用SeaWinds散射计数据反演海面风矢量的神经网络模型  

A neural network-based model for retrieval of the ocean surface wind vector from SeaWinds scatterometer data

作　　者：解学通^② 方裕[1] 陈克海[2] 解学通[2] 黄舟[1] 陈斌[1] 

机构地区：[1]北京大学地球与空间科学学院,北京 100871 [2]中山大学遥感与地理信息工程系,广州 510275

出　　处：《高技术通讯》2008年第2期184-189,共6页Chinese High Technology Letters

基　　金：863计划(2002AA134100);985工程(105203200400006)资助项目

摘　　要：针对构建用 SeaWinds 散射计数据反演海面风矢量的神经网络模型所面临的主要问题,结合 SeaWinds 散射计的几何观测特征,提出了适合 SeaWinds 散射计旋转圆锥扫描方式的先风向后风速的两步神经网络反演模型及相应算法,并采用两组不同的 L2A 和相应 L2B 数据及浮标数据对该神经网络反演模型进行了初步验证。实验结果证明了该神经网络反演模型的可行性。与最大似然估计(MLE)反演方法相比,该神经网络反演模型在能够保证反演精度的情况下,运行效率提高了约5倍,从而为海面风矢量的实时反演提供了可能性。Aiming at the main problems in establishing a neural network model for retrieval of the ocean surface wind vector from SeaWinds data, and in consideration of the SeaWinds scatterometer＇ s geometry observation characteristics, the paper presents a neural network retrieval model and a corresponding algorithm, which are suitable to the conical scanning mode of the scatterometer. The retrieval of the model consists of two steps the wind direction retrieval and the wind speed retrieval. Two sets of data, level 2A and corresponding Level 2B, and buoy data were selected to validate the model preliminarily. The experimental results demonstrated the feasibility of the retrieval model. The retrieval efficency of the model was approximately five times higher than that of the maximum likelihood estimation （MLE） method without obviously reducing the retrieval precision. Thereby, the model provides the possibility for real-time retrieval of ocean surface wind vectors.

关 键 词：SeaWinds 散射计 风矢量 神经网络 反演模型 

分 类 号：P732[天文地球—海洋科学] TP183[自动化与计算机技术—控制理论与控制工程]