Study On Spatio-temporal Changes Of Snow Cover By Remote Sensing In Jilin Province

Snow is one of the essential land covers, which plays an important role in the global climate system and is also the main source of available water source. Especially in north Hemisphere 8 percent of the total terrain surface is covered by snow. Snow cover is one of the most significant factors to indicate the seasonal change and to influence the interaction between the global atmosphere and the ground surface. Thus, conbine with the climate model to the influence of the snowmelt flooding can be predicted, for the irrigatation in agriculture.In China, snow cover is distributed in Xinjiang province, Tibetan Plateau and the northeast region in the winter. Jilin province is located in northeast China, which south is locates Liaoning province, the west of Inner Mongolia Autonomous Region and Heilongjiang Province in the north. The elevation of the terrain is apparently higher in the southeast than that in the northwest. It is temperate continental monsoon climate and has the four distinct seasons. Especially in winter, it is cold and long, the average temperature is about-11℃.Furthermore, Jilin province is one of the main grain production area in our country and is vital important to the grain supply in our country. Thus, it is chosen as study area in our study. Several factors that can affect the snow cover extraction is firstly studied, including cloud cover in the remote sensing image, the terrain and the vegetation. Finally, the spatial and temporal changes of the snow cover in the year from 2000 to 2010 and its relationship with the temperature, precipitation and the terrain are analyzed in the paper. The results are as follows:1. Elimination of cloud cover in the remote sensing image. MODIS (Moderate Resolution Imaging Spectroradiometer) reflection data and the MOD35_L2 cloud mask data in Jilin Province on Nov.23, Nov.24 and Nov.25 in 2008 are selected to remove the cloud cover on Nov.24 data. Then the snow cover is extracted from cloud-reduced data by the Normalized Difference Snow Index (NDSI) algorithm. Finally, EOS Aqua (AMSR-E) snow water equivalent data from Advanced Microwave Scanning Radiometer (AMSR) is employed to verify the snow cover undercloud. The results indicate that snow cover fraction of cloud -reduced data is much closer to the AMSR-E snow cover fraction.2. The terrain effect is corrected by using the cosine topography correction method. Changbai mountain is taken as a topography test area due to its complex terrain and high altitude. After the adjusted snow cover index is lower than its original values on the sunny slope, the snow cover index is reduced after adjustment but the change of shady slope is not remarkable. The sensitive analysis indicates the different solar zenith angle and the slope has very tremendous influence to the extraction of snow. The aspect is not sensitive to the snow cover fraction but slop shows a good positive correlation. The comparation between snow cover derived from before and after correction shows the good result.3. The snow cover is derivation at a forest area.Based on GORT model asimulated hybrid relfection of plant canopy and snow. The result are as follows:The simulation results show an obvious hot spot and the trend is fit the measured curve. when the canopy has snow, the underlying surface is all snow cover and the underliying suface is snow free has the same sensitive and trend to the NDSI,but when the canopy has no snow, the underliying sucface is all snow cover appeared different character. The snow cover fraction isn't increasing when the radius of plant canopy increase until the radius reach 1.5.The LAI and the snow cover fraction is negative related, but the solar zenith angle is opposite.4. In order to determine the threshold of the NDSI, under different parameters of the model in deferent situations extraction algorithm of the snow cover are simulated. Then the threshold of the snow cover can be verified through the analysis of the scatter map of the parameters. The algorithm separates the region to forest area and terrain area by NDVI and slop parameter. The new method not only derived more snow pixels, but also reduced the confused pixels of snow and forest. 5. Spatial-temporal changes of the snow cover in Jilin province are generated. The data used is from the day October 1st,2000 to April 30th,2010. The parameters include the average snow cover duration time (SCD), snow cover onset day (SCOD), snow cover melting day (SCMD),snow cover index(SCI) and snow cover ratio(SCR). The SCD presents the same obvious changes with altitude.The average SCOD in Jilin province is at mid-November, but it will be in the beginning of November in eastern mountain area. But the SCMD is in the early February, and the mountain area will be in the late February or later.The duration between 2006 to 2007 has the largest SCI, but the year 2001 to 2002 has minmum value.The largest SCR is appereant in 2000 to 2001,2001 to 2002 has the smallest SCR in the ten years.6. Relationships between the snow cover indexes and the terrain,the climate are analysised. With increasing of the elevation, the snow cover duration becomes longer and longer. The temperature also has an obvious impact on the snow cover duration. The snow cover duration will be long if the temperature is low in the year. While the snow cover duration will be short if the annual mean temperature is relatively high. The same correlation occurs to the relationship between the precipitation and the snow cover duration. The research on the influence of temperature and precipitation on the SCMD indicates that the temperature has apparently influence on the SCMD, while the influence from the precipitation is irregular. SCI and SCR have positive correlation to temperature, but the precipitation is only relative to SCI. The croplands, grass, artificial land, bare land and forest are use for this study. The SCD is long in the bare land and forest, the cropland is the lowest. The SCOD is earlier in bare land and forest, but SCMD in these two types are late. The SCOD of cropland is late, and SCMD is earlier. The result indicted the snow cover distribution is not only related to altitude and climate but also associated with the human activities.