Ocean Wind Speed Retrieval From FY-3E GNSS-R Data Using ResAtt-Net With Fine-Tuning: A Method Optimized for High Wind Speeds
- 27 November 2025
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 23 (1545598X) , 1-5
Abstract
Global Navigation Satellite System Reflectometry (GNSS-R), as a novel retrieval technique, can be effectively used for the retrieval of ocean wind speed. However, existing wind speed retrieval methods have not fully solved the problems of inaccurate reference values and sparse samples in high wind speed areas, resulting in insufficient accuracy in high wind speed retrieval. To tackle this issue, this letter proposes a multimodal wind speed retrieval network based on ResAtt-Net, which further improves the retrieval accuracy in high wind speed areas by combining fine-tuning strategies and wind speed fusion strategies. Experiments based on Fengyun-3E (FY-3E) GNSS Occultation Sounder II (GNOS-II) level 1 data at different wind speed ranges show that our model exhibits superiority in both overall and high wind speed retrieval, especially in the high wind speed regime (>17 m/s), where the retrieval error is significantly reduced. The root-mean-square error (RMSE) decreases from 4.32 to 2.65 m/s, and the correlation coefficient increases to 0.897. This method provides an efficient and accurate solution for ocean high wind speed remote sensing retrieval tasks based on FY-3E data, and has broad application prospects.Keywords
Funding Information
- Beijing Municipal Education Commission Research Plan General Project (KM202411232014)
- Young Backbone Teacher Support Plan of Beijing Information Science and Technology University (YBT 202424)
- Anhui Provincial Natural Science Foundation (2208085UQ01)
This publication has 20 references indexed in Scilit:
- DDM-Former: Transformer networks for GNSS reflectometry global ocean wind speed estimationRemote Sensing of Environment, 2023
- Information fusion for GNSS-R wind speed retrieval using statistically modified convolutional neural networkRemote Sensing of Environment, 2022
- GNSS reflectometry global ocean wind speed using deep learning: Development and assessment of CyGNSSnetRemote Sensing of Environment, 2022
- TDS-1 GNSS reflectometry wind geophysical model function response to GPS block typesGeo-spatial Information Science, 2022
- Analysis of coastal wind speed retrieval from CYGNSS mission using artificial neural networkRemote Sensing of Environment, 2021
- The ERA5 global reanalysisQuarterly Journal of the Royal Meteorological Society, 2020
- Wind Speed Estimation From CYGNSS Using Artificial Neural NetworksIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2020
- Development of the CYGNSS Geophysical Model Function for Wind SpeedIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2018
- The GNSS Reflectometry Response to the Ocean Surface Winds and WavesIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016
- Deep Residual Learning for Image Recognition2016