Seasonal and Geographical Differences of Accuracy for SAR Sea Wind Retrieval Using Deep Neural Networks in Coastal Waters of Korea

A study was conducted to estimate sea wind in coastal waters using SAR satellite images and deep neural network (DNN) models. The Sentinel-1 SAR images from 2015 to 2020 were collected that cover the coastal waters of Korean south-east regions. Predicted wind data from meteorological forecast model results and observed wind data from sea buoys were matched-up with the SAR images to establish a training dataset for wind retrieval DNN model build-ups. Two DNN models, for eastward wind speed and northward wind speed, were created and optimized through training processes. The final results of the DNN models showed correlation coefficient (CC) of 0.901 and MAE 1.467 for eastward wind speed, CC 0.826 and MAE 1.586 for northward wind speed. The eastward wind DNN model showed a better accuracy than the northward. The difference in accuracy between the eastward wind DNN model and the northward wind DNN model was considered due to the low correlation between north wind and wave in southern sea area of Korea, especially in winter season when north wind dominates. For better accuracy, it is recommended to classify coastal regions and seasons along common marine meteorological characteristics, and to construct a deep neural network model for each classified case.