Issue link: http://read.dmtmag.com/i/29869
You Really Can Compare Apples and Oranges BEYONDMAPPING H ow many times have heard someone say “you can’t compare apples and oranges; they are totally different things”? But in GIS, we see it all the time when a presenter projects two maps on a screen and uses a laser pointer to circle the “obvious” similarities and differences in the map displays. But what if there was a quantitative BY JOSEPH BERRY technique that would objectively compare each map location and report metrics describing the degree of similarity? … For each map location? … For the entire map area? Because maps have been “numbers first, pictures later” for a couple of decades, “ocular subjectivity” should have been replaced by “numeri- cal objectivity” in map comparison a long time ago. A few years back, a couple of “Beyond Mapping” columns described grid-based map-analysis techniques for comparing discrete and continuous maps (see “Statistically Compare Discrete Maps,” GeoWorld, July 2006, and “Statistically Compare Continuous Map Surfaces,” GeoWorld, September 2006). An earlier column described procedures for normalizing mapped data (see “Normalizing Maps for Data Analysis,” GeoWorld, September 2002). Given these conceptual footholds, I bet we can put the old “apples and oranges” quandary to rest. Don’t Believe Your Eyes Consider the maps of elevation and slope shown in Figure 1. I bet your eyes are quickly assessing the color patterns and “seeing” what you believe are strong spatial relationships—dark greens in the northwest and warmer tones in the middle north- east. But how “precise and consistent” can you be in describing the similarity or in delineating the similar areas? And what would you do if you needed to assess a thousand of these patches? In practice, SNV comparison maps can be generated for the same variables at different locations or different variables at the same location. Obviously, elevation (measured in feet) and slope (measured in percent) aren’t the same thing, but they’re sort of related. It wouldn’t make sense to directly compare the map values; they’re apples and oranges after all, so you can’t compare them … right? Joseph Berry is a principal in Berry & Associates, consultants in GIS technology. He can be reached via e-mail at jkberry@du.edu. 10 G E O W O R L D / A P R I L 2 O 1 1 Figure 1. Elevation and slope, like apples and oranges, can’t be directly compared. Mobility/GPS Special Issue