About six months have passed since I first started using a geographic information system (GIS) and now I can't remember how I survived without it. If you haven't discovered this relatively new and amazingly powerful species of PC software, allow me to convert you. I've also used a product called MapInfo, which is cheaper but less powerful than Atlas* GIS. PC Magazine just gave a thorough treatment to GIS packages in the July issue; you should turn there for a detailed comparison of prices and features.
I'm a big fan of a package called Atlas*GIS, which I got last winter in preparation for this spring's tidal wave of census data. Census counts are by definition geographic, and unless you're very quick with crayons you'll need a program to map the data. For this article, I'll limit myself to describing some uses of a GIS in the newsroom. Our first census story, for example, looked at population changes in Missouri from 1980 to 1990.
We got the raw data in the form of an alphabetical listing of the state's 115 counties. It was hard to discern any trends from looking at the list; some counties showed dramatic population gains while others showed equally dramatic losses. But as soon as I brought the data into Atlas* GIS, stories started jumping off the computer screen. By having Atlas color each county to reflect the degree and direction of population change, we could see that northern Missouri - farm country - was dying while southwestern Missouri - lakeside resort country - was thriving.
This is called a thematic map, in which a region's color indicates a specific range of data. The story was equally easy to ascertain when we turned our attention to St. Louis, making thematic maps of racial population changes in the area's 300 census tracts. Black families who had the resources were abandoning the city's decaying North Side in droves, moving to surrounding counties and to the city's South Side.
At the same time, white families from the South Side were moving farther south into solidly white areas of the city and nearby counties. Because I was using the Census Bureau's complete digital map of the St. Louis area (called a TIGER file), I was able to zoom in on specific tracts and print out a street map showing the tract boundaries. This made it easy for my colleague Bill Smith to hop in his car and get interviews that would illustrate the trends we'd identified.
In the midst of all these census stories, we also found Atlas* GIS to be indispensable for an analysis of real estate in St. Louis. We obtained a copy of the assessor's records for all 140,000 parcels in the city and used Atlas to 'geocode' the data, assigning latitude and longitude coordinates to each record based on its street address. This allowed us to generate 'pin maps' showing thousands of abandoned buildings, vacant lots and tax-abated properties.
By overlaying a digital map of the city's 28 political wards, we were able to tally all these items by ward. With a great deal of precision, we identified vast areas of the city where the tax base had all but disappeared - the same areas that had been neglected by the mayor's redevelopment initiatives of the 1980s. Atlas* GIS has also proved its worth for smaller projects. For example, the Post-Dispatch has been publishing an intermittent series profiling city neighborhoods.
To get a demographic profile of a neighborhood, I first need to spend about 20 minutes building the neighborhood outline from street segments on the TIGER digital map. From there, it's a simple matter to have Atlas*GIS select the census blocks inside the neighborhood and aggregate the data. These are only a few of the uses we've found for our GIS. We have several other projects in the pipeline, some of them quite ambitious, that also will depend on geographic analysis.
I promise that if you can just get your hands on a good GIS, you'll never want to let go. - George Landau has been doing computer-assisted reporting for the St. Louis Post-Dispatch since August, 1989.