Brains vs. Brawn – Comparative strategies for the calibration of a Cellular Automata – Based Urban Growth Model

The need for good modelling tools to simulate urban growth are a planning tool necessary in today’s age of widespread urban growth and the natural and human disasters that quickly follow. The SLEUTH urban growth model is a CA-based urban change model that simulates urban growth according to a calibrated set of parameters. The following work compares two methods of calibration. The first, “Brute Force” method uses a predetermined order of stepping through the “coefficient space”. The second method, fully introduced here, uses a Genetic Algorithm (GA) to search through the coefficient space in an adaptive manner. This work compares the computational speed and modelled accuracy of the two methods, using the city of Sioux Falls, South Dakota (US) as a testbed. The results show that the GA method of calibration is superior to the Brute Force method, because of improved model fits, as well as superior computational needs. The forecasted model runs, to the year 2015 show different results from the GA and Brute Force methods, possibly due to the data used in model preparation. Improvements on the GA method of calibration are suggested as well as a hybrid approach to SLEUTH calibration.