WSPTGAN for Global Ocean Surface Wind Speed Generation with High Temporal Resolution and Spatial Coverage

• Published in: IEEE transactions on geoscience and remote sensing 2024-01, Vol.62, p.1-1
• Main Authors: Li, Menglong, Hou, Yonghong, Song, Xiaowei, Hou, Chunping, Wang, Zhipeng, Xiong, Zixiang, Ma, Dan
• Format: Article
• Language: English
• Subjects: Adaptation models; Algorithms; Data models; defective data; Generative adversarial networks; Mathematical models; Measurement methods; Ocean surface; Oceans; Prediction algorithms; Prediction models; Predictive models; Root-mean-square errors; Scatterometers; self-attention; spatial coverage; Spatial data; Surface wind; Temporal resolution; Transformer; Transformers; Weather forecasting; Wind; Wind forecasting; Wind speed
• ISSN: 0196-2892; 1558-0644
• DOI: 10.1109/TGRS.2024.3369640

Obtaining global ocean surface wind speed data with high temporal resolution and spatial coverage is a challenging task. Due to the lack of widely applicable direct measurement methods and algorithms, current research and data products can only achieve good performance in a small spatial range or at low temporal resolution. In this article, a generative adversarial network with Transformer structure called Wind Speed Prediction Transformer-GAN(WSPTGAN) is proposed to generate wind speed data with good spatial coverage and high temporal resolution for areas. The WSPTGAN is trained with the proposed image-like wind speed data combined partial missing dataset, which is combined of fifth generation of the European Center for Medium-Range Weather Forecast reanalysis data and advanced scatterometer data from Meteorological Operational satellites. Thanks to the Defective Data Learning Mechanism, Sequential-wise Multi-head Self-attention Mechanism and Sequence Feature Adaptive Verification Mechanism in the proposed algorithm, the obtained model has good wind speed prediction accuracy with root mean square error of 0.8984 m/s and can achieve multi-step 10-minute wind speed data generation within the global ocean. After comparison with five state-of-the-art prediction models, it is confirmed that the algorithm in this article is able to make better use of the defective data for learning and prediction of wind field trends in global ocean regions.