A modeling approach to path dependent and non-path dependent urban allocation in a rapidly growing region

Abstract

The present paper combines the strengths of path dependent and non-path dependent modeling approaches to detect sustainable urban development profiles in a rapidly expanding region in Iran. In doing so, the Cellular Automata-Markov Chain (CA-MC) model is calibrated and validated using an integrative application of Multi Criteria Evaluation (MCE) method and Multi Layer Perceptron Neural Network (MLPNN) algorithm in addition to applying different Kappa-based metrics. The CA-MC model is then employed to project urban growth trajectories in the targeted research location. By combining the results of different urban allocation approaches, highly suitable zones for future development of the area are detected. These lands were not identifiable if each approach was implemented alone. The potential lands for future urbanization are evaluated and ranked in terms of their connectivity and compactness as well as their effect on other land-use categories such as forest and agriculture. The suggested methodology not only improves the calibration of a spatially-explicit model, but also it provides a wider range of options for policy makers. The findings of this study are useful to municipal decision makers in order to protect against further ecological consequences of ill-planned and uncontrolled urbanization.