Multi-year El Niño events tied to the North Pacific Oscillation

The El Niño/Southern Oscillation (ENSO) is the dominant mode of tropical ocean–atmosphere variability, influencing global climate and extremes. Typical ENSO events develop in boreal summer, peak in early winter, and decay rapidly the following spring. However, some events persist into a second year, resulting in multi-year El Niño or La Niña episodes that amplify and prolong climate impacts. Forecasting these multi-year events is challenging; for instance, the 2014/15/16 multi-year El Niño was poorly predicted by most operational models. While many hypotheses explain asymmetries in the duration of El Niño vs. La Niña, few address the mechanisms causing multi-year persistence. Most existing theories emphasize tropical coupled processes within the Pacific, Indian, or Atlantic Oceans and generally neglect extratropical forcings. Recent work suggests extratropical atmospheric variability, notably the North Pacific Oscillation (NPO), can affect ENSO development, pattern, transitions, and decadal variability. This study asks whether the NPO is dynamically linked to multi-year ENSO events and, if so, by what mechanism. Using observations and climate model experiments, the authors propose that two-way teleconnections between the NPO and the tropical Pacific, mediated by the North Pacific meridional mode (NPMM) and Central Pacific (CP) El Niño, are a key source of multi-year El Niño persistence.

Prior studies highlight several pathways by which extratropical variability can influence ENSO. The NPO can trigger tropical responses via the North Pacific meridional mode (NPMM) through the seasonal footprinting mechanism and wind–evaporation–SST (WES) feedback, reducing trade winds and warming the subtropical NE Pacific, with anomalies propagating equatorward to favor El Niño onset. Additional mechanisms include off-equatorial trade wind weakening and Rossby wave excitation that precondition the equatorial Pacific. The extratropical influence has been linked to ENSO development, spatial pattern (CP vs EP types), phase transition, and decadal variability. Nonetheless, most work has focused on event initiation rather than on prolongation to multi-year states. There is also evidence that CP El Niño elicits a distinct North Pacific SLP response (e.g., negative anomalies over Hawaii) different from EP El Niño, suggesting sensitivity of extratropical teleconnections to the ENSO flavor. Gaps remain regarding whether and how NPO variability can maintain El Niño conditions across successive years to produce multi-year events.

Observations: Monthly SST datasets (HadISST, ERSSTv4, Kaplan SST, COBE SST) and atmospheric fields (NCEP/NCAR Reanalysis 1; HadSLP2 for SLP) were analyzed. Anomalies are relative to 1981–2010 climatology. Definitions: Multi-year El Niño/La Niña are defined using the Niño3.4 index (5°S–5°N, 120°–170°W), detrended and 3-month smoothed. A multi-year El Niño occurs when Niño3.4 > 0.75 SD during Oct(0)–Feb(1), remains > 0 thereafter, and exceeds 0.5 SD during Oct(1)–Feb(2); analogous thresholds apply with opposite sign for La Niña. NPO is defined as EOF2 of JFM SLP anomalies poleward of 15°N in the North Pacific; the NPO index is PC2. NPMM index is normalized SST anomalies over 15°–25°N, 150°–120°W. CP (and EP) ENSO indices are derived via EOFs of tropical (20°S–20°N) SST anomalies after regressing out the canonical (Niño1+2) or CP (Niño4) signals, respectively. CMIP5/6 model analysis: Historical (1900–1999) simulations from CMIP5/6 were screened for a significant JFM(0) NPO to JFM(1) Niño3.4 linkage and for realistic representation of CP/EP El Niño (via DJF Niño3–Niño4 correlation). Twenty-nine models meeting criteria were used to assess NPO’s role in multi-year El Niño/La Niña. For future projections, 23 of these with RCP8.5 (2000–2099) runs were analyzed. Relationships between changes in multi-year El Niño frequency and NPO variance were evaluated. Significance testing: Bootstrap resampling (10,000 realizations) assessed the significance of differences in NPO indices preceding multi-year events and changes in event counts; correlations/regressions used two-tailed Student’s t-tests. Targeted CGCM experiments: Using FGOALS-g2 coupled model with a weakly coupled data assimilation system, positive NPO forcing (monthly U, V, T anomalies at 1000/925/850 hPa regressed on NPO) was assimilated during Nov(-1)–May(0) into the atmosphere to constrain circulation (EXP_NPO). A control (EXP_CTRL) featured climatological atmosphere. CTRL ran 100 years; EXP_NPO comprised 60 ensemble members initialized from the last 60 CTRL years. Differences in Niño3.4 evolution, SLP over Hawaii, NPO/NPMM indices, and frequency of multi-year El Niño were computed.

