Development of better predictions of stress-equivalent winds (ocean forcing) based on machine learning algorithms using scatterometer data.