Resumen
Urban landscapes are heterogeneous mosaics that develop via significant land-use and land cover (LULC) change. Current LULC models project future landscape patterns, but generally avoid urban landscapes due to heterogeneity. To project LULC change for an urban landscape, we parameterize an established LULC model (Dyna-CLUE) under baseline conditions (continued current trends) for a sub-tropical urban watershed in Tampa, FL. Change was modeled for 2012?2016 with observed data from 1995?2011. An ecosystem services-centric classification was used to define 9 LULC classes. Dyna-CLUE projects change using two modules: non-spatial quantity and spatial reallocation. The data-intensive spatial module requires a binomial logistic regression of socioecological driving factors, maps of restricted areas, and conversion settings, which control the sensitivity of class-to-class conversions. Observed quantity trends showed a decrease in area for agriculture, rangeland and upland forests by 49%, 56% and 27% respectively with a 22% increase in residential and 8% increase in built areas, primarily during 1995?2004. The spatial module projected future change to occur mostly in the relatively rural northeastern section of the watershed. Receiver-operating characteristic curves to evaluate driving factors averaged an area of 0.73 across classes. The manipulation of these baseline trends as constrained scenarios will not only enable urban planners to project future patterns under many ecological, economic and sociological conditions, but also examine changes in urban ecosystem services.