On the Divergent Evolution of ENSO After the Coastal El Niños in 2017 and 2023

Coastal El Niño is an extreme situation of El Niño‐Southern Oscillation (ENSO) with sea surface temperature warming confined in the far‐eastern equatorial Pacific. Some coastal El Niños evolve into a basin scale El Niño, and some don't, implying a diversity in ENSO evolutions after a coastal El Niño event. In this study, the coastal El Niños in 2017 and 2023 are selected to examine their subsequent evolution. Both coastal El Niños developed after a La Niña, with the former followed by a La Niña and the latter by a basin‐scale El Niño. The cold (warm) subsurface temperatures in 2017 (2023) were key factors leading to the divergent ENSO evolution. Convection over the western tropical Pacific and the atmospheric circulation anomalies across the equatorial Pacific also contributed to the differences. Model predictions suggest that differences in ENSO evolution after a coastal El Niño are associated with differences in ENSO predictability.


Introduction
El Niño-Southern Oscillation is the strongest year-to-year fluctuation of the tropical air-sea interaction on the Earth, and also the most important origin of predictability in global climate on seasonal-to-interannual time scales (e.g., Hu et al., 2022;McPhaden et al., 2020;National Research Council, 2010).Geographically, the largest warm or cold sea surface temperature (SST) anomalies concentrated in the central or eastern equatorial Pacific result in what is termed ENSO diversity (e.g., Capotondi et al., 2015;Hu et al., 2012;Kao & Yu, 2009;Kug et al., 2009).In some extreme situations though, ocean warming is confined to the far-eastern equatorial Pacific and along the South American coast in what are called coastal El Niño events (Garreaud, 2018;Hu et al., 2019;Peng et al., 2019;Takahashi & Martínez, 2019).To distinguish these coastal El Niño events with warming confined to the far-eastern equatorial Pacific, the conventional ENSO events (El Niño and La Niña) are referred to as basinscale ENSO (El Niño and La Niña) events.
Coastal El Niño is different from conventional ENSO in several aspects.Takahashi and Martínez (2019) indicated the importance of local air-sea interaction and excluded the role of downwelling equatorial Kelvin waves in initiating the coastal El Niño in 1925.They argued that the coastal warming was associated with an abrupt onset of strong northerly winds across the equator and the strengthening (weakening) of the intertropical convergence zones (ITCZ) south (north) of the equator.The coupled ocean-atmosphere feedback dynamics associated with the southward shift of ITCZ, enhanced convergence along ITCZ, and enlarged north-south SST anomaly (SSTA) asymmetry in the far-eastern Pacific and produced the coastal El Niño event.Later, Hu et al. (2019) emphasized the variety of coastal El Niño in terms of evolution, mechanisms, and timing.They noted that some coastal El Niños occurred due to an equatorially centered ITCZ during the decaying phase of a basin-scale El Niño.Some coastal El Niños resulted from westerly wind bursts in the western Pacific that forced downwelling Kelvin waves and a thermocline depression in the far-eastern equatorial Pacific.Some coastal El Niños were associated with surface westerly wind anomalies in the eastern equatorial Pacific and largely driven by ocean surface heat flux anomalies, which extended the warm phase of the seasonal cycle and amplified the positive SSTAs along the South American coast.Some early observational studies argued that warming along the South American coast (coastal El Niño) was later accompanied by a basin-wide warming (El Niño).For instance, Rasmusson and Carpenter (1982) noted that the warming near the South American coast preceded the warming in the central equatorial Pacific from the 1950s to 1970s, indicating that a coastal El Niño was a precursor to a subsequent basin-scale El Niño.However, this relationship did not hold since the 1982-1983strong El Niño (McPhaden & Zhang, 2009;Rasmusson & Wallace, 1983), implying an interdecadal variation in the connection between coastal El Niños and the following evolution of ENSO (Hu et al., 2019).
A coastal El Niño event is defined when both the 3-month running mean Niño1+2 index is equal to or larger than one STD (0.8°C) and ENSO-adjusted 3-month running mean Niño1+2 index is equal to or larger than one STD (0.6°C) with anomalies of this magnitude persisting for at least three consecutive 3-month running mean overlapping seasons (Hu et al., 2019).The ENSO-adjusted Niño1+2 index is referred to the residual of the index after eliminating the portion associated with basin-scale ENSO through linear regression of the Niño1+2 index into the Niño3.4index.According to the definition, in addition to the two latest coastal El Niños in 2017 and 2023 (Figures S1b and S1c in Supporting Information S1), Hu et al. (2019) identified six coastal El Niños during 1979-2016: 1983, 1987, 1997, 2008, 2014, and 2015.Among the six coastal El Niños, 1983 and 2008 were followed by La Niña, and 1987Niña, and , 1997Niña, and , 2014Niña, and , and 2015 1b and 1g) developed after La Niña (Figures 1a and  1f), but the coastal El Niño in the spring of 2017 (Figure 1b) was followed by a La Niña (Figures 1d and 1e) and the coastal El Niño in 2023 (Figure 1g) was followed by an El Niño (Figures 1h-1j).This is another example demonstrating the diversity of ENSO evolution after coastal El Niños.Deser and Wallace (1987) noted that what happens near the coast is often but not always related to what happens in the interior ocean and that the dynamics can be quite different in the two regions.However, the physical connection of a coastal El Niño to the following ENSO evolution has not been well documented.Examining such connections may benefit the understanding of ENSO evolution and prediction.That is the focus of this study.
The data used in this work are introduced in Section 2. Section 3 shows the results, including the temporal evolution of the coastal El Niños in 2017 and 2023, the role of subsurface ocean conditions, and the influence of western Pacific convection and off-equatorial atmospheric circulation.A summary is given in Section 4 with a discussion about the prediction of ENSO evolution after the coastal El Niños in 2017 and 2023.

