Super-resolution reconstruction of subsurface temperature field in South China Sea using satellite observations DOI
Hao Wu, Zijian Zhao, Dan Zhang

и другие.

Physics of Fluids, Год журнала: 2025, Номер 37(3)

Опубликована: Март 1, 2025

The accurate reconstruction of high-resolution sea subsurface temperature structures is essential for comprehending meteorological models and evaluating climate change impacts. However, the diversity environment complex physical processes make it challenging to directly reconstruct data from low-resolution satellite observations at once with high accuracy. This study proposes an indirect two-phase transformer-based model achieve super-resolution temperatures a 1/12° resolution, utilizing 1/4° resolution South China Sea. method decouples task reduce complexity optimization, leading more result. In first phase, inverts depth profiles. second performs based on inverted in Experiments are conducted using Copernicus Marine Environment Monitoring Service dataset, performance proposed compared against Attention U-net, Very Deep Super-Resolution, Super-Resolution Convolutional Neural Network. results indicate superior model, achieving root mean square error 0.3524 °C, structural similarity index 0.9854, peak signal-to-noise ratio 42.5031 27-layer layer profile, covering depths 0 200 m. demonstrates model's effectiveness enhancing data, which crucial improving understanding marine environments dynamics.

Язык: Английский

Super-resolution reconstruction of subsurface temperature field in South China Sea using satellite observations DOI
Hao Wu, Zijian Zhao, Dan Zhang

и другие.

Physics of Fluids, Год журнала: 2025, Номер 37(3)

Опубликована: Март 1, 2025

The accurate reconstruction of high-resolution sea subsurface temperature structures is essential for comprehending meteorological models and evaluating climate change impacts. However, the diversity environment complex physical processes make it challenging to directly reconstruct data from low-resolution satellite observations at once with high accuracy. This study proposes an indirect two-phase transformer-based model achieve super-resolution temperatures a 1/12° resolution, utilizing 1/4° resolution South China Sea. method decouples task reduce complexity optimization, leading more result. In first phase, inverts depth profiles. second performs based on inverted in Experiments are conducted using Copernicus Marine Environment Monitoring Service dataset, performance proposed compared against Attention U-net, Very Deep Super-Resolution, Super-Resolution Convolutional Neural Network. results indicate superior model, achieving root mean square error 0.3524 °C, structural similarity index 0.9854, peak signal-to-noise ratio 42.5031 27-layer layer profile, covering depths 0 200 m. demonstrates model's effectiveness enhancing data, which crucial improving understanding marine environments dynamics.

Язык: Английский

Процитировано

0