Tropical Cyclone Wind Direction Retrieval Using Histogram of Oriented Gradients on Dual-Polarized Synthetic Aperture Radar Images

Accurate knowledge of wind directions plays a critical role in atmospheric dynamics exploration, numerical weather prediction and Tropical Cyclone (TC) research. This study proposes a new method for wind direction retrieval from TC Synthetic Aperture Radar (SAR) images. Unlike conventional approaches, which estimate wind directions from singlepolarization imagery, the method utilizes dual-polarized (VV and VH) signals to obtain continuous wind directions across moderate and extreme wind speed regimes. The technique is developed based on the Histogram of Oriented Gradient descriptor and the Hann window function. In addition, the neighbouring information is introduced to alleviate sharp directional variations. As case studies, the wind directions in TCs Karl and Maria are derived and subsequently verified by simultaneous dropsonde and ASCAT (ambiguity-removed) measurements. The encouraging results suggest that the wind direction retrieval method based on dual-polarization SAR imagery can be useful in extracting wind direction and contribute to further exploitation of SAR images in TC studies.