Extraction And Temporal Variation Of Snow Information And Influencing Factors During The Spring Flood

Snow widely exist in high altitude mountain and North,South polar regions, is one of the main types of earth surface. It can roughly be divided into seasonal snow and permanent snow, in middle and high latitudes, in addition to the permanent snow, seasonal snow cover distribution more widely. Accordingly, the snow producing the profound influence on the local and global water resources management and utilization, climate change, sustainable development of agriculture and animal husbandry industry, natural disaster prediction and stability of ecology and environment. Xinjiang is located in arid and semi arid region, climate is very dry and have little rain, the seasonal snow melt water is an important way to the many rivers of Xinjiang. At the same time, the snow is also the main cause of flood disaster. Accurate extraction of the snow cover information is an important direction of the study of snow cover in arid areas. Therefore, fast, accurate and convenient to get and distribution of snow cover information for the sustainable development of agriculture and animal husbandry, melting process simulation prediction in Xinjiang has an important significance.Snow melt water in agricultural irrigation in arid and semi-arid region has an important significance, and semi arid region of Xinjiang is located in the arid area, snow resources mainly distributed in three mountains, the seasonal snow pack is very rich. Xinjiang seasonal snow melt water is contributing to the numerous rivers supply. Therefore, Xinjiang’s agricultural and animal husbandry production activities of a high degree of dependence on snow melt water, but snow are the cause of flood and ice disaster. Therefore, it is very important to accurately and comprehensively reflect the distribution of the snow, which is very important for the hydrological forecasting and Simulation of snow disaster in Xinjiang. Remote sensing technology can make up some of the lack of monitoring in remote mountainous areas, remote sensing technology to its macro, comprehensive, rapid, multi temporal advantages become the main means of large-scale monitoring of snow.Remote sensing technology can obtain the surface information in a macroscopic in a rapid and accurate way, and has been widely used in the research of snow cover. At this stage, based on multi-source remote sensing data has developed the normalized difference snow index(NDSI), the algorithm Snowmap, snow cover rate, mixed pixel decomposition method, a series of snow information extraction method and so on. The MODIS and Landsat TM remote sensing data is plot snow information extraction methods are more commonly used type, MODIS using its high time resolution to extract large scale snow cover information.Based on MODIS images as the main data source, feature space as the theoretical basis of construct NDSI-Albedo feature space and NDSI-NDVI feature space in the study area snow area extraction; use NWFE method to extract the image texture characteristics of the study area, select the optimal texture window and step size, the optimal texture parameters are input to support vector machine(SVM) classifier, extraction of snow information. Landsat5 TM image data of NDSI as "true value" and the three kinds of snow information retrieval results for accuracy comparison. The MOD10C2 and AMSR-E snow product as the data source, to Xinjiang from 2002 to 2011 spring freshet period of snow information as temporal and spatial variation analysis, described the Xinjiang snow cover information space variation, and combined with the surface temperature, terrain, altitude, vegetation and other factors to be explained clearly. The results of the study show that:(1) compared to the two eigen space method, combined with texture features of support vector machine(SVM) classification results low precision, two method of feature space of snow information retrieval results have little difference. NDSI-NDVI feature space inversion of the best results.(2) through the analysis of 10 years during the spring freshet of snow information changes, to typical study area of the middle section of the Tianshan Mountains as an example analysis, northeast to the higher surface temperature found that snow melt faster, snow covered less; low temperature of the whole southern region, snow cover has a large range. Due to the terrain of sunny slope, shady received solar radiation of different, shady slope temperature lower than sunny slope temperature, snow melt slow than sunny slope; elevation in the study area undulating region of the larger, accept the solar radiation energy, snow melting speed fast and smooth regions; high vegetation covered area of snow temperature is higher than that of non vegetation area, in a certain extent accelerate or slow down the snow melt.