Abstract
Ancillary data were compared with ground-truth data for use in separating saline and fresh wetland-landcover classes exhibiting similar spectral signatures. Two ancillary datasets, United States Fish and Wildlife Service (USFWS) National Wetland Inventory (NWI) and United States Geological Service (USGS) 5-foot contour, were evaluated for integration within the National Oceanic and Atmospheric Administration (NOAA) Coastal Change Analysis Program (CCAP) classification model. Comparison through 2-way matrix analysis showed that the USGS dataset was the superior dataset for separating saline and fresh wetland-landcover classes and was subsequently integrated into the classification model. The resulting wetland-landcover classification using Landsat Thematic Mapper (TM) imagery of West Galveston Bay, Texas was validated using 407 ground-truth locations. It was concluded, that integration of ancillary data within the NOAA C-CAP classification model can produce map classification accuracies in excess of 85%, thereby meeting the NOAA C-CAP objective.
Hinson, James Mithland (1993). Evaluation and integration of ancillary datasets for coastal wetland landcover classification using Landsat TM Imagery. Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -1993 -THESIS -H665.