| 本文已被：浏览 747次 下载 619次 |
码上扫一扫！ |
| 基于新型观测算子的双偏振雷达雨滴谱变分反演 |
|
陈垚1,2,3,4, 寇蕾蕾1,5, 蒋银丰1, 杨春生1, 林正健1, 楚志刚1,5
|
|
1. 南京信息工程大学大气物理学院，江苏 南京 210044;2. 福建省平潭综合实验区气象局，福建 平潭 350400;3.4. 平潭海洋气象野外科学观测研究站，福建 平潭 350400;4.5. 福建省灾害天气重点实验室，福建 福州 350001;5.3. 南京信息工程大学气象灾害预警与评估协同创新中心，江苏 南京 210044
|
|
| 摘要: |
| 提出一种基于变分理论的双偏振多普勒天气雷达反演雨滴谱方法，最优反演雨滴谱的同时可实现衰减订正。反演过程中使用一种新型观测算子，利用滴谱仪实测数据计算状态变量和双偏振参量，使观测算子更能代表本地降水特性。将新型观测算子、误差协方差矩阵和状态变量的先验估计用于代价函数中，基于高斯-牛顿迭代方法求解代价函数得到最优雨滴谱。利用理想模拟试验和南京信息工程大学C波段双偏振多普勒天气雷达实测个例对算法进行验证和评估。结果表明：算法反演得到的状态变量(液态水含量LWC和质量权重平均直径Dm)与滴谱仪数据计算结果的相关系数达到了0.96和0.80，相对偏差为25.19%和10.63%，均方根误差比常规反演结果改善了50%左右，比基于模拟观测算子的变分反演结果改善了30%左右，最优反演雨滴谱得到的降雨率R和雨量计数据相关系数达到0.89，相对偏差为14.78%。 |
| 关键词: 变分反演 观测算子 双偏振多普勒天气雷达 雨滴谱反演 衰减订正 |
| DOI：10.16032/j.issn.1004-4965.2022.077 |
| 分类号: |
| 基金项目: |
|
| VARIATIONAL RAINDROP SIZE DISTRIBUTION RETRIEVAL FROM DUAL-POLARIMETRIC RADAR BASED ON A NEW OBSERVATION OPERATOR |
|
CHEN Yao1,2,3,4, KOU Leilei1,5, JIANG Yinfeng1, YANG Chunsheng1, LIN Zhengjian1, CHU Zhigang1,5
|
|
1. School of Atmospheric Physics, Nanjing University of Information Science and Technology, Nanjing 210044, China;2. Pingtan Meteorological Bureau, Pingtan, Fujian 350400, China;3.4. Pingtan Marine Meteorological Field Scientific Observation and Research Station, Pingtan, Fujian 350400, China;4.5. Fujian Provincial Key Laboratory of Disaster Weather, Fuzhou 350001, China;5.3. Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters, Nanjing University of Information Science and Technology, Nanjing 210044, China
|
| Abstract: |
| This research proposes a method of retrieving raindrop size distribution by using dual-polarimetric Doppler weather radar based on variational theory, which can achieve attenuation correction while optimally retrieving raindrop size distribution. A new type of observation operator is used in the retrieval process, which employs the measured data from the disdrometer to calculate the state variables and dual-polarimetric radar parameters, so that the observation operator can better represent the characteristics of local precipitation. The new observation operator, error covariance matrix, and prior estimation of state variables are used in the cost function, and the optimal raindrop size distribution is obtained by solving the cost function using the Gauss-Newton iteration method. The algorithm in this paper is verified and evaluated by using an ideal simulation test based on actual data from the C-band dual-polarimetric Doppler weather radar of Nanjing University of Information Science and Technology. The results show that the correlation coefficients between the state variables including the liquid water content and the mass weight average diameter obtained by the algorithm in this paper and the calculated results of the disdrometer data reach 0.96 and 0.80, and the relative deviations are 25.19% and 10.63%, respectively. The root mean square error is improved by about 50% compared with that of the conventional retrieval and about 30% compared with that of the variational retrieval based on the simulated observation operator. The correlation coefficient between the rainfall rate obtained by the optimal retrieval of the raindrop size distribution (DSD) and the rain gauge data reaches 0.89, and the relative deviation is 14.78%. |
| Key words: variational retrieval observation operator dual-polarimetric Doppler weather radar DSD retrieval attenuation correction |