We need a thread to discuss future trends in the ENSO cycle
e,g, this new paper,,,
https://www.nature.com/articles/s41586-023-06236-9Increased occurrences of consecutive La Niña events under global warming
AbstractMost El Niño events occur sporadically and peak in a single winter1,2,3, whereas La Niña tends to develop after an El Niño and last for two years or longer4,5,6,7. Relative to single-year La Niña, consecutive La Niña features meridionally broader easterly winds and hence a slower heat recharge of the equatorial Pacific6,7, enabling the cold anomalies to persist, exerting prolonged impacts on global climate, ecosystems and agriculture8,9,10,11,12,13.
Future changes to multi-year-long La Niña events remain unknown. Here, using climate models under future greenhouse-gas forcings14, we find an increased frequency of consecutive La Niña ranging from 19 ± 11% in a low-emission scenario to 33 ± 13% in a high-emission scenario, supported by an inter-model consensus stronger in higher-emission scenarios. Under greenhouse warming, a mean-state warming maximum in the subtropical northeastern Pacific enhances the regional thermodynamic response to perturbations, generating anomalous easterlies that are further northward than in the twentieth century in response to El Niño warm anomalies. The sensitivity of the northward-broadened anomaly pattern is further increased by a warming maximum in the equatorial eastern Pacific. The slower heat recharge associated with the northward-broadened easterly anomalies facilitates the cold anomalies of the first-year La Niña to persist into a second-year La Niña.
Thus, climate extremes as seen during historical consecutive La Niña episodes probably occur more frequently in the twenty-first century.Observed features and model selection
a, Skewness of historical (1900–1999) Niño3.4 SST anomaly in observation (black bar) and CMIP6 models (coloured bars). The vertical line separates selected models with positive skewness (orange bars) from non-selected models with negative skewness (blue bars). The error bar denotes 1.0 s.d. of the inter-model spread in the selected (non-selected) MME.
b, Temporal evolution of Niño3.4 SST anomaly composited for multi-year (red) and single-year (blue) La Niña events in the selected models over 1900–1999. Solid lines and shading indicate multi-model mean and 1.0 s.d. of a total of 10,000 inter-realizations based on a bootstrap method, respectively. Dashed lines indicate observations. The time series are smoothed with a three-month running-mean filter before analysis. The vertical grey shading denotes the time (October to February) when ENSO typically matures.
c,d, Multi-model mean composite map of anomalous SST (°C; colouring) and surface wind stress (N m−2; vectors) for single-year (c) and multi-year (d) La Niña events during D(1)JF(2) in 1900–1999. Shown are values at which the ensemble mean exceeds 1.0 s.d. of the inter-model spread using a bootstrap method. Selected models simulate reasonably the observed evolution and pattern of multi-year La Niña.
Fig. 2: Projected increase in frequency of multi-year La Niña events.
a, Comparison of multi-year La Niña numbers (events per 100 years) over 1900–1999 (blue bars) and 2000–2099 (red bars) in the selected models under SSP585 (left of the vertical line). Multi-model mean results from other emission scenarios are also provided for the selected ensembles. Models that simulate a decrease are greyed out. Shown in the last four columns are the MME results of non-selected models under SSP585 and of selected models under low-emission scenarios. Note that not exactly the same set of models is used under different scenarios owing to data unavailability. The horizontal dashed line indicates observation.
b, Evolution of multi-year La Niña occurrence (events per 100 years) diagnosed in a 60-year sliding window that moves separately in the past 500 years of piControl (black) and from 1850 (the start of historical run; blue) to the end of the twenty-first century under SSP585 (red). Years on the x axis denote the end year of the sliding window. Solid lines and shading indicate multi-model mean and 95% confidence intervals based on a Poisson distribution, respectively. The dashed black line indicates the mean level of piControl.
c, As in a but for proportions (as a percentage) of multi-year La Niña occurrences in different situations under SSP585 (see letters on the x axis and corresponding descriptions at the bottom). Error bars on the multi-model mean in a and c are calculated as 1.0 s.d. of 10,000 inter-realizations of a bootstrap method. Disproportionally more frequent multi-year La Niña events occur after a strong El Niño during the 2000–2099 period than during the 1900–1999 period.
Summary and discussionOur finding of an increase in the occurrence of consecutive La Niña events under greenhouse warming is underpinned by northward-broadened easterly anomalies in the subtropical North Pacific in response to equatorial eastern Pacific warm anomalies. The northward broadening and its increased occurrences are—in turn—a consequence of a faster mean-state warming in the subtropical northeastern Pacific that induces a further northern and more sensitive response to El Niño convective anomalies, which are—per se—intensified by a faster warming in the equatorial eastern Pacific. The consequence of the northward-broadened easterlies is a slower heat recharge of the equatorial Pacific, leaving a colder upper-ocean condition after the first-year La Niña to persist into the second year. Our discovery of a two-way interaction between the tropics and subtropics that intensifies under greenhouse warming represents an advance beyond recent findings of a one-way warming-induced enhancement of the NPMM influence on ENSO50,51.
Our result of a probable future increase in multi-year La Niña frequency strengthens calls for an urgent need to reduce greenhouse-gas emissions to alleviate the adverse impacts.
