Information transfer between dissimilar areal units: a comparison between areal weighting and the dasymetric method

Abstract

When information is available in areal units not compatible with the study area boundaries, the values must be transferred between these dissimilar polygons. Areal interpolation is the technique used to calculate these values and the most common interpolation method areal weighting, while simple and integrated into most GIS packages, is the least accurate for human population distributions. As shown in previous studies, the more accurate alternative to areal weighting is the dasymetric method. This paper describes a variation of the dasymetric, called Land Use Apportionment, as used within a GIS system. In the example of transferring human population totals from Texas county outlines to Texas major river basins, it is shown that differences between interpolation calculations do arise. Factors accounting for these differences can include surface characteristics of the area, scale between the source and target zones, geometric features of the target area and the quality of information. In this study, the surface characteristics and information quality had the most effect on calculations. Considerations when deciding which method to use are level of acceptable error, surface characteristics, scale, information quality and procedural requirements of the methods.