Wisconsin Geospatial News

Data-driven maps take cartography a step forward

I am sometimes asked for recommendations about what software to use for making publication-quality maps. In particular, people often want to know when to use GIS and when to use vector graphics editing software like Adobe Illustrator. For years it has been common practice to use GIS for data processing and then export the results to a graphics editor for final map production. I sometimes field questions about how to facilitate this export process.

In my view, this process is often unnecessary, and frequently introduces unintended consequences. In many cases maps can be made without exporting to a graphics editor with no serious loss of cartographic fidelity. This includes sophisticated maps suitable for publication in books and atlases. More important, GIS-based data-driven maps have some inherent advantages over “dumb” vector graphics. Specifically, the connection between the map and the underlying intelligent geospatial database allows GIS-based maps to be more easily and efficiently created, updated, and repurposed. When users export their maps out of GIS, this connection is lost, creating a cartographic cul-de-sac; the resulting maps may serve their specific purpose quite well but cannot always be easily adapted for other purposes. For these reasons I often advise against using graphics editing software, and instead recommend workflows that make use of the mapping capabilities within GIS itself.

There is obviously nothing new about the idea of using GIS to make maps. Furthermore, most cartographers know that modern GIS software offers a variety of tools to produce high-quality map output. So it is surprising to see how much resistance there still is to the idea of a mapmaking workflow based entirely on GIS. This distrust probably arose initially due to the limited mapmaking capabilities of early GIS packages, and the poor quality of many early GIS-based maps.

Another factor is the tendency to think of maps as unique one-of-a-kind objects. This tendency is reinforced by cartography’s emphasis on thematic maps, those special-purpose statistical maps focusing on a specific topic and often designed to deliver a particular rhetorical statement. For these maps, graphics editing software is often a logical choice, given the strong emphasis on the design elements of the map.

General-purpose reference maps, found in atlases or produced under national mapping programs, have quite different characteristics. These maps usually have high feature densities, a variety of often competing geospatial themes, and complex selection and symbology rules. As a result, these maps can be costly to produce and can benefit enormously from software tools found in GIS such as automated feature generalization and label placement. Reference maps are often created as a series; there may be multiple maps at the same scale, or maps at different scales, all with essentially the same symbology. Hundreds or even thousands of such maps may be required to fill an atlas or cover an entire country.

For reasons of production efficiency and flexibility (the ability to update and repurpose), these maps can make use of GIS capabilities in ways that single thematic maps simply cannot. A commercial example is the 22nd edition of Goode’s World Atlas. I served as editor of this atlas while at Rand McNally prior to my appointment as State Cartographer. For the 22nd edition the team at Rand McNally created a new reference map series of the world spanning several hundred pages at a variety of scales. All of these maps were created with commercial GIS software (ArcMap) without any use of graphics editing software.

It is mostly production efficiency that is driving GIS use in such cases. Producing an extensive map series would be impractical to say the least if each map had to be individually hand-crafted. Another driver is the ease with which the maps can be updated and repurposed. Map symbology is dependent on underlying geospatial data values. For example, on most reference maps, the size of city symbols and text is proportional to city population or importance. What happens when new population statistics are available and the maps need to be updated? GIS-based mapping makes this process comparatively easy, due to the existence of an underlying intelligent geospatial database coupled with stored rules about how these attributes drive map symbology. A simple database join operation is all it takes to update city populations and regenerate symbols of the appropriate size. The same is true for changes to other attributes like feature names, highway numbers, and so on. And the process can be particularly efficient when all maps in a series, or at least all maps at a particular scale, are linked to a single underlying geospatial database, such that a single database update causes all maps to be refreshed.

Repurposing of maps is also easier, since, by a simple change of symbology rules, a map intended for one purpose can be redesigned for another. This leverages the initial map development effort, increasing return on investment. It also extends the life of the map by adaptively reusing it for a new purpose. This means that the knowledge and effort involved in producing the original map have not gone to waste.

Many of the same arguments also apply to thematic maps, especially if a series of maps is being created with essentially the same specifications, or if there is the intention to update the maps periodically with new data. In these situations, the economies of GIS-based mapmaking come into play.

These ideas imply that what is critical in many situations is the creation of mapmaking capability rather than the production of a single unique map. Mapmaking capability within GIS refers to a framework or template based on an attribute data specification coupled with a set of rules that define a desired cartographic expression of the data. A particular map is just one of many possible representations of the data, and other representations are easily created by modifying how the rules are applied. The emphasis on a single map as the culmination of the mapmaking process is out of synch with this technology, since its power lies in the ability to generate different representations so easily. It bears mentioning that GIS-based approaches are also more amenable to online interactive mapping, which, like GIS-based mapping, generates map symbology based on rules applied to the underlying geospatial data.

So, what software should you be using? There are certainly cases where graphics editing software like Illustrator is the logical choice. But it should not be the automatic choice. We need to get over the idea that GIS is incapable of producing high-quality maps, and we need to acknowledge the advantages of GIS-based mapmaking. Critics will point out that GIS does not have all of the cartographic bells and whistles they are used to, but that’s a problem with a solution, and that solution involves asking more of GIS software, not less. If we want to leverage all of GIS’s advantages and capabilities, we should be influencing the evolution of GIS as a mapmaking tool and actively pushing the technology forward to ensure it meets our mapmaking needs.

Note: Some of these comments appeared earlier in an article entitled “GIS and geoenabled cartography” in Cartography and Geographic Information Science, Vol. 38, 2011.