The network interpolation of population for flow modeling using dasymetric mapping

Abstract

In spatial analysis, population frequently is aggregated into source units having an areal extent. When using such data in a flow model, distances are calculated as an average between each source unit and a set of destinations. In a network, this average distance might be the shortest path between a destination and the centroid of a source unit. However, population is never concentrated at centroids nor is it uniformly distributed within each spatial unit. In urban areas, it is more likely located proximate to the road system that traverses most areal units. This article presents a method for interpolating areally aggregated data to the segments of the road network bounding an areal unit using a dasymetric approach. A case study for Phoenix, Arizona, compares using a network-interpolated population distribution with the area-based approach for the problem of defining service areas for a given set of facilities. A comparison of the road network used, the total demand within each service area, and the total weighted travel distance to facilities shows that the areal-based method underestimates the portion of road network used in travel and misestimates both the expected demand of each service area and the overall travel distance to a facility. © 2013 The Ohio State University.