| Abstract: |
Land-use/cover datasets derived from Landsat satellite imagery have been extensively used to calibrate or parameterize several regional land change models (e.g., SLEUTH, LEAM, and L-THIA) and hydrologic models (HSPF/BASINS, SPARROW, GIS-Hydro). For such models, researchers often rely on publicly available land-use/cover data sources such as the USGS’ National Land Cover Dataset (NLCD) or NOAA’s Coastal Change Analysis Program (CCAP) dataset. In some areas, researchers have developed sub-pixel measures of change in impervious surfaces and annual land-change products such as MacDonald Dettwiler and Associates’ (MDA Federal) Correlated Land Change dataset. Researchers have also used supplementary parcel or cadastral land-use datasets in their models. Interpretations of the rates, patterns, and drivers of urban change based on these datasets can differ significantly and affect both researchers’ interpretations of regional phenomena and the results of land-change and hydrologic models, particularly when interpreting the extent of low-density residential development because the spatial resolution (30m) of Landsat imagery is too coarse to resolve isolated dwellings and the spectral signatures of residential lawns and trees are often confused with agricultural and forest land uses respectively.
This presentation explores potential interpretations of the rates, patterns, and drivers of urbanization in the greater Washington, D.C. - Baltimore region (Landsat path-row 15/33) between 1992 and 2001 using the new Chesapeake Bay Land Cover Data series, University of Maryland’s impervious surface datasets, MDA Federal’s Correlated Land Change product (1982 – 2009), and land use data from the Maryland Department of Planning and local governments. Differing interpretations of urban land use and change derived from these data are illustrated, along with the potential affects on forecasts generated by the SLEUTH urban-growth and HSPF hydrologic models. Land-change and hydrologic modelers relying on Landsat-derived land-use/cover classifications should be aware of these modeling limitations and ramifications.
|
