我国主要积雪区AMSR-E被动微波雪深算法对比验证研究

采用高级微波扫描辐射计(AMSR-E)亮温数据, 选取Chang算法、GSFC 96算法、AMSR-E SWE 算法、青藏高原改进算法和Savoie算法等5种雪深反演算法, 利用2010年2月10-12日3 d的气象站台雪深观测数据, 对比分析了5种雪深反演算法在新疆地区、青藏高原、内蒙古地区、东北地区、西北地区和华北平原的精度和适用性. 结果表明:总体验证中, 青藏高原改进算法3 d的结果均优于其他算法, 其均方根误差(RMSE)为9.16 cm、9.96 cm和9.63 cm, 平均相对误差(MRE)分别为59.77%、52.79%和48.47%. 分区验证中, 结果最佳的算法分别为:在新疆地区, GSFC 96算法RMSE为6.85~7.48 cm;内蒙古地区, 青藏高原改进算法的RMSE分别为5.9 cm、6.11 cm和5.46 cm;东北地区, 青藏高原改进算法RMSE为6.21~7.83 cm;西北地区和华北平原5种算法的适用性不佳;青藏高原由于缺乏实测数据, 无法得到该区验证统计结果.

Comparative Validation of Snow Depth Algorithms Using AMSR-E Passive Microwave Data in China

Using passive microwave data to derive snow depth has become a fast and effective method, but regional accuracy of global snow algorithms is limited, especially in the regions where ground-based measured data are short, such as west China and the Tibetan Plateau. In this study, for the sake of evaluating the accuracy of snow inversion algorithms by using passive microwave data from the Advanced Microwave Scanning Radiometer-Earth Observing System (AMSR-E) in China, five different snow algorithms (Chang algorithm, Goddard Space Flight Center, GSFC, 96 algorithm, AMSR-E SWE algorithm, improved Tibetan Plateau algorithm and Savoie algorithm) are selected, making use of AMSR-E brightness temperature data and meteorological station data in February 10-12, 2010, to validate the accuracy of snow depth algorithms in Xinjiang, the Tibetan Plateau, Inner Mongolia, Northeast China, Northwest China, the North China Plain, and then compared the accuracies of the five snow inversion algorithms. It is found that in the overall validation, the results of improved Tibetan Plateau algorithm are better than other algorithms, of which the root mean square error (RMSE) is 9.16, 9.96 and 9.63 cm;the average relative error (MRE) is 59.77%, 52.79% and 48.47%. In the partition validation, the results of the best algorithms are: in Xinjiang, with the RMSE of GSFC 96 algorithm ranging from 6.85 to 7.48 cm;in Inner Mongolia, with the RMSE of the improved Tibetan Plateau algorithm of 5.9, 6.11 and 5.46 cm;in Northeast China, with the RMSE of improved Tibetan Plateau algorithm ranging from 6.21 to 7.83 cm;in the Tibetan Plateau, owing to the lack of measured data, the verify statistical result is unable to obtain;and the existing algorithms have a poor applicability in Northwest China and the North China Plain.