4.1 What is a Geographic Information System?

A geographic information system (GIS) is generally described as a collection of various tools and practices that work together to analyze spatial data. At its root, the power of GIS comes from the fact that it combines both spatial and attribute data allowing us to not only see where things are, but also describe what they are in great detail. This spatial database approach helps to expose patterns and links that might otherwise not be visible in a nonspatial context. Esri, the creators of the industry-standard ArcGIS software, describes a GIS as:

You may have noticed that we have described GIS as a geographic information system in the singular, as opposed to describing the field as geographic information systems in the plural. This distinction comes in part from the early days of GIS in the 1960s and 1970s, when computer-aided spatial analysis necessarily relied on mainframe computer hardware and often proprietary command-line software for analyzing data (Coppock & Rhind, 1991). An individual setup could be referred to as a geographic information system. Most spatial analysis carried out today does not rely on the mainframe model, although a specific collection of hardware, software, and data can still be referred to as a geographic information system. Goodchild (1992) described a growing disconnect between the practice of using a GIS and the science that drives GIS technology. He coined the term geographic information science (GISci) as both a way of making a distinction between the two and pointing a spotlight on some of the major theoretical hurdles facing the GIS world.

Today a GIS is most often a combination of a desktop or notebook computer using GIS software with a graphical user interface, while accessing data stored locally, on a centralized server, or in the cloud. The GIS software is often Esri’s ArcGIS, although other commercial and open-source packages such as QGIS are in use, see Chapter 7 for a discussion of available software packages. Data are frequently combined with locally hosted information collected in the field via Global Positioning System (GPS) units for analysis. If all the talk of definitions and distinctions is confusing, do not panic! Colloquially, the software is simply referred to as GIS software, while the practice of working with a GIS is commonly known as doing GIS. While GISci is an important component to the field, many users never come into contact with this element of GIS in their day-to-day activities.

4.2 Layering the Data

GIS is powerful because it can tie spatial vector data to nonspatial database information, allowing us to visualize this information. Spatial vector data are the locational infrastructure; nonspatial database information, or attribute data, refers to features in a table such as schools or types of crime within a particular city. Each database feature corresponds with a coordinate-based vector feature and is mapped within a geographic coordinate space. This results in separate maps or layers of information. While looking at one layer of information can expose spatial patterns not visible from the ground, one of the ways that GIS lets us explore more complex questions is by layering multiple sources of information. By taking multiple layers of data representing natural and human-built features, GIS can create a model of portions of the Earth’s surface, see Fig. 4.1.

These models can be incredibly powerful, allowing us to see previously unknown connections between disparate systems and predict how changes in human behavior may affect the natural environment. Some models only require a few layers of information while others can be quite complex, factoring in many layers of information. One model might show the location of schools in a city relative to crime events. Another example could layer data describing elevation, soil, surface cover, and precipitation information to explore urban flooding. By modifying the data in the surface cover layer we could then determine what impact a proposed parking lot for a new shopping center might have on flash flooding in a city.

Another example of how layering data can be used to answer complex questions is a site suitability scenario. Imagine that you have been tasked with finding areas where an endangered species lives in order to better protect it. This species has certain requirements for life, including the presence of particular plant types for food, a specific type of soil, average temperature range, and amount of annual rainfall. Finding the possible habitat would involve four different layers of information, each describing the requirements above. When the four layers are overlaid, some areas will meet only some of this species’ habitat needs, but other locations will meet all four. In this way, you have discovered the suitable sites for this species to live, see Fig. 4.2. A similar example based in the human world would be choosing a site to build a new factory. The factory would need to be close to major transportation routes, large enough population centers to gather employees, and have suitable terrain for the building. By layering information about the natural and human environments, suitable locations for the factory could be discovered.

These examples describe relatively simple GIS operations, but by layering GIS data, we can discover a great deal of information. Combining this layering approach with more advanced techniques, such as spatial statistical analyses, the power of GIS has made exploring and understanding the world more accessible and manageable.

