A tropical cyclone (TC) is an intense cyclonic vortex with a warm core structure, originating from warm tropical or subtropical oceans. TCs play an important role in sea-air coupling interactions. On one hand, they can drive mesoscale ocean circulation and intensify the turbulent mixing within specific regions, resulting in alterations to the temperature and salinity structure of the upper ocean. The passage of TCs involves the input of a large amount of mechanical energy into the ocean, leading to the loss of local sea surface water vapor and heat due to the high wind stress and the associated wave action. Furthermore, TCs can induce Ekman pumping and heat pumping, causing the cooling of the surface’s warm water, and the heating of the subsurface’s cold water. With the passing of TCs, the surface water returns to the normal thermal condition of the Pre-TCs, and the subsurface water remains warm (Emanuel, 2001). Additionally, strong turbulent mixing and upwelling can be caused in the subsurface water due to the entrainment of TCs (Greatbatch, 1984; Monaldo et al., 1997; Gierach and Subrahmanyam, 2008; Hung et al., 2013; Chang et al., 2014). On the other hand, the energy input to the ocean from TCs propagates as near-inertial internal waves and interacts nonlinearly with local internal waves and near-inertial oscillations in the ocean, ultimately affecting energy distribution in the ocean basin and even globally (Quanan et al., 2006; Furuichi et al., 2008; Elipot et al., 2010). Since temperature anomalies of surface and subsurface seawater are closely related to global climate (Wyrtki, 1985; Zebiak and Cane, 1987), TCs exert significant impacts on climate-scale events.

TCs commonly occur in the equatorial Pacific, particularly in the northwestern region of the western equatorial Pacific, which accounts for approximately one-third of the global total (Wang and Chan, 2002; Chan, 2005). Additionally, almost all equatorial Kelvin waves originate in the western equatorial Pacific. Previous studies (Sriver et al., 2008; Fedorov et al., 2010; Sriver and Huber, 2010) indicated that TCs can affect equatorial Kelvin waves and thereby influence ocean dynamics. Therefore, examining the potential relationship between TCs and equatorial ocean dynamics can further deepen our understanding of TC generation and climate processes.

Previous studies have explored the forcing effect of cross-equatorial TCs and TC pairs on equatorial Pacific dynamics based on buoy data and found that individual synoptic events may influence the entire El Niño–Southern Oscillation (ENSO) cycle (Keen, 1982; Nitta, 1989). Based on a statistical analysis of observations of multiple TC events at an interannual timescale, Gao et al. (1988) found that the near-equatorial tropical cyclones developed equatorward of 10°N can intensify the equatorial westerlies and produce the Kelvin waves, which propagate to the South American coasts in about 2-3 months, inducing the rise of Sea Surface Temperature (SST) there. Chen and Gao (1994) pointed out that the durations of TCs (in days) in the near-equatorial and tropical areas of the Northwest Pacific are significantly higher in El Niño than in non-El Niño years, and concluded that the development of TCs in the near-equatorial and tropical areas of the Northwest Pacific strengthens the equatorial westerly wind pulse and excites the Kelvin waves, contributing to the development and maintenance of El Niño. Sriver et al. (2013) used the Community Climate System Model to prove that three successive TC events may excite equatorial Kelvin and Yanai waves, and they found the wave amplitude is directly proportional to TCs wind speed. They also suggested that the wind forcing of TCs would affect the period and amplitude of ENSO events. Wang et al. (2019) proposed that TC events in the Western Pacific could significantly affect the development of ENSO events, mediated by Walker circulation and equatorial Kelvin waves. Previous modeling studies on the effects of TCs on Kelvin waves mainly focused on multiple TC events at interannual scales. However, the response of equatorial Kelvin waves to an individual TC remains unknown. To our knowledge, the forcing effect of a tropical storm (TS), which is a low-intensity TC, on Kelvin waves has never been studied.

In this study, we concentrate on TS Dujuan, which lasted for approximately 10 days (from the tropical disturbance stage) in the western equatorial Pacific in February 2021. The primary reason we select it is that it was active near the equator for a very long time ( Figure 1A ). TS Dujuan occurred when the Eastern Pacific was transitioning from a La Niña to a neutral state ( Figure 1B ), which is more conducive to the development of Kelvin waves (McPhaden, 1999). For the first time, the forcing effect of individual TS on the equatorial Kelvin wave and the associated ocean dynamics are investigated in this study.

