Mapping snowmelt sequence at Ivotuk, Alaska, using RADARSAT SAR images acquired with different standard beams

End-of-winter snow depth distribution in tundra, steppe, and prairie can be recovered from a sequence of optical images acquired during the snow melt if an energy balance model is used to calculate the melt rate. However, frequent occurrences of clouds over tundra areas like Alaskan North Slope reduce the probability of obtaining a suitable set of visible and infrared images during the rapid melt, which often takes less than 10 days. Sequence of images acquired by microwave sensors are ideal for catching such rapid changes because microwave sensors are capable of mapping the Earth's surface even under overcast conditions and during periods of dark. A sequence of RADARSAT full-resolution SAR images acquired with different standard beams were collected to maximize the temporal coverage of a study site at Ivotuk, Alaska, during the melt season in May 1999. The wide range of incidence angles (31/spl deg/-49/spl deg/) of the radar beams of the available RADARSAT SAR images makes it necessary to normalize radar backscatter between different beams. The rolling surface of the hilly terrain in the study area requires further corrections of geometric and radiometric distortions caused by slant range mapping and variations of local incidence and azimuth angles. For terrain correction of RADARSAT SAR images, a fine resolution of digital elevation model (DEM) is generated by averaging DEMs derived from interferometric pairs of ERS-1 and ERS-2 tandem SAR images. The resulting sequence of terrain-corrected and incidence angle normalized RADARSAT standard beam SAR images with a spatial resolution of 20 meters reveal backscatter variations caused by increase of snow wetness and exposure of bare ground due to snow melt, thus depicting the details of the snowmelt process in the study area.