Snow depth and snow water equivalent estimation from AMSR-E data based on a priori snow characteristics in Xinjiang, China

Snow depth and snow water equivalent estimation from AMSR-E data based on a priori snow characteristics in Xinjiang, China

Remote Sensing of Environment

Static snow depth retrieval algorithms tend to underestimate the snow depth at the beginning of the snow season and overestimate the snow depth at the end of the snow season because the snow characteristics vary with the age of snow cover. A novel snow depth/water equivalent (SWE) data retrieval algorithm from passive microwave brightness temperature is proposed based on a priori snow characteristics, including the grain size, density and temperature of the layered snowpack. The layering scheme was established based on the brightness temperature difference (TBD) at two different frequencies, which indicates volume scattering, whereas the snow grain size and density of each layer were parameterized according to the age of the snow cover, and the snow temperature and temperature at the snow/soil interface were determined by the air temperature and snow depth. Furthermore, the microwave emission model of layered snowpacks (MEMLS) was used to simulate the brightness temperature at 10 GHz, 18 GHz and 36 GHz based on the a priori snow characteristics including snow grain size, density, depth and snow layering. Finally, three look-up tables (one layer, two layers and three layers) were generated for each day, which represent the relationship between the brightness temperatures at 10 GHz, 18 GHz and 36 GHz and snow depth. To avoid underestimation caused by the saturation of the microwave signal at 36 GHz, the TBD1 (the difference of brightness temperature at 18 and 36 GHz) was used to estimate the snow depth if TBD1 was less than 40 K, and TBD2 (the difference of the brightness temperature at 10 and 18 GHz) was used if TBD1 was greater than 40 K. The snow depth and SWE determined by this new algorithm were validated by snow measurements at thirteen meteorological stations in Xinjiang, China from 2003 to 2010 and compared with existing SWE products from the National Snow and Ice Data Center (NSIDC), the Environmental and Ecological Science Data Center for West China (WESTDC), the European Space Agency (ESA) and measurements with a snow course. The results showed that the root mean squared error (RMSE) and the bias from this new algorithm were greatly reduced compared to NSIDC, moderately reduced compared to ESA and slightly reduced compared to WESTDC. The understanding of a priori local snow characteristics can improve the accuracy of snow depth and snow water equivalent estimation from passive microwave remote sensing data.