MONTHLY SURFACE WATER MAPPING USING MULTIPLE INCIDENCE ANGLES AND MULTIPLE TEMPORAL ENVISAT ASAR GLOBAL MONITORING MODE DATA

摘要

Synthetic Aperture Radar (SAR) is a useful and efficient technology for monitoring the water extent. However, the return signal of SAR is most likely affected by the wind, vegetation, and dry sand distributed around the water, which will reduce the separability of water and land using SAR. In order to overcome the mentioned problems, two method with the ratio of the standard deviations of backscatter (σ~0) and local incidence angle (α) over the window sample (SDR, SDR=sd(σ~0) / sd(α) ) and, the slope of a linear model fitting backscatter against local incidence angle (S,σ~0=S×α+b), are primarily calculated using multiple incident angles and multiple temporal ENVISAT ASAR Global Monitoring Mode data, and are subsequently to obtain monthly water extent through the year 2011. Compared with MODIS derived reference maps, Slope S provides evident advantage to segregate land from water with an average pixel accuracy of 57.1%, while SDR merely has an average pixel accuracy of 38.35%.