Central-Pacific El Niño-Southern Oscillation less predictable under greenhouse warming

El Niño-Southern Oscillation (ENSO) is the dominant mode of interannual climate variability in the tropical Pacific, significantly impacting global weather patterns, ecosystems, and agriculture. ENSO events are categorized into Eastern Pacific (EP) and Central Pacific (CP) types, distinguished by their spatial patterns and climate impacts. While numerous studies have examined the effects of climate change on ENSO characteristics, such as intensifying sea surface temperature anomalies (SSTAs), less attention has been given to the potential changes in ENSO predictability. A few studies have indicated a decrease in ENSO predictability in recent decades, but the role of global warming remains unclear due to short observational periods and potential influence of multi-decadal climate variability. ENSO predictability depends on factors beyond SSTA persistence, including the cross-correlation between SSTA and subsurface ocean heat content. Extratropical processes, such as the North Pacific Meridional Mode (NPMM), and inter-basin interactions from the Atlantic and Indian Oceans, can also modulate ENSO dynamics and predictability. Furthermore, the spring predictability barrier (SPB), a sharp decline in ENSO prediction skill during boreal spring, is a crucial yet understudied aspect in the context of climate change. This study aims to address the fundamental question of how greenhouse warming will affect ENSO predictability, focusing on the changes in the predictability of both CP and EP ENSO and the strength of the SPB under future warming scenarios. The research utilizes the latest CMIP6 climate models to analyze ENSO persistence and predictability, employing both Linear Inverse Model (LIM) and a two-box recharge oscillation model (ROM) to gain a comprehensive understanding of the underlying mechanisms.

Prior research has extensively explored the effects of future climate change on ENSO characteristics. A consistent finding is the projected intensification of ENSO's magnitude in the next century under various emission scenarios. ENSO's response to global warming has far-reaching global implications, including modulating the rate of future mid-latitude Southern Ocean warming and accelerating Antarctic shelf ocean warming. Extratropical climate change also plays a crucial role in modulating the tropical climate, impacting the Walker circulation. However, studies investigating changes in ENSO predictability under warming are less numerous. Some studies have observed a decrease in ENSO predictability in recent decades, but the role of global warming is ambiguous due to short observational records and internal multi-decadal climate variability. Previous work has indicated a slight reduction in the 6-month persistence of ENSO in an equilibrated future warmer climate. However, a complete understanding of ENSO predictability requires considering factors beyond SSTA persistence, including subsurface ocean heat content and extratropical influences, such as the strengthening NPMM, potentially enhancing predictability. Interactions between the Atlantic and Indian Oceans and sub-seasonal variability (like the Madden-Julian Oscillation) also affect ENSO dynamics and predictability. The response of the SPB to global warming has yet to be investigated thoroughly.

This study utilizes data from the CMIP6 multi-model ensemble, including monthly SST, subsurface temperature, currents, and downward net heat flux. The models were forced with historical and future greenhouse-gas forcing under the SSP585 emission scenario. The analysis compares ENSO predictability between the present-day (1900-1999) and future (2000-2099) climates. Principal component (PC) time series derived from SSTAs in the equatorial Pacific were used to define CP and EP ENSO indices (C-index and E-index). The study assesses changes in ENSO persistence, using autocorrelation, and focuses on the spring persistence barrier (SPB), defined by the maximum autocorrelation decline during boreal spring. The strength of the SPB is calculated as the sum of the maximum gradient of the autocorrelation across the months. The timing of the SPB is estimated by calculating persistence month. ENSO predictability was evaluated using a Linear Inverse Model (LIM) of tropical Pacific SST, quantifying predictability using the anomaly correlation coefficient (ACC). The two-box recharge oscillation model (ROM) was used to identify key factors affecting the change of SPB strength. The ROM uses a simplified representation of the tropical Pacific, incorporating two boxes for SST and thermocline depth anomalies. Parameters of the ROM were fitted to each CMIP6 model, and sensitivity experiments were performed to isolate the effects of individual parameters on persistence and predictability. Bootstrap methods were employed for statistical significance testing. The study also investigated the contribution of thermodynamic damping (TD) and dynamic damping (DD) to the change in ENSO damping rate and their spatial patterns. The North Pacific's contribution to ENSO prediction was analyzed by constructing a coupled LIM framework that included variables from both tropical and northern Pacific regions, comparing results with a decoupled LIM. A stochastic time series of SST was constructed by integrating the LIM with random noise/forcing to evaluate the model's capability of simulating and predicting ENSO. Finally, the study analyzed individual feedbacks (zonal advective feedback, thermocline feedback, Ekman feedback, thermodynamical damping and dynamical damping) to isolate the impact on the CP ENSO damping rate.

