论文部分内容阅读
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 the neural network model of sea surface wind vector inversion using SeaWinds scatterometer data and combining with the geometric observation features of SeaWinds scatterometer, a two-step neural Network inversion model and corresponding algorithm. The neural network inversion model was preliminarily verified by using two different sets of L2A and corresponding L2B data and buoy data. Experimental results show that the neural network inversion model is feasible. Compared with the maximum likelihood estimation (MLE) inversion method, the neural network inversion model can improve the operating efficiency about 5 times when it can guarantee the accuracy of inversion, which makes it possible for the real-time inversion of the sea surface wind vector Sex.