How certain are El Nino-Southern Oscillation frequency changes in Coupled Model Intercomparison Project Phase 6 models?

El Nino-Southern Oscillation (ENSO) is one of the most important modes of climate variability on interannual timescales. We aim to find out whether a change in ENSO frequency can be predicted for the nearer future. We analyse the unforced pre-industrial control run and the forced 1%/year CO2 increase run for an ensemble of 43 general circulation models that participated in the Coupled Model Intercomparison Project Phase 6 (CMIP6). We assume that the uncertainty of ENSO frequency trend estimates from an ensemble is caused by apparent trends as well as model differences. The part of the uncertainty caused by apparent trends is estimated from the pre-industrial control simulations. As a measure for ENSO frequency, we use the number of El Nino- and La Nina-like months in a moving 30-year time window. Its linear decadal trend is calculated for every member. The multimember mean of the trend for both experiments is less than 0.7 events per decade. Given that the standard error is of the same order of magnitude, we consider this a negligible trend. The uncertainties are large in both experiments and we can attribute most of the intermember variability to apparent trends due to natural variability rather than different model reactions to CO2 forcing. This means that the impact of intermodel differences might have been overstated in previous studies. Apparent trends make it very difficult to make reliable predictions of changes in ENSO frequency based on 120-year time series.

Automated summary

El Nino-Southern Oscillation (ENSO) is one of the most important modes of climate variability on interannual timescales. This study aims to investigate the predictability of ENSO frequency change in the near future. The analysis of various climate models suggests that the uncertainty in ENSO frequency trends is mainly due to apparent trends and natural variability, rather than different model reactions to CO2 forcing. The study highlights the challenges in making reliable predictions of ENSO frequency changes based on limited time series data.