Snow Depth Retrieval Study Based On The Passive Microwave Remote Sensing In Northern Xinjiang, China

Snow is a special kind of underlaying surface, with high reflectivity, low thermal conductivity and snowmelt hydrological. It has important influence on earth's energy, radiation balance and water cycle. Xinjiang is one of the three snow area in China. To study the distribution characterictics and variation regularity of snow based on the monitoring of snow has an important significance. It promoted the agriculture production and ecological environment protection. Snow is one of the most important elements in the cryosphere, and it is being increasingly widely attention. With the advent of the earth observation system (EOS) age, the study of snow passive microwave remote sensing has already become a hot research field. The more accurate fitting effect and the snow depth retrieval result was acquired on the basis of NASA snow depth retrieval algorithm in Northern XinJiang by attemptting to set up respectively snow depth retrieval model according different height.Combining AMSR_E 19GHz and 37GHz bands horizontal polarization brightness temperature data with the measured 60 meteorological stations in Northern Xinjiang in 2009-2010 snow season (Decimember - February), on the basis of analyzing snow depth retrieval impact factor of study area and the 1865 valid samples after eliminated unreasonable data, the snow depth retrieval model in Northern XinJiang was established based on AMSR_E brightness temperature data by dividing these valid samples into two groups according to altitude and picking the altay and tacheng area for alone regression analysis.The precision of the model was evaluated, and the result shows the negative average error of retrieval snow depth is -4.8 cm when snow depth was between 3 and 10cm, and RMSE is 4.1 cm. When snow depth was between 11 and 30cm, the average error of retrieval snow depth is only -0.2 cm, and its RMSE, positive average error, absolute average error are all less. When snow depth was larger than 30cm, the negative average error of retrieval snow depth is small, but other errors were larger. By comparing the retrieval snow depth with the observed data, the simulation results are substantial agreement with observation of meteorological station in Northern XinJiang. It basically reflected the snow depth distribution in Northern XinJiang. Comparing Chang algorithm to the retrieval model in this paper, it is superior to Chang algorithm, and can reflect snow depth variation characteristics in Northern XinJiang. Meanwhile, the average snow depth distribution map and most heavy snow depth map were inversed by using synthetic method in five snow seasons(2006-2010) in Northern XinJiang. The average depth distribution map and most heavy snow depth map show that snow mainly distributed in northern Altai Mountains and southern tianshan area, and the proportion of altai region is largest. The snow layer is shallower or no snow in the hinterland of Junggar basin, kelamayi regions in middle of Xinjiang. Although the accuracy of the snow depth retrieval model in north of Xinjiang is better than Chang algorithm, the error of local areas is larger and remains to further study. Especially when establishing snow depth retrieval model the filter conditions and methods of AMSR_E bright temperature data were deeply analyzed. More channels brightness data and more accurately threshold value of every indicator were combined, and the influence factors of more multifaceted were considered, to develop new snow depth retrieval algorithm and to provide the scientific basis for husbandry production and disaster prevention and reduction in north of Xinjiang.