Predicting future urban impervious surface distribution using cellular automata and regression analysis

Abstract

Urban impervious surfaces are considered as key indicator of urbanization intensity and environmental quality. Due to their significant impact on surface runoff, flood frequency, and water quality, impervious surfaces have been identified as an important indicator for examining the hydrological impact of urbanization. The amount and distribution of impervious surfaces have been estimated using remote sensing and geographic information system (GIS) techniques. Little research, however, has been conducted to predict future impervious surface distributions. To address this problem, we developed an integrated residential/commercial growth and impervious surface distribution model to predict urban impervious surface distribution. Taking Milwaukee River Basin, Wisconsin as a case study, we simulated future residential and commercial developments using a CA model. Further, we developed a linear regression model to predict impervious surface distributions in residential and commercial land uses. Analysis of results suggests that the proposed model performs significantly better than the traditional approaches.