光学与微波遥感的新疆积雪覆盖变化分析

利用2002-2013年冬季的MODIS光学遥感数据,以及AMSR-E、AMSR2与MWRI被动微波遥感数据,建立了新疆地区冬季每日积雪分布遥感反演模型。首先,将Terra与Aqua双星MODIS的积雪产品融合,初步去云并最大化积雪信息;然后,利用AMSR-E/AMSR2和MWRI被动微波数据进行每日雪盖提取;最后,利用被动微波遥感数据反演得到的每日雪盖结果对双星融合后依然有云的像元进行替换,得到每日积雪分布情况。据此模型提取了11年间冬季的积雪天数信息,结合气象台站观测数据,分析了新疆冬季积雪的年内和年际变化规律。结果表明,新疆地区积雪主要分布在北部新疆,积雪天数与地形关系密切,山区积雪天数较多,盆地及城市区积雪天数较少;积雪天数年内变化是从11月到次年1月随温度降低逐渐增加,从1月到3月积雪天数则逐渐减少。新疆地区积雪天数在这11年中存在一定的波动,积雪天数与该年的平均气温,以及月低于0℃的天数存在显著相关性,与降雪量关系不明显。新疆地区近年来积雪天数重心有向西向南移动的趋势,这可能与全球气候变暖导致多年积雪融化有关。

Retrieval and Analysis of Snow-covered Days in Xinjiang Based on Optical and Microwave Remote Sensing Data

As an important land cover type of the Earth’s surface, snowpack has a huge effect on water cycle, climate change, environment and human activities. The snow covers are sensitive in indicating climate change and extreme weather events. Thus, it is significant to map the snowpack’s distribution and study the change trend of its time series. This study is focused on the use of the MODIS, AMSR-E, AMSR2 and MWRI data in winter (from November to March) from 2002 to 2013 to develop a retrieval model of daily snow distribution in Xinjiang. The technical process of establishing this model is divided into the following steps. First, data fusion was employed to handle the Terra and Aqua MODIS snow cover products, reduce the proportion of cloud pixels, and maximize the percentage of snow pixels, because the cloud has more mobility and it can change rapidly during the Terra and Aqua transit time. Second, AMSR-E, AMSR2 and MWRI data were used to retrieve daily snow cover map. Third, the two snow cover products were fused to produce cloud-free snow cover products. Based on the proposed approach, daily snow cover data of the entire Xinjiang were extracted for the winter days from 2002 to 2013. After that, the snow-covered days were extracted from the daily snow cover data using a statistical approach. For example, if a pixel location on the images was identified “snow” in three consecutive days, it was recognized as the type of “snow” and the count of days is “3”; otherwise, it was recognized as “non-snow” type. This algorithm was adopted to avoid mistakes in counting snow-covered days when the pixel location was only covered by snow within a short period that is less than a day. The method complies with the continuity aspect of the “snow cover” definition. The number of days covered with snow was finally retrieved for each pixel location in every winter using this proposed algorithm. The map of the averaged snow-covered days shows that the snow-covered days of Xinjiang have a close relationship with terrain, and the number of the snow-covered days in the mountainous areas is bigger than that in the basin and urban areas. The snow packs were mainly distributed in the northern part of Xinjiang. From November to January, snow-covered days increased because the temperature decreased, while from January to March, the number of the snow-covered days decreased with the rise of temperature. The snow-covered days of Xinjiang reveal some fluctuations from 2002 to 2013. The monthly snow-covered days showed significant correlation with the monthly average temperature and the number of days in a month having temperature lower than 0℃. The snow-covered days have no obvious relationship with snowfall. In recent years, the gravity centre of the snow-covered days in Xinjiang area has a southward and eastward movement trend, which may relate to the global warming. Further studies are needed to explain this issue in future.