TY - JOUR
T1 - Sea surface wind speed retrieval based on ICESat-2 ocean signal vertical distribution
AU - Xu, Jinghong
AU - Liu, Qun
AU - Liu, Chong
AU - Chen, Yatong
AU - Xu, Peituo
AU - Ma, Yue
AU - Chen, Yifu
AU - Zhou, Yudi
AU - Zhang, Han
AU - Sun, Wenbo
AU - Yang, Suhui
AU - Lv, Weige
AU - Wu, Lan
AU - Liu, Dong
N1 - Publisher Copyright:
© 2025
PY - 2025/5/1
Y1 - 2025/5/1
N2 - Accurate retrieval of sea surface wind speed is crucial for ecological research and marine resource development. The advent of satellite technology provides a feasible approach for global wind speed retrieval. As a photon-counting lidar, ICESat-2 provides unparalleled details of the sea surface and has the potential for sea surface wind speed retrieval. To facilitate the retrieval of sea surface wind speed from ICESat-2, a vertical ocean signal distribution model of ICESat-2 was established, and then training samples were collected by changing the parameters and inputted into the back propagation neural network to fit the relationship between the ICESat-2 vertical distribution signal and the sea surface wind speed. The model considered both environmental factors (solar noise, atmospheric absorption, sea surface reflection, water backscattering, etc.) and hardware characteristics (the spatial and temporal distribution of laser energy, dead time, and dark noise of the detectors, etc.). The validation against MERRA-2 data revealed that the RMSE is 1.57 m/s for nighttime and 1.89 m/s for daytime, while buoy comparisons showed RMSE values of 1.53 m/s for nighttime and 1.82 m/s for daytime. Additionally, comparisons of global monthly mean results also agree well, underscoring the capability of ICESat-2 in sea surface wind speed retrieval.
AB - Accurate retrieval of sea surface wind speed is crucial for ecological research and marine resource development. The advent of satellite technology provides a feasible approach for global wind speed retrieval. As a photon-counting lidar, ICESat-2 provides unparalleled details of the sea surface and has the potential for sea surface wind speed retrieval. To facilitate the retrieval of sea surface wind speed from ICESat-2, a vertical ocean signal distribution model of ICESat-2 was established, and then training samples were collected by changing the parameters and inputted into the back propagation neural network to fit the relationship between the ICESat-2 vertical distribution signal and the sea surface wind speed. The model considered both environmental factors (solar noise, atmospheric absorption, sea surface reflection, water backscattering, etc.) and hardware characteristics (the spatial and temporal distribution of laser energy, dead time, and dark noise of the detectors, etc.). The validation against MERRA-2 data revealed that the RMSE is 1.57 m/s for nighttime and 1.89 m/s for daytime, while buoy comparisons showed RMSE values of 1.53 m/s for nighttime and 1.82 m/s for daytime. Additionally, comparisons of global monthly mean results also agree well, underscoring the capability of ICESat-2 in sea surface wind speed retrieval.
KW - Altimetry
KW - BP neural network
KW - ICESat-2 ocean detection model
KW - Photon-counting lidar
KW - Radiometric characteristics
KW - Sea surface wind speed retrieval
KW - Significant wave height
UR - http://www.scopus.com/inward/record.url?scp=85218862721&partnerID=8YFLogxK
U2 - 10.1016/j.rse.2025.114686
DO - 10.1016/j.rse.2025.114686
M3 - Article
AN - SCOPUS:85218862721
SN - 0034-4257
VL - 321
JO - Remote Sensing of Environment
JF - Remote Sensing of Environment
M1 - 114686
ER -