This study proposes a deep learning (DL) approach for retrieving high wind speeds during tropical cyclones using a random forest algorithm applied to RADARSAT and Sentinel-1A/B Synthetic Aperture radar (SAR) images. The effectiveness of the proposed DL-based model is then demonstrated through a comprehensive validation of the results. Statistical analysis of the results showed that the proposed method performs well, with a low root-mean-square error, mean bias, and high correlation coefficient when compared to SFMR (Stepped-Frequency Microwave Radiometer) winds. The reconstructed wind speeds and inner-core structures were found to be in good agreement with surface wind measurements from SFMR. These findings could have significant implications for improving our understanding and prediction of tropical cyclone dynamics, as well as for operational forecasting and disaster management.

Synthetic-aperture radar (SAR) plays a crucial role in monitoring the fine structure of tropical cyclones, but its effectiveness is constrained by limitations such as signal degradation and saturation. To address this challenge, we proposed a transfer learning-based generative adversarial network (GAN) framework with a dilated convolution and attention mechanism for reconstructing inner-core high winds from SAR images. We have employed the principles of transfer learning to adapt pre-trained models developed by the HWRF (Hurricane Weather Research and Forecasting model) winds to SAR images during tropical cyclone events for reconstruction. The proposed model can effectively capture the relationship between features in the low-precision areas and global features from SAR images, facilitating tropical cyclone wind speed reconstruction. The utilization of Global Precipitation Measurement (GPM) Level 3 rainfall data facilitates the identification of rainfall regions in 89 SAR images obtained from Radarsat-2 and Sentinel-1A/B missions. Comparison with Stepped Frequency Microwave Radiometer (SFMR) data reveals that the model exhibits a bias of –0.69 m/s, an RMSE of 4.08 m/s, and anRvalue of 0.91 under heavy rainfall conditions (>7.62 mm/hr). Remarkably, the GAN model exhibits excellent performance compared with measurements from the Soil Moisture Active Passive (SMAP) L-band radiometer, achieving an RMSE of 3.78 m/s. Our findings indicate that deep learning technology holds significant promise for the reconstruction and monitoring of tropical cyclones through the utilization of SAR imagery.

In this study, we propose a method for wind speed retrieval using a random forest (RF) algorithm for Cyclone Global Navigation Satellite System (CYGNSS) data. We first compared CYGNSS data with Soil Moisture Active Passive (SMAP) data and found a certain deviation in the CYGNSS ”young sea, limited fetch” (YSLF) data product for high winds. Then, we used SMAP as the ”ground truth” to train an RF model and applied it to the wind speed retrieval of CYGNSS data. The experimental results show that using the RF algorithm for wind speed retrieval can eliminate noise in the CYGNSS YSLF wind speed data and improve retrieval accuracy. In addition, we explored the impact of different input parameter combinations on model performance and found that using an 11-parameter model in CYGNSS wind speed retrieval can achieve optimal performance. This can provide valuable reference for rapid near-real-time retrieval of tropical cyclones using CYGNSS.

The surface waves investigation and monitoring (SWIM) instrument onboard the China–France Oceanography Satellite (CFOSAT) can retrieve directional wave spectra with a wavelength range of 70–500 m. This study aims to validate the partitioned integrated wave parameters (PIWPs) from SWIM, including partitioned significant wave height (PSWH), partitioned peak wave period (PPWP), and partitioned peak wave direction (PPWD), against those from National Data Buoy Center (NDBC) buoys. With quasi-simultaneous spectra from two NDBC buoys 13 km away from each other near Hawaii, the methods of comparing PIWPs from two sets of spectra were discussed first. After cross-assigning partitions according to the spectral distance, it is found that wrong cross-assignments lead to many outliers strongly impacting the estimate of error metrics. Three methods, namely comparing only the best-matched partition, changing the threshold of spectral distance during cross-assignment, and maximum likelihood estimation of root-mean-square error (RMSE) of PIWPs, were used to reduce the impact of potential wrong cross-assignments. Using these methods, the SWIM PIWPs were validated against NDBC buoys. The results show that SWIM performs well at finding the spectral peaks of different partitions with the RMSE of PPWPs and PPWDs of 0.9 s and 20°, respectively, which can be a useful complement for other wave observations. However, the accuracy of PSWH from SWIM is not that good at this stage, probably because the high noise level in the spectra impacts the result of the partitioning algorithm. Further improvement is needed to obtain better PSWH information.

The potential mymargin for improving storm surge simulation is demonstrated by using winds derived from ground-based Global Navigation Satellite System Reflectometry (GNSS-R) that uses BeiDou geostationary Earth orbit (GEO) satellite signals. We reconstruct wind fields by blending GNSS-R coastal winds with the European Center for Median Weather Forecasts (ECMWF) reanalysis product. The reconstructed winds agree well with the weather station data collected at Yangjiang in Guangdong, China. The ECMWF winds and the reconstructed winds are used to force a storm surge model off the Chinese coast during typhoon Utor 2013, respectively. The model storm surges forced by the reconstructed winds agree substantially better with tide-gauge observations than those forced by the ECMWF winds. The average error has been reduced by 30.5% from 24.3 cm with the ECMWF winds to 16.9 cm with the reconstructed winds. This letter suggests that GNSS-R coastal winds can have a positive impact on the accuracy of storm surge hindcasting directly and forecasting indirectly by improving the initial conditions.