| Abstract: |
One of the things done well and often by many of today’s geographic information systems is to measure the planar distance to each location on a map from the nearest of a specified set of locations. A related technique that is done less well but used with increasing frequency measures distance not in terms of meters or miles but in terms of more generally defined costs that reflect the time, the effort, the expense, and/or the impact of some sort of travel. This technique, developed by the author in the 1980s and now implemented in a number of geographic information systems including the Spatial Analyst extension of ArcGIS, accumulates increments of travel cost in orthogonal and diagonal directions through a rectilinear grid. In regions of uniform travel cost, that results in zones of travel-cost distance that are octagonal rather than circular. The presentation proposed introduces a new algorithm that corrects this problem in ways that also hold promise for significant improvement in areas ranging from dispersion modeling and shape recognition to the delineation of minimum-cost paths. |