摘要：The need for models that forecast land use change spans many disciplines and encompasses many approaches. Pattern-based models were the first in which projections of change at specific locations in actual landscapes could be predicted. In contrast, recent economic models have modeled the underlying behavioral process that produces land use change. This paper combines attributes from each approach into a hybrid model using a multiple discrete continuous extreme value formulation that allows for multiple conversion types, while also estimating the intensity of each type of conversion, which is an important but often overlooked dimension. We demonstrate the simulation routine, which successfully predicts a majority of growth by type, time, and location at a disaggregated scale, for a three-county region in Maryland.
