Exploring Geographic Information Systems, 2nd ed.

Citation

Ashford, J. (2002), "Exploring Geographic Information Systems, 2nd ed.", Journal of Documentation, Vol. 58 No. 3, pp. 339-342. https://doi.org/10.1108/jd.2002.58.3.339.9

Geographic information systems (GIS), in the sense of computer‐supported processes and presentations, have been growing in scope and practical utility for 25 years or more. Spatial data for even a flattish area are more complex and more abundant than can be readily represented on 2D map sheets, and much more so for a mountainous terrain or a city and its infrastructure or a water resource survey. Traditional maps as sheets of paper are selective, they depend on inflexible choices of measurement and representation standards, and they are labour intensive and costly to maintain and revise. The attractions of being able to derive application‐specific graphics from data sets representing all knowledge about a space, selecting and setting aside, rescaling and recoding to suit the user, and to remake the presentation as the data change, have been evident since minicomputers and relational databases became “normal” development platforms. Good applications can offer lucid and comprehensible presentations of complex issues in spatial domains and it is sometimes difficult for their users to conceive how they might have done their jobs before GIS was invented.

What was not so obvious in the early days was the scale of effort involved in data collection, validation and conversion, and the intrinsic complexity of many of the operations which, though well known to specialist cartographers and geodesists, proved difficult to embed in reliable and comprehensive software. This integration of existing geodata techniques and the development of new approaches particularly useful for GIS took a long time, and research in the field is still active. Nicholas Chrisman is one of the academic geographers who has been involved throughout much of the formative period – 1982 was a good time to present a PhD thesis on spatial analysis to the University of Bristol. He has been a member of staff of the University of Wisconsin‐Madison, and latterly at the University of Washington.

The second edition has been revised and much extended, taking account of the author’s experience of teaching GIS, and so of finding ways to elucidate those aspects which have at times been seen by students and computing specialists alike as wilfully complex. “Exploratory” in the title seems to mean “avoiding too rigid or impersonal an approach” and does not imply any lack of direction, and the text deals with complex and often non‐intuitive material in a commendably readable style. Most of the examples are North American, although the main case study which recurs throughout the book to illustrate specific points follows a University of Washington project in the La Selva Nature Reserve in Costa Rica.

The author envisages the chapters as concentric rings, from an inner kernel of data acquisition and quality control to an outer halo of “social and cultural context”. After a chatty “this is what GIS feels like” introduction, three chapters – necessarily somewhat dense and formal – deal with the fundamental topics of “Reference systems for measurement”; “Measurement frameworks”; and “Representation”. No quick miracle is available to make this stuff easy, and no naïve user of a GIS software package who has not got a firm grip on these basics is likely to present the truth except by accident. Geodetic and projective systems get clear but fairly brief treatment; this branch has been made much easier by the publication of standards and their incorporation into many GIS software packages. More space is devoted to levels of measurement (nominal, ordinal, interval, ratio and so on) and Chrisman is more explicit and more precise than the usual social science approach. This care continues with the discussion of attributes as controls for spatial measurement. Here are object frameworks, topological relationships, point‐ and area‐based frameworks and diverse examples of their application. “Relationships” include TINs (triangulated irregular networks) and “composite frameworks” covers chloropleths[1]. “Representation” builds on the previous chapters and goes from the primitives of attributes, of space and of time, to vector and raster models and database architectures. Slivers, the spurious offspring of merged data sources[2], are first encountered here, and reappear under post‐processing of new data and under quality control. The section ends with a succinct but practical account of data conversion by digitisation from existing map sources and the necessary clean up and quality control activities which follow.

The middle section of the book deals with the processes taking the data which have been collected and verified against “real world” sources, and go on to build the constructs which will satisfy the goals of the outer layer, the social and cultural context in which the GIS will be applied. The basic transformations on attributes are generally straightforward – grouping, isolation, scaling of data, ranking and tabulation and ratio operations are briefly but comprehensively described, followed by an account of the more difficult topic of aggregation (generalisation) and its consequences illustrated in the working out of partially overlapping classification land‐use schemes. Overlays, which in essence associate information about the same space from different sources, are a key part of GIS analysis and receive appropriately extended treatment. The traps and hazards of this phase (slivers, again, fictitious boundaries and others) are well, though probably not exhaustively, covered. The effect of a successful overlay operation is to make all attribute values related to a location or object available together. The attributes may again be combined to yield further interpretations, selecting dominant values, or combinations of values, or pair‐wise interactions, as well as retaining the original detail. Examples of studies using such processes include land use changes over time in Cwmbran, Wales; selection criteria against radioactive waste disposal sites in Pennsylvania; weighted combinations used in land evaluation and site assessment in Kansas; and contingent weighting (rock type and depth to bedrock) in a study of susceptibility to groundwater contamination in Wisconsin. Distance relationships, considering mainly buffers but with some attention to Voronoi networks, and general surface operations extend point‐based analyses to neighbourhoods and topographies. The combination of neighbourhood attributes is open‐ended and the locus of current research, and the author provides here rather more of an overview, with pointers to the literature. Examples include analysis of views from roadways and minimum cost analysis for location of a water pipeline. A final chapter covers transformations of surfaces, including standard material on interpolation and slope metrics, and a perceptive introduction to the atypical transformations which depend on attributes other than distance, angle or area. Examples are drawn from remote sensing for forest mapping; wetland assessment; and utility planning.

The final section, “The broader arena”, outlines the planning and implementation of GIS projects and considers their place in institutional contexts. There is a useful bibliography and a reference to a dedicated Web site http://www.wiley.com/college/chrisman, where resources to carry forward study and exploration are promised. At the time of this review (December, 2001) it contained notes for teachers, but not much else that is not in the printed edition.

No textbook covering a subject as complex and awkward as quantitative geography is likely to please everyone, and GIS practitioners are likely to have personal preferences and differences with the author. I would have liked more examples from an earth science or engineering aspect (volcanic hazard mapping; aquifer supply and depletion), but the mainly social science approach is appropriate for many students. The bibliography really ought to have included E.R. Tufte (1983), for his classic work on the visual presentation of statistical data. Indeed, Chrisman’s figure 1‐8 showing the use of scaled pie charts dotting a map of California as an example of derived scales is remarkably similar to Tufte’s (1983, p. 178) example from Japan, which he takes apart to show just why not to use this form of presentation. If the online database had had a functional feedback link, I could have made this point to the author more directly.

This is a good book, and should appeal to geographers, geologists, computer programmers who have to deal with cartographers and planners, and information scientists who wish to look beyond texts to information in spaces.

Notes

1. 1.

   Chloropleth framework: a measurement framework whose spatial units (derived from a categorical coverage of named objects) serve as a control for attribute measurements (e.g. census data). Newspapers are fond of them, especially at General Election time.

2. 2.

   Slivers: mall objects created by differences arising from error or from divergent practice, in the representation of boundaries that should in reality have been coincident.