Data
Monthly mean SSTs on a 1°by 1°grid are computed from the latest version of daily Optimum Interpolation SSTs on a 0.25°by 0.25°grid (OIv2.1;Huang et al., 2021), which are available from September 1981 onward.The Niño1+2 and Niño3.4 indices (see the rectangles in Figure 1a; Barnston et al., 1997;Li, Hu, Ding, & Liu, 2023) are defined as the averaged SSTA in (10°S 0°, 90°W 80°W) and (5°S 5°N, 170°W 120°W), respectively.The monthly mean depth of the 20°C isotherm (D20), subsurface ocean temperature, and surface wind stress from the Global Ocean Data Assimilation System (GODAS; Behringer, 2007) in January 1979 December 2023 are also analyzed.The recharge/discharge oscillator paradigm for the phase transition of the ENSO cycle (Jin, 1997) is quantified by the warm water volume (WWV) index which is defined as the D20 anomalies averaged in (5°S 5°N, 120°E 80°W) following Meinen and McPhaden (2000).Sea level pressure (SLP) is from the reanalysis of the National Center for Environmental Prediction and the Department of Energy (NCEP/DOE; Kanamitsu et al., 2022), and monthly means of outgoing long-wave radiations (OLR) on a 1°× 1°grid from January 1991 onward (Guo et al., 2024) are used to represent deep convection variation.The OLR data is the National Oceanic and Atmospheric Administration's Climate Prediction Center (NOAA/CPC) blended level 2 OLR retrievals.
To assess the prediction skill of the SSTA evolution after the coastal El Niños in 2017 and 2023, we utilize the North American Multi-model Ensemble (NMME; Becker et al., 2022).The six models in NMME are the National Centers for Environmental Prediction (NCEP) Climate Forecast System version 2 (CFSv2), the National Aeronautics and Space Administration NASA_GEOS5v2, the National Center for Atmospheric Research NCAR_CCSM4, the Geophysical Fluid Dynamics Laboratory GFDL_SPEAR, the Environment and Climate Change Canada CanCM4i and GEM5_NEMO.The predictions (hindcasts and real-time forecasts) start from January 1982 to the present with lead times extending to 9 months.The numbers of ensemble members range from 4 to 24 and the spatial resolution of the NMME reprocessed data is 1°× 1°.Details about the models and data can be found in Becker et al. (2022) and https://www.cpc.ncep.noaa.gov/products/NMME/.To calculate the anomalies, lead time and target month-dependent climatologies were applied for each model.
The anomalies in the observations and NMME predictions are computed as the departures from climatologies over January 1991-December 2020.

