Hierarchical modeling of urban growth across the conterminous USA: developing meso-scale quantity drivers for the Land Transformation Model

The Land Transformation Model (LTM) is hierarchically coupled with meso-scale drivers to project urban growth across the conterminous USA. Quantity of urban growth at county and place (i.e., city) scales is simulated using population, urban density and nearest neighbor dependent attributes. We compared three meso-scale LTMs to three null models that lack meso-scale drivers. Models were developed using circa 1990-2000 data and validated using change in the 2001 and 2006 National Land Cover Databases (NLCD). LTM and null models were assessed using the mean difference in quantity between simulated and actual growth measured at multiple spatial scales. We found that LTM models performed relatively well at spatial scales as small as 450 m, and that the mean difference between the NLCD and LTM with meso-scale drivers at 900 m was 2-3%, whereas null models produced a mean difference of similar to 5%. Thus, introducing meso-scale modules into large-scale LTM simulations significantly increases model accuracy.