Application of the polarimetric matched image filter to the assessment of SAR data from the Mississippi flood region

The optimal use of polarimetric scattering information provided by multiband synthetic aperture radar (SAR) is examined for its utility to improve the assessment of damage caused by the Mississippi flood of 1993. This natural flood disaster submerged or saturated large geographical regions including extensive farmland, and deposited large amounts of sandy sediment. A polarimetric matched image filter technique based on eigenanalysis is applied to enhance image contrast in selected regions for improved detection and classification of flooded regions. A subset of L-Band SAR data is first radiometrically and polarimetrically calibrated, and then coarsely segmented for the extraction of spatial statistics useful for filter application. The polarimetric matched image filter is applied to increase the contrast between dry and flooded land regions that may include stressed embankments (dikes) and levees, while also reducing speckle. The processing procedure is described, and the potential utility of the results is examined. The results are compared to the span image and verified with ground truth information