Editorial

Abstract

Introduction

Software

Software Evaluations

Assessment

Summary

Conclusions

Acknowledgements

References

Case Studies Index

2. Software for the Visualization of Area-based Data

The development of visualization techniques for area-based spatial data has its roots in the dissatisfaction expressed by a number of authors about the poor spatial analytical tools available in Geographic Information Systems (Goodchild 1987, Burrough 1990, Goodchild, Haining and Wise 1992) Most GIS provide analytical capabilities based on the idea of Map Algebra (Tomlin 1990) in which operations on layers of data could be used to identify spatial queries such as the amount of land satisfying a set of criteria, but were unable to provide any statistical operations beyond the calculation of descriptive statistics.

A number of researchers began to explore different means by which extra analytical functionality could be added to such systems, and this work is reviewed in Haining et al (1996). One of the problems however, was that many of the standard techniques of spatial analysis (as reviewed in Haining (1990) for example) are highly numerical and rooted in the statistical tradition of model building and hypothesis testing , and would not necessarily be of interest to many GIS users. Therefore a good deal of work also focused on the potential of exploratory methods, in which the aim is to use a range of largely graphical methods in order to explore properties of data sets and suggest hypotheses, but without the need for formal model building. EDA had been developed in the context of non-spatial data, and using traditional pen and ink graphics, but it was realised that the ideas could be extended to spatial data (thus producing ESDA) and implemented using interactive graphical methods.

One of the key early pieces of software in this field was SPIDER (Haslett et al 1990) which demonstrated how highly visual exploratory methods could produce useful insights into spatial data. SPIDER later became REGARD, and was a forerunner of MANET, one of the packages we review in this report. Table 1 contains a list of all the software packages of which we are aware which provide facilities for ESDA - for a review of the field see Haining et al. (1996)

Four of these software packages (cdv, MANET, SAGE and SpaceStat) have been evaluated. All are available to the social science community to varying degrees. They were chosen because, of the packages which were available within the rather limited time available for this study, they seemed to span the range of relevant software for spatial statistical analysis in terms of statistical capability and visualization capability. The non-inclusion of other packages should not be taken to imply that they are less effective visualization packages.

Cartographic Data Visualiser (cdv) developed by Dykes (1996) from the Tcl and Tk toolkit of Ousterhout (1994) is principally a map visualization toolkit with relatively modest computational and statistical capability. MANET, developed by Unwin and Theus builds on the earlier, innovative ESDA packages of SPIDER and REGARD. MANET is highly visual but contains a number of tools, drawn from the statistics literature, with a central objective of providing the user with a suite of tools for exploring multivariate spatial data and dealing with missing values (Unwin et al 1996). SAGE, developed by Haining, Ma and Wise (1996), adds ESDA capability to the ARC/INFO GIS but also includes spatial data modelling capability so that ESDA can be applied to raw data and model residuals generated within the package. Finally SpaceStat, developed by Anselin, is principally designed for spatial modelling but has been enhanced recently to include a visualization capability through an ArcView extension (Anselin and Bao 1997).

Reference

Description

Anselin, L. and S. Bao. (1997)

ArcView extension for the visualization of results from SpaceStat.

Bailey T.C. and Gatrell A.C. (1995)

INFOMAP. PC-based mapping and Spatial Analysis package distributed on disk with textbook on spatial analysis..

Brunsdon C and M.E.Charlton (1996)

Prototype spatial analysis system written using public domain software system XLisp-Stat.

Cook, D., J.J.Majure, J.Symanzik and N.Cressie (1996)

Link between ArcView GIS and the Xgobi graphics system. Has ability to handle both vector and raster data in graphical windows linked to statistical graphs.

Dykes J. (1996)

cdv.

Haining R.P., Wise S.M. and Ma J.(1996)

SAGE.

Haslett J., Wills G. and Uniwn A.R. (1990)

SPIDER (later called REGARD).

MacDougall E.B. (1992)

Polygon explorer. Mac-based software for map display and analysis.

Unwin A., Hawkins G. Hofman H and Siegl B. (1996)

MANET

Nagel M. (1996)

System built using data analysis language ISP. Includes a range of map types (Choropleth, isoline) and spatial analysis techniques.

Table 1: Software developed for the visualization of area-based spatial data. (Packages in bold are discussed in the report).