| 基于变分多参数正则化方法融合卫星散射计资料 |
| |
| 引用本文： | 张凯峰,邓婉月,王挺,王慧鹏,项杰,宋清涛,刘春霞.基于变分多参数正则化方法融合卫星散射计资料[J].海洋学报,2017,39(12):122-135. |
| |
| 作者姓名： | 张凯峰 邓婉月 王挺 王慧鹏 项杰 宋清涛 刘春霞 |
| |
| 作者单位： | 1.解放军理工大学 气象海洋学院, 江苏 南京 211101 |
| |
| 基金项目： | 国家自然科学基金（41275113）；全球变化与海气相互作用专项。 |
| |
| 摘 要： | 基于常规三维变分同化（3DVAR）思想和反问题中的正则化技术，提出了适用于风场融合的带正则化约束项的3DVAR方法，在南海海域开展数据融合试验，同时采用模型函数方法确定合理的正则化参数，针对一次台风个例进行了QuikSCAT散射计海面风场数据和华南中尺度模式海面风场数据的融合试验，结果表明采用带正则化约束的3DVAR融合方法，明显消除了常规3DVAR方法融合风场时带来的虚假信息，融合后分析风场以及涡度场和散度场分布均匀，结构清晰，气旋中心显著，且分析场中观测起主导作用；采用信号自由度（DFS）方法对融合方法进行定量评估，发现相对常规3DVAR方法，带正则化约束的3DVAR融合系统中观测数据提供的DFS较多，同时提高了观测场对分析场的影响；基于独立观测资料对融合结果进行检验发现相对华南中尺度模式和常规3DVAR方法的统计结果，带正则化约束的3DVAR方法得到的风场具有最小的均方根误差和最大的相关系数。
|
| 关 键 词： | 3DVAR 正则化 QuikSCAT 信号自由度 模型函数 |
| 收稿时间： | 2017/1/23 0:00:00 |
| 修稿时间： | 2017/3/16 0:00:00 |
Blending satellite scatterometer data based on variational with multi-parameter regularization method |
| |
| Zhang Kaifeng,Deng Wanyue,Wang Ting,Wang Huipeng,Xiang Jie,Song Qingtao and Liu Chunxia.Blending satellite scatterometer data based on variational with multi-parameter regularization method[J].Acta Oceanologica Sinica (in Chinese),2017,39(12):122-135. |
| |
| Authors: | Zhang Kaifeng Deng Wanyue Wang Ting Wang Huipeng Xiang Jie Song Qingtao and Liu Chunxia |
| |
| Institution: | Institute of Meteorology and Oceanography, PLA University of Science and Technology, Nanjing 211101, China,Xinjiang Uygur Autonomous Region Environmental Monitoring Station, Urumqi 830011, China,Institute of Meteorology and Oceanography, PLA University of Science and Technology, Nanjing 211101, China,Unit No. 75839 of PLA, Guangzhou 510510, China,Institute of Meteorology and Oceanography, PLA University of Science and Technology, Nanjing 211101, China;Key Laboratory of Mesoscale Severe Weather (Nanjing University), Ministry of Education, Nanjing 210023, China,National Satellite Ocean Application Service, State Oceanic Administration, Beijing 100081, China and Guangzhou Institute of Tropical and Marine Meteorology, China Meteorological Administration, Guangzhou 510080, China |
| |
| Abstract: | A 3DVAR method with regularization constraints is proposed to blend sea surface wind data in the South China Sea based on the traditional 3DVAR and regularization technology of the inverse problem, and the model function method which is used to determine the reasonable regularization parameters and then the blended experiments of the satellite scatterometer (QuikSCAT) and Guang Zhou Mesoscale Model (GZMM) sea surface wind field data are carried out for a typhoon case. Results show that when we use the regularization method for experiments, the false information caused by the traditional 3DVAR is eliminated obviously and the noise is almost disappeared, at the same time, the wind field and vorticity field as well as divergence field are distributed evenly, and the structure is clear, more importantly, it is clear that the cyclone center is remarkable, and observation is dramatic in the analysis field. Besides, the degrees of freedom for signal (DFS) method is used to evaluate blended systems quantitatively, it is found that the regularized constraint 3DVAR system has a higher DFS and observation influence related to traditional 3DVAR. The blended results are tested based on the independent observation data, it indicates that the result of regularized constraint 3DVAR method has the smallest root mean square error and maximum correlation coefficient, which is better than the statistical result of GZMM and the conventional 3DVAR method. |
| |
| Keywords: | 3DVAR regularization QuikSCAT the degrees of freedom for signal model function |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《海洋学报》浏览原始摘要信息 |
| 点击此处可从《海洋学报》下载免费的PDF全文 |
|
