Deep ocean warming-induced El Niño changes

Human-induced warming has accelerated over recent centuries, raising urgent concerns and motivating the Paris Agreement’s goal to limit warming below 2°C (ideally 1.5°C). Even if atmospheric CO₂ is reduced to preindustrial or current levels, the climate system will not fully return to its prior state due to irreversibility, driven in part by the deep ocean’s slow heat uptake and release. The deep ocean stores large amounts of heat and releases it over centuries, potentially producing regional irreversibility in sea surface temperature (SST) and precipitation. This release is expected to be more effective where stratification is weak (e.g., equatorial eastern Pacific), suggesting an El Niño-like mean-state warming pattern. ENSO is the leading mode of interannual climate variability with substantial global impacts, and its future changes under greenhouse forcing have been widely studied. However, how delayed deep ocean warming during a CO₂ stabilization or recovery period specifically affects ENSO characteristics remains unclear. This study investigates whether deep ocean warming, through changes in the tropical Pacific mean state, causes an eastward shift and intensification of ENSO, leading to more frequent and extreme Eastern Pacific (EP) El Niño events.

Idealized CO₂ mitigation and removal experiments (e.g., CDRMIP) have identified irreversibility across multiple climate components, including surface temperature, sea level, precipitation, AMOC, Arctic ice, and ITCZ. The deep ocean’s role has been emphasized as a slow reservoir whose heat release can drive regional SST and precipitation irreversibility, especially in high latitudes and the equatorial eastern Pacific where weak stratification and upwelling facilitate heat transfer to the surface. Prior work on ENSO under greenhouse warming documents changes in intensity, asymmetry, duration, spatial pattern, diversity, and teleconnections, with tendencies toward more Central Pacific (CP) events and enhanced EP extremes due to nonlinear atmospheric responses. Increased eastern Pacific SST variability has been linked to southward ITCZ shifts under certain scenarios. Yet, a direct link between deep ocean heat release during recovery periods and systematic ENSO changes had not been explicitly tested, motivating the present analysis.

Models and configuration: The study employs CESM v1.2 comprising CAMS (atmosphere), CLM4 (land), CICE4 (sea ice), and POP2 (ocean). Atmospheric and land components use ~1° horizontal resolution with 30 vertical levels; ocean/ice use nominal 1° resolution (meridional ~1/3° near equator) with 60 ocean levels.

Baseline and CO₂ pathway experiments: A present-day (PD) control with constant CO₂ (367 ppm) was integrated for 900 years. A ramp-up/ramp-down experiment used 20 ensemble members with differing initial conditions: CO₂ increased by 1% yr⁻¹ from a preindustrial level (284.7 ppm) to 4× (1469 ppm) over 140 years, then decreased symmetrically by 1% yr⁻¹ back to 367 ppm over 140 years, followed by a 220-year recovery with constant CO₂ at 367 ppm (net-zero phase). Each ensemble member spans 500 years.

Initial warming (IW) experiments to isolate deep ocean heat effects: Three sets of experiments were branched from year-2000 initial conditions of each ensemble member and integrated with constant CO₂ (367 ppm) for 150 years with nine ensemble members. Horizontally uniform vertical profiles of ocean temperature and salinity anomalies were added to the initial ocean state to mimic deep ocean warming while preventing artificial stratification effects. Perturbation depths: entire depth (IW_whole), below 100 m (IW_be100), and below 700 m (IW_be700). Mean-state changes are assessed as differences between the average of years 51–150 and PD.

Diagnostics: ENSO amplitude is quantified using standard deviations of Niño3 (5°S–5°N, 150–90°W) and Niño4 (5°S–5°N, 160°E–150°W) SSTA after removing linear trends and monthly climatology. ENSO events are defined by DJF Niño3 exceeding 1 STD (PD-based). The spatial diversity (CP vs EP flavors) is diagnosed by detecting the longitude of peak DJF SSTA along the equator; CP events are those with peaks between 165°E–145°W and EP events between 125°W–150°W, and a CP ratio is computed as CP/(CP+EP). Extreme and convective extreme El Niño are identified following previous studies. To examine feedbacks and dynamics, linear regressions of ocean currents onto zonal wind stress anomalies (120°E–90°W, 5°S–5°N, DJF) are analyzed, along with precipitation and wind regressions onto Niño3 SSTA. Thermocline depth, upwelling, and stratification (vertical temperature gradient) diagnostics complement the analysis. Statistical significance is assessed with a bootstrap test at 95% confidence. Pattern correlations between restoring/IW responses and PD are computed to assess similarity.

