NEURAL NETWORK WIND RETRIEVAL FROM SCATTEROMETER DATA
LIN Ming-sen 1,3, SONG Xin-gai 2,1, PENG Hai-long 1, FENG Qian 1
1.National Satellite Ocean Application Service, Beijing 100081, China; 2.Ocean University of China, Qingdao 266003, China; 3.The Third Istitute of Oceanography, SAO,Xiamen 361005, China
This paper presents a neural network method for retrieving wind vectors from ERS-1/2 scatterometer data, which resolves wind directional ambiguities for scatterometer derived winds by a circular median filter algorithm. Learning data set and test data set come from ERS-1/2 scatterometer data collocated pairs with ECMWF vectors. A comparison with COMD4 and ECMWF wind vector shows that the result is good and the performance is quicker than any other methods. The good performance of the neural network method suggests the possibility of wind retrieval from ERS-1/2 scatterometer.