Scholarlink

<p>The China-France Oceanography Satellite (CFOSAT), equipped with the wind scatterometer (SCAT) and the surface waves investigation and monitoring (SWIM) instrument, achieves the first simultaneous global observation of ocean surface wind fields and wave spectra. SCAT retrieves wind fields based on the Bragg scattering mechanism, while SWIM derives wave spectrum parameters near nadir through quasi-specular reflection. However, SCAT’s backscatter signal is often disrupted by large-wave slope effects, and SWIM measurements can be influenced by small-scale wave interference under strong wind conditions. To address these limitations, this study proposes a novel dual-instrument joint inversion method. The proposed approach leverages the XGBoost algorithm for feature importance evaluation and selection, and develops a hybrid convolutional neural network (CNN)-long short-term memory (LSTM) deep learning model to achieve the simultaneous retrieval of wave and wind parameters. Validation against European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis shows root mean square errors (RMSEs) of 26.28°(wind direction), 0.94 m/s (wind speed), 0.22 m (significant wave height), and 0.58 s (mean wave period). Further comparisons with the National Data Buoy Center (NDBC) buoy data revealed RMSEs of 1.21 m/s for wind speed and 0.37 m for significant wave height. Furthermore, the joint inversion model demonstrates superior accuracy over single-instrument retrievals, reducing RMSE by approximately 37% for significant wave height compared to SWIM Level-2 products and by 46% for wind speed compared to SCAT MLE retrievals. It also outperforms standalone CNN and LSTM models consistently. The results consistently demonstrated the superiority of the proposed joint inversion model.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A novel joint inversion method for ocean wave and wind fields using CFOSAT SWIM and SCAT data

  • 期刊论文

  • 发表日期：

    2026年7月4日

查看全文
  • Aiying Wu,
  • Yong Wan,
  • Ligang Li,
  • Xiangying Miao,
  • Yuan Zhu

摘要

The China-France Oceanography Satellite (CFOSAT), equipped with the wind scatterometer (SCAT) and the surface waves investigation and monitoring (SWIM) instrument, achieves the first simultaneous global observation of ocean surface wind fields and wave spectra. SCAT retrieves wind fields based on the Bragg scattering mechanism, while SWIM derives wave spectrum parameters near nadir through quasi-specular reflection. However, SCAT’s backscatter signal is often disrupted by large-wave slope effects, and SWIM measurements can be influenced by small-scale wave interference under strong wind conditions. To address these limitations, this study proposes a novel dual-instrument joint inversion method. The proposed approach leverages the XGBoost algorithm for feature importance evaluation and selection, and develops a hybrid convolutional neural network (CNN)-long short-term memory (LSTM) deep learning model to achieve the simultaneous retrieval of wave and wind parameters. Validation against European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis shows root mean square errors (RMSEs) of 26.28°(wind direction), 0.94 m/s (wind speed), 0.22 m (significant wave height), and 0.58 s (mean wave period). Further comparisons with the National Data Buoy Center (NDBC) buoy data revealed RMSEs of 1.21 m/s for wind speed and 0.37 m for significant wave height. Furthermore, the joint inversion model demonstrates superior accuracy over single-instrument retrievals, reducing RMSE by approximately 37% for significant wave height compared to SWIM Level-2 products and by 46% for wind speed compared to SCAT MLE retrievals. It also outperforms standalone CNN and LSTM models consistently. The results consistently demonstrated the superiority of the proposed joint inversion model.