中文摘要：

本研究利用青藏高原地区2002-2008年MODIS/Terra-Aqua逐日雪被产品（MOD10A1及MYD10A1）和AMSR-E/Aqua每日雪水当量产品AE_DySno,研究了MODIS和AMSR-E逐日数据的融合算法,合成出逐日无云积雪分类图像MATS10A1,并利用气象台站提供的雪情数据验证了合成图像的积雪分类精度。研究结果表明：1）在青藏高原地区,虽然在晴空时MODIS积雪分类精度较高（当雪深〉3cm时达到80.82%）,但MOD10A1和MYD10A1图像中的平均云量比分别达到39.74%和48.74%,无法对牧区雪情进行实时监测。2）MOD10A1和MYD10A1的合成图像（MOYDTS10A1）云量比为24.13%,不但消除了大部分云的影响,而且提高了积雪分类精度（积雪分类精度为81.67%）。3）合成图像MATS10A1结合了AMSR-E资料不受天气影响和MODIS雪被产品较高空间分辨率的优点,不仅完全消除了云的干扰,同时具有较高的积雪分类精度（79.36%）。因此,这种改进型算法生成的逐日无云图像,在青藏高原牧区雪灾监测与预警研究中将具有重要的应用前景。

英文摘要：

The 2002 to 2008 MODIS/Terra-Aqua daily snow products （MOD10A1 and MYD10A1） and AMSR- E/Aqua daily snow water equivalent product AE DySno from the Tibetan Plateau were used to study the daily composition algorithm of MODIS and AMSR-E data. The accuracies of composited daily cloud-free snow classi- fication image MATS10A1 were validated based on snow depth data from climate stations. 1） In the Tibetan Plateau, snow classification accuracy of MODIS daily snow product is high （reaching 80.82 % when the snow depthS3 cm） under clear sky conditions, but MOD10A1 and MYD10A1 are not suitable for monitoring the snow distribution in real time in pastoral areas because the average cloud ratios are 39.74 % and 48.74 %, re- spectively; 2） Composited images （MOYDTS10A1） of MOD10A1 and MYD10A1 not only eliminate most of the cloud＇s impact （the average cloud ratio is 24. 13%0） but also improves the snow classification accuracy （reaching 81. 67% when the snow depth 〉3 cm）; 3） Daily cloud-free snow classification image MATS10A1 combines the advantages of AMSR-E （cannot be affected by weather conditions） and MODIS （relatively high resolution）, completely eliminates the interference of clouds, and also has a relatively high snow classification accuracy （reaching 79. 36% when the snow depth 〉3 cm）. Therefore, the daily cloud-free images generated by the improved algorithm in this study will play an important rote in snow disaster monitoring and evaluation in the Tibetan Plateau pastoral areas in the future.