Tuesday, 24 January 2012
Inundation Extent Mapping with Soil Moisture Active-Passive (SMAP) Mission L-Band SAR Data
Hall E (New Orleans Convention Center )
Frequent, spatially precise inundation maps are needed for a diverse set of applications, including flood disaster response, seasonal inundation monitoring, and mapping long-term changes in wetlands extent. NASA's planned Soil Moisture Active-Passive (SMAP) mission will combine 1- to 3-km resolution synthetic aperture radar (SAR), 40-km-resolution L-band radiometry, and 3-day revisit period to make a novel dataset that can provide frequent inundation maps potentially superior to alternative methods. This potential has been investigated using a combination of SMAP-like datasets derived from NASA's Advance Microwave Scanning Radiometer-EOS (AMSR-E) and JAXA's Phased Array L-Band Synthetic Aperture Radar (PALSAR) and a multi-scale Bayesian inundation extent mapping framework. The framework uses flood plain topography, hydrology, land cover, historical flood extents, and other factors to make relative flood potential index (RFPI) maps at the resolution of digital elevation models (e.g., 30-100 m). A downscaling and data-merging algorithm uses the RFPI maps to map flood extent from lower-resolution areal flooded fraction (FF) or flood detection data retrieved from near-real time remote sensing observations. The algorithm takes into account retrieval uncertainties, sensor footprint sampling, and prior retrievals to produce a flood extent estimate and an associated error model. The role of SMAP L-band SAR data in this framework has been investigated from two perspectives. First, PALSAR scenes from four test regions have been used to simulate SMAP observations at 1- to 10-km spatial resolution in order to understand the dependence of L-band SAR inundation mapping errors on biome type and the trade-offs between resolution and error in the retrieval process. Second, twice-daily AMSR-E brightness temperatures have been used to exercise the inundation mapping framework with frequent, lower-resolution (~15 km) FF observations. Flood maps derived from Landsat (30-m resolution) and Moderate-resolution Imaging Spectroradiometer (MODIS, 500-m) scenes were used for flood map validation but may in the future be incorporated into the flood mapping process.