**1. Introduction**

As the prominent atmospheric mode occurring in the North Atlantic, the North Atlantic Oscillation (NAO) features an alternation of air mass over Iceland and the Azores [1–3]. Changes in the polarity and intensity of NAO exert profound influences on the surface air temperature and precipitation across the broad areas of North America and Eurasia [4]. In the positive phase, the pressure difference between the Icelandic Low and the Azores High is stronger than average, resulting in stronger westerly winds and storm tracks over the North Atlantic. This can lead to wetter and stormier conditions in Western Europe, while eastern parts may experience milder and drier weather. Conversely, the negative phase of the NAO exhibits almost the opposite characteristics. It is commonly recognized that the formation of NAO arises from the internal stochastic processes of the atmosphere [5,6]. One thing that has become clear is that the forcing stemming from the stratosphere [7–9] and sea surface temperature (SST) anomalies can also generate NAO-like anomalies [10–12].

The El Niño-Southern Oscillation (ENSO) is the dominant interannual air-sea coupled mode in the topics, which exerts salient impacts on global climate via localized forcing or remote teleconnections [13–17]. Whether the ENSO SSTA can excite the NAO-like atmospheric pattern or not, however, is still debated. Early research [18,19] believed that ENSO-related climate variability is hard to observe over the North Atlantic-Eurasia sectors. Despite the strong intrinsic variability of NAO, more and more observational and numerical

**Citation:** Zhang, P.; Wu, Z. Insight into Asymmetry in the Impact of Different Types of ENSO on the NAO. *Climate* **2023**, *11*, 136. https:// doi.org/10.3390/cli11070136

Academic Editors: Xiaolei Zou, Guoxiong Wu and Zhemin Tan

Received: 23 May 2023 Revised: 22 June 2023 Accepted: 23 June 2023 Published: 27 June 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

experimental evidence has proven that a negative NAO-like anomaly sometimes occurs in the El Niño winter; by contrast, a positive NAO-like pattern coincides with the La Niña events [20–22]. The ENSO-NAO relationship, therefore, may provide a potential source for seasonal climate prediction in North America and Eurasia [23]. Nevertheless, many studies also found that the atmospheric signals of ENSO over the North Atlantic exhibit salient inter-event variability [24–28]. This unstable behavior may be attributed to tropical volcanic eruptions, other non-ENSO factors [23,29], the interdecadal variation of ENSO itself [30], and the diversity of ENSO [24,26].

The ENSO diversity involves varying amplitudes and SSTA patterns [31]. The strong ENSOs with SSTAs larger in amplitude usually induced enormous property loss and casualties, grabbing the close attention of the scientific community [32,33]. However, a growing number of studies noticed the impact of moderate ENSOs, emphasizing that atmospheric responses to strong and moderate ENSOs sometimes show patterns with opposite signs rather than reduced amplitudes [27,34,35]. Diverse ENSO patterns have been apparent in recent decades when central Pacific (CP) ENSOs have occurred frequently, exerting impacts that differ from the conventional (eastern Pacific, EP) ENSO in terms of their disparate atmosphere-ocean coupling processes [36–39]. This classification, however, does not distinguish moderate ENSOs from strong ones, although strong El Niños usually exhibit an EP type, making it difficult to explain the climatic impact and evolution of relatively weak ENSOs. Wang et al. [40] classified El Niño events as strong basin-wide, moderate eastern Pacific, moderate central Pacific, and successive events. They revealed that the more frequent occurrence of extreme ENSOs in the past 40 years might be attributed to a background warming in the equatorial western Pacific and associated enhanced zonal SST gradients in the equatorial central Pacific.

The zonal SST gradients reflect the SSTA contrast between the western and eastern Pacific, which accelerates the equatorial zonal wind and increases the tilt of the thermocline, favoring the maintenance of an ENSO event [41]. As an indispensable segment of the Bjerknes positive feedback, zonal SSTA gradient largely furnishes violent tropic deep convections [42] and determines extratropical teleconnections [43–45]. Referencing Wang et al. [40], Zhang P et al. [27] classified ENSO events according to their amplitude and maximum SST, divided the winter La Niña events, based on the ENSO intensity and east-west gradient in the Pacific basin, into three groups: strong intensity La Niñas, weak intensity La Niñas with strong or weak gradients (their features are summarized in Table 1), and those impacted by the three flavors of La Niña on the East Asian winter monsoon. Whether or not El Niño events can be divided into three similar categories can be debated.

