Improvement Of Passive Microwave Snow Algorithm Over Qinghai-Tibet Plateau

This paper describes the significance of the study that the Qinghai-Tibet Plateau wascalled as the "roof of the world", its complex terrain and the special climate affect thedistribution and characteristic of the snow, which made it more difficult on snow research.At the same time, snow over the Qinghai-Tibet Plateau has a profound impact on regionaland global climate change and energy balance. Therefore, it has a very high scientific valueto obtain the snow information over the Qinghai-Tibet Plateau. The paper also summarizespassive microwave snow products, introduces snow passive microwave remote sensingtheory. Furthermore, AMSR-E L2A brightness temperature data, meteorological stationdata and ancillary data are used to improve snow depth and snow water equivalentalgorithm over Qinghai-Tibet Plateau. This paper focuses on validation and algorithmscomparative study for microwave remote sensing of snow depth over the Qinghai-TibetPlateau, time series analysis of snow algorithm and brightness temperature gradient basedon different underlying surface, snow water equivalent inversion algorithm improvementand accuracy evaluation over the Qinghai-Tibet Plateau.In this study, research work carried out as follows:1. Validation and algorithms comparative study for microwave remote sensing ofsnow depth over the Qinghai-Tibet Plateau. In this study, we have selected five differentsnow algorithms, making use of AMSR-E brightness temperature data and meteorologicalstation data during2009/2010winter, to validate the accuracy of snow depth algorithmsover the Qinghai-Tibet Plateau, and then compared accuracy of five snow inversionalgorithms. The results show that, RMSE of five snow depth algorithms are8.10cm~20.35cm,5.59cm~15.74cm,8.58cm~24.63cm,3.65cm~8.52cm,4.38cm~11.80cmrespectively. According to this study, the existing passive microwave snow depthalgorithms overestimated snow depth over Qinghai-Tibet Plateau, their regional accuracyand applicability are limited.2. Single point time series analysis of snow algorithm and brightness temperaturegradient. Time series analysis of snow algorithm and brightness temperature gradientcarried out based on different underlying surface. The results of spatial-temporal analysis show that, in the forest area (such as Nyalam), five SWE algorithms underestimated SWE.In other regions, the Improve Tibet Plateau algorithm has underestimated SWE, and otherfour SWE algorithms overestimated SWE. Point was selected from each underlying surface,extracting and analyzing brightness temperature gradients information and characteristic.Because of the t1019v/h without scattering frequency, it fluctuated around0k and had noobvious trendency. High frequency brightness temperature is more sensitive, due to thehigh frequency are susceptible to the influence of atmospheric absorption, if improperhandling with high frequency, it will affect the precision of inversion algorithm. Wholewave form and trendency of t1037v/h and t1937v/h is similar, brightness temperaturegradients were greater than0k which explain the existence of scatterer.3. Improvement of passive microwave SWE algorithm over the Qinghai-Tibet plateau.Using AMSR-E brightness temperature data and meteorological station data during2009/2010winter. Analyzing relationship between measured snow depth and t1937v andt1037v/h for selecting suitble brightness temperature gradient based on diffirent underlyingsurface. After extractes six dynamic coefficients, analyzed the relationship betweendynamic coefficients and snow depth to selected appropriate dynamic coefficient. Byregression analysis, obtained improved separate snow depth algorithms for forest, shrub,grass and barren. The research has established snow distinguishing process, combined withsnow density data and land cover reclassified data to get snow water equivalent algorithmfor each underlying surface. Then, making use of land coverage data, a final SWEalgorithm represents the area-weighted based on the proportional land cover within eachpixel over the Qinghai-Tibet Plateau. When compared with NASA AMSR-E SWE prodectsby using ground measured data, the results show that, accuracy qualification rate rangesfrom46.38%to75.00%for snow water equivalent obtained in this study, while it is5%~31.51%for NASA AMSR-E SWE prodects, which indicated that improved SWEalgorithm in this study is superior to AMSR-E SWE algorithm.