BUILDING A GEOSPATIAL/TEMPORAL DATABASE FOR ASSESSING SEDIMENT EROSION IN THE SUSQUEHANNA WATERSHED: CONTRIBUTIONS OF ASTER, LANDSAT-7 ETM AND ALI IMAGERY

Sediment erosion caused by agricultural and/or urban land use has a significant impact on the health and sustainability of the Chesapeake Bay ecosystem. The Susquehanna River watershed is the second largest contributor of suspended sediment to the Bay, averaging 1.11 million tons per year. Within the Susquehanna River basin, rivers draining the Conestoga River basin show the highest sediment yields (174 to 1050 tons/mi²/yr), possibly reflecting high agricultural land use and/or urban development.

In order to identify the main sources and factors contributing to current sediment erosion in the watershed and to model the processes of sediment mobilization, a GIS database is being compiled at four different scales using data from a variety of sources and formats, including remote sensing imagery. The four scales of interest are, from largest to smallest: the Susquehanna watershed, Lancaster County, Little Conestoga Creek watershed and smaller tributary basin(s).

Multispectral sensors such as the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), Landsat-7 ETM and the Advanced Land Imager (ALI) provide useful raster formatted data for the spatial characterization of agriculture soils and urban land cover. Clay minerals can be used as tracers for identifying stored and transported sediment sources. Calibrated ASTER VNIR-SWIR (0.56 – 2.40 microns) reflectance spectra were used to map soils containing variable amounts chlorite, smectite, kaolinite, illite or muscovite. Field sampling of soils from within the Little Conestoga watershed confirms the presence of smectite in those areas mapped as such using ASTER.

Temporal analysis of changes in Vegetation-Impervious-Soil (VIS) cover is also being done using multispectral classification methods and soil-adjusted normalized difference vegetation indices (NDVI) applied to cloud-free ASTER, ALI and Landsat imagery acquired since 11 May 2000. Of particular interest is a comparison of the utility of these datasets for land cover mapping and change detection. For example, urban land cover classification using the most recent ASTER scene available, 02 April 2003, shows an estimated 6% increase in developed areas throughout the Little Conestoga watershed since completion of the Landsat derived national land cover dataset (NLCD) in 1992.