Urban growth forecast using segmented and complete maps with the SLEUTH simulator

Abstract

Commercial, industrial and public administration activities depend on projecting how cities will evolve. One of the main characteristics of a city is its internal complexity, which makes it difficult to make any planning that depends on their understanding. Computational simulation, exemplified by the growth model SLEUTH, is a possible way to help the study of this problem. Its usage, however, depends on multiple data source and several parameters that have an impact on the quality of results. The objective of this dissertation was to perform simulation studies of the city of Ponta Grossa - Brazil, using the SLEUTH model, and analyze its behavior under the use of different parameters and approaches for data input. Experiments were planned according to different partitioning of the data; simulations were performed with each of the scenarios constructed, and the outputs were compared. Till the Final calibration, it was possible to observe that the model adapts to the way the city growths, although the outputs indicated a smaller expansion than expected; but the prediction results were lower than expected. One of the regionalization schemes presented a slightly better performance, but very near to the other approaches used, not justifying the time to spend in the calibration process. The results are analyzed and possible explanations, involving the model and the data, were discussed.