Temporal Evolution of the Coastal El Niños in 2017 and 2023
For the coastal El Niño in 2017, SSTA warming in the Niño1+2 region began in late 2016 and was larger than 0.5°C during January-April 2017 with a peak value of 1.8°C in March 2017 (the black line in Figure 2a).The Niño3.4 index (the shading in Figure 2a) was negative in July-December 2016, consistent with the La Niña conditions, Similarly, SSTA warmed up quickly in the Niño1+2 region beginning in late 2022 after the triple-dip La Niña during 2020-2023 (Hasan et al., 2022;Jiang et al., 2023;Li et al., 2022Li et al., , 2023b;;Zhang et al., 2022) (the black line in Figure 2b).The peak of the coastal El Niño appeared in August 2023 with the value of 3.30°C for the Niño1+2 index, exceeding 3 standard deviations (Figure S1a in Supporting Information S1).Meanwhile, the SSTA in the Niño3.4region increased gradually with values equal to or larger than 0.5°C beginning May 2023 (the shading in Figure 2b), indicative of emerging of El Niño conditions.The Niño3.4 was 1.6°C in October 2023, becoming a strong El Niño event with a peak value of 2.0°C in December 2023.The positive SSTA first developed along the South American coast in late 2022, grew in spring 2023, and then expanded westward into the central and eastern tropical Pacific Ocean.Thus, the evolution of both the coastal El Niños in 2017 and 2023 occurred after La Niñas.However, the coastal El Niño in 2017 was followed by a La Niña, while in contrast, the coastal El Niño in 2023 evolved into an El Niño.In the following, we examine the possible reasons for the divergent outcomes after the coastal El Niños in 2017 and 2023.

Role of Subsurface Ocean Conditions
The phase transition of the ENSO cycle links to the recharge/discharge processes of the equatorial Pacific (Jin, 1997;Meinen & McPhaden, 2000).Statistically, the WWV index leads the Niño3.4index by 1-3 seasons and can serve as an ENSO precursor (Hu et al., 2017;Meinen & McPhaden, 2000).In 2016-2017, the WWV index was below average (the green line in Figure 2a), suggesting that the subsurface ocean heat conditions were unfavorable for a transition to El Niño.In contrast, the WWV index in 2023 was well above average with a peak value of 1.8 standard deviations appearing in April 2023 (the green line in Figure 2b).Consistently, warm subsurface temperature anomalies were strong and propagated eastward along the thermocline in the equatorial Pacific during the onset and development stages of El Niño in 2023 (right panel in Figure S2 in Supporting Information S1); their arrivals to the east induced warm SSTA over the eastern tropical Pacific and surface westerly wind anomalies in the western and central tropical Pacific (Gao et al., 2022).In contrast, the warm subsurface temperature anomalies in 2017 were weaker and replaced by cold anomalies in late 2017 (left panel in Figure S2 in Supporting Information S1).Therefore, from the recharge/discharge perspective, it was more favorably disposed to transition to a basin-scale El Niño in 2023, but it was more likely to transition to a La Niña in 2017.
According to Hu et al. (2019), there were another six coastal El Niños during 1979-2017 (see supplement Figure S3 in Supporting Information S1).For the cases of 1997, 2014, and 2015, the coastal El Niños were accompanied by positive WWV, followed by an El Niño, similar to 2023.The situation in 1983 was similar to that in 2017, that is, despite the coastal El Niño in 1983, the negative WWV index favored a transition to La Niña.However, such a relationship seems indistinct for the cases in 1987 and 2008, implying that the subsurface ocean heat content is not the sole factor determining post-coastal El Niño evolution (Kim et al., 2023).

