Introduction

The western North Pacific (WNP) is the most active basin for tropical cyclone (TC) activity. As such, TCs that form in WNP during June–November often result in massive fatalities and economic losses in densely populated East Asian countries. Previous studies have extensively explored the relationship between WNP TC activity and climate modes of variability at various timescales1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22. At the interannual timescale, in particular, it is well-established that WNP TC activity is largely modulated by the El Niño-Southern Oscillation (ENSO)9,10,12,23,24,25,26,27. Specifically, during the positive phase of ENSO (i.e., El Niño), the genesis location of WNP TCs shifts southeastward, and the likelihood of intense TCs forming in the deep tropics increases. During the negative phase of ENSO (i.e., La Niña), on the other hand, the genesis location of WNP TCs shifts northwestward and the likelihood of TC landfalls over the East Asian countries increases9,10,12,23,25.

A number of studies have shown that the Pacific Meridional Mode (PMM)28,29,30,31,32 and sea surface temperature (SST) variability in the tropical North Atlantic (TNA)13,14,17, are also linked to WNP TC activity, especially during ENSO-neutral years. For example, a positive PMM tends to favor TC genesis in the southeastern WNP, whereas a negative PMM tends to inhibit TC genesis in the same region29,30,31. Similarly, warm TNA SSTs tend to suppress WNP TC activity via a change in the Atlantic-Pacific Walker circulation pattern, whereas cold TNA SSTs tend to enhance WNP TC activity13,14,17,33. However, the influences of PMM and TNA SSTs on WNP TC activity are quite incoherent between different years. For example, the early TC season (June–August, JJA) in 2018 was extremely active (i.e., 18 named TCs) with a clear northward shift in TC genesis34. Several studies identified a positive phase of the PMM that co-occurred as the major contributor to the extremely active early TC season since ENSO and TNA SSTs were neutral during that period34,35,36,37. However, WNP TC activity shifted northward during the early TC season in 2018 (Supplementary Fig. S1), which is inconsistent with a southward shift in TC activity that would result from a positive PMM according to previous studies29,30. This suggests the existence of other potential modulators of the early TC activity in 2018.

During the early TC season of 2018, Atlantic Niño/Niña, the leading mode of boreal summer tropical Atlantic SST anomalies (SSTAs), was in a strong positive phase (i.e., Atlantic Niño). Atlantic Niño/Niña is characterized by warm/cold SSTAs and low-level westerly/easterly wind anomalies in the eastern equatorial Atlantic, and is known to influence rainfall in the West African countries bordering the Gulf of Guinea and in northeastern South America38,39,40,41,42,43,44,45,46,47 as well as the formation of Cape Verde hurricanes48. Previous studies have suggested that Atlantic Niño/Niña influences atmospheric circulations over the tropical Pacific via an interbasin tropical teleconnection49,50,51,52,53,54,55,56,57. Thus, it is reasonable to hypothesize that Atlantic Niño/Niña events are linked to WNP TC activity, as might have been the case for the 2018 WNP TC season.

As briefly reviewed above, the correlation between global JJA SSTAs and JJA TC genesis over the WNP (5°–40°N, 100°E–160°W) shows well-established SSTA patterns related to WNP TC activity, such as positive PMM, El Niño, and cold SSTAs over the TNA (Fig. 1a). However, separating TCs formed in the northern and southern halves of the WNP, two distinctive global SSTA patterns emerge. TC genesis in the northern half of WNP (22.5°N-40°N, 100°E–160°W) is significantly correlated with equatorial SSTAs in the Pacific and Atlantic. Specifically, La Niña and Atlantic Niño-like patterns are linked to TC genesis in the northern WNP (Fig. 1b). On the other hand, the global SSTA pattern linked to TC genesis in the southern half of WNP (5°N–22.5°N, 100°E–160°W) appears to closely resemble that of TC genesis in the entire WNP, indicating that SSTAs in this region play a dominant role in determining to the total number of TCs that form in WNP (Fig. 1c). However, there is a clear difference in SSTAs over the equatorial Atlantic region. Specifically, there is a significant negative correlation between TC genesis in the southern WNP and the equatorial Atlantic SSTAs, indicating that TC genesis in the southern WNP is linked to an Atlantic Niña-like pattern. Therefore, the appearance of Atlantic Niño in JJA 2018 is consistent with the concurrent northward shift in WNP JJA TC activity (Fig. 1b). This further suggests that Atlantic Niño/Niña could modulate or contribute to the meridional shift in the location of WNP TC activity. The main objective of this study is to explore the relationship between Atlantic Niño/Niña and WNP TC activity (e.g., TC genesis and track density), and the associated atmospheric environments, focusing on the early TC season (i.e., JJA). The relative roles of Atlantic Niño/Niña versus ENSO on WNP TC activity are also investigated in this study.

