Calibrating a Parallel Geographic Cellular Automata Model

As a consequence of years of study, geographic Cellular Automata (Geo-CA) models have become far more complex than the original basic CA concepts. In many Geo-CA applications, calibration processes are needed to determine the appropriate model parameter values so that CA models can produce more realistic simulation results. For a multi-parameter Geo-CA model, due to the large number of combinations of parameter values and the vast volume of geospatial data, the calibration is usually extremely computationally intensive. This paper investigates the possibility and feasibility of improving the performance of Geo-CA models, especially the calibration processes, by deploying parallel computing technologies. More importantly, parallel computing is likely to allow the removal of the simplifying assumptions during the calibration processes. Thus, the comprehensive (full) calibration processes might produce different best-fit parameter combination(s) other than the one(s) produced by simplified calibration processes, hence alter the final simulation results.