This study attempts to determine what types of El Niño and La Niña can profoundly impact NAO. Moreover, owing to the asymmetries that exist in amplitudes [46–48], evolutions [49–51], and effects [52,53], the warm and cold ENSOs, therefore, are not simply mirror images of each other. A systematic contrastive study of the physical process and mechanism for the impact of diverse warm and cold ENSO events on NAO is necessary. To discuss the above questions, Section 2 displays the datasets and model. Section 3 shows the diverse ENSOs and their impacts on NAOs. The possible mechanisms are discussed in Section 4. In Section 5, the major conclusions and discussions are exhibited.


**Table 1.** The years of different ENSO types.

Central-Pacific ENSO years are marked with bold fonts.

#### **2. Materials and Methods**

The merged Extended Reconstructed SST version 5 (ERSSTv5) [54] and the Hadley Centre Sea Ice and Sea Surface Temperature dataset (HadISST) [55] with the 2◦ × 2◦ horizontal resolutions are used in this study. Atmospheric reanalysis datasets include ERA-40 [56] and ERA-Interim [57] data with a horizontal resolution of 1.5◦ × 1.5◦. Precipitation Reconstruction (PREC) data were provided by NOAA [58]. The differences between the western Pacific K-shape and eastern Pacific triangle SSTs are defined by the mega-ENSO index [59]. The Niño-3.4 index is obtained from Climate Prediction Center (CPC). The NAO is defined using Hurrell's station-based index [1]. The winter refers to the period from December to the next February. The data used for compositing analysis are de-trended to exclude the potential influence of linear trends.

We normalized the two ENSO indices to contrast them under the same criteria (Figure 1). The present study primarily employs composite analysis; the anomaly refers to the difference between the salient ENSO and ENSO-neutral years. "ENSO-neutral" indicates the normalized Niño3.4 index between −0.5 and 0.5 standard deviations (STD). ENSO events refer to the years with the absolute values of the Niño3.4 index greater than 0.5 STD. Considering that the average intensity of the selected warm and cold ENSO events is around ±1.1, we classify ENSO events as of weak intensity when the absolute value of the Niño3.4 index falls within 0.5 and 1.1 standard deviations and as strong intensity when the absolute value of the Niño3.4 index exceeds 1.1. To separate strong and weak gradient ENSOs, we set ±0.7 STD of the mega-ENSO index as the criteria. A strong gradient ENSO needs to satisfy the absolute value of the mega-ENSO index greater than 0.7 STD. If not, we define it as the weak gradient ENSO. Altering the criteria to ±0.8 STD, the qualitative results will not change.

**Figure 1.** Time series of the normalized December–January–February (DJF) Niño3.4 (red curve) and mega-ENSO (black curve) indices for the period of 1957–2018. The mega-ENSO index is multiplied by −1 for comparison purposes. The red, black, and purple dashed lines denote ±0.5, ±0.7, and ±1.1, which refer to the thresholds for dividing ENSO events, SG/WG ENSO events, and strong/moderate ENSO events, respectively.

The European Center-Hamburg (ECHAM 5.4) [60] model from the Max Planck Institute is applied to illuminate a possible mechanism. The resolution is T63L19 (horizontal grid of 1.875◦ and 19 vertical levels). Forced by the observational AMIP II SST, the model was integrated from 1950 to 2010. To reduce the potential impact of ENSO, we picked out twenty neutral or weak ENSO winters from 1955 to 2010 as the samples for the control experiments. Next, sensitivity experiments consisted of nine groups of simulation. Each experiment was forced by climatological monthly mean SSTs (same as in the control run) and observed DJF SST anomalies. The initial conditions for the sensitivity simulations were acquired from the control run. The detailed experiment design is shown in Table 2.

**Table 2.** List of SST perturbation experiments conducted in this study.

