Calibrating a CA-Based Urban Growth Model Using Evolutionary Computation

Yun Wang, Department of Urban and Regional Planning, University of Illinois at Urbana-Champaign

A variety of applications that are based on computer models still rely on incomplete data. How to transform unstructured raw data into knowledge and how to extract information from incomplete data remains a challenge for researchers. An urban growth model coupled with Cellular Automata (CA) and built on the theory of Urban Ecology needs to be calibrated to incomplete historical urban extent data by optimizing simulation errors. Traditional optimization approaches cannot satisfactorily calibrate CA-based dynamic model of complex systems, particularly with incomplete data with considerable noise. Advances in computational and analytical efficiencies have directed researchers to Evolutionary Computation (EC) that adopts the idea of evolution. The broad goal of this research is to advance multi-scalar spatial dynamic modeling by calibrating a CA-based urban growth model with the EC technology. If successful, the calibrated urban growth model will be widely applied to analyze the interactions between humans and the environment and hence provide a better decision support system. More specifically, we propose to:
1)advance the understanding of the theoretical issues of EC and evolution theory by applying evolutionary algorithms (EAs) to a real world simulation model built on the theory of Urban Ecology;
2)uncover the power of EC, compared to the other calibration methods, in optimizing the behavior of a complex system simulation model with incomplete and noisy data; and
3)assist urban planning for sustainable development with a better calibrated model and enhance educational perception of EC in nonlinear dynamic modeling.