• Deep ocean heat release produces an El Niño-like tropical Pacific mean-state response with stronger eastern Pacific SST warming, increased tropical precipitation, weakened equatorial trade winds, and a southward ITCZ shift. The restoring period patterns closely match those from the IW_be700 experiment despite uniform deep warming only below 700 m; pattern correlations are ~0.95 (SST) and ~0.97 (precipitation) over the tropical Pacific.
• The mean-state changes are robust across initial warming depths (IW_whole, IW_be100, IW_be700), with magnitudes scaling with the total imposed deep warming.
• ENSO amplitude: Niño3 SSTA standard deviation increases significantly in restoring and IW experiments, while Niño4 variability remains largely unchanged (except for some increase in IW_whole). The increases in Niño3 variance rank IW_whole > IW_be100 > IW_be700, proportional to the mean-state warming magnitude.
• ENSO flavor and diversity: The distribution of SSTA peak longitudes shifts eastward, yielding more frequent EP El Niño events and a reduced CP ratio during the restoring period; IW_be700 and other IW runs reproduce these changes. La Niña peak distributions show similar eastward shifts.
• Extreme events: Deep ocean warming leads to an increase in convective extreme El Niño events by approximately 40–80% relative to the current climate.
• Dynamics and feedbacks: Upper-ocean stratification is enhanced (positive change in vertical temperature gradient in the upper layer), intensifying surface-layer coupling. Regressions show stronger eastward surface current responses to zonal wind stress (enhanced zonal advective feedback), particularly in the eastern Pacific; strengthened equatorial upwelling and thermocline responses accompany this. Precipitation and wind responses regressed onto Niño3 SSTA shift eastward, enhancing Ekman and thermocline feedbacks and favoring EP-type variability.
• Causality: The close replication of restoring-period ENSO changes by the idealized deep-warming IW experiments indicates deep ocean heat release is the principal driver of the observed mean-state shift and ENSO changes during CO₂ recovery to present-day levels.

The results demonstrate that delayed heat release from the deep ocean fundamentally alters the tropical Pacific mean state toward an El Niño-like configuration and shifts the ENSO feedback system eastward. This transition enhances eastern Pacific SST variability, increases the frequency of EP El Niño, and raises the likelihood of convective extreme El Niño events, even when atmospheric CO₂ is stabilized at current levels. The enhanced upper-ocean stratification and the eastward-shifted atmospheric and oceanic responses (winds, precipitation, thermocline, currents) reinforce the Bjerknes, thermocline, and Ekman feedbacks in the eastern Pacific, explaining the amplified Niño3 variability and altered ENSO diversity. These findings have implications for decadal predictability and risk assessment, as variability and extremes become more pronounced in the eastern Pacific under deep ocean heat release. The study also provides a lens to interpret discrepancies between model projections and recent observed warming patterns, suggesting that modeled deep ocean heat uptake/release may modulate SST patterns and ENSO behavior during mitigation and recovery scenarios.

Deep ocean warming is identified as a key driver of El Niño-like tropical Pacific mean-state changes during CO₂ stabilization or recovery, promoting an eastward shift of ENSO variability, increased Niño3 variance, more frequent EP El Niño, and a 40–80% rise in convective extreme El Niño events relative to the present climate. Idealized initial warming experiments that impose uniform deep ocean temperature anomalies reproduce restoring-period changes with high pattern correlations, underscoring the causal role of deep heat release. These insights highlight the long-lived influence of anthropogenic heat stored in the deep ocean on ENSO, even under net-zero conditions. Future research should reassess ENSO projections in light of deep ocean heat dynamics, evaluate model representations of deep heat uptake/release, and further constrain feedback shifts and stratification changes with observations.

The study relies on a single modeling framework (CESM v1.2) and idealized initial warming experiments that impose horizontally uniform deep ocean temperature/salinity anomalies, which may not capture all regional complexities of deep heat distribution. The paper notes discrepancies between model projections and recent observed warming patterns, potentially related to common model biases (e.g., cold-tongue bias, inter-basin warming contrast, and Walker circulation representation). It also suggests that models may overestimate deep ocean heat accumulation relative to observations, which could bias SST and ENSO projections. The exact causes of model–observation discrepancies remain unclear.