USING EXPLORATORY FACTOR ANALYSIS TO IMPROVE THE CALIBRATION OF SLEUTH URBAN GROWTH MODELS

Abstract

Simulation models based on cellular automata (CA) are often used to track changes in land cover caused by urban growth. The main program for these models is the SLEUTH Urban Growth Model, which is open source and free. This program uses Monte Carlo (MC) iteration in three phases: testing, calibration, and prediction. Calibration is based on 13 variables. The goal of this study is to test the effectiveness of exploratory factor analysis (EFA), a previously untested method, to improve the calibration of CA-based urban growth models. EFA was evaluated against two commonly-used methods, Lee-Sallee and Optimum SLEUTH Metrics (OSM), using synthetic test data from the Project Gigalopolis website. The results show that EFA had highest regression scores for six of the 13 metrics, performing better than OSM with high scores in four variables and Lee-Sallee with high scores in just three variables. Improving calibration with EFA will result in more accurate models and predictions of urban growth.