| This thesis focuses on developing algorithms for radar-based natural disaster response. We demonstrate that fully-polarimetric Synthetic Aperture Radar (SAR) can be used to assess natural disasters involving terrain resurfacing such as landslides, volcanic eruptions and fires. These hazards often temporarily remove a natural vegetation cover and, in doing so, modify the physical properties of the land surface. This land-cover disturbance causes a detectable change in dominant microwave scattering mechanism for the areas affected. SAR has operational advantages over optical sensors for rapid disaster assessment because of its day/night acquisition capability; its ability to "see through" smoke, clouds and dust; and its side-looking viewing geometry, which is an advantage whenever data collection directly above the site would prove dangerous. To assess the usefulness of fully-polarimetric SAR, we apply a uniform approach to map (1) landslides resulting from the 1999 Mw 7.6 Chi-Chi earthquake in Taiwan, (2) volcanic flows from the major 1996 eruption of Manam Volcano in Papua New Guinea, and (3) the extent of damage from the summer 2002 Rodeo-Chediski wildfire in Arizona, USA. We then reexamine the data from Manam Island and Taiwan to determine the effects of a polarization preserving speckle reduction filter. Our results demonstrate the potential utility of fully polarimetric SAR for hazard mapping and disaster response. |
