中文摘要：

针对单一被动微波遥感反演雪深的精度和空间分辨率不足的问题,提出了一种星-地多源数据融合的雪深反演方法。以北疆每日站点观测雪深、AMSR-E遥感反演雪深和SSM/I遥感反演雪深数据为研究对象,首先利用地统计方法结合地面站点观测数据估计北疆区域的雪深,然后采用Triple-Collocation方法分别估计三个雪深数据的误差方差,最后结合最小二乘原理实现星-地雪深观测数据的融合。对融合雪深与地面雪深观测数据进行验证分析,结果显示,与AMSR-E和SSM/I遥感反演雪深相比,融合的雪深与地面观测雪深的相关性更高;融合的雪深的精度有一定程度的提高。实验结果证明了多源数据融合方法的有效性。

英文摘要：

Because of the insufficient accuracy and spatial resolution of snow depth products retrieved by passive microwave remote sensing,a new multi-sources data fusion approach is developed for retrieving snow depth.The data from different sources contains visible,passive microwave satellite and in-situ data.The daily in-situ,AMSR-E and SSM/I retrieved snow depth products are used in this study.First,combining in-situ snow depth,the snow depth of Northern Xinjiang is estimated through geostatistical analysis.Then the error variances of each product are calculated using a triple collocation（TC）method.Finally,the new snow depth products are obtained by merging in-situ,AMSR-E and SSM/I snow depth data in a least squares criterion where the optimal weights of each product are determined with the TC-based error variances.The merged snow depth is validated against in-situ snow depth and exhibits a higher correlation with in-situ observations than that with original AMSR-E and SSM/I snow depth.The results with higher accuracy demonstrate the effectiveness of our approach.