El Niño-Southern Oscillation (ENSO) is the strongest climate variability at the interannual timescales, and places significant impacts on global climate (Rasmusson et al,
1983). Understanding the underlying dynamics of ENSO is one of the hot topics in climate research. The classic ENSO theory attributes the growth of ENSO to positive feedbacks among sea surface temperature (SST), trade wind, and upwelling in the equatorial eastern Pacific (Bjerknes,
1969). To the phase transition between El Niño and La Niña, it’s proposed that the tropical upwelling Rossby wave, equatorial Kelvin wave and poleward transport of upper ocean heat content along with El Niño development are the two dominant processes (Schopf et al,
1988; Jin,
1997).
Above theories based on the low-frequency ocean-atmosphere coupling perfectly explain the growth of SST anomaly(SSTA) during ENSO developing phase, as well as the quasi-periodicity of ENSO. The spatial patterns of El Niño and La Niña predicted by these classic theories are expected to be symmetric in space with comparable magnitudes. However, the observed ENSO events exhibit strong irregularity, asymmetry, and diversity. The strength of El Niño is in general greater than the La Niña (Burgers et al,
1999), whereas the La Niña lives longer than the El Niño (Wolter et al,
2011). In addition, El Niño usually peaks inboreal winter, but this phasing locking behavior is weaker for La Niña (Galanti et al,
2000).
More importantly, the El Niño exhibits strong diversity both in magnitude and spatial distribution than the La Niña (Kug et al,
2009). All of the extreme ENSO events in history are El Niño. The El Niño events are thus can be categorized into the extreme type and the moderate type (Takahashi et al,
2011). On the other hand, although the maximum SST anomaly during the La Niña events is confined in the eastern equatorial Pacific, the center of SST anomaly during the El Niño events can be found in either the equatorial central Pacific or the eastern Pacific. Therefore, the El Niño can be also grouped into another two types according to the longitude of the maximum SST anomaly. The one with maximum warming in the eastern equatorial Pacific, as predicted by the classic ENSO theory, is called the eastern-Pacific (EP) type, whereas the one with the SST warming center near the equatorial dateline is termed the central-Pacific (CP) El Niño (Ashok et al,
2007; Kao et al,
2009). Chen et al (
2015) showed that the El Niño in fact has three distinct flavors. The moderate El Niño which locates in the EP area is symmetric to La Niña, both of which contribute the regularity of ENSO. The CP and extreme El Niño are the irregular ones, and contribute to ENSO nonlinearity.
Clearly the ENSO nonlinearity cannot be explained by the classic ENSO theory. The tropical-subtropical Pacific interactions, pantropical teleconnections, and multi-scale interactions are thus proposed to supplement the low-frequency ENSO dynamics (Fang et al,
2020). In particular, previous studies showed that the strong equatorial westerlies from the synoptic to intraseasonal timescales are crucial to the genesis of El Niño diversity. For example, Yu et al (
2003) found that the strong equatorial westerlies in early 1997 are originated from the East Asian cold surge in late 1996, and this westerly event is important to the onset of 1997/98 El Niño. Harrison et al (
1997) pointed out that the intensity of the springtime westerly wind bursts (WWBs) in the tropical western Pacific is significantly related to El Niño strength. Chen et al (
2015) indicated that almost every El Niño in history are proceeded by the WWBs, and the interplay between WWBs and ocean heat content determines the pattern and magnitude of El Niño. Sobel et al (
2005) suggested that tropical cyclone in the northwestern Pacific excites strong equatorial westerlies, and the intensity of tropical cyclone significantly leads ENSO by about 5 months. Lian et al (
2018) further confirmed the possible influence of tropical cyclone on ENSO using numerical experiment. Wang et al (
2019) also showed that summer-autumn tropical cyclone intensity leads the upcoming El Niño by 3 months.
The aforementioned works show that ENSO is resulted from both the low-frequency ocean-atmosphere coupling and atmospheric perturbation at the synoptic scales within the tropical Pacific. However, the nonlinearity in ENSO dynamics prevents any quantitative estimation of the two to ENSO evolution. Kang et al (
2000) and Syu et al (
2000) constructed a statistical atmospheric model using the singular vector decomposition (SVD) between SST and surface wind. They found that the main ENSO features can be reproduced using the first two coupled modes. Nevertheless, as the SVD technique cannot restrict the spectrum of the modes explicitly, the results from Kang et al (
2000) and Syu et al (
2000) cannot be fully interpreted as the results of the pure low-frequency coupling. To this end, Lian et al (
2021) introduced a new method named as the online low-frequent filtering method (OLF). The OLF method was designed to exclude the high-frequency part of wind stress during ocean-atmosphere coupling in model, thus can be used to scale the impact of the pure low-frequent ocean-atmosphere interaction on ENSO development. Using this new technique, Lian et al (
2021) showed that the strong WWB in March 1997 is the necessary condition for the genesis of the 1997/98 extreme El Niño.
Although Lian et al (
2021) validated the role of WWB in the 1997/98 El Niño event, the impact of high-frequency wind stress forcing on ENSO from the statistical point of view is lacking. In this study, we aim to estimate the influence of the pure low-frequency coupling to the development of ENSO using the OLF method. The rest of the manuscript is arranged as follows: model and method are presented in Section 2. In Section 3, we compare model simulations with and without the OLF module, and analyze the change of ENSO caused by removing high-frequency wind, followed by the conclusion and discussion in Section 4.