Development of Snow Retrieval Algorithm Using AMSR-E for the BJ Ground-Based Station on Seasonally Frozen Ground at Low Altitude on the Tibetan Plateau

Abstract

The influences of biomass, snow grain size and frozen ground covered with snow are problems that should be solved in the estimation of snow quantity when using a snow retrieval algorithm. This paper focuses on the latter problem in that algorithm: the influence of frozen ground covered with snow. The algorithm is developed for the ground station BJ, located in a region of seasonally frozen ground at low altitude on the Tibetan Plateau. An improved radiative transfer model (Improved RTM) with a frozen soil layer is developed for the station BJ on the seasonally frozen ground at low altitudes on Tibetan Plateau. On the other hand, the detection method of the apparent freezing signal of soil and the signal of apparent snow based on the microwave were developed because two problems, which are the estimation error caused by the incorrect detection of snow in the period without snow cover and the estimation error caused by the apparent snow cover being undetectable, were found out. These errors were discovered upon comparative verification of estimated snow depth, based on the normal snow retrieval algorithm, and in situ (measured) snow depth. An enhanced retrieval algorithm for BJ station is then created, by introducing the improved RTM and the detection method into the normal retrieval algorithm. Output from this improved algorithm for BJ was then compared with in situ snow depth. It was found that the latter depth was much larger than estimated snow depth. The former was too large because of snowdrift; it was therefore corrected to a more realistic depth, using a decrement proportion based on the in situ observation. A comparison of estimated snow depth from the improved snow retrieval algorithm and the more realistic in situ snow depth was then done, thereby evaluating the effectiveness of the algorithm. This verification showed good agreement.