A Study Utilizing Automatically Processed Multispectral Data to Delineate a Flood Plain

LARS/Purdue University West Lafayette, Indiana The use of computer-aided analysis techniques (CAAT) and multispectral scanner data from aircraft and satellite altitudes has shown considerable promise for water resources applications. Two areas of particular interest are: (1) temperature mapping of surface water to study the environmental impact of thermal effluents and (2) snow-cover mapping in mountainous regions to improve water yield predictions. The conventional LARSYS classifier has proven to be inadequate for these purpose~ Therefore, a layered classification approach was utilized to overcome the lirnit~ tions of the conventional LARSYS classifier. If the primary objective is to map the thermal characteristics of water bodies only, the ability to discriminate between water and all other earth surface features becomes important. In some cases however, the radiant temperatures given off by vegetation and soils are the same as those given off by the water bodies. Therefore, such discrimination cannot be accomplished using only a thermal band. A new procedure involving a layered classification was developed to separate water from all other caver types by utilizing a near or middle infrared band, and then to calibrate the thermal data of only those elements classified as water. Thus, the result is a temperature map of water only. In a second water reSOurces study, tne areal extent of snow cover in a mountainous region of Colorado was mapped by using LANDSAT-1 and SKYLAB S-192 multispectral data. The results indicated that it was not possible to obtain a multispectral classification of the snow covered area through the conventional LARSYS classifier because several of the wavelength bands on the LANDSAT and SKYLAS data saturated whenever the scanner was collecting data over the snowpack. The saturated detectors produced a singular covariance matrix which precluded classification. However, 2B-38 a layered classifier approach was developed which could be effectively utilized to obtain a snow cover map. MAPPING WETLANDS FROM THEIR SPECTRAL PROPERTIES RECORDED BY ERTS SYSTEM S. J. Kristof, R. A. Weismiller