One area of data analysis that's gotten a lot of attention is Geographic Information Systems (GIS). GIS is a system that is designed to store, manage, manipulate, and analyze geographic data. As such, GIS sits at the intersection of cartography, computers, statistics, and information science.
GIS is applied to fields as diverse as military planning, epidemiology, architecture, urban planning, archaeology, and many other fields. Basically, any domain or problem that involves location or topology can use GIS techniques or methods.
As you can imagine from this very brief description, we won't even scratch the surface of GIS in this chapter. However, we'll apply it to a small problem to see how it can help us understand the way climate change affects the continental United States in a better manner.
While the preceding description is accurate, it doesn't really help us much. As befits a field concerned with the lay of the land, GIS really begins in the field. Data is gathered using aerial and satellite photography, and it is also gathered from people on the ground using GPS, laser range finders, and surveying tools. GIS can also make use of existing maps, especially for historical research and to compare time periods. For example, this may involve studying how a city has evolved over time or national boundaries have changed. A lot of time and energy in GIS goes into gathering this data and entering it into the computer.
Once the data is in the computer, GIS can perform a wide range and variety of analyses on the data, depending on the questions being asked and the task at hand. For example, the following are some of the many things you can do with GIS:
- View-shed analysis: This attempts to answer the question, "What can someone standing right here at this elevation (and perhaps at a second story window) see?". This takes into account the elevation and slope of the terrain around the viewer.
- Topological modeling: This combines the GIS data with other data in the data mining and modeling to add a geospatial component to more mainstream data mining and modeling. This allows the models to account for the geographical proximity.
- Hydrological modeling: This models the way in which water interacts with the environment through rainfall, watershed, runoff, and catchment.
- Geocoding: This involves associating human-readable addresses with their geospatial coordinates. When you click on a Google Map or Bing Map and get the business or address of a location, it's because it's been geocoded for the coordinates you tapped on.
The primary tool for most GIS specialists is ArcGIS by ESRI (http://www.esri.com/). This is a powerful, full-featured GIS workbench. It interoperates with most data sources and performs most of the analyses. It also has an API for Python and APIs in Java and .NET to interact with ArcGIS servers. We'll use ArcGIS at the end of this chapter to generate the visualization.
However, there are other options as well. Most databases have some GIS capabilities, and Quantum GIS (http://www.qgis.org/) is an open source alternative to ArcGIS. It isn't as polished or as fully featured, but it's still powerful in its own right and is freely available. GeoServer (http://geoserver.org/) is an enterprise-level server and management system for the GIS data. There are also libraries in a number of programming languages; Geospatial Data Abstraction Layer, also known as GDAL, (http://www.gdal.org/) deserves special mention here, both in its own right and because it serves as the foundation for libraries in a number of other programming languages. One of the libraries for Java is GeoTools (http://www.geotools.org/), and part of it calls GDAL under the table.