4.3 What is Remote Sensing?

“Remote sensing describes the collection of data about an object, area, or phenomenon from a distance with a device that is not in contact with the object” (U.S. Army Corps of Engineers, 2003, p. 2-1). This is a broad description, but it generally refers to the use of aerial platforms such as planes, drones, kites, blimps, and satellites for gathering raster imagery. Raster data define space with a continuous series of rows and columns of cells or pixels each with its own attribute value. While remote sensing (RS) is its own field, it often acts as a complement to GIS analyses, adding unique information and analysis techniques to the GIS toolbox. For example, most GIS software packages contain common RS tools for working with raster imagery.

There are two types of RS, active and passive, and they are generally used for different applications. Active RS involves sending out a signal and waiting for its return to the sensor. RADAR and LIDAR are examples of active RS, as they send out energy, microwave and laser pulses respectively, and record the signals as they bounce back (Derr & Little, 1970). Since this effectively measures the distance between the sensor and the target, one of the major uses of active RS is to generate three-dimensional models of surfaces and elevation. RADAR RS also has the advantage that it passes through cloud cover, allowing for imaging even in cloudy atmospheric conditions (ESA Earthnet Online, 2014).

Passive RS does not send out a signal to be returned; rather, it records information using energy already present in the environment. This means that passive imagery is generally collected during the day, when the sun provides plenty of incoming radiation to reflect off the Earth’s surface. This type of RS can be in the form of aerial photographs, but like the active approaches, it can go beyond what we think of as pictures. One of the most powerful elements of remotely sensed imagery is that it lets us see information outside the visible spectrum. Human eyes can see only a narrow portion of the electromagnetic radiation (EMR) spectrum, see Fig. 4.3, but wavelengths that fall outside our range of vision can tell us a great deal about the natural world.

Using information from multiple bands of the EMR spectrum, remotely sensed imagery can help us to identify objects and materials on the surface of the Earth. Every material will respond uniquely to incoming solar radiation, absorbing, transmitting, and reflecting EMR in differing amounts depending on the physical properties of that material and the incoming radiation’s wavelength (Natural Resources Canada, 2015). Using this knowledge, we can look at an image showing the volume of different wavelengths reflected back from a surface, known as the spectral response, and know that one portion of the surface is covered in asphalt while another is a field of grass. That example may sound a bit obvious to the point of not needing a satellite, but RS can also help us to distinguish between much subtler features, differences that oftentimes cannot be determined using our eyes.

One classic example is the use of the infrared portion of the EMR spectrum to monitor vegetation. Not only will different species of plants have different spectral responses at a given time in their lifecycle, the health of a particular species can also be determined based on its spectral response (Tucker, 1979). Because vegetation monitoring often uses a nonvisible portion of the spectrum, it is displayed using false color imagery. This shifts the primary colors of the visible spectrum into the nonvisible portion, allowing us to see how intense the infrared response is in the case of vegetation. An example of false color imagery can be seen in Fig. 4.4; in this example, the colors pink and red indicate healthy green vegetation. This kind of information has a variety of practical uses, from monitoring for drought conditions, tracking responses to climate change, and following the health of individual fields for precision agriculture.

4.4 The Difference Between Vector and Raster Data

Digital geospatial data are generally stored in two different forms: raster or vector. The two formats are fundamentally different from one another in their structures, and each one has strengths and weaknesses regarding their ability to represent the world. Vector data are good at representing discrete objects and features with high levels of precision. A vector file is made of a series of points, lines, and polygons existing on a Cartesian coordinate system, typically a coordinate system tied to the Earth’s grid, as discussed in Chapter 3. Points are quite simple, consisting of a set of X/Y coordinates defining the location, while lines are made up of a series of points that are connected. Polygons are a series of lines that form an enclosed feature; examples of vector data can be seen in Fig. 4.5.

Individual vector features are tied to tabular attribute data representing information about the feature, and each vector feature can be connected to any amount of tabular data. For example, a single point in a vector file might represent a city; querying the point would show a table with fields representing the city name, the population, the demographic breakdown, economic information, or any number of pieces of information tied to that particular point object. In this way, vector data allow us to take advantage of the spatial database structure of GIS. However, because of the discrete nature of vector geometry it is not particularly good at representing continuous features such as elevation. Additionally, the math involved in vector spatial analysis tends to be more complex than that employed in raster analyses.

