CoAuthors

Jingsong Yang (State Key Laboratory of Satellite Ocean Environment Dynamics, Second Institute of Oceanography, Ministry of Natural Resources, China); Jiuke Wang (National Marine Environmental Forecasting Center, China); Guoqi Han (Fisheries and Oceans Canada, Institute of Ocean Sciences, Canada); Lin Ren (State Key Laboratory of Satellite Ocean Environment Dynamics, Second Institute of Oceanography, Ministry of Natural Resources, China)

The scatterometer onboard China-France Oceanic SATellite (CFOSAT) can provide simultaneous observations of the wide swath wind fields. However, it has limitations in high wind speed retrieval due to the complexity of remote sensing mechanism and the influence of rainfall on radar cross section under the condition of tropical cyclones. This paper evaluates the performance of Chinese scatterometer operational wind products from CFOSAT over a period of 2019 to 2021 with respect to SMAP (Soil Moisture Active Passive) L-band radiometer remotely sensed measurements in Tropical Cyclones. The temporal and spatial differences between the CFOSAT and in situ measurements were limited to less than 1 hour and 12.5 km. The bias, RMSE and correlation coefficient in tropical cyclones are 1.74 m/s, 4.06 m/s and 0.83, which show that the underestimation of high wind speed occurs in the CFOSAT wind speed products. The machine learning algorithm is explored to improve the performance of tropical cyclone winds from CFOSAT scatterometer. Comparisons show that neural network algorithm shows an outstanding performance with a small bias of 0.08 m/s and a small RMSE of 3.50 m/s, demonstrating an improvement of 13.7% in RMSE (root-mean-square error) compared to the CFOSAT wind operational products. The evaluation results show that the performance of CFOSAT wind product under the condition of tropical cyclones is significantly improved by the machine learning method, which are useful for tropical cyclone disaster prevention and mitigation.