COASTAL LANDCOVER CLASSIFICATION USING NASA'S AIRBORNE TERRESTRIAL APPLICATIONS SENSOR (ATLAS) DATA

摘要

Impervious surface is a key indicator of the extent of urbanization within a given geographic area. Extensive impervious surface area can reduce quality of nearby waterways by increasing runoff volume, increasing peak flow rates, and reducing rainwater infiltration and pollutant filtering by subsurface flow (Corbett et al., 1997). Thus, relatively easily attained estimates of impervious surface area would allow both a measure of urbanization and risk to receiving waters. In comparison, vegetated surface area slows runoff and traps pollutants better than open land. Estimates of percent impervious surface, vegetated, and open land, along with morphology of urban land use, where shape and density are key elements, can be measured and analyzed with the use of Remote Sensing and Geographic Information Systems (GIS). NASA's Airborne Terrestrial Applications Sensor (ATLAS) data were used to classify areas of Murrells Inlet, South Carolina into three land-cover classes: impervious surfaces, open land, and vegetation. The spectral range of ATLAS is 0.45 - 12.2 um and is displayed in 14 channels with a 3 meter (m) Ground Spatial Resolution (GSR). The ATLAS data were rectified, transformed using ENVI's Principal Components Analysis (PCA), classified using a parallelepiped classifier from ERDAS, Inc. Image Analysis extension for Arcview, and converted to vector format for use with the GIS. The accuracy of the classification was estimated using a hybrid approach of ground-truthing and a visual examination of the National Aerial Photography Program's (NAPP) Color Infrared (CIR) photography with a GSR of 1 m. Remotely sensed impervious, vegetated, and open surfaces are being used in empirical relationships to predict risks to and impacts upon the receiving estuary.