Influence of Western Pacific Convection and Off-Equatorial Circulation
In addition to the differences in the subsurface ocean heat content between 2017 and 2023, the atmospheric circulation anomalies were dissimilar, especially in the western and southeastern Pacific.Specifically, enhanced convection persisted in the western tropical Pacific in 2017, which maintained the easterly wind anomalies in the western and central tropical Pacific (Figure 3a), leading to the emergence of La Niña in 2017.In contrast, convection was mostly suppressed in the western tropical Pacific since March 2023, and associated zonal wind anomalies were variable and weak in the western and central tropical Pacific (Figure 3b).The suppressed convection became more obvious since September 2023, accompanied with gradually strengthened westerly wind anomalies (Figure 3b).This pattern is consistent with the statistical relationship (Figure 3c; Hu et al., 2014) of suppressed convection in the western tropical Pacific associated with westerly wind anomalies and positive SSTA in the central and eastern tropical Pacific and opposite tendencies for enhanced convection.Thus, the difference in convection anomalies in the western tropical Pacific seems to be an additional factor resulting in the divergent evolution of ENSO in 2017 and 2023.
With strong subsurface warming in the equatorial Pacific and suppressed convection in the western equatorial Pacific, positive SSTAs grew and strengthened in the central and eastern equatorial Pacific in 2023 (Figures 1g, 1j  and 2b).However, both easterly and westerly wind anomalies in the equatorial Pacific were present (Figure 3b), which was not consistent with the SSTA zonal gradient along the equator (Figures 1g-1j).The variable and small zonal wind anomalies may be associated with the across-equatorial atmospheric circulation.Compared with the mean in January-October 2017 (Figure S4a in Supporting Information S1), the meridional gradients of SLP anomalies across the equatorial Pacific averaged in January-October 2023 (Figure S4b in Supporting Information S1) were more favorable for maintaining the southeasterly wind anomalies.That may counter the expected westerly wind anomalies along the equator associated with the zonal SSTA gradient in 2023 (Lindzen & Nigam, 1987).Thus, there may be a competition between the westerly wind anomalies associated with the SSTA zonal gradient along the equator and easterly wind anomalies linked to the meridional gradient of SLP anomalies in the tropical Pacific (Figure 4b; Chen et al., 2016).The across-equator meridional gradient in SLP anomalies in 2023 was stronger than that during 2017 with positive values in the tropical Northern Hemisphere and negative values in the Southern Hemisphere during May-October 2023 (Figure 4).Both the positive gradients in the tropical Northern Hemisphere and negative gradients in the tropical Southern Hemisphere favor easterly wind anomaly in the tropical Pacific.As a result, the zonal wind anomalies in 2023 were small in the central equatorial Pacific and easterly wind anomalies predominated in the eastern equatorial Pacific, meaning weak atmosphere and ocean coupling and suggesting the importance of extratropical circulation in ENSO evolution (e.g., Ding et al., 2015;You & Furtado, 2017;Zhao & Di Lorenzo, 2020;Zheng et al., 2024;Zhu et al., 2016).
From this comparison, we note that a coastal El Niño-induced zonal SSTA gradient and westerly wind anomalies along the equator in the Pacific, combined with warm subsurface ocean temperatures and suppressed convection over the western tropical Pacific, may lead to basin-scale El Niño development as observed in 2023.However, even with the strong zonal SSTA gradient along the equator in the Pacific associated with a coastal El Niño,

Summary and Discussion
Both the coastal El Niños in 2017 and 2023 developed after La Niñas.However, the former was followed by a La Niña and the latter evolved into an El Niño.Here we examined factors leading to the opposite phases of ENSO after the two coastal El Niños.A deficit of subsurface heat content dominated in 2017 while in contrast, elevated subsurface ocean heat content was evident in 2023.Thus, from a recharge/discharge perspective, persistence of the deficit of subsurface heat content favored the cold phase of ENSO in 2017, while high subsurface heat content favored a transition to a warm phase of ENSO in 2023.
In addition, convection over the western tropical Pacific and the atmospheric circulation anomalies in the offequatorial Pacific contributed to the divergent evolution.The persistent enhanced convection over the western tropical Pacific and the emergence of strong easterly wind anomalies led to negative SST and D20 anomaly development that resulted in the development of La Niña in the second half of 2017.In 2023, westerly wind anomalies in March and April led to rapid warming in the tropical and southeastern Pacific, followed by westward expansion of positive SSTAs from the coastal areas to the central and eastern equatorial Pacific.The competition of the westerly wind anomalies associated with the zonal SSTA gradient along the equator with easterly wind anomalies linked to the across-equatorial circulation in the tropical Pacific weakened the atmosphere and ocean coupling, so that development of the El Niño in the second half of 2023 was mainly due to the strong subsurface ocean warming.NMME predictions of the 2017 and 2023 coastal El Niños from initial conditions in January 2017 and 2023, respectively, captured the coastal warming but with underestimated amplitude (Figure 5).For the 2017 case, the NMME failed to predict the development La Niña conditions in the second half of 2017 but instead predicted continued warming, likely due to the weakly elevated subsurface WWV (green line in Figure 2a).In contrast, the NMME captured El Niño conditions in the second half of 2023, which was favored by high subsurface heat content (green line in Figure 2b), a precursor for El Niño development.Thus, different evolution of climatic conditions in the tropical Pacific following a coastal El Niño is associated with differences in ENSO predictability.
evolved into El Niño, implying a diversity of evolutions after a coastal El Niño.It is unclear what factors led to this diversity in evolution.