hand, is available from another source: https://www.pgadmin.org/download/pgadmin-4-macos/. Finally, the GeoServer macOS image can be downloaded from http://geoserver.org/release/stable/, while its dependency of Java 8 can be downloaded from https://www.java.com/en/download/.

The only thing left is configuring the QGIS Server. As the OS X and macOS operating systems are shipped with an Apache web server, we don't have to install it. However, we have to make some configurations manually due to the lack of the FastCGI Apache module, on which QGIS Server relies. This configuration can be made based on the official guide at http://hub.qgis.org/projects/quantum-gis/wiki/QGIS_Server_Tutorial.

Congratulations! You're through the hardest part of this chapter. The following step is to make some initial configurations on the installed software to make them ready to use when we need them. First of all, let's open QGIS. At first glance, it has a lot of tools. However, most of them are very simple and self-explanatory. We can group the parts of the GUI as shown in the following image:

We can describe the distinct parts of the QGIS GUI as follows:

The only thing we will do now without having any data to display, is customizing the GUI. Let's click on Settings and choose the Options menu. In the first tab called General, we can see some styles to choose from. Don't forget to restart QGIS every time you choose a new style.

The next piece of software we look at is PostGIS via pgAdmin. If we open pgAdmin, the least we will see is an empty Server Groups item on the left panel. If this is the case, we have to define a new connection with the plug icon and fill out the form (Object | Create | Server in pgAdmin 4), as follows:

The Name can be anything we would like, it only acts as a named item in the list we can choose from. The Host, the Port, and the Username, on the other hand, have to be supplied properly. As we installed PostgreSQL locally, the host is 127.0.0.1, or simply localhost. As the default install comes with the default user postgres (we will refer to users as roles in the future due to the naming conventions of PostgreSQL), we should use that.

Upon connecting to the server, we can see a single database called postgres. This is the default database of the freshly installed PostgreSQL. As the next step, we create another database by right-clicking on Databases and selecting New Database. The database can be named as per our liking (I'm naming it spatial). The owner of the database should be the default postgres role in our case. The only other parameter we should define is the default character encoding of the database:

Choosing the template0 template is required as the default template's character encoding is a simple ASCII. You might be familiar with character encoding; however, refreshing our knowledge a little bit cannot hurt. In ASCII, every character is encoded on 8 bits, therefore, the number of characters which can be encoded is 2⁸ = 256. Furthermore, in ASCII, only the first 7 bits (first 128 places) are reserved, the rest of them can be localized. The first 7 bits (in hexadecimal, 00-7F) can be visualized as in the following table. The italic values show control characters (https://en.wikipedia.org/wiki/C0_and_C1_control_codes#C0_.28ASCII_and_derivatives.29):

Open source GIS software offer a very high degree of freedom. Their license types can differ; however, they are all permissive licenses. That means we can use, distribute, modify, and distribute the modified versions of the software. We can also use them in commercial settings and even sell the software if we can find someone willing to buy it (as long as we sell the software with the source code under the same license). The only restriction is for companies who would like to sell their software under a proprietary license using open source components. They simply cannot do that with most of the software, although some of the licenses permit this kind of use, too.

There is one very important thing to watch out for when we use open source software and data. If somebody contributes often years of work to the community, at least proper attribution can be expected. Most of the open source licenses obligate this right of the copyright holder; however, we must distinguish software from data. Most of the licenses of open source software require the adapted product to reproduce the same license agreement. That is, we don't have to attribute the used software in a work, but we must include the original license with the copyright holders' name when we create an application with them. Data, on the other hand, is required to be attributed when we use it in our work.

There are a few licenses which do not obligate us to give proper attribution. These licenses state that the creator of the content waives every copyright and gives the product to the public to use without any restrictions. Two of the most common licenses of this kind are the Unlicense, which is a software license, and the Creative Commons Public Domain, which is in the GIS world mostly used as a data license.

The first data we will download is the administrative boundaries of our country of choice. Open data for administrative divisions are easy to find for the first two levels, but it becomes more and more scarce for higher levels. The first level is always the countries' boundaries, while higher levels depend on the given country. There is a great source for acquiring the first three levels for every country in a fine resolution: GADM or Global Administrative Areas. We will talk about administration levels in more details in a later chapter. Let's download some data from http://www.gadm.org/country by selecting our study area, and the file format as Shapefile:

In the zipped archive, we will need the administrative boundaries, which contain our division of choice. If you aren't sure about the correct dataset, just extract everything and we will choose the correct one later.

The second vector dataset we download is the GeoNames archive for the country encasing our study area. GeoNames is a great place for finding data points. Every record in the database is a single point with a pair of coordinates and a lot of attribute data. Its most instinctive use case is for geocoding (linking names to locations). However, it can be a real treasure box for those who can link the rich attribute data to more meaningful objects. The country-level data dumps can be reached at http://download.geonames.org/export/dump/ through the countries' two-letter ISO codes.

Data acquisition with instruments mounted on airborne vehicles is commonly called remote sensing. Mounting sensors on satellites is a common practice by space agencies (for example, NASA and ESA), and other resourceful companies. These are also the main source of open source data as both NASA and ESA grant free access to preprocessed data coming from these sensors. In this part of the book, we will download remote sensing data (often called imagery) from USGS's portal: Earth Explorer. It can be found at https://earthexplorer.usgs.gov/. As the first step, we have to register an account in order to download data.

