SST subseasonal-to-seasonal forecasting: A heterogeneous three-path fusion model with wavelet decomposition
Xiao Chen, Yihao Wang, Jing Chen, MingXin Liu · 2026
Read SST subseasonal-to-seasonal forecasting: A heterogeneous three-path fusion model with wavelet decomposition by Xiao Chen, Yihao Wang, Jing Chen, MingXin Liu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0351923.txt.
Description
Xiao Chen and three co-authors present this 2026 PLOS One climate paper. It forecasts Pacific Ocean surface heat weeks to months ahead. That window ties to the El Niño cycle, which shapes weather worldwide. Their model first splits daily heat records into two parts. One part is a smooth, slow trend. The other is a fast-changing remainder. Each part gets a branch suited to it. A learned gate then blends the outputs.
Tested on daily data from five watch regions, the model beat several deep learning rivals. It won at most one-to-three-month ranges. It also won under every El Niño phase, from onset to decay. Simple baselines still won at very short ranges, which the authors admit. They propose the approach as a base for richer ocean forecasts.
Catalog record
| CREATOR | Xiao Chen, Yihao Wang, Jing Chen, MingXin Liu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 37792 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |