For the application of soil moisture and ocean salinity（SMOS） remotely sensed sea surface salinity（SSS） products,SMOS SSS global maps and error characteristics have been investigated based on quality control infor...For the application of soil moisture and ocean salinity（SMOS） remotely sensed sea surface salinity（SSS） products,SMOS SSS global maps and error characteristics have been investigated based on quality control information.The results show that the errors of SMOS SSS products are distributed zonally,i.e.,relatively small in the tropical oceans,but much greater in the southern oceans in the Southern Hemisphere（negative bias） and along the southern,northern and some other oceanic margins（positive or negative bias）.The physical elements responsible for these errors include wind,temperature,and coastal terrain and so on.Errors in the southern oceans are due to the bias in an SSS retrieval algorithm caused by the coexisting high wind speed and low temperature; errors along the oceanic margins are due to the bias in a brightness temperature（TB） reconstruction caused by the high contrast between L-band emissivities from ice or land and from ocean; in addition,some other systematic errors are due to the bias in TB observation caused by a radio frequency interference and a radiometer receivers drift,etc.The findings will contribute to the scientific correction and appropriate application of the SMOS SSS products.展开更多
This paper proposes a new method to retrieve salinity profiles from the sea surface salinity （SSS） observed by the Soil Moisture and Ocean Salinity （SMOS） satellite. The main vertical patterns of the salinity prof...This paper proposes a new method to retrieve salinity profiles from the sea surface salinity （SSS） observed by the Soil Moisture and Ocean Salinity （SMOS） satellite. The main vertical patterns of the salinity profiles are firstly extracted from the salinity profiles measured by Argo using the empirical orthogonal function. To determine the time coefficients for each vertical pattern, two statistical models are developed. In the linear model, a transfer function is proposed to relate the SSS observed by SMOS （SMOS_SSS） with that measured by Argo, and then a linear relationship between the SMOS_SSS and the time coefficient is established. In the nonlinear model, the neural network is utilized to estimate the time coefficients from SMOS_SSS, months and positions of the salinity profiles. The two models are validated by comparing the salinity profiles retrieved from SMOS with those measured by Argo and the climatological salinities. The root-mean-square error （RMSE） of the linear and nonlinear model are 0.08-0.16 and 0.08-0.14 for the upper 400 m, which are 0.01-0.07 and 0.01-0.09 smaller than the RMSE of climatology. The error sources of the method are also discussed.展开更多
海表面盐度是描述海洋状态、模拟海洋循环和检测气候变化的重要指标,对海洋研究意义重大。土壤湿度与海水盐度(soil moisture and ocean salinity,SMOS)卫星为全球海表面盐度分析提供了重要数据,但其整体精度尚未达到预期要求。基于海...海表面盐度是描述海洋状态、模拟海洋循环和检测气候变化的重要指标,对海洋研究意义重大。土壤湿度与海水盐度(soil moisture and ocean salinity,SMOS)卫星为全球海表面盐度分析提供了重要数据,但其整体精度尚未达到预期要求。基于海表面盐度遥感机理和SMOS卫星盐度反演基础理论,选取海表面盐度敏感因子,建立随机森林(random forest,RF)模型,并基于网格搜索算法优化模型参数,辅助提高SMOS卫星产品精度。其中基础RF得到的海表面盐度与Argo(array for real-time geostrophic oceanography)数据之间的平均绝对误差为0.08,均方根误差为0.15。而经网格搜索算法优化后的随机森林模型精度稍有所提升,其与Argo数据的绝对平均误差为0.08,均方根误差仅为0.14,且误差分布范围较小。两种模型均显著优于SMOS卫星Level 2级盐度产品。从机器学习与统计学理论出发,建立的高精度、高适应性的随机森林海表面盐度反演模型大幅提高了盐度精度,能够为相关海洋研究提供数据支撑。展开更多
基金The National Natural Science Fund of China under contact No.41276088the National Natural Science Fund for Young Scholars of China under contact Nos 41206002 and 41306010
文摘For the application of soil moisture and ocean salinity（SMOS） remotely sensed sea surface salinity（SSS） products,SMOS SSS global maps and error characteristics have been investigated based on quality control information.The results show that the errors of SMOS SSS products are distributed zonally,i.e.,relatively small in the tropical oceans,but much greater in the southern oceans in the Southern Hemisphere（negative bias） and along the southern,northern and some other oceanic margins（positive or negative bias）.The physical elements responsible for these errors include wind,temperature,and coastal terrain and so on.Errors in the southern oceans are due to the bias in an SSS retrieval algorithm caused by the coexisting high wind speed and low temperature; errors along the oceanic margins are due to the bias in a brightness temperature（TB） reconstruction caused by the high contrast between L-band emissivities from ice or land and from ocean; in addition,some other systematic errors are due to the bias in TB observation caused by a radio frequency interference and a radiometer receivers drift,etc.The findings will contribute to the scientific correction and appropriate application of the SMOS SSS products.
基金The Startup Foundation for Introducing Talent of Nanjing University of Information Science and Technology under contract No.S8113078001
文摘This paper proposes a new method to retrieve salinity profiles from the sea surface salinity （SSS） observed by the Soil Moisture and Ocean Salinity （SMOS） satellite. The main vertical patterns of the salinity profiles are firstly extracted from the salinity profiles measured by Argo using the empirical orthogonal function. To determine the time coefficients for each vertical pattern, two statistical models are developed. In the linear model, a transfer function is proposed to relate the SSS observed by SMOS （SMOS_SSS） with that measured by Argo, and then a linear relationship between the SMOS_SSS and the time coefficient is established. In the nonlinear model, the neural network is utilized to estimate the time coefficients from SMOS_SSS, months and positions of the salinity profiles. The two models are validated by comparing the salinity profiles retrieved from SMOS with those measured by Argo and the climatological salinities. The root-mean-square error （RMSE） of the linear and nonlinear model are 0.08-0.16 and 0.08-0.14 for the upper 400 m, which are 0.01-0.07 and 0.01-0.09 smaller than the RMSE of climatology. The error sources of the method are also discussed.
