Exurban development derived from Landsat from 1986 to 2009 surrounding the District of Columbia, USA

People's preference for living in rural areas is converting rural landscapes into low-density residential development (i.e., exurban development). To assess the environmental impacts of exurban development (e.g., habitat fragmentation, threats to wildlife, and increased demand for natural resources), accurate maps of its spatial extent and change over time are needed. Mapping technologies that are based on remote sensing spectral data alone have generally failed to separate exurban development from the surrounding landscape and from other mixed pixels with similar spectra. Although deciduous forests in the eastern United States are thought to have experienced a significant increase in exurbanized area, a rigorous assessment of exurban trends has yet to be undertaken. The purpose of this study was to develop a novel analytic approach to map exurban development and to assess its magnitude and rate in north-central Virginia and western Maryland. We applied spectral mixture analysis to Landsat TM images from 1986 to 2009 at 4 time steps to estimate the fractional cover of vegetation, shade, substrate, and non-photosynthetic vegetation endmembers within each image. Using training data based on aerial photos, we classified the resulting endmember fraction images using a decision tree. Finally, terminal nodes from the decision tree that did not differentiate between exurban and urban areas were analyzed using morphological spatial pattern analysis to assess the shape and form of landscape elements. Scattered, isolated pixels were considered representative of exurban development. Overall classification accuracies ranged from 93 to 98%, an improvement of up to 34% over the decision tree alone. Our mapping approach effectively identified 7.3% of north-central Virginia and western Maryland as exurban development. Exurban development had a substantial expansion in the region, increasing on average 6.1% per year between 1986 and 2009. The mapping of land-cover changes beyond urban fringe provides valuable information for policymakers, planners, and land managers tasked with managing and mitigating the potential adverse consequences of this increasingly common form of landscape change.

► Identified exurban areas using spectral and spatial characteristics.
► Overall classification accuracies ranged from 93 to 98%.
► In 2009, 7.3% of north-central Virginia and western Maryland was exurban.
► Exurban development increased on average 6.1% per year between 1986 and 2009.