Radar Polarimetry: Analysis Tools and Applications

Abstract

We have developed several techniques to analyze polarimetrie radar data from the NASA/JPL airborne SAR for earth science applications. Specifically, we determine the heterogeneity of scatterers within subregions, optimize the return power from these areas, and identify probable scattering mechanisms for each pixel in a radar image. We then apply these techniques for discrimination and characterization of geologic surfaces and vegetation cover, and find that their utility varies depending on the terrain type. For example, scattering from different types of sedimentary rocks in a semi-arid environment is very homogeneous, with the apparent exception of units with greater than 10-percent vegetation cover. In this particular application, changing the transmit polarization does not appreciably change the output image. However, units with greater than 10-percent vegetation cover are easily identified in classification maps depicting probable scattering mechanisms. In the case of vegetation discrimination, the contrast between vegetated and clear-cut areas changes much more dramatically with polarization, and in some cases clear-cut areas that were not discernable on HH and VV images are enhanced significantly. We also find that the classification technique introduced here is very useful for interpreting scattering from forested areas since clear-cut and tree-covered areas are classified as having different scattering mechanisms. We have also found that polarimetrie signatures of lava flows change with flow age in a way that is consistent with the physical mechanisms of lava flow weathering in arid areas. These processes, such as rubbling and filling, smooth the surfaces of lava flows over time. Polarimetrie radar sensors may therefore provide a quantitative means for relative dating of lava flows in arid areas and for study of the processes that modify their surfaces. We conclude that there are several classes of problems amenable to single-frequency polarimetrie data analysis, including characterization of surface roughness and vegetation structure, and estimation of vegetation density. Polarimetrie radar remote sensing can thus be a useful tool for monitoring a set of earth science parameters. © 1988 IEEE