Across the CMIP6 models, the study reveals a robust decrease in the year-round persistence and predictability of CP ENSO in the future climate, primarily driven by a significant increase in the spring persistence barrier strength (approximately 21% increase in multi-model mean). This is confirmed by a significant reduction in the anomaly correlation coefficient (ACC) of CP ENSO predictions, particularly during boreal spring (approximately 25% increase in multi-model mean SPB strength). Analysis of individual models shows that a substantial majority (over 89% for persistence and 76% for ACC) exhibit a significantly enhanced SPB of CP ENSO in the future. In contrast, no robust changes were observed for EP ENSO persistence or predictability. The ROM analysis indicates that the enhanced SPB is mainly due to a decrease in the growth rate (α₂₁) of CP ENSO, reflecting a stronger damping rate in the future climate. This damping rate increase is primarily driven by a net increase in the negative feedbacks (TD and DD) outweighing the enhanced positive feedbacks under warming conditions. The stronger TD in the future is evident from the spatial patterns of thermodynamic damping, calculated by regressing net surface heat flux anomalies onto the C-index. Analysis using the coupled LIM framework suggests that the North Pacific impact on ENSO predictability does not change significantly in the future, indicating that the enhanced SPB of CP ENSO is largely intrinsic to the tropical Pacific dynamics. Both ROM and LIM analyses consistently show the reduced predictability of CP ENSO, primarily through boreal spring, highlighting the challenges in future CP ENSO forecasting.

The findings demonstrate that future greenhouse warming will likely lead to a substantial decrease in the predictability of CP ENSO, particularly during boreal spring. This is a significant result, as CP ENSO events have substantial global impacts. The increased damping rate, primarily driven by thermodynamic damping, is a key factor underlying the reduced predictability. This suggests that the future climate system, under enhanced warming, may exhibit a reduced capacity to sustain CP ENSO events, leading to more unpredictable interannual climate variability. While the study focused on the SPB, the reduced year-round persistence of CP ENSO further supports the conclusion of decreased predictability. The consistency between the LIM and ROM results strengthens the robustness of the findings. The unchanged predictability of EP ENSO suggests that the mechanisms governing predictability differ between CP and EP ENSO, highlighting the complexity of ENSO behavior under climate change. The increased challenge in forecasting CP ENSO has considerable implications for improving global climate prediction models and for developing effective strategies for mitigating and adapting to the effects of ENSO.

This study provides compelling evidence from CMIP6 models showing that the predictability of CP ENSO will significantly decrease under future greenhouse warming, mainly due to an intensified spring predictability barrier. This reduction is linked to an enhanced thermodynamic damping rate caused by faster surface ocean warming in the tropical Pacific. The findings highlight the increasing challenge of accurately predicting CP ENSO in the future, underscoring the need for improved climate models and forecasting techniques. Future research should focus on investigating the detailed mechanisms of the enhanced thermodynamic damping, particularly the regional differences and physical processes involved, and improving the representation of these processes in climate models. Furthermore, exploring the implications of decreased CP ENSO predictability on global climate impacts and developing strategies for adaptation and mitigation is crucial.

The study relies on CMIP6 model outputs, which have inherent limitations. Model biases and uncertainties may affect the results. The study uses a specific emission scenario (SSP585), and the results might vary under different scenarios. The ROM is a simplified representation of the complex ENSO dynamics, and some processes may not be fully captured. The analysis focuses primarily on the SPB, and further research is needed to examine other aspects of ENSO predictability.