The structure of raster data is one that most people are likely familiar with, as it is the basis for most of the electronic displays that we use today. Rasters operate in the same way that a cell phone, computer, or television screen does: they are a continuous grid of cells (or pixels), each with its own single attribute value. In the case of a digital photograph, these values represent the colors that form the overall image. Rasters can be photographs, but they can also display nonphotographic information. Fig. 4.6 shows an example of a nonphotographic raster conception of the world where the Earth’s surface has been classified into land-cover categories. Each cell has a single value representing what is on the ground in that grid space and no empty cells exist in the grid. Because of this continuous nature rasters are good for representing data such as elevation or surface temperature.

The single variable per cell is an obvious limitation of the raster format, as natural features are rarely if ever laid out in neat, evenly distributed square cells of material. In reality, nature is not grid-friendly, with uneven distributions of materials and fuzzy boundaries between land-cover types. Related to the issue of the artificially imposed grid is the question of resolution. As illustrated in Chapter 3, the resolution of a raster image indicates how much surface area is described by an individual cell. The lower the resolution, the more generalization is being made about the surface. Higher resolutions are generally preferable, as an image with 1-m resolution will show much greater detail in the scene than one with 1-km-sized cells. Unfortunately, as the resolution increases, so too do storage requirements, and large, high-resolution raster datasets can be slow to display and analyze, not to mention how quickly they can fill computer storage.

4.5 Sources of Raster Data

A great deal of raster data comes from the remote sensing field, in the form of aerial photographs and satellite imagery. Aerial photographs have been taken nearly as long as the photographic process has existed. Today air photos can be found in black and white, color, and color infrared, see Fig. 4.4. Although it might seem simple, aerial photography is not as straightforward as taking a picture from a plane or kite. The surface of the Earth is not flat, and all camera lenses introduce distortion to the images they collect. Orthophotos are aerial images that have been corrected to remove these distortions from the photo, thus representing ground features in their accurate locations from a vertical perspective (Southard, 1958). The process of this transformation is known as image rectification or georectification. By taking photos and digitally georectifying them to remove distortion and apply geographic coordinates, it allows a RS or GIS user to make accurate measurements from the photo, making them suitable for advanced spatial analysis techniques.

Satellite imagery comes from a variety of sources, some public, others private. The topic of choosing appropriate satellite imagery involves many factors; chiefly, these revolve around cost and resolution. Some satellite data are freely available, such as that generated by the Landsat program, while other sources charge for access to imagery. Ideally, freely available data can be used, but sometimes it may not meet all the needs of a particular project, necessitating a purchase of data. As previously mentioned, resolution refers to the scale at which data are collected, and in the context of satellites, it could be in reference to cell size, scene size, return time, or spectral coverage. The cell size is the ground area covered by an individual cell in the image. For example, imagery in the red/green/blue visible spectrum collected by the GeoEye-1 sensor has a resolution of 1.84 m meaning each cell in the raster covers 3.4 m², while Landsat 8’s imagery in the visible spectrum has a resolution of 30 m, covering 900 m² (e-geòs, n.d.; Garner, 2013). If a project needs high levels of detail, the GeoEye imagery would likely be better suited to the task.

Related to resolution is the scene size, or how much surface area is covered in a single image. Generally speaking, satellites with higher cell resolution will cover less surface area in a single scene than those with lower spatial resolutions. Looking at GeoEye and Landsat 8 again, the swath widths of their imaging sensors are 15.2 km and 185 km, respectively. Satellites with smaller scenes will require more images to be combined to cover larger areas, whereas lower resolution imagery can cover the same ground in a single image. Regarding return time, imaging satellites orbit the Earth in such a way that they will be able to return to the same piece of ground every few days or weeks. GeoEye’s return time is less than 3 days, while Landsat 8’s is 16 days. Some projects may require frequent data updates, while others may have no problem waiting a few weeks or months between images for comparison. Keep in mind that cloud cover can render a satellite pass useless if it is heavy enough, so not every return pass will generate usable imagery.

