Evaluation of the Real-time El Niño Forecasts by the Climate Network Approach between 2011 and Present

TL;DR

This paper assesses a climate-network–based method for real-time El Niño forecasting, demonstrating the ability to predict events about a year in advance. By tracking the mean network link strength $S(t)$ and triggering alarms when it crosses a fixed threshold $\Theta$, the authors achieve several correct onsets (e.g., 2013, 2017, 2023) with a single false alarm (2019) and report a highly significant $p$-value ($p\ll 0.01$) against random guessing. The work also introduces statistical refinements and a two-step verification to reduce false alarms, and discusses integration with methods that estimate event magnitude and type. The findings underscore the practical potential for early ENSO mitigation, while noting open questions about extending the approach to La Niña and further reducing false alarms.

Abstract

El Niño episodes are part of the El Niño-Southern Oscillation (ENSO), which is the strongest driver of interannual climate variability, and can trigger extreme weather events and disasters in various parts of the globe. Previously we have described a network approach that allows to forecast El Niño events about 1 year ahead. Here we evaluate the real-time forecasts of this approach between 2011 and 2022. We find that the approach correctly predicted (in 2013 and 2017) the onset of both El Niño periods (2014-2016 and 2018-2019) and generated only 1 false alarm in 2019. In June 2022, the approach correctly forecasted the onset of an El Niño event in 2023. We show how to determine the $p$-value of the 12 real-time forecasts between 2011 and 2022 and find $p\cong 0.005$, this way strongly rejecting the null hypothesis that the same quality of the forecast can be obtained by random guessing. We also discuss how the algorithm can be further improved by reducing the number of false alarms in the network model forecast. When combined with other statistical methods, a more detailed forecast, including the magnitude of the event and its type, can be obtained. For 2024, the method indicates the absence of a new El Niño event.

Figures (3)

• Figure 1: The structure of the climate network. Each of the 14 grid points in the "El Niño basin" (red circles) is linked to each of the 193 grid points outside this domain (blue circles). The green rectangle denotes the Niño3.4 region.
• Figure 2: The forecasting scheme. We compare the average link strength $S(t)$ in the climate network (red curve) with the decision threshold $\Theta=2.82$ (horizontal line) and the ONI (right scale), between January 1981 and December 2011. When the link strength crosses the threshold from below and the last available ONI is below 0.5° C, we give an alarm and predict that a new El Niño episode will start in the following calendar year. Periods with an ONI greater or equal 0.5° C are displayed in blue. The El Niño episodes (when the ONI is greater or equal 0.5° C for at least 5 months) are displayed in dark blue. Correct predictions are marked by green arrows and false alarms by dashed arrows. Note that the early false alarms in February 1994 and July 2004 are followed by at least one ONI value equal or above 0.5° C in the same year.
• Figure 3: The real-time forecasts. Same as Fig. 2, but for the period between January 2011 and December 2023. As in Fig. 2, the false alarm (in 2019) is followed by at least one ONI value equal to or above 0.5° C in the same year. Only alarms until 2022, where the outcome is known, are marked by arrows.