The Temporal-spatial Distribution Of Snow Properties In North Xinjiang

There is a vast tract of snow distributed in the surface of earth, Which exhibits considerable variation. Snow and ice melt water raises about one-sixth of the people in the world, and one- fifth of the people in China. In addition, Snow has an important influence on climate change, the surface energy balance, national economy and people’s life and property. Therefore the research of snow is of great importance. North Xinjiang is one of the main snow region in China, Snow and ice melt water is very important for the stock raising, agriculture and road transport. Based on the above understanding, This article makes a thorough inquiry into the temporal-spatial distribution of Snow properties in North Xinjiang.The data used to the research of snow including station data, experimental data and remote sensing data. Station data is consisted of air temperature,soil temperature, snow depth and snow density, We got the experimental data in south of the Altai Mountains and north of the Tianshan Mountains, Which includes the snow depth, snow density, snow water content and snow grain size. The remote sensing data includes SSM/I and AMSR-E, The former is used to evaluate the accuracy of snow depth inversion algorithm, While the latter is used to inverse the snow depth and snow water equivalent of North Xinjiang. This paper tries to explore the temporal-spatial distribution of Snow properties(snow depth,snow days, snow density) recorded by station data in North Xinjiang and describe the snow layers based on the field work, hence count the snow properties of each layer. Remote sensing is a quick and large-area observation method for the research of snow. By combining the long time series of station data and passive microwave remote sensing data, We evaluated the accuracy of snow inversion algorithms, Which can be a reference for the inversion of snow in North Xinjiang. Based on this, We inversed the snow depth and snow water equivalent by AMSR-E in North Xinjiang.The results of the study show that:(1) Most of the stations in North Xinjiang have an increasing tendency in max snow depth and the growth of max snow depth accelerated up to 0.52 cm/y since 1990. There is a good linear relationship between annual snow growth rate and max snow depth. The South of Altai Mountains has the largest snowday, There is a increasing trend of the snowday, Which has a good logarithmic relationship with and max snow depth. Similarly, There is a good logarithmic relationship between annual snowday and elevation. (2)The investigation of snow properties in typical melting period indicate that the snow density and snow grain size increase from ground surface to the snow surface, the ground surface and the snow surface are found to have greater liquid water content. In addition, There is a positive linear relationship between elevation and snow depth(including the mean snow depth, max snow depth, min snow depth),The mean snow depth increases 6.7cm every 100m. (3) the Chang algorithm underestimates snow depth over the former USSR and overestimates snow depth in China and Mongolia, the Che algorithm has relatively better results in China and Mongolia than the Chang algorithm, Accordingly, there is no single algorithm which can be used for global application in snow depth estimation, The traditional method, namely using the difference of brightness temperature of 19GHz and 37GHz to inverse snow depth is not a ideal method. For the six snow types, the tundra snow and prairie snow, which always has less vegetation and relatively flat surface,has relatively better results, For ephemeral snow,marine snow and alpine snow, the results do not look good. Thus the underlying surface and climate can effect the accuracy of snow depth inversion, In addition, We found that the snow depth inversed from passive microwave change from upvaluation to underestimate along with the latitude and snow depth.(4) We constructed models to reverse the snow of North Xinjiang based on the above work.To begin with, We found that the difference of brightness temperature of lOGHz and 23GHz can be used to improve the accuracy of deep snow inversion. Thus this factor can be induced to the snow inversion models, We divided the models into four cases by extracting the pure snow pixels and the classification of elevation, And finally we constructed the snow depth and snow water equivalent inversion models by AMSR-E. The RMSE of snow depth inversion model is lower than 10cm, while lower than 23mm for snow water equivalent.By the combination of station data, the experimental data of field work and the remote sensing data, We discussed the distribution of Snow properties in a large temporal-spatial scale,In the meantime, We figured out the snow properties of each layer in a much smaller scale. After the evaluation of accuracy of snow depth inversion algorithms, We constructed a set of new snow depth and snow water equivalent algorithms, By this way, We can improve the inversion accuracy of snow on the consideration of heterogenicity of snow both on the horizontal direction and vertical direction. This method can be used to study the distribution of Snow properties in a large temporal-spatial scale of North Xinjiang, and extend this study from points to surface.