Fig. 1: Relationship between sea surface temperature (SST) and western North Pacific (WNP) tropical cyclones (TC) by genesis locations during June–August (JJA).
figure 1

Correlation between JJA SST and JJA TC genesis over (a) the WNP basin (5°–40°N, 100°E–160°W; green box), (b) the northern part of the WNP (22.5°N-40°N, 100°E–160°W; green box), and (c) the southern part of the WNP basin (5°N-22.5°N, 100°E–160°W; green box) during 1979–2022. Gray dots indicate significance above the 90% confidence level based on a Student’s t test.

Full size image

Results

WNP TC activity during Atlantic Niño/Niña and ENSO

The spatial patterns of JJA TC genesis linked to Atlantic Niño/Niña are examined using a composite analysis for the 44-year observation record (1979–2022). During Atlantic Niño years, TC genesis increases significantly over the East China Sea between 20°N–40°N and decreases over the South China Sea and the Philippine Sea (0°–20°N, Fig. 2a). In contrast, during Atlantic Niña years, TC genesis increases over the South China Sea and decreases over the East China Sea (Fig. 2b). This meridional dipole pattern of TC genesis, as depicted in the composite maps, aligns with the positive (negative) correlation of SSTAs over ATL3 with TCs generated in the northern half (southern half) of the WNP, as shown in Fig. 1b, c. Consistent with the spatial pattern, the number of TCs generated in the northern half of the WNP (100°E–160°W, 22.5°N–40°N) is significantly higher during Atlantic Niño years (4.57 ± 0.33 per year) and lower during Atlantic Niña years (3.00 ± 0.29 per year), as shown in Fig. 2c. The robustness of this relationship is tested by performing additional composite analyses using two other TC best-track datasets (Supplementary Fig. S2). Despite some differences in the number of TCs formed in the northern WNP, all three TC best-track datasets used in this study consistently show an increase in TC genesis in the northern WNP and a decrease in the southern WNP during Atlantic Niño years, and almost the opposite pattern during Atlantic Niña years.

Fig. 2: Composite of TC genesis during Atlantic Niño/Niña and El Niño-Southern Oscillation (ENSO) years.
figure 2

Spatial patterns of anomalous JJA TC genesis (per year) composites during (a) Atlantic Niño and (b) Atlantic Niña. c Number of JJA TCs over the northern WNP (100°E–160°W, 22.5°N-40°N) during Atlantic Niño (red bar), neutral (gray bar) and Atlantic Niña (blue bar) years. The error bars indicate the 95% confidence level based on a two-tailed Student’s t test. d–f Are the same as (a–c) but for ENSO years. Note that TC genesis is spatially smoothed to aid visual comparison. Purple dots indicate significance above the 90% confidence level based on a Student’s t test.

Full size image

On the other hand, ENSO, the primary modulator of TC activity over the WNP, is associated with a slightly different spatial pattern of TC genesis compared to that associated with Atlantic Niño/Niña. During El Niño, TC genesis increases greatly in the southeast flank of the WNP and decreases slightly in the northwest flank of the WNP, while during La Niña (Fig. 2d, e), TC genesis increases slightly in the northwest flank of the WNP and decreases greatly in the southeast flank of the WNP, consistent with previous studies10,11,12. These variations in TC genesis in the northwest and southeast flanks of the WNP between El Niño and La Niña years lead to a significant difference in the number of TCs generated in the southern half of the WNP. Although weaker, a similar difference is found in the northern half (Fig. 2f). Specifically, the number of TCs generated in the northern half of the WNP is lower during El Niño years (3.11 ± 0.21 per year) and significantly higher during La Niña years (4.07 ± 0.32 per year).

