An interpreter in a conference room hears not just words but intonations, pauses, and gestures. From these cues, they anticipate where the conversation will turn. The Pacific climate speaks the language of wind, heat, and currents, and PTSTnet listens like an experienced simultaneous interpreter, grasping the grammar of ocean and atmosphere.
The model looks at 12 months of global data: sea surface temperature, upper-ocean heat content, and winds. Then it builds a forecast step by step for the next 24 months, using its own previous outputs as context. Physical laws of waves and diffusion are embedded in the architecture itself, so the network does not merely memorize statistics but learns real dynamics.
When the model errs, its errors propagate from the western equatorial Pacific toward the eastern part. This aligns with what happens in nature: strengthening trade winds and upwelling off the coast of South America. PTSTnet accurately reproduced the super El Niño of 2015–2016 and mitigated the spring predictability barrier, when accuracy typically drops sharply.
Now operational climate forecast centers can receive warnings about El Niño two years in advance. Agriculture, water resources, and the insurance sector, which depend on seasonal forecasts, will gain a more reliable tool. And researchers get a transparent model to test hypotheses about ENSO mechanisms. The interpreter who understands the grammar of the ocean has finally started talking about the future.
The model was able to predict the 2015–2016 super El Niño, initializing as early as February 2015, when most dynamical models gave large uncertainty.
In Kim Stanley Robinson's novel 'Forty Signs of Rain', scientists struggle for accurate prediction of climate shifts; PTSTnet resembles a step toward such a system, where physics and data merge into a transparent tool.