AIGan-based Eddy-resolving Reconstruction of Subsurface Temperature and Salinity in the South China Sea
-
Graphical Abstract
-
Abstract
The inversion of ocean subsurface temperature and salinity (TS) is a hot and challenging problem in the ocean. In this study, a new method for inversion of underwater TS based on an improved GAN model is proposed in the South China Sea. The proposed model can derive the underwater TS with an eddy-resolving horizontal resolution of 1/12° from the sea surface information. For comparison, a robust statistics-based model MODAS is also carried out to invert the subsurface TS in this study. The results show that the RMSE of TS inversions from GAN-based model are significantly smaller than those from MODAS especially in the thermocline, where the RMSE of temperature can be reduced by up to 21.7%, and the subsurface salinity RMSE is smaller than 0.32. Especially, the inversion results via the proposed model perform more accurate in either the season-scale or the synoptic-scale analysis. Firstly, GAN-based model is more effective for the season-scale extraction and diagnosis of the subsurface stratification, especially in the near Luzon Strait and coastal shelf sea areas. Secondly, effects of vertical heat pump and cold suction in the ocean upper layers induced by the passage of Typhoon can be reflected more reasonable based on the synoptic-scale analysis with the proposed model. Further, the underwater 3D structure of mesoscale eddies can be well captured by AIGAN, which can extract finer eddy patterns compared with MODAS. The present study can be extended to further explore the subsurface characteristics of the internal variability in the South China Sea.
-
-