As shown in Fig. 2, despite some differences in the spatial patterns of TC genesis modulated by ENSO and Atlantic Niño/Niña, the numbers of TCs generated in the northern flank of the WNP during La Niña (El Niño) and Atlantic Niño (Atlantic Niña) are similar. This suggests that ENSO and Atlantic Niño/Niña may not be entirely independent. Supporting this conjecture, the correlation between Niño3.4 and ATL3 indices in JJA during 1979–2022 is −0.37, which is statistically significant based on a two-tailed Student’s t test. Consistent with this, previous studies have shown that both ENSO and Atlantic Niño/Niña can influence each other with a time lead-lag44,50,51,53,56,58,59,60,61. For instance, a developing La Niña/El Niño in boreal winter can lead to Atlantic Niño/Niña in the following spring and summer59,61, and Atlantic Niño/Niña in boreal summer influences the development of ENSO in the subsequent winter months51,53. Therefore, to properly examine the individual impacts of ENSO and Atlantic Niño/Niña on WNP TC activity, we carry out a partial regression analysis of WNP TC onto JJA ATL3 and Niño34 indices. Note that the PMM and ATL3 are not correlated (r = −0.02). Thus, we focus on the impacts of Atlantic Niño/Niña and ENSO on the WNP TC.

TCs and associated environments in the WNP during Atlantic Niño and La Niña

During Atlantic Niño, strong anomalous upper-level divergence prevails over the tropical Atlantic, which is accompanied by anomalous upper-level convergence over the western Pacific warm pool region (Fig. 3a). Meanwhile, during La Niña, strong anomalous upper-level convergence appears over the central tropical Pacific and western Pacific warm pool region, together with compensating anomalous upper-level divergence over the Indian Ocean (Fig. 3c). These anomalous upper-level atmospheric circulation patterns decrease precipitation over the western Pacific warm pool region, similar to what occurs during Atlantic Niño (Fig. 3b). The anomalous upper-level convergence and dry conditions over the central tropical Pacific, driven by Atlantic Niño and La Niña, are also reasonably simulated in CESM-LENS (Methods), despite subtle differences in the zonal position and extent of the anomalous upper-level convergence associated with Atlantic Niño compared to the reanalysis (Supplementary Fig. S3). These results suggest that the anomalous upper-level subsidence over the western Pacific warm pool region during La Niña and Atlantic Niño may work in tandem to suppress low-latitude TC genesis in the WNP, consistent with the composite of TC genesis illustrated in Fig. 2a, e.

Fig. 3: Response of June–August (JJA) upper-level atmospheric circulation and precipitation to Atlantic Niño and La Niña.
figure 3

Partial regression of JJA velocity potential at 200 hPa (shaded, 10−6 m2 s−1) and sea surface temperature (contours, interval 0.3 K) on (a) ATL3 and c NINO3.4 indices for 44 years (1979–2022). Vectors are divergent winds at 200 hPa (m s−1; omitted below 0.4 m s-1). b, d Are the same as (a) and (b) but for precipitation (mm d−1). Note that the sign of NINO3.4 is reversed. Purple dots indicate significance above the 90% confidence level based on a Student’s t test.

Full size image

To separate the influences of Atlantic Niño/Niña from those of ENSO on the WNP TC activity, we next explore the partial regressions of TC genesis, track density, and TC-related environments onto the ATL3 and Niño3.4 indices. Atlantic Niño decreases TC genesis in the southern half of the WNP, including the South China Sea and the Philippine Sea (Fig. 4a). In these regions, anomalous subsidence in the western Pacific warm pool region decreases convection and induces pronounced low-level easterly wind anomalies, which decrease both mid-level relative humidity and low-level relative vorticity east of around 130°E between 0°–20°N (Fig. 4c, e), producing positive lifted index (LI) anomalies in that region (Supplementary Fig. S4a). These large-scale atmospheric conditions decrease the likelihood of TCs forming in the southern flank of the WNP, decreasing TC track density around the Philippines and Southeast Asia (Fig. 4a). In contrast, Atlantic Niño increases TC genesis in the northern half of the WNP, along the southern seaboard of Japan, increasing TC track density and the probability of TC landfall along Korea, Japan, and the eastern seaboards of China (Fig. 4a). The enhanced TC activity is driven by anomalous subsidence motion in the tropical WNP (0°–10°N), which leads to anomalous ascending motion in the subtropical belt of WNP (15°–30°N, Fig. 4g). In addition, the anomalous subsidence in the tropical WNP produces an anomalous cyclonic circulation in the subtropics, which can be described as the Rossby wave response to anomalous diabatic cooling induced by suppressed convection in the tropical WNP62,63,64,65,66. It leads to significant positive low-level relative vorticity anomalies in the subtropical belt (Fig. 4e), largely consistent with the results derived from CESM-LENS (Supplementary Fig. S5a, c). As a result, the likelihood of TCs forming in the northern flank of WNP increases, which in turn increases TC track densities near Korea and Japan (Fig. 4a).

