Wave retrieval from quad-polarized Chinese Gaofen-3 SAR image using an improved tilt modulation transfer function

Self Cite

This study introduces a new machine learning-based algorithm for the retrieving significant wave height (SWH) using synthetic aperture radar (SAR) images. This algorithm is based on the azimuthal cut-off wavelength and was developed in quad-polarized stripmap (QPS) mode in coastal waters. The collected images are collocated with a wave simulation from the numeric model, called WAVEWATCH-III (WW3), and the current speed from the HYbrid Coordinate Ocean Model (HYCOM). The sea surface wind is retrieved from the image at the vertical–vertical polarization channel, using the geophysical model function (GMF) CSARMOD-GF. The results of the algorithm were validated against the measurements obtained from the Haiyang-2B (HY-2B) scatterometer, yielding a root mean squared error (RMSE) of 1.99 m/s with a 0.82 correlation (COR) and 0.27 scatter index of wind speed. It was found that the SWH depends on the wind speed and azimuthal cut-off wavelength. However, the current speed has less of an influence on azimuthal cut-off wavelength. Following this rationale, four widely known machine learning methods were employed that take the SAR-derived azimuthal cut-off wavelength, wind speed, and radar incidence angle as inputs and then output the SWH. The validation result shows that the SAR-derived SWH by eXtreme Gradient Boosting (XGBoost) against the HY-2B altimeter products has a 0.34 m RMSE with a 0.97 COR and a 0.07 bias, which is better than the results obtained using an existing algorithm (i.e., a 1.10 m RMSE with a 0.77 COR and a 0.44 bias) and the other three machine learning methods (i.e., a >0.58 m RMSE with a <0.95 COR), i.e., convolutional neural networks (CNNs), Support Vector Regression (SVR) and the ridge regression model (RR). As a result, XGBoost is a highly efficient approach for GF-3 wave retrieval at the regular sea state.

Section: Auxiliary Datamentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A Technique for SAR Significant Wave Height Retrieval Using Azimuthal Cut-Off Wavelength Based on Machine Learning

Leng,

Hao,

Shao

et al. 2024

Self Cite

“…Traditionally, two methods are used to invert waves from an SAR intensity image: theoretically based algorithms and empirical models. The methodology of the theoretically based algorithm is exploited through the SAR wave mapping mechanism [21], e.g., the Max-Planck-Institut (MPI) algorithm [22], the semi-parametric retrieval algorithm (SPRA) [23], the partition rescaling and shift algorithm (PARSA) [24], the parameterized first-guess spectrum method (PFSM) (e.g., [25,26]) and the quad-polarized technique [27]. In order to deal with the complex modulation transfer function (MTF) of the SAR mapping mechanism, empirical models have also been developed, e.g., CWAVE_ERS for ERS-1/2 [28], CWAVE_ENVI for ENVISAT-ASAR [29], CWAVE_S1 for S-1 [30,31] and CSAR_WAVE for GF-3 [32].…”

Section: Introductionmentioning

confidence: 99%

Validation of Surface Waves Investigation and Monitoring Data against Simulation by Simulating Waves Nearshore and Wave Retrieval from Gaofen-3 Synthetic Aperture Radar Image

Hao,

Shao,

Shi

et al. 2023

Self Cite

The Chinese-French Oceanography SATellite (CFOSAT) jointly developed by the Chinese National Space Agency (CNSA) and the Centre National d’Etudes Spatiales (CNES) of France carries a wave spectrometer (Surface Waves Investigation and Monitoring, SWIM). SWIM has one nadir and five off-nadir beams to measure ocean surface waves. These near-nadir beams range from 0° to 10° at an interval of 2°. In this work, we investigated the performance of wave parameters derived from wave spectra measured by SWIM at off-nadir beams during the period 2020 to December 2022, e.g., incidence angles of 6°, 8° and 10°, which were collocated with the wave simulated by Simulating Waves Nearshore (SWAN). The validation of SWAN-simulated significant wave heights (SWHs) against National Data Buoy Center (NDBC) buoys of National Oceanic and Atmospheric Administration (NOAA) exhibited a 0.42 m root mean square error (RMSE) in the SWH. Our results revealed a RMSE of 1.02 m for the SWIM-measured SWH in the East Pacific Ocean compared with the SWH simulated by SWAN, as well as a 0.79 correlation coefficient (Cor) and a 1.17 squared error (Err) for the wave spectrum at an incidence angle of 10°, which are better than those (i.e., the RMSEs were > 1.1 m with Cors < 0.76 and Errs > 1.2) achieved at other incidence angles of SWH up to 14 m. This analysis indicates that the SWIM product is a relevant resource for wave monitoring over global seas. The collocated wave retrievals for more than 300 cases from Gaofen-3 (GF-3) synthetic aperture radar (SAR) images in China Seas were also used to verify the accuracy of SWIM-measured wave spectra. The energy of the SWIM-measured wave spectra represented by SWH was found to decrease with an increasing incidence angle in a case study. Moreover, the SWIM-measured wave spectra were most consistent with the SAR-derived wave spectra at an incidence angle of 10°, yielding a 0.77 Cor and 1.98 Err between SAR-derived and SWIM wave spectra under regular sea state conditions (SWH < 2 m). The error analysis indicates that the difference in SWH between SWIM at an incidence angle of 10° and SWAN has an increasing tendency with the growth in sea surface wind and sea state and it stabilizes to be 0.6 m at SWH > 4 m; however, the current and sea level have less influence on the uncertainties of the SWIM product.

“…The spaceborne synthetic aperture radar (SAR) has the capability to acquire backscatter from the ocean surface in all-weather conditions [1,2]. This imagery has been widely used to detect ocean dynamic phenomena, such as wind vectors [3,4], surface waves [5][6][7][8], surface currents [9,10], and internal waves [11,12]. In general, ocean surface roughness is positively dependent upon wind speed.…”

Section: Introductionmentioning

confidence: 99%

A New Approach for Ocean Surface Wind Speed Retrieval Using Sentinel-1 Dual-Polarized Imagery

Gao,

Wang,

Wang

2023

A synthetic aperture radar (SAR) has the capability to observe ocean surface winds with a high spatial resolution, even under extreme conditions. The purpose of this work was to develop a new method for wind speed retrieval with the combination of SAR dual-polarized signals. In this study, we collected 28 tropical cyclone imageries observed using the Sentinel-1 dual-polarization mode. These imageries were collocated with radiometer wind speed measurements and reanalysis of wind vector products. In the new method, the wind speed was set as the output. VV-polarized (vertical transmitting–vertical receiving polarized) normalized radar cross section (NRCS), incident angle, VH-polarized (vertical transmitting–horizontal receiving polarized) NRCS, and wind direction were set as the inputs. Based on different output combinations, wind retrieval models were developed with multiple linear regression (MLR). According to the validation and comparison, the proposed models performed better than the traditional piecewise VH-polarization geophysical model functions (GMFs). The impact of thermal noise on the retrieval of low wind speeds (<10 m/s) could be partially reduced. The input of wind direction is unnecessary if the combination of VV- and VH-polarized imageries has been utilized. These results suggest that the use of MLR and the dual-polarization combination can improve SAR wind retrieval accuracy. Compared with SMAP measurements, our SAR retrievals can provide fine structures of TC wind fields.