The data and the anomalous ocean signals during TS Dujuan are described and discussed in Section 2. The configuration of the global ocean circulation model and the model verification are given in Section 3. The diagnostic modeling results are analyzed in Section 4, and the discussion and conclusions are presented in Section 5.

Based on the analysis of the anomalous ocean signals, the existence of a Kelvin wave in the equatorial Pacific between February 1 and April 30, 2021 is inferred. Although easterly anomalies dominate over the Pacific during La Niña events, more than 50% of the tropical cyclones in the western Pacific reverse the direction of local trade winds by enhancing the equatorial westerly anomalies (Lian et al., 2019) ( Figure 2A ), which is conducive to the generation of downwelling Kelvin waves.

Kelvin wave has been identified above, but the effects of TS Dujuan on the Kelvin wave during the La Niña event remain unclear. Therefore, here we perform diagnostic experiments to systematically identify the specific effects of TS Dujuan forcing on the equatorial Kelvin wave.

Here we compare the differences in detail between the simulations of the TSF and NONTSF experiments. The oceanic responses, including SSH, temperature, and zonal current, to the forcing of TS Dujuan are analyzed and discussed in the following.

The longitude-time plots for the results of TSF and NONTSF from February to May 2021 are shown in Figure 11 . The locations of the alternating positive and negative differences in the western equatorial Pacific correspond to those of the zonal wind stress anomalies in late February ( Figure 2A ), indicating that TS Dujuan causes the change in SSH in the western Pacific. The intensity of TS Dujuan is low, and consequently so is the accumulated cyclone energy (ACE; only 0.54×10⁴ m² s^-2). Therefore, the SSH characteristics in the TSF experiment do not differ considerably from those in the NONTSF experiment, which is depicted visually in Figures 11A, B . Compared to the NONTSF case, TS Dujuan winds raise the SSH on the east side of the forcing region (by more than 2.1 cm) and lower it on the west side. The differences between the positive and negative increase with proximity to the TS track (Sun et al., 2014). In the Northern Hemisphere, the asymmetrical nature of the SSH (i.e., raised on the east side of the TS track and lowered on the west side) is probably due to the local winds. The direction of the local winds on the east side of the TS track is consistent with that of the Ekman drift. For a slow-moving TS, the wind enhances Ekman pumping and induces stronger upwelling on the east side of the TS than that on the west side, driving sea surface water masses away from the center of the TS. As a result, the water accumulates on the east side of the TS, and the SSH rises. In addition to local differences, the eastward propagation phenomenon also occurs. The negative SSH difference between the TSF and NONTSF in the Western Pacific caused by TS Dujuan propagates westward in the form of Rossby waves and remains in the Western Pacific. Over the next ~60 days, the positive SSH difference propagates eastward at a speed consistent with Kelvin waves ( Figure 11C , yellow dashed line). The signal raises the SSH in the eastern equatorial Pacific and ultimately reaches the eastern boundary.

Compared to the NONTSF, the TSF has positive temperature differences; the maximum value is about 0.34°C between 130°E and 135°E on February 21. The characteristics of the temperature differences between the TSF and NONTSF experiments are also consistent with those of the SSH differences between the two experiments ( Figure 11F ).

The characteristics of the temperature differences between the TSF and NONTSF experiments along the TS track could also be explained by the fact that the direction of wind stress on the east side of the TS is aligned with that of the Ekman drift. As the TS moves from south to north, the energy input to the ocean on the east side of its track will be greater than that on the west side. The vertical mixing, entrainment, and Ekman pumping that occur on the east side can be more intensive and more likely to transport the subsurface low-temperature seawater up to the sea surface. Due to the heat-pumping effect of TS, more warm water masses at the surface can be transported downward by entrainment and mixing, warming the subsurface seawater. Moreover, the negative wind stress curl is input by the TS on the left side of the track ( Figure 5 ), causing the equator-ward Sverdrup transport. The subsurface seawater is conveyed to the equator and accumulated, resulting in a change in the equatorial D20. The temperature differences in the upper ocean then propagate eastward with Kelvin waves. Since the temperature differences between the TSF and NONTSF experiments are small and the upper ocean temperature difference can only reach ~120°E, we speculate that the effect of TS Dujuan on the equatorial Kelvin wave is an amplification rather than an excitation. The differences between the TSF and NONTSF highlight the fact that the temperature change in the upper ocean of the western Pacific is caused by TS Dujuan rather than the climatological winds ( Figures 11C, F, I ). The TS winds induce the vertical mixing and intensify the air-sea heat exchange, causing the decrease of the temperature in the upper ocean on the west side of the forcing area, whereas the increase on the east side. The negative temperature differences on the west side present in the western Pacific, whereas the positive temperature difference on the east side moves eastward to a certain extent. TS Dujuan causes much more local changes than remote changes in SST ( Figure S2A ). In order to verify the negative feedback of Dujuan on La Niña events, the Niño3.4 index is calculated. The ENSO cycle in TSF enters the neutral state 1 to 2 days before that in the NONTSF ( Figure S2B ), and the difference between the two is slight due to low ACE, reflecting the contribution of TS-forced Kelvin waves in the La Niña event. However, for a significant impact on a La Niña event to be observed, the interactions and actions of multiple tropical cyclones are necessary (Sriver et al., 2013; Wang et al., 2019; Chen and Lian, 2020; Tong et al., 2023).

