Home | Business News | Browse by Publication | C | Cartography and Geographic Information Science

Automated cartography in a Bush of Ghosts.

Article Excerpt
A Bush of Ghosts

Research into automated cartography has been going on for about fifty years, and it is interesting to see how the techniques and methodologies have changed during that time. That richness in variety, and the exciting breadth of cognate disciplines drawn into this area of research reflects 1) the challenges of modeling the art and science of cartography, and 2) the changing way in which we use maps. Some of our ideas have foundered but this has always led to new innovations and the lateral application of new methodologies. This talk will review the past as a way of explaining current thinking and argue that cartography has much to do with reasoning about space and that "design is all about making sense of things." Perhaps these should be the goals of automated cartography?

The title of this piece is inspired by the book of Amos Tutuola from 1954, My Life in a Bush of Ghosts. Tutuola writes of an underworld odyssey in which an eight-year-old boy, abandoned during a slave raid, flees into the bush, "a place of ghosts and spirits" (Tutuola 1954). I think automated cartography is in a bush of ghosts--trapped in tradition and in a bit of a wilderness. There are three ideas I want to explore here:

* Firstly, I think we have built our mapping systems on sand--that sand is the paper map.

* Secondly, I do not think we have properly understood what automated cartography is trying to achieve.

* Thirdly there are, however, many exciting research activities that are promoting and making the art and science of automated cartography more relevant.

I want to talk about efforts to automate cartography. Before I start, I want to begin by saying that in discussing ideas of automated cartography and, in particular, generalization, I feel it pertinent to mention where my current research interests lie. The focus of my research is on the derivation of very small-scale mapping automatically from very large-scale mapping (Figure 1).

[FIGURE 1 OMITTED]

The idea is that we record only at the fine detail, and taking advantage of the interdependence of geographic phenomena we somehow automatically derive higher-order phenomena. I am particularly interested in modelling large changes in the level of detail at which we portray geographic information. My interests lie in modelling what the wonderful Jean Claude Muller referred to as "conceptual cusps" in map generalization--points in the representational continuum where the role of the map and the way in which phenomena are represented fundamentally changes (Muller 1991). And I think trying to find answers to this type of problem sheds light on problems of small changes in scale where there are relatively small changes in the level of detail (Figure 2).

[FIGURE 2 OMITTED]

Given the complexity of this task, and the absence of users with cartographic skills, it is assumed that the human should be largely absent from this process. In terms of an automated solution, what is required is an autonomous self-evaluating system able to create maps in digital or analog form that vary widely both in scale and thematic content. This is touted as the "holy grail" of automated cartography.

Failure

I want to try and explain why attempts at automated cartography (as defined by my research goal) have, by and large, failed. I want to use those explanations to reason about what I feel we should be doing in order to move on from where we currently are. For some, failure is an emotive term, but there is no need for any gnashing of teeth. On the contrary, it is often through failure that we learn most. So why do I have this feeling? My first observation is that cartographic methodologies have been poorly incorporated within commercial GIS. It seems to me that even after 40 years of development, most GIS systems deal with the cartographic component as some sort of "end process" via simple data reduction algorithms, colour ramps, symbol defaults, and choropleth maps (choropleth mapping being a fine example of a digital fossil--that is to say, an outdated way of mapping, sadly made permanent in every GIS system on this Earth). The incarceration of these inadequate and anachronistic ways of doing cartography within GIS systems to me signals our failure to convince the community of the importance of cartography and a failure to innovate cartographic science.

I think we have also failed to design interfaces and interaction models that make it easy for users to specify their map requirements; the cartographic process seems absent in most geographic information systems; we seem mesmerised by the linguistic intoxication associated with "intelligent visualisation" and complex interaction models that afford real-time, high dimensional fly-throughs in space and time.

I think the problems have arisen for two main reasons: The first is that cartography is a poorly understood art and science; and when it came to automation, we failed to develop methodologies that captured the process of design (namely, the idea of creating, and evaluating different cartographic solutions). Secondly, we failed to anticipate the disruptive nature of technology and the changing context in the use of maps--maps as interface, maps as product. Let me elaborate a little...

View this article FREE - Now for a Limited Time, try Goliath Business News
Free for 3 Days!