Deep learning-based sea surface wind speed retrieval method for Sentinel-1 dual-polarimetric SAR under typhoon sea state

It is of great significance to realize high-precision observation of the sea surface wind field under typhoon sea conditions for disaster prevention and mitigation. Traditional methods rely on geophysical model functions to retrieve wind speed, which necessitates the input of external wind direction information. The accuracy of wind direction directly affects the accuracy of wind speed retrieval. The deep neural network method combines traditional methods with data mining, and uses this method to retrieve the sea surface wind speed without the input of external wind direction information, which simplifies the retrieval process and opens up a new development direction of sea surface wind speed retrieval for synthetic aperture radar (SAR). However, its fitting ability is limited. To achieve high-precision sea surface wind speed retrieval without the need for external wind direction information input, this paper proposes a deep learning-based method for typhoon sea state and sea surface wind speed retrieval from Sentinel-1 dual-polarimetric SAR. Based on the DenseNet deep learning model, this method achieves high-precision sea surface wind speed retrieval without external wind direction input. Experimental results show that the root mean square error of wind speed inversion by the proposed method can reach 1.7418 m/s, and the correlation degree can reach more than 0.9, which is better than the inversion results of traditional methods and deep neural network methods. The proposed method further demonstrates the effectiveness of deep learning techniques in the field of SAR sea surface wind field retrieval for sea surface wind field retrieval.