Snow Mapping And Spatial-temporal Changes With Multi-source Remote Sensing Data In China

Snow-covered area(SCA) is widely distributed in China. It is very important to obtaine the accurate SCA and snow depth information for further understanding the climate change, water cycle, and ecological system in our country. Accurate monitoring the snow cover also can provide contribute information for the development of animal husbandry, and the snow disaster prevention prediction. MODIS data with its high spatial and temporal resolution, is widely used in the fields of ecological, atmospheric and hydrological fields and other. Owing to the strong interference of cloud, the optical sensor cannot be directly used to obtain the SCA for effective statistics. Passive microwave can penetrate clouds, and it is not affected by the weather condition. However, the coarse resolution of passive microwave products greatly limits the accuracy of regional snow monitoring. Therefore, cloud removal and downscaling are major technical problems for snow monitoring using optical and passive microwave products respectively.Combining different cloud removal methods, this study developed a daily cloud-free fractional snow cover(FSC) product and 500 m resolution snow depth product(SNDsp) by combining optical remote sensing data and passive microwave data. The accuracy of the two products was also tested. On this basis, using MODIS cloud-free binary snow product and SNDsp by downscaling methodology, the spatial and temporal change and tendency of snow were analyzed during December 2000 to November 2014 in China. The results showed that:(1) Daily cloud-free fractional snow cover product developed in this study completely removes the interference of clouds, thereby improving the accuracy of snow area monitoring. The study produced a high-resolution fractional snow cover map using a Landsat 8 OLI image as a true value to test the accuracy of the cloud-free fractional snow cover product. The mean absolute error of our product is 0.19, and its root mean square error is 0.27. The error analysis shows that the land cover type and terrain factors are the main factors that limit the accuracy of the daily cloud-free fractional snow cover product which developed in this study.(2) The produced SNDsp product can significantly eliminate 10 × 10 km pixel block structure of AMSR2 standard snow depth product. The snow depth data observed from 21 climate stations are used to validate the new products. The results showed that the root mean square error was decreased about 1.78 cm and 2.68 cm after downscaling in the descending and ascending orbit of AMSR2 snow depth product, respectively. And the relationship also improved between observations and downscaling products in descending orbit of AMSR2. The AMSR2 standard snow depth products show a minimum error when snow depth was between 7 and 9 cm, and the accuracy of downscaling products reached to the best when snow depth between 16 and 18 cm. When snow depth is bigger than 12 cm, the accuracy of AMSR2 standard snow depth product became worse. But the new product seems much better than the standard product, and the accuracy was improved in whole study area.(3) In recent 14 years, the average annual snow-covered days and average snow depths of China followed an increasing tendency in the majority of areas in Chian. Overall, the annual average snow-covered area of China had no obvious change tendency. In China, the snow-covered days exhibited an increasing tendency in the three seasons of spring, autumn, winter but decreasing in summer; the average snow depth showed a decreasing trend in summer, autumn, winter but increasing in spring; the average snow-covered area decreasing in summer and winter, but increasing in spring and autumn.(4) The tendency of average annual snow-covered day changes showed high consistency with annual average snow depths in terms of spatial distribution, and the spatial distribution of tendency in different seasons also showed high consistency. The snow change tendency of China had a huge different in spatial distribution. Snow showed the significant increasing trend in southern and northeast China, but Xinjiang region was characterized by significant decreasing tendency, and the southeast and southwest margin of Tibetan Plateau showed an increasing tendency, but decreasing in the north and northwest over the China.