基于Triple-Collocation方法的微波遥感土壤水分产品不确定性分析及数据融合

微波遥感可以获取大范围的地表土壤水分信息，以及由此得到全球尺度的土壤水分产品。但由于传感器观测配置和反演方法等诸多因素的影响，使得不同的土壤水分产品在精度和可靠性方面存在差异。基于Triple-Collocation（TC）方法，在青藏高原那曲地区的0.25°×0.25°和1.0°×1.0°两个空间尺度上对AMSR2、SMAP和SMOS 3种土壤水分遥感产品进行不确定性分析，开展基于随机误差的数据融合算法研究。研究结果表明:不同遥感产品间的随机误差在空间分布上存在显著的不一致性，使得应用传统的算术平均方法进行数据融合不具有普适性。基于此不确定性，对3种产品配赋相应的权重进行融合，相比于3种土壤水分原始数据集，融合产品不仅具有更丰富的数据量，也会对数据精度有所改善。当遥感产品间的随机误差接近时，等权重和优化权重的融合结果非常接近；当遥感产品间的随机误差差异较大时，基于不确定性的数据融合方法相比等权重方法可以明显的提高融合数据的精度。

Uncertainty Analysis and Data Fusion of Microwave Remote Sensing Soil Moisture Products based on Triple-Collocation Method

Microwave remote sensing can provide large scale soil moisture information, and even further derive soil moisture products at global scale. Due to impacts of observation configuration and retrieval method etc., different soil moisture products feature different accuracies and reliabilities. Based on Triple-Collocation method, this study analyzes the uncertainties among AMSR2, SMAP and SMOS soil moisture products at two spatial scales in Naqu study area, i.e., 0.25°× 0.25° and 1.0°×1.0°, and further performs data fusion based on analyzed random errors to obtain more reliable soil moisture product. The uncertainty analysis indicates that the three products have distinct random error distribution in spatial. In this case, the traditional arithmetic mean method may not be appropriate. Hence, data fusion is performed by proposed optimizing weighting method based on the analyzed uncertainties. In comparison with the three original soil moisture products, the fusion result shows a more effective data size and improved accuracy. When different original products present similar errors, the fusion products of equal weighting and optimized weighting methods show the similar performance. Oppositely, the uncertainties analysis based fusion method is superior to equal weighting method in terms of effective data size and accuracy.