Spatial-temporal variations analysis of snow cover in China from 1992−2010

Abstract

Snow cover is an important component of land cover on the Earth’s surface and is also a good indicator of climate change. Hence, monitoring the spatial distribution and temporal variation of snow cover is of great significance to the study of the global water cycle and climate change. Traditional snow cover monitoring has been primarily based on in situ observations; however, the uneven and low-density distribution of meteorological stations made it difficult to reflect the overall picture of snow cover in some regions. To solve this problem, we used the Presence and Background Learning (PBL) algorithm to estimate the snow cover in China and obtained the 5 days (5 d) snow cover maps of the Special Sensor Microwave/Imager (SSM/I). The PBL model is a type of one-class classifier that needs no negative training sample in the training set. The cornerstone of this method is to combine SSM/I Brightness Temperature data and in situ observations to estimate the probability of the existence of snow cover based on the Artificial Neural Network (ANN). The estimation result indicates that the average annual overall accuracy of the PBL model in China is 0.88, which shows good agreement with the Interactive Multisensor Snow and Ice Mapping System (IMS) snow cover products. Compared with other snow cover products, such as MODIS/Terra Snow Cover 8-Day L3 Global 500 m Grid (MOD10A2) and AMSR-E/Aqua 5-Day L3 Global Snow Water Equivalent EASE-Grids(AE_5DSno), the performance of the PBL model is better at predicting the snow cover distribution in China. After obtaining the SSM/I 5d snow cover maps of China, we analysed the temporal and spatial variations of snow cover in China during 1992−2010 using the Mann-Kendall test, which included the variation of annual snow cover days, seasonal snow cover distribution and change characteristics, and the variation of the stable snow cover area. The results suggest that China’s snow cover is primarily distributed in the Tibetan Plateau, Xinjiang, northeast China, and Inner Mongolia, which are all high altitude or high latitude regions. From 1992−2010, the following occurred. (1) The number of snow cover days decreased significantly in the three major snow cover regions of China due to rising temperatures, while there was an observably upward trend in the northwestern Tibetan Plateau with the increase of precipitation. (2) The snow cover area of Xinjiang and Northeast-Inner Mongolia reached its maximum in winter and was relatively small in the spring and autumn. The Tibetan Plateau’s snow cover area is generally the largest in the spring, and it is also very large in the autumn and winter. Thus, the seasonal variation characteristics of snow cover there is not as obvious as that of the other two major snow regions. (3) In the four years of 1996, 2003, 2004 and 2006, the snow cover area in the spring in Xinjiang was much lower than that in the autumn, which indicates a “warmer winter” event caused by warmer than usual temperatures in the winter. (4) Although no significant change in the snow cover area was found in all regions during the study period, the change in snow cover on the Tibetan Plateau is particularly noteworthy due to its area having increased greatly since 2005, which might be the beginning of the increase of snow cover area in this region. (5) The stable snow cover area of China is 3.39 million km2. In the three major snow cover regions, the stable snow cover area of the Tibetan Plateau is the largest (1.68 million km2), the Northeast-Inner Mongolia region is the second largest (1.05 million km2), and the Xinjiang region is the smallest (0.63 million km2). The stable snow cover area had no significant variation trend in China during 1992−2010—only the snow cover days of the Tibetan Plateau had a larger interannual fluctuation.