- Observational linkage: All five observed multi-year El Niño events (since 1950) were preceded by a pronounced positive NPO during JFM(0), with NPO index > 1.0 SD and SLP patterns significantly correlated (>95% confidence) with the canonical NPO dipole. Example: 2018/19/20 event had NPO index 1.27 and spatial correlation 0.81. - Two-way feedback: Positive NPO in JFM(0) excites an NPMM-like SST footprint via WES feedback, propagating warm anomalies equatorward to yield central equatorial Pacific warming and a CP El Niño in JFM(1). The CP El Niño forces negative SLP anomalies over the Hawaiian region, re-energizing NPO-like variability and weakening off-equatorial trades, thereby reactivating the NPMM in MAM(1). This sustains and re-intensifies warming into year (1), producing multi-year El Niño persistence. Composite indices show two peaks of SLP over Hawaii and NPO in JFM(0) and JFM(1). - Event-type dependence: NPO-preceded CP El Niños tend to have SST anomalies extending west of 170°E into the warm pool, eliciting stronger Hawaiian SLP responses and stronger NPMM, favoring multi-year evolution. Non-NPO-preceded CP El Niños have SST confined east of 170°E, weaker Hawaiian SLP/NPMM, and evolve as single-year. - CMIP5/6 historical support: Across 29 models, 71% of multi-year El Niño events are preceded by positive NPO, versus 41% for all El Niño and 27% for single-year El Niño. Multi-year El Niño shows a pronounced shift toward positive JFM(0) NPO; the composite JFM(0) NPO index for multi-year El Niño is +0.95 SD and significant (>95% via bootstrap). - Targeted CGCM experiments: Imposed positive NPO forcing prolongs El Niño into the subsequent winter, increases NPO-like SLP variability near Hawaii, strengthens the NPMM, and raises the frequency of multi-year El Niño events from 18% in CTRL to 39% in EXP_NPO (significant at >95%). - Future projections: In 23 CMIP5/6 models (RCP8.5), multi-year El Niño frequency increases by an ensemble mean of 43% (from ~7 to ~10 events per 100 years; significant via bootstrap). Inter-model relationships show increases in NPO variance are significantly correlated with increases in the number of multi-year El Niño events (R = 0.76, p < 0.01) and with increases in the ratio of multi-year to total El Niño events (R = 0.79, p < 0.01).

The study directly addresses the question of what sustains multi-year El Niño events by identifying a two-way extratropical–tropical feedback loop linking the NPO, NPMM, and CP El Niño. A positive NPO in boreal winter initiates subtropical SST anomalies that favor a CP El Niño the following winter; in turn, the CP El Niño’s atmospheric teleconnections re-intensify NPO-like SLP anomalies over the North Pacific (notably near Hawaii), reactivating the NPMM and fostering another year of El Niño conditions. This mechanism differs from earlier explanations centered on purely tropical processes by highlighting extratropical atmospheric variability as a dominant source of multi-year El Niño persistence. The modeling evidence (both CMIP ensembles and targeted assimilation experiments) corroborates the observational findings and demonstrates causality: assimilating NPO-like forcing increases the likelihood and duration of El Niño recurrence. The results imply that correctly representing NPO variability, its coupling to the NPMM, and the CP vs EP ENSO pattern dependence can improve prediction of multi-year El Niño events. The paper also notes asymmetries with multi-year La Niña: while many are linked to strong preceding El Niño via ocean heat recharge, negative NPO can also contribute via analogous feedbacks, albeit less dominantly. Projected increases in NPO variance under anthropogenic warming suggest multi-year El Niño events will become more frequent and comprise a larger fraction of El Niño occurrences, with important implications for long-lived hydroclimate extremes.

This work identifies extratropical NPO variability, operating through NPMM dynamics and CP El Niño teleconnections, as a key driver of multi-year El Niño persistence. Observations, CMIP5/6 historical simulations, and targeted CGCM experiments consistently show that positive NPO in boreal winter often precedes and sustains El Niño across years via a two-way feedback with the tropical Pacific. Future projections indicate a significant increase in multi-year El Niño frequency, linked to enhanced NPO variance in a warmer climate. These insights underscore the need to account for NPO influences in seasonal-to-multiyear ENSO prediction systems and climate projections. Future research should clarify how NPO dynamics evolve with warming, refine understanding of CP versus EP ENSO teleconnections, quantify the interplay with other basins (Indian/Atlantic, South Pacific), and assess predictability gains from assimilating extratropical precursors.

- Limited observational sample: Only five multi-year El Niño events since 1950 constrain statistical robustness; some precursor patterns also show tropical cooling whose dynamical significance remains unclear. - Model dependence and biases: Although many CMIP5/6 models reproduce key linkages, representation of CP/EP ENSO diversity, NPMM, and NPO variance varies across models, affecting quantitative estimates. - Attribution complexity: While the NPO–NPMM–CP El Niño feedback is central, other processes (e.g., Indo-Pacific/Atlantic coupling, South Pacific influences, equatorial ocean dynamics) can modulate ENSO duration and may confound simple attributions. - Future mechanism uncertainty: The precise reasons for projected increases in NPO variance are not fully established; causal pathways (e.g., changes in WES feedback strength or Kuroshio Extension forcing) require further investigation.