While those factors are important to consider, one of the most crucial elements to understand is the spectral range and resolution of a satellite. The imagery collected by satellites is a record of the EMR that was reflected from the Earth’s surface at the time of the satellite’s pass. Satellite sensors classify specific wavelengths of reflected EMR energy, see Fig. 4.3, into segments and measure their intensity, generating multispectral data. For example, band 2 of Landsat 8’s Operational Land Imager sensor collects information between 436 and 528 nm, corresponding to blue visible light (Taylor, 2016). Multiband imagery is created using this multispectral data by combining different bands to create a composite image. As an example, if one were to display bands 2, 3, and 4 from a Landsat 8 image and display them as blue, green, and red respectively, they can be combined to create a so-called true color image. We can also generate false color images, as bands outside the visible spectrum may also be displayed. Fig. 4.7 shows both true color and false color images of the Murfreesboro, Tennessee (TN) region derived from Landsat 8 OLR data side-by-side. The left image shows true color data (bands 2, 3, and 4) while the image on the right displays a false color near-infrared image (bands 3, 4, and 5). Much like Fig. 4.4, the near-infrared portion of the EMR spectrum is displayed in red in the false color image, indicating healthy green vegetation.

Multispectral imagery has allowed us to learn a great deal about the Earth and its natural processes, but the spectral resolution of satellite sensors can be increased to create what is known as hyperspectral data. Instead of breaking down the EMR spectrum into a dozen bands, hyperspectral data take the same total range of the spectrum and divide it into as many as hundreds of bands (Landgrebe, 2003). This higher spectral resolution allows for a much more precise knowledge of the surface, to the point of being able to distinguish between different mineral content in exposed rock material based on spectral response. Just like the consideration of spatial resolution, spectral resolution is important to consider when choosing a source of RS imagery. Hyperspectral imagery may be needed, but it often provides far more precision than is actually necessary to answer research questions.

4.6 Web GIS as a Component of NeoGeography

The Internet has changed many aspects of our daily lives, and GIS has not been immune to its influence. At its simplest, Web GIS is similar to any other web application: it involves a server hosting content and an end user who accesses the content via hypertext transfer protocol (HTTP) (Fu & Sun, 2010). What distinguishes Web GIS from other websites or Internet-enabled applications is that the content served is geospatial in nature. Web GIS does not necessarily look like desktop GIS software, in part because it tends to operate either through a web browser or a mobile application format (e.g., Android, iOS, etc.). In general, Web GIS is more limited in capabilities when compared to a desktop GIS software package, but this is by design. Most Web GIS users need a fairly small range of tools, most commonly the ability to query locations, create navigation routes, and take simple measurements of distance. All the major commercial mapping applications provide these tools, including Google Maps, Microsoft’s Bing Maps, Yahoo Maps, and MapQuest.

While these services may be invaluable to many, they generally do not provide any specialized GIS tools to end users, particularly analysis-related functions. Many businesses, government agencies, and research organizations have a need for more advanced GIS capabilities in their Web GIS applications, and these are provided by software such as Esri’s ArcGIS for Server. ArcGIS for Server can host interactive map services resembling the interfaces of the large commercial map outfits while also providing some GIS analysis capabilities. For example, a map server may be hosting a raster layer representing elevation. Using one of these advanced tools, an end user can click on a location and the server will analyze the elevation layer, then draw the boundaries of the watershed in which the point resides. While still limited when compared to the capabilities of desktop GIS, this is a step beyond the analysis capabilities of most online mapping applications. Many organizations have Web GIS applications built including tools related to the needs of their field. These services are often for internal use rather than public facing, but some organizations use specialized applications to display data to the public, such as the U.S. Geological Survey’s Earthquake Hazards Program, which displays the locations of detected seismic activity (U.S. Geological Survey, 2016). These web platforms are one component of NeoGeography, discussed in Chapter 1.

4.7 Volunteered Geographic Information

While today’s Web GIS applications may not have the same level of analysis capability as a desktop GIS package, they do have one major feature that desktop GIS lacks: the ease of participation for the public. Desktop GIS can be quite daunting to the novice user, and a good deal of training is generally required to gain the level of knowledge necessary to successfully carry out GIS analyses. Not only are they more user friendly, platforms such as Google Maps, OpenStreetMap, and Wikimapia invite users to assist with data collection and quality control, by adding points of interest, photos, and reporting errors in data throughout the world. Many geospatially enabled mobile apps rely on this user participation as a core component to their operation, such as Yelp, Foursquare, and countless other GPS-enabled services. This kind of interactivity is called volunteered geographic information or VGI (Goodchild, 2007). VGI is not limited to restaurant reviews and vacation photos; it can involve natural hazard warnings and response, scientific data collection, and up-to-the-minute reporting of global events. For example, geotagged Twitter content is commonly mined for event-tracking purposes, both commercial and scientific, although data from these sources are typically analyzed in a more traditional desktop GIS environment. Just as the web enables NeoGeography, VGI is an essential social component to the mix providing a source of data.

