星载微波辐射计的非线性反演算法

Non-linear retrieval algorithm for passive satellite microwave radiometers

为了提高用星载微波辐射计观测的数据反演大气海洋参数的精度和效率,提出了使用非线性的Nelder-Mead算法求解星载微波辐射计的辐射传输模型(RTM)的新的反演算法。与目前常用的多元线性回归(MLR)算法相比,Nelder-Mead算法保证了整个反演过程具有明确的物理意义,避免了引入MLR模型带来的误差。反演结果表明,与传统的M氓算法相比,Nelder-Mead算法具有较好的反演效率和较高的精度。

To improve the accuracy and efficiency of the retrieval of oceanic and atmospheric parameters from the ob- served data of passive satellite microwave radiometers, a new retrieval method using the non-linear Nelder-Mead al- gorithmis to solve the radiative transfer model （RTM） of a passive satellite microwave radiometer is proposed. Com- pared with the multiple linear regression （MLR） algorithm, commonly used in the operational retrieval systems, the Nelder-Mead algorithm has definite physical meanings in the whole retrieval process and avoids the errors caused by the use of the MLR model. The retrieval results indicate that the Nelder-Mead algorithm has the better computation- al efficiency and the higher retrieval accuracy compared with the MLR algorithm.