基于风带的海表面亮温增量统计模型

Statistical Models of Brightness Temperature Incremen Based on Wind Belts

海表面盐度(Sea Surface Salinity,SSS)是研究大洋环流和海洋对全球气候影响的关键参数之一。目前借助卫星遥感技术获取全天候和连续的SSS是最有效的方法,但是SSS的反演精度在大部分海域达不到预期目标。众所周知,海表面亮温是反演SSS的关键因素之一,海面粗糙度导致了亮温增量的产生,亮温正演模型的误差会影响盐度反演的精度。本文首次提出了依据6个风带划分全球海域,利用Argo实测盐度数据、SMOS卫星数据和相关辅助数据,通过LASSO统计方法在各风带覆盖的海域构建了一个全新的二次曲线亮温增量模型,再通过贝叶斯迭代反演算法计算出了各个海域的SSS产品。与Argo实测SSS对比,新模型下6部分海域反演SSS的绝对平均误差分别为0.76、0.88、0.93、0.92、1.28和1.21,均显著优于修正前(SMOS L2 SSS)产品的误差(0.98、1.61、2.82、1.50、2.35和3.13)。

Sea surface salinity(SSS)is one of the key parameters to study ocean circulation and ocean’s impact on global climate.Although satellite remote sensing technique for all-weather and continuous SSS is the most effective method,the precision of SSS in most sea areas is not able to achieve the expected goal.It is well known that brightness temperature of sea surface is one of the key factors for the inversion of SSS,and the roughness of sea surface results in the production of brightness temperature increment.The precision of inversed SSS is largely affected by the error of the brightness temperature forward model.For the first time,we build the brightness temperature increment model for six wind belts respectively.In each area,a new quadratic curvilinear model for the brightness temperature increment is established with the Least Absolute Shrinkage and Selection Operator(LASSO)algorithm,using the Argo measured data,Soil Moisture and Ocean Salinity(SMOS)satellite data and auxiliary data.Then the SSS product of individual parts is calculated by the inversion algorithm of Bayesian iterative method.Compared with Argo salinity data,the mean absolute error of inversed SSS in the six parts derived from our model is 0.76,0.88,0.93,0.92,1.28 and 1.21 respectively,while the values are 0.98,1.61,2.82,1.50,2.35 and 3.13 for SMOS Level 2 SSS products respectively.The errors of inversed SSS from new model are dramatically smaller than those of SMOS Level 2 SSS products.