Sea Surface Wind Retrieval Algorithms And Applications For Dual-pol SAR Images

Sea surface wind is an essential parameter which is strongly related to climate change,maritime surveillance,and the marine environment.Highresolution synthetic aperture radar(SAR)has become an invaluable asset for monitoring ocean since it is capable of providing continuous all-weather day/night imagery with wide spatial coverage.SAR images can achieve sea surface wind retrieval with higher spatial resolution and near the coastal areas.Nowadays,with more available polarimetric SAR systems,sea surface wind field retrieval from dual-polarization(dual-pol)SAR images has been a topic of active research.However,there are several issues regarding the sea surface wind speed retrieval from dual-polarization SAR images.Specifically,formulating major factors(such as wind speed,relative direction,and incidence angle)into a consistent function for wind retrieval is a question yet to be solved.The appropriate polarimetric feature that can be used as a proxy of the wind speed needs to be investigated.The objective of this research is to achieve a high accuracy wind retrieval result using dual-pol SAR images.From the view of SAR backscattering intensity,polarimetric characterization,and statistical modeling of sea clutter,this research proposes two novel features to characterize wind speed and develops two novel algorithms for wind speed retrieval.Furthermore,the retrieved wind speeds are applied to the fields of ship detection and tropical cyclones monitoring.Significant contributions can be summarized as follows:(1)Based on the SAR normalized radar cross section(NRCS)value,we propose a novel wind retrieval algorithm consists of three components: the polynomial function of incidence angle,the linear function of wind speed,and the wind direction constant.We also specify the speed threshold above which the VH polarization can be used for wind speed retrieval.For the dataset with low-to-medium wind speed(0-17.9 m/s),the proposed algorithm achieves a high retrieval accuracy with root mean square error(RMSE)below 1.26 m/s.(2)Based on the polarimetric covariance matrix and the reflection symmetry property,we calculate the cross-polarization(cross-pol)correlation between VV and VH polarizations.The cross-pol correlation’s sensitivity to factors such as wind speed,relative direction,and incidence angle is investigated.A wind retrieval algorithm using the cross-pol correlation is developed and validated with the medium-to-high wind speed(6-30.9 m/s)dataset.The experiments illustrate a satisfied retrieval performance with RMSE below 1.70 m/s for low incidence angle cases.(3)Based on the statistical modeling of sea clutter from VV and VH polarization,the sensitivity of statistical distribution features of sea clutter to wind vectors is investigated.For the dataset with low-to-medium wind speed,the homogeneous-degree parameter from the Generalized Gamma Distribution(GΓD)is proposed to be a proxy of wind speed.For the dataset with medium-to-high wind speed,the shape parameter from the K-distribution(KD)is recognized to characterize wind speed.This study bridges the gap between the physical interpretation of SAR speckle and the application of sea surface wind speed.(4)We construct the Open SARShip,an open-access dataset dedicated to ship interpretation under low wind speed conditions near coastal regions.The Open SARShip contains SAR ship sub-images integrated with automatic identification system(AIS)messages.The objective of Open SARShip is to provide a benchmark dataset for developing adaptive and robust ship interpretation algorithms.(5)An algorithm is designed to predict the ship detectability from SAR image under low wind speed condition.Based on the SAR wind retrieval algorithm and the ship detection algorithm,an empirical function is developed to estimate the minimum detectable ship length for a specific SAR acquisition under particular wind condition.The minimum detectable ship length assesses the ship detectability of SAR images.It is a practical way to apply sea surface wind speed to marine object detections.(6)The SAR-retrieved wind speed is applied to characterize the roll wavelength under extreme weather condition.We observe that the high wind speed inside the tropical cyclone dominates both the roll wavelength and the roll spatial distribution.This study can be applied to the typhoon forecast algorithm and plays an essential role in monitor coastal regions of our nation.