Global and local analysis of fragmentation in a mountain region of Colorado

Although relationships between fragmentation of urban development and other forms of administrative and land cover fragmentation are important, they are poorly understood. This research aimed to better understand these relationships in order to inform land use planning in the Roaring Fork/Colorado River Corridor of Colorado. Change in fragmentation of urban development between 1985 and 1999 was modeled as a function of other forms of administrative and land cover fragmentation using two different regression specifications. While a standard "global" regression provided a good averaged model of change for the entire study area, a geographically weighted regression (GWR) demonstrated how the process changed locally over space. Results of the global regression showed that the intercept was close to zero and therefore the fragmentation of urban development was expected to be close to zero in the absence of other forms of fragmentation. Results of the GWR showed that the relationships between change in fragmentation of urban development and other fragmentation variables (initial edge density of urban development, edge density of public/private interface, farmland density and road density) varied significantly within the study area. By modeling this variation, GWR helped to identify ways to reduce fragmentation of urban development in two different regions of the study area. The analysis suggested that fragmentation of urban development in one area, Aspen-Basalt, was more strongly driven by amenity-related variables, while in the New Castle area fragmentation was more strongly driven by infrastructure-related variables. Ultimately, local analysis may help fine-tune "one-size-fits-all" land use polices for specific regions. (C) 2003 Elsevier B.V. All rights reserved.