Fig. 4: Response of TC-related environments to Atlantic Niño and La Niña.
figure 4

Partial regressions of TC genesis (shaded) and track density (contours) onto (a) Atlantic Niño (ATL3) and (b) Niño 3.4 (NINO3.4) indices. c, d Are the same as (a, b) but for relative humidity (%) at 700 hPa. e, f Are the same as (a, b) but for relative vorticity (shaded, 106 s−1) and wind vectors (m s−1, omitted below 0.5 m s−1) at 850 hPa. g, h Are the same as (a, b) but for zonally averaged (100°E–160°W) omega (shaded, hPa s−1, contour lines are climatology). Note that TC genesis and track density are spatially smoothed to aid visual comparison. Purple dots indicate where TC genesis, relative humidity at 700 hPa, relative vorticity at 850 hPa, and omega regressions are significant above the 90% confidence level based on a Student’s t test.

Full size image

Similarly, La Niña decreases TC genesis in the southeast WNP and increases TC genesis in the northwest WNP. This generates a significant decrease in TC track density over most of the southern half of the WNP and an increase in TC track density near the east coast of China (Fig. 4b)10,12,23,25,67. During La Niña, anomalous subsidence motion appears in the central Pacific, decreasing mid-level relative humidity in the southeast WNP (Fig. 4d), and producing positive LI anomalies in the same region (Supplementary Fig. S4b). Negative (i.e., anticyclonic) low-level relative vorticity anomalies are also observed in the southern half of WNP (Fig. 4f). Strong anomalous subsidence motion appears in the tropical North Pacific in the latitude band of 10°–20°N (Fig. 4h), thus showing a slight northward shift from the Atlantic Niño case. The anomalous subsidence motion in turn produces a significant decrease in TC genesis and TC track density in the southern WNP (Fig. 4b). The low-level relative vorticity and zonally averaged omega anomalies associated with La Niña are well reproduced in CESM-LENS (Supplementary Fig. S5b, d), showing the robustness of the results.

We also carry out several AGCM sensitivity experiments (Methods) to further test the results from the observational and reanalysis datasets (Supplementary Fig. S6). An experiment with prescribed Atlantic Niño SSTAs (+ATL3 experiment) shows that anomalous ascending motion in the subtropical belts of WNP (15°N–35°N) leads to anomalous low-level cyclonic circulation and relative vorticity in that region. On the other hand, another experiment with prescribed La Niña SSTAs (-NINO34 experiment) shows anomalous subsidence motion in the southern WNP (0°–20°N), and a decrease in relative humidity in that region. These, in turn, increase low-level easterlies and decrease relative vorticity in the southern WNP, supporting the observational results shown in Fig. 4.

Impact of Atlantic Niño/Niña on TC landfall in far eastern Asia

Figure 5a shows the spatial distribution of the total number of landfalling TCs in the WNP. TC landfalls are widespread across South China, Taiwan, and Luzon Island (i.e., the northern part of the Philippines), yet limited in far eastern Asian countries including Korea and Japan. As shown in Fig. 4a, Atlantic Niño increases TC genesis in the northern WNP while Atlantic Niña decreases TC genesis in that region. Therefore, it is reasonable to further hypothesize that Atlantic Niño increases the likelihood of TC landfall in far eastern Asian countries, particularly Korea and Japan, whereas Atlantic Niña decreases TC landfall in these countries. Consistent with this hypothesis, there is a statistically significant difference in the number of TC landfalls in Korea and Japan (green box in Fig. 5a: 125°–150°E, 30°–50°N) between Atlantic Niño and Niña years. Specifically, the number of TC landfalls in Korea and Japan is significantly higher during Atlantic Niño years (2.0 ± 0.22 per year) than during Atlantic Niña years (1.57 ± 0.13 per year).

Fig. 5: Difference of landfalling TC in Korea and Japan between Atlantic Niño/Niña and ENSO years.
figure 5

a Spatial pattern of the total number of landfalling TC generated in the WNP basin for 44 years (1979–2022). b Number of landfalling TC in Korea and Japan (125°E–150°E, 30°N–50°N, green box in Fig. 5a) during Atlantic Niño (red bar), neutral (gray bar) and Atlantic Niña (blue bar) years. c Is the same as (b) but during El Niño (red bar), neutral (gray bar), and La Niña (blue bar) years. The error bars indicate the 90% confidence level based on a two-tailed Student’s t test. Note that TC landfall is counted when the TC center passed within 60 miles of land, which is the same method used by IBTrACS.

