Modeling urban dynamics with artificial neural networks and GIS

The modeling of dynamic urban systems has been of interest to spatial analysts for the better part of the past four decades, and the development of geographic information systems (GIS) has sparked continued interest in spatial process modeling. Recent research in a number of far reaching disciplines has shown artificial neural networks (ANN) to be powerful tools for modeling many dynamic systems (Vemuri and Rogers, 1994). This research investigated the possibilities for ANN's as spatial analytic tools. To this end, an artificial neural network was linked with a GIS for the purpose of modeling urban growth in sub-regions of a metropolitan area. The validity of the ANN model was tested against a linear regression model. The results of this research support the hypothesis that ANN are in fact useful spatial analytic tools and can be used to accurately model dynamic urban systems