The effects of TS Dujuan on zonal currents in the upper ocean are analyzed based on our model simulations as well as a conceptual model (See Supplementary Material 2 ). A westward current is presented in both the TSF and NONTSF experiments ( Figures 11G, H ), which is in good agreement with the results shown in Figure 12A , further verifying the reliability of the model results. The distribution pattern of the speed differences between the TSF and NONTSF is similar to that of the SSH and temperature differences. TS Dujuan causes the zonal current in the equatorial region to produce the extreme value of the speed difference between TSF and NONTSF. The zonal current on the western side of the extreme value accelerates westward, and that on the eastern side accelerates eastward. The equatorial wind pulse induced by TS Dujuan produces the same acceleration characteristics in u ¯ and τ x ( Figures 12A–C, E ), and also results in alternating positive and negative changes in the zonal pressure gradient ( Figure 12F ), causing the Kelvin wave pulse during this period. Meanwhile, it is found that the TS-induced extreme values shown in Figures 11C, F, I lag behind the TS Dujuan activity. The cross-correlation analysis of Δ u ' and τ x indicates that the shear current lags behind the pulsing wind for ~6 days. Based on this, we calculate the vertical eddy viscosity coefficients using the e-folding scale and regression fitting and estimate the values to be 132 cm² s^-1 and 64 cm² s^-1 ( Figure 12D ), respectively. The latter is roughly half of the former, similar to the relation obtained by McPhaden et al. (1988). However, the estimated value is lower than that in previous studies. Because the selected point is placed at the edge of TS Dujuan and the duration of the wind pulse is relatively short, and the wind intensity is relatively low.

In TSF, TS Dujuan induces the increases in local SSH, upper ocean temperature, and zonal current speed. Those signals propagate eastward as Kelvin waves. Nevertheless, as seen in Figure 13 , the difference between TSF and NONTSF declines rapidly. The SSH, temperature, and zonal current speed are decreased to 33%, 30%, and 53% of their maximum values at the start of eastward-propagation, and then to only 11%, 3%, and 35% in the eastern Pacific. We speculate that the weak propagation of the temperature gradient to the eastern boundary is due to the strength of TS Dujuan and thermal damping. Consequently, we suggest that TS Dujuan is not the primary source of the Kelvin waves activity shown in Figures 2 – 4 , given the fact that the amplitude of SSH in NONTSF is lower than that in TSF by<4%.

In this study, the equatorial Kelvin waves are primarily excited by the reflection of Rossby waves at the western boundary of the Pacific Ocean. The diagnostic experiments allow us to detect the forcing effect of Dujuan on the equatorial Kelvin waves. Since the difference between the TSF and NONTSF rapidly decays during the eastward-propagation process, the temperature difference between the TSF and NONTSF does not propagate to the eastern boundary of the equatorial Pacific ( Figures 11F , 13B ). Moreover, the difference between the TSF and NONTSF experiments is much smaller than the change from the long-term average (anomalous signals), so we infer that TS Dujuan amplifies the Kelvin waves.

The 2020/2021 La Niña event began in August 2020 and ended in May 2021. TS Dujuan, which was active in the northwest Pacific in mid-February 2021, caused significant changes in the upper ocean in the equatorial Pacific, and the timing of the eastward propagation of the TS-forced wave coincided with the transition from the La Niña to a neutral state, which provided negative feedback to the development of La Niña.