When we have an account, we should proceed to the Earth Explorer application and select our study area. We can select an area on the map by holding down the Shift button and drawing a rectangle with the mouse, as shown in the following screenshot:

Active remote sensing

As the next step, we should select some data from the Data Sets tab. There are two distinct types of remote sensing based on the type of sensor: active and passive. In active remote sensing, we emit some kind of signal from the instrument and measure its reflectance from the target surface. We make our measurement from the attributes of the reflected signal. Three very typical active remote sensing instruments are radar (radio detection and ranging) using radio waves, lidar (light detection and ranging) using laser, and sonar (sound navigation and ranging) using sound waves. The first dataset we download is SRTM (Shuttle Radar Topographic Mission), which is a DEM (digital elevation model) produced with a radar mounted on a space shuttle. For this, we select the Digital Elevation item and then SRTM. Under the SRTM menu, there are some different datasets from which we need the 1 Arc-Second Global. Finally, we push the Results button, which navigates us to the results of our query. In the results window, there are quite a few options for every item, as shown in the following screenshot:

The first two options (Show Footprint and Show Browse Overlay) are very handy tools to show the selected imagery on the map. The footprint only shows the enveloping rectangle of the data, therefore, it is fast. Additionally, it colors every footprint differently, so we can identify them easily. The overlay tool is handy for getting a glance at the data without downloading it.

Finally, we download the tiles covering our study area. We can download them individually with the item's fifth option called Download Options. This offers some options from which we should select the BIL format as it has the best compression rate, thus, our download will be fast.

Note

If you have access to lidar data in your future work, don't hesitate to use it. Up to this time, it offers the most accurate results.

Passive remote sensing

Let's get back to the Data Sets tab and select the next type of data we need to download--the Landsat data. These are measured with instruments of the other type--passive remote sensing. In passive remote sensing, we don't emit any signal, just record the electromagnetic radiance of our environment. This method is similar to the one used by our digital cameras except those record only the visible spectrum (about 380-450 nanometers) and compose an RGB picture from the three visible bands instantly. The Landsat satellites use radiometers to acquire multispectral images (bands). That is, they record images from spectral intervals, which can penetrate the atmosphere, and store each of them in different files. There is a great chart created by NASA (http://landsat.gsfc.nasa.gov/sentinel-2a-launches-our-compliments-our-complements/) which illustrates the bands of Landsat 7, Landsat 8, and Sentinel-2 along with the atmospheric opacity of the electromagnetic spectrum:

From the Landsat Archive, we need the Pre-Collection menu. From there, we select L8 OLI/TIRS and proceed to the results. With the footprints of the items, let's select an image which covers our study area. As Landsat images have a significant amount of overlap, there should be one image which, at least, mostly encases our study area. There are two additional information listed in every item--the row number and the path number. As these kinds of satellites are constantly orbiting Earth, we should be able to use their data for detecting changes. To assess this kind of use case (their main use case), their orbits are calculated so that, the satellites return to the same spot periodically (in case of Landsat, 18 days). This is why we can classify every image by their path and row information:

Note

To make sure the images are illuminated the same way every time on a given path/row, this kind of satellite is set on a Sun-synchronous orbit. This means, they see the same spot at the same solar time in every pass. There is a great video created by NASA visualizing Landsat's orbit at https://www.youtube.com/watch?v=P-lbujsVa2M.

Let's note down the path and row information of the selected imagery and go to the Additional Criteria tab. We feed the path and row information to the WRS Path and WRS Row fields and go back to the results. Now the results are filtered down, which is quite convenient as the images are strongly affected by weather and seasonal effects. Let's choose a nice imagery with minimal cloud coverage and download its Level 1 GeoTIFF Data Product. From the archive, we will need the TIFF files of bands 1-6.

Note

The tar.gz extension is a shorthand for a gzipped tape archive. It is by far the most common compressed archive type on Unix-like operating systems and any decent compressing software can handle it.

The last dataset we put our hands on is the swiss army knife of open source GIS data. OpenStreetMap provides vector data with a great global coverage coming from measurements of individual contributors. OpenStreetMap has a topological structure; therefore, it's great for creating beautiful visualizations and routing services. On the other hand, its collaborative nature makes accuracy assessments hard. There are some studies regarding the accuracy of the whole data, or some of its subsets, but we cannot generalize those results as accuracy can greatly vary even in small areas.

One of the main strengths of OpenStreetMap data is its large collection and variety of data themes. There are administrative borders, natural reserves, military areas, buildings, roads, bus stops, even benches in the database. Although its data isn't surveyed with geodesic precision, its accuracy is good for a lot of cases: from everyday use to small-scale analysis where accuracy in the order of meters is good enough (usually, a handheld GPS has an accuracy of under 5 meters). Its collaborative nature can also be evaluated as a strength as mistakes are corrected rapidly and the content follows real-world changes (especially large ones) with a quick pace.

Accessing OpenStreetMap data can be tricky. There are some APIs and other means to query OSM, although either we need to know how to code or we get everything in one big file. There is one peculiar company which creates thematic data extracts from the actual content--Geofabrik. We can reach Geofabrik's download portal at http://download.geofabrik.de/. It allows us to download data in OSM's native PBF format (Protocolbuffer Binary Format), which is great for filling a PostGIS database with OSM data from the command line on a Linux system but cannot be opened with a desktop GIS client. It also serves XML data, which is more widely supported, but the most useful extracts for us are the shapefiles.

Due to various reasons, open source shapefiles are only exported by Geofabrik for small areas. We have to narrow down our search by clicking on links until the shapefile format (.shp.zip) is available. This means country-level extracts for smaller countries and regional extracts for larger or denser ones. The term dense refers to the amount of data stored in the OSM database for a given country. Let's download the shapefile for the smallest region enveloping our study area: