THE OVERLAID NETWORK ALGORITHMS FOR AREAL INTERPOLATION PROBLEM

In economic, social, and urban studies, areal units under analysis fre quently differ from areal units over which data are compiled Most area -based analyses, hence, face an unavoidable problem of transferring da ta across different zonal systems. This paper introduces a novel appro ach, the overlaid network algorithm, to develop a series of improved m ethods for tackling the population interpolation problem based on curr ent GIS techniques and available digital information. The term network means that the partitioning of population for source zones is carried our over street segments. People are sheltered by houses which are lo cated along the sides of streets or connected by roads. Thus the stree t network provides an important information about the spatial distribu tion of population. The network length method discerns an even populat ion distribution along one-dimensional line. The network hierarchical weighting method observes variations of residential density among diff erent classes of streets. The network house bearing method breaks the assumptions of even population distribution over one and two dimension s and provides an automatic way of enumerating population over space. The application of these techniques to the interpolation of population in Erie County, New York improves the performance of areal interpolat ion significantly when compared with traditional methods.