中国海洋大学 博士学位论文 全极化微波辐射成像仪海面风矢量遥感模型和反演算法研究 姓名：赵屹立 申请学位级别：博士 专业：海洋信息探测与处理（海洋物理学） 指导教师：贺明霞 2011-08-10 ? 1? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? 2? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? 3? ? Modeling and Inversion Algorithm for Sea Surface Wind Vector from Polarimetric Microwave Radiometer Measurements Abstract Traditional microwave radiometers measure the brightness temperature of orthogonal polarization, from which sea surface wind speed could be retrieved. In contrast, polarimetric microwave radiometers are new-generation satellite borne passive microwave instruments that measure the multi-frequency four Stokes parameters, from which sea surface wind vector could be retrieved. The first satellite borne polarimetric microwave radiometer developed by the Naval Research Laboratory (NRL) of America under the NPOESS program, the Windsat, was launched successfully in 2003 onboard the satellite Coriolis and has been operated since then. The retrieval of the sea surface wind vector from WindSat measurements is based on the two-scale sea surface emission model and a parameterized atmospheric microwave radiative transfer model with a nonlinear optimization wind-vector retrieval algorithm. Both the two-scale model and the nonlinear optimization algorithm are relatively complex. Typically, a combination of polarimetric microwave radiometer and microwave scatterometer is regarded as the best approach for sea surface wind vector monitoring, which may be adapted for the Chinese HY-2 ocean satellite series. Thus, the objective of this work is to explore a simplified forward microwave radiative transfer model and a rapid retrieval algorithm using the WindSat dataset. In this paper, a simplified polarimetric microwave radiometer forward model was developed using the Stokes parameters provided by the WindSat project, satellite-based observations of sea surface parameters (sea surface wind from QuikScat, sea surface temperature, water vapor, and cloud liquid water from TRMM), and statistical analyses. The simplified model was based on the vertical/horizontal polarization brightness temperature model developed from the Wentz ocean-atmosphere microwave radiative transfer theory, and on the third and fourth Stokes parameters model. The differences between the simplified model results and ? 4? ? WindSat measurements at 10.7GHz, for example, are 0.42 K, 0.52K, 0.33 Ks and 0.22 K for the four Stokes parameters. Comparison with another simplified model by S. H. Yuel (2006) showed that the new forward model yielded more accurate results in the Stokes parameters. The new simplified forward model was applied to a simulated dataset of the 5 parameters as the model input (sea surface wind speed and direction, SST, water vapor, and cloud liquid water; all varied randomly within the given ranges) to simulate the four Stokes parameters, two of which were used in a new retrieval algorithm to derive the wind vector. The retrieval algorithm was based on the multiple linear regression between the two Stokes parameters and the to-be-retrieved parameters, including sea surface wind speed, SST, columnar water vapor, columnar cloud water. Multiple ambiguous wind directions were estimated with the Maximum Likelihood Method by using the retrieved sea surface wind speed and the other parameters, and a median filter was used to remove the ambiguity of the retrieved wind direction. Using the U.S. NDBC (National Data Buoy Center) sea surface wind speed and direction data as the ground truth, the root-mean-square (RMS) errors of the retrieved sea surface wind speed and di