In this study, the responses of the ocean to TS Dujuan are quantified and the effects of TS Dujuan on the equatorial oceanic Kelvin wave are analyzed and discussed based on satellite data, TAO buoy observations, and model simulations. According to both the satellite observations and simulations, the strong zonal wind anomaly which links to TS Dujuan elevates the local sea level, and pushes warm water from the sea surface into the subsurface, thereby deepening the D20, and accelerating zonal currents. These anomalies propagate eastward along the equator as Kelvin waves, the equatorial wave decomposition method is adopted to detect the wave. In addition to the zonal redistribution of heat, the spatial-temporal patterns of the surface GVAs reveal that the TS-forced Kelvin waves also redistribute the zonal current at the surface. We assume that there might be a connection between the TS Dujuan and the Kelvin wave because TS Dujuan happened near both the location and time of the onset of the equatorial Kelvin wave. To elucidate the forcing effect of TS Dujuan, we designed the diagnostic experiments using recursive filters to modify the surface wind forcing. The simulated SSH, upper ocean temperature, and zonal current are generally consistent with the observations, indicating that the modeling results are reliable.

Our simulation results suggest that even low-intensity cyclones like TS Dujuan can have a significant effect on the dynamics and propagation of Kelvin waves. The modeled Kelvin wave propagates faster than the observed one. Compared with the case in which the wind forcing of TS Dujuan is removed, TS Dujuan wind forcing induces a rise in the local SSH by ~2.1 cm and an increase in the upper ocean temperature by ~0.34°C, exhibiting a strong positive correlation. The simulation results also show that TS Dujuan promotes the acceleration of the zonal current. This acceleration lags behind the TS winds, and the acceleration effect spreads to the eastern Pacific. The equatorial upper ocean is impacted by TS Dujuan, causing Ekman suction to cool the surface seawater and heat the subsurface seawater (Emanuel, 2001). Combined with the negative wind stress curl input by TS Dujuan on the left side of its track, the warm water in the upper ocean is transported to the equator and further propagates eastward. Our results demonstrate that the transient wind forcing of TS Dujuan can redistribute the zonal heat and current of the upper ocean in the equatorial Pacific through the equatorial Kelvin waves. The outcomes of our model bear a striking resemblance to those derived from prior research (Sriver et al., 2013). However, a noteworthy deviation from other studies is observed in that the positive discrepancy rapidly descends to 30-53% of its peak. Additionally, the difference in the upper ocean temperature fails to extend to the eastern boundary of the equatorial Pacific. Also, the signal of SSH difference is significantly smaller than the SLA signal. Therefore, we infer that the forcing effect of TS Dujuan on the equatorial Kelvin wave is an amplification. In conclusion, the enhanced equatorial Kelvin wave propagates to the eastern Pacific, deepening the thermocline and inducing warm temperature anomalies, thus reducing the zonal thermal gradient of the equatorial Pacific during the La Niña phase and to a small extent accelerating the transition from the cold phase of ENSO to a neutral state in May 2021.

As the global warming, more strong tropical cyclones can be generated in the near-equatorial Pacific. It is estimated that before 2100 the continuous SST rise in the low-latitude area of the northwest Pacific will increase the intensity of tropical cyclones by ~14% (Emanuel, 2005; Webster et al., 2005; Mei et al., 2015). Further understanding of the forcing effects of near-equatorial TCs on equatorial Kelvin waves would help improving the predictions of ENSO events. Different from the previous researches that focused on asymmetric multi-TC forcing and symmetric twin-TC forcing (Keen, 1982; McPhaden et al., 1988; Nitta, 1989; Sriver et al., 2013), our study for the first time concentrated on the effect of a single TS. Based on the simulated results, we validly confirm the amplification effect of an individual TS on equatorial Kelvin waves from the perspectives of SSH, temperature, and zonal current. Owing to the limit of the TS wind stress intensity, the forcing effect is not very great. However, it can still be used as a reference for the study of the cross-scale influence of TCs on ENSO events. In addition, this study only investigated the forcing effect of the atmosphere on the ocean, without the coupling interactions between the ocean and the atmosphere (Boucharel et al., 2016). The impacts of ocean variability on the generation and development of TS Dujuan have not been considered.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ZD conceived and designed the study. YC conducted the numerical experiments and wrote the original draft. TY carried out the data analysis and language review. All authors contributed to the manuscript and agreed to the submitted version of the manuscript.