Not only are Web GIS applications designed with user friendliness and interactivity in mind, they often take advantage of open-source technologies and focus on software extensibility and data interoperability. By allowing users to freely modify and embed Web GIS technology and spatial data into websites and apps, these services have expanded far beyond their original functionality. Google Maps and Google Earth are good examples of this. The application program interfaces, or APIs, provided for both Maps and Earth have allowed countless users to take advantage of interactive spatial data who would otherwise not be involved in Web GIS. Other open-source technologies like the JavaScript-based D3 library (Bostock, 2013) and the GeoJSON format (Butler et al., 2008) have empowered users to explore and embed geospatial data on the web with an ease unthinkable at the turn of the century.

4.8 The Role of GPS in VGI

One of the factors that has enabled this high level of public participation is the broad reach of GPS technology. Today, anyone with a smartphone can get highly accurate location information about the world around them, enabling the use of geospatial applications. It is difficult to stress just how transformative GPS technology has been for the human experience, but it has changed virtually every aspect of our lives from the supply chain of food we consume to our day-to-day navigational behavior. Although some individuals still consult paper maps for navigation today, the ubiquity of handheld navigation units and GPS-enabled cell phones has changed our entire mode of transport. While there is an argument that reading a paper map is becoming a sadly lost activity, the benefits that GPS has provided to our lives are undeniable, and many would be lost without GPS navigation and restaurant reviews, both literally and figuratively. Between the explosion of GPS usage and the open-source, extensibility-focused software movement, user involvement in Web GIS and VGI has never been greater than today.

For all the benefits that Web GIS, VGI, and NeoGeography have given us, the field still faces some challenges. On the VGI side of things, volunteered information circumvents traditional Old Media quality barriers. Using Wikipedia as an example, it is clear that user-generated content can be incredibly useful but must be approached with a skeptical eye. Both innocent mistakes and outright vandalism occur in VGI, and because this is a spatial context, the added factor of positional accuracy of data can become a serious issue. The idea of a gatekeeper to knowledge also comes into play in regards to GIS and GISci education. NeoGeography practitioners may have little or no background in geography or GIS, and mistakes can unintentionally render data misleading or even dangerous. For example, issues related to coordinate systems and projections can distort spatial data, such as misregistration of aerial imagery in Google Earth, as described by Goodchild (2007). Poorly applied data generalization or classification approaches can lead, intentionally or otherwise, to faulty conclusions about data (Monmonier, 1996). Remember that much like Wikipedia, users often look to Web GIS applications as a source of authority, and errors can quickly propagate thanks to the ease of sharing that the Internet enables.

Additionally, VGI can become embroiled in issues of privacy and power imbalances. Google has a procedure for removing or obscuring personal information in their street view application, but many may not be aware of this ability, or even that their personal information may be publicly available in this format (Google Maps, n.d.). On a broader level, NeoGeography remains largely in the realm of those with access to technology and education. While technology access and VGI participation is often strong, albeit uneven, in developed parts of the world, other regions may be lacking in access, participation, and educational opportunities. This can lead to the misrepresentation and skewed perspectives of events and places provided through VGI. In many ways, NeoGeography has increased the number of voices involved in GIS activities to previously unimaginable levels and helped to level social and political powers, but uneven access to technology and spatial education remains a serious concern of GIScience and Web GIS (Elwood, 2006). Given that public participation in GIS can help alter major public policy decisions, these issues of access and education are quite concerning.

4.9 Conclusions

Over the past half-century, GISs and RS have completely changed the way we track, manage, and make decisions about spatial information. These technologies assist us in countless ways, yet knowledge of them remains somewhat limited amongst the general public. In part, this is due to the complexity of the systems and their operation. Library patrons have often heard of these technologies and are interested, but may not have much understanding of what terms like GIS actually describe. It is imperative that librarians be familiar with geospatial technologies in order to assist clients in finding resources and creating instructional services for online mapping programs. While learning to use GIS may be daunting, the broad overview of geospatial technology described in this chapter should help provide a context for their uses.