Modeling exurban development near Washington, DC, USA: comparison of a pattern-based model and a spatially-explicit econometric model

Abstract

The development of private rural lands can significantly fragment landscapes, with potentially negative consequences on ecosystem services. Models of land-use trends beyond the urban fringe are therefore useful for developing policy to manage these environmental effects. However, land-use change models have been primarily applied in urban environments, and it is unclear whether they can adequately predict exurban growth. This study compared the ability of two urban growth models to project exurban development in north-central Virginia and western Maryland over a 24-year period. Pattern-based urban growth models (such as SLEUTH) are widely used, but largely mimic patterns that emerge from historic conditions rather than allowing landowner decision-making to project change. In contrast, spatially-explicit econometric models (such as the complementary log–log hazard assessed in this study) model landowner choices as profit-maximizing behavior subject to market and regulatory constraints. We evaluated the two raster-based models by comparing model predictions to observed exurban conversion at pixel and county scales. The SLEUTH model was more successful at matching the total amount of new growth at the county scale than it was at the pixel scale, suggesting its most appropriate use in exurban areas is as a blunt instrument to forewarn potential coarse-scale losses of natural resources. The econometric model performed significantly better than SLEUTH at both scales, although it was not completely successful in fulfilling its promise of projecting changes that were sensitive to policy. The lack of significance of some policy variables may have resulted from insufficient variation in drivers over our study area or time period, but also suggests that drivers of land use change in exurban environments may differ from those identified for urban areas.