CaGIS vol. 30, No. 4

Evaluating the Usability of a Tool for Visualizing the Uncertainty of the Future Global Water Balance

Terry A. Slocum, Daniel C. Cliburn, Johannes J. Feddema, James R. Miller

We describe the development of software that is intended to enable decision makers (and their scientific advisors) to visualize uncertainties associated with the future global water balance. This is an important task because the future water balance is a function of numerous factors that are not precisely known, including the historical climatology, the model of potential evapotranspiration, the soil water holding capacity, and the global circulation models (GCMs) used to predict the effect of increased CO2 in the atmosphere. In developing the software, we utilized the principles of usability engineering. In our case, we utilized six steps: prototype development, evaluation by domain experts, software revision, evaluation by usability experts, software revision, and evaluation by decision makers. Although this approach led to an improved piece of software, decision makers should have been involved earlier in the software design process, possibly at step two (instead of the domain experts). Decision makers found the notion of uncertainty discomforting, but their positive comments regarding the software suggest that it could prove beneficial, especially with improvements in spatial and temporal resolution. One interesting characteristic of our approach was the utilization of a wall-size display measuring 25 x 6 feet. The wall-size display engendered great interest, but determining whether it is truly effective will require a study that directly compares it with more traditional approaches.

KEYWORDS: Usability, uncertainty, decision-making, visualization, wall-size display

Representing Datum-level Uncertainty in Historical GIS

Brandon S. Plewe

Geographic information systems have great potential as a tool for studying and teaching historical geography. However, using traditional GIS data models, even spatio-temporal forms, has been difficult due to the prevalence of uncertainty—both ambiguity and fuzziness—in source information concerning space, time, and theme. Explicitly uncertain assertions of a geo-historical datum can be modeled as an Evidentiary Set, a hybrid of a fuzzy set with probability and Dempster-Shafer evidence theory. This set formalism is designed to represent continuous and discrete value ambiguity (e.g., “about 10”), and fuzzy membership (e.g., “somewhat Central European”), including ambiguous membership and other fuzzy-ambiguous combinations. The formal set structure can be stored in GIS by representing continuous variation with a patch model, producing logical models for object-oriented and relational GIS databases. The relational implementation was tested in two GIS databases focused on human historical geography, showing the potential for the model to represent explicit datum-level uncertainty in a wide variety of GIS applications.

Computing Distance to the Nearest Utility: As the Crow Flies vs. As the Gas Flows

Arthur J. Lembo, Jr., Rachel A. Davidson, Thomas D. O’Rourke, and Linda K. Nozick

The purpose of this paper is to describe an automated geographic information system (GIS) process to estimate the length of pipeline required to provide utility service to a potential customer base using the street rights-of way. Using a case study application for a large gas utility company, we compare this method to the traditional method based on straight-line Euclidean, or as-the-crow-flies, distances. The GIS method using the street rights-of-way provides a substantially more accurate estimate of the length of required pipeline than the as-the-crow-flies method does. The improvement is particularly significant when the potential customers are located a substantial distance from existing pipelines. With improved estimates of required pipeline length, utilities can better pre-qualify potential customers for marketing opportunities.

Scale-independent Land-use Allocation Modeling in Raster GIS

Robert G. Cromley and Dean M. Hanink

A common application of raster-based geographic information systems (GIS) is as an aid in multi-criteria, multi-objective land-use decision problems. However, as the cell resolution increases by reducing cell size, the number of rows and columns in the raster representation also increases. The size of raster representations of land-use problems is often a determining factor in the type of methodology used in solving such problems. Previous land-use allocation models integrated with a raster GIS have used either decision heuristics or exact methods based on linear programming models. The former is fairly scale independent but produces only approximate answers, whereas the latter produces optimal solutions but remains more scale dependent. This paper presents a specialized dual simplex method adapted to the generalized assignment problem that can be used to solve large-scale land-use allocation problems. The dual approach only requires that information for one pixel be stored at a time thus allowing the solution of problems based on any size raster database.

KEYWORDS: Raster GIS, land use, linear programming, Von Thunen model.