This work stands apart: its nearest neighbors in the archive deal with predicting El Niño via climate networks, early warning signals in climate, power grid resilience, and the ocean conveyor belt. None of them tackles the problem of determining the El Niño type with a long lead time. Traditional models hit the spring predictability barrier and only yield the type about a month ahead. The authors propose a simple index that works a year in advance.
The researchers used monthly sea surface temperature data (ERSSTv5) from 1950 to the present. They computed temperature anomalies relative to a sliding 30-year climatological mean to account for climate change. Then, for each December, they calculated the difference between the average anomaly in the western equatorial Pacific (120°E–165°E, 5°N–5°S) and that in the central equatorial Pacific (165°E–150°W, 5°N–5°S). The sign of this difference became the precursor: positive indicates an EP El Niño the following year, negative a CP. For validation, they used a consensus classification of types based on 11 independent methods. They also verified that the same precursor works for sea surface height anomalies and surface air temperature anomalies.
From 1950 to 2021, there were 21 El Niño events with an unambiguous type. The December sign of the temperature difference correctly predicted the type in 18 out of 21 cases: all 9 CP events and 9 of the 12 EP events. The three false EP predictions occurred in 1962, 2008, and 2017, when CP events happened instead. The probability of guessing 18 out of 21 correctly by chance, given the base rates of the types, is p = 9.1·10⁻⁴, so the result is highly significant. When combined with a climate network that predicts the occurrence of El Niño itself, the method correctly identified the type for 2/3 of EP forecasts and for all CP forecasts. For comparison, operational models on average achieve only about 50% correct type forecasts at a lead time of one month.
Unlike neighboring studies, which either predict only the onset of El Niño or use complex models with short lead times, this work provides a simple and physically interpretable index for the event type with an unprecedented one-year horizon. The method requires no parameter tuning: a single number in December suffices. This is the first demonstration that the zonal temperature gradient between the western and central tropical Pacific can serve as an early discriminator of El Niño type, and it succeeds across seven decades of data.
The topic could evolve toward clarifying the physical mechanism: why the sign of the gradient is linked to the type. Westerly wind bursts likely play a role, as they interact with the ocean differently depending on the prior state. One could also test whether the precursor remains valid under climate change, as the background state of the Pacific shifts. The index might be improved by adding information on sea surface height or upper-ocean heat content.
Knowing the El Niño type well in advance would allow preparation for specific impacts: floods in Peru and Ecuador for EP events, or droughts in India for CP events. This is important for agriculture, water resource management, and humanitarian planning.
The immediate next step is to test the precursor on independent data, such as paleoreconstructions or forecasts for future decades. It would also be worth comparing it with indices based on other variables and trying to combine it with more sophisticated models for predicting El Niño onset.
This work directly addresses an unsolved problem in climate science: why El Niño diversity arises and whether it can be predicted beyond the spring barrier. The role of stochastic forcings, such as westerly wind bursts, in selecting the event type remains an open question. It is also unclear how the ratio of EP to CP events will change under global warming.
In October 2022, the temperature difference was +1.6°C, pointing to a likely EP El Niño if one began in 2023. And indeed, the 2023–2024 event was classified as eastern Pacific.
In Kim Stanley Robinson's novel 'Forty Signs of Rain', climate scientists try to predict abrupt climate changes, and the accuracy of long-term forecasts becomes a matter of life and death. This work on an early precursor for El Niño type resonates with that theme: a small but reliable gain in lead time can change how we prepare for a disaster.