Abstract
Urban development is a principal driver of landscape change affecting the integrity of ecological systems and the capacity of the landscape to support species. We developed an urban growth model (SPRAWL), evaluated it with hindcasting, and used it to simulate urban growth across the northeastern United States between 2010 and 2080 under four alternative scenarios. In the model, urban growth is constrained by demand for new development for each time step at the subregional scale. Demand is subsequently allocated to local application panes (5 km on a side within 15 km window) using a unique landscape context matching algorithm, such that the more historical development that occurred in the matched training windows the higher the proportion of future demand assigned to the pane. Lastly, demand in each pane is allocated among development types and then allocated to individual patches based on suitability surfaces unique to that landscape context. SPRAWL has a multi-level, multi-scale structure that captures urban growth drivers operating at multiple scales and, when combined with the unique matching and suitability algorithms, induces non-stationarity in urban growth across time and space. Our evaluation indicated that SPRAWL was highly discriminatory, well-calibrated, and highly predictive of new development, but performed weakly for redevelopment transitions. We evaluated the ecological impacts of four alternative urban growth scenarios varying in total demand for new development and “sprawliness” of new development relative to historical patterns using an ecological integrity index. The results were consistent with expectations and demonstrated the potential of SPRAWL for scenario analysis.