Full size image

On the other hand, the difference of the number of TC landfalls in Korea and Japan between El Niño (1.9 ± 0.24 per year) and La Niña (2.1 ± 0.16 per year) is not statistically significant based on a two-tailed Student’s t test at a 90% significance level (Fig. 5c). This may be due to the fact that the influence of ENSO on WNP TC activity is weaker during the early TC season compared to that during the late TC season (September-November, SON)25,68.

Interactive impacts of ENSO and Atlantic Niño/Niña on the WNP TC activity

Although Atlantic Niño/Niña modulates the location of WNP TC genesis during the early TC season, ENSO is still the primary modulator of WNP TC activity, as shown in Fig. 2. Additionally, previous studies44,48,53,58 and the significant correlation between Niño3.4 and ATL3 indices suggest that the superposition of ENSO and Atlantic Niño/Niña events can further amplify or suppress WNP TC activity. Therefore, it is important to examine the interactive effects of ENSO and Atlantic Niño/Niña on WNP TC activity.

Figure 6 shows the linear combinations of the influences of Atlantic Niño/Niña and ENSO on TC genesis and TC track density, based on the partial regression analysis. Overall, WNP TC activity is largely determined by ENSO, while Atlantic Niño/Niña modulates ENSO’s impacts on WNP TC activity. For example, during La Niña and Atlantic Niño (Fig. 6a), significant increases in TC genesis and TC track density are observed over the East China Sea, Korea, and Japan, while decreases are seen in the southern flank of the WNP. On the other hand, overall TC activity decreases during La Niña and Atlantic Niña, except over the South China Sea (Fig. 6b). Note that the patterns for El Niño combined with Atlantic Niño and for El Niño combined with Atlantic Niña are the reverse of Fig. 6a, b, respectively. These suggest that the meridional dipole pattern of TC activity is enhanced when ENSO and Atlantic Niño/Niña are out of phase and suppressed when they are in phase. Despite the small sample size, this relationship is also evident in the composite analysis (Supplementary Fig. S7). This indicates that Atlantic Niño/Niña can either amplify or diminish the impacts of ENSO on WNP TC activity during the early WNP TC season, even though the impact of Atlantic Niño/Niña on WNP TC activity is weaker than that of ENSO.

Fig. 6: Interactive influences of Atlantic Niño and La Niña on TC genesis and TC track density.
figure 6

Partial regression analysis showing the interactive impacts of Atlantic Niño/Niña and El Niño/La Niña on WNP TC activity. a Linear summation of the partial regression of TC genesis (shaded), and track density (black contours) onto -NINO3.4 and ATL3 indices. b Is the same as (a) but for linear subtraction. TC genesis and track density are spatially smoothed to aid visual comparison. Purple dots indicate where TC genesis regressions are significant above the 90% confidence level based on a Student’s t test.

Full size image

Discussion

In this study, we showed that Atlantic Niño/Niña modulates early-season (JJA) WNP TC activity by shifting the location of TC genesis meridionally through Atlantic-Pacific interbasin teleconnections. During Atlantic Niño years, anomalous upper-level convergence develops in the tropical Pacific. The associated anomalous subsidence motion decreases precipitation in the western Pacific warm pool region, causing low-level easterly wind anomalies in the southern flank of the WNP. This results in negative (i.e., anticyclonic) low-level relative vorticity anomalies and also reduces mid-level relative humidity in that region. Conversely, anomalous ascending motion occurs in the northern flank of the WNP, producing positive (i.e., cyclonic) low-level relative vorticity anomalies. These environmental conditions lead to an increase in TC activity over the northern flank of WNP and a decrease over the southern flank. The increased TC genesis and track density in the northern WNP heighten the risk of landfalling TCs in far eastern Asian countries, particularly Korea and Japan. Further analysis shows that although overall WNP TC activity is largely determined by ENSO, the impact of ENSO on WNP TC activity can be amplified when ENSO and Atlantic Niño/Niña are out of phase and suppressed when they are in phase.

There are several key points that may benefit from further discussion and clarification. First of all, this study focused solely on the influence of