| As one of the storage methods of water resources on earth,snow accumulation has an important impact on soil moisture and freshwater distribution.With a latitude difference of about 50° between the north and south ends of China,spanning several temperature zones,the spatial and temporal distribution of snow in the south and north varies significantly,and it is important to grasp the spatial and temporal dynamics of snow accumulation for the development and utilization of snow water resources.In this study,we use the latest observation data of Fengyun-3(FY-3),the snow cover product(SNC)generated by MULSS,and evaluate the accuracy of SNC by batch pre-processing the data with Arc GIS and Python and introducing the snow depth data synthesized by passive microwave remote sensing The snow cover information extracted from the SNC was analyzed using statistical analysis,meteorological analysis,and deep learning methods,and the spatial and temporal dynamics of snow cover in China over a long period of time and at a large scale were studied with the advantage of data in the field of remote sensing,and the link between snow cover and impact factors was explored.The main conclusions of this study are as follows:(1)The accuracy of binary snow map made based on SNC data is evaluated.83.20%accuracy of snow discrimination is achieved by FY-3A and 85.92% by FY-3C,the wind and cloud snow products have higher accuracy in high latitude areas and meet the application requirements.(2)Based on the binary snow map of SNC,the snow cover frequency(SCF)and snow cover rate(SCR)in China were calculated from 2010 to 2019,and the spatial and temporal dynamic characteristics of both were analyzed.the SCF and SCR are characterized by spatial heterogeneity and seasonal differences,with the SCF in northern Xinjiang,northeastern The SCF in northern Xinjiang,northeastern China and central-eastern Inner Mongolia reached the maximum in January,and the Qinghai-Tibet Plateau reached the maximum in April.Northeast SCF showed a decreasing trend,decreasing by 1.0-1.6 d per year.the SCR of Inner Mongolia,Tibet and China as a whole showed a slow and steady decreasing trend,and the SCR of typical snow cover areas in China all went from increasing to decreasing from 2015-2017.Combining SCF and SCR,it is concluded that Xinjiang is the most stable snow cover region in China between 2010 and 2019.(3)Based on the Chinese Snow Depth(SD)long time series dataset,the trend and distribution characteristics of snow depth in China from 2000 to 2019 were analyzed,and the first Chinese daily snow depth prediction model was constructed.SD in the northern northeast,Inner Mongolia,northern Xinjiang and the abdomen of Qinghai-Tibet Plateau has a significant decreasing trend,and SD in the Tarim Basin and Junggar Basin has an increasing trend.The most prominent characteristics of snow distribution in China are "less in the south and more in the north" and "more in the east and more in the west and less in the center".The daily snow depth prediction model based on the Conv LSTM network in China has the best results at the time step of 3.The structural similarity between the model output snow depth distribution and the original map is 0.937 at the highest.(4)The spatial and temporal dynamics of snow cover,temperature and precipitation are discussed on a Chinese scale,and the response of snow cover index to temperature and precipitation changes is explored.The distribution of Start of snow cover(SCS)and Melt of snow cover(SCM)is characterized by spatial heterogeneity,and they keep synchronous changes.There is a significant trend of delay in SCS and SCM in eastern Inner Mongolia,Qinghai-Tibet Plateau and eastern Xinjiang,and a trend of advance in SCM and SCS in North China Plain,western Inner Mongolia and southern Xinjiang.Snow accumulation index is more sensitive to air temperature,and SCF and SCR in most areas have negative correlation with air temperature and positive correlation with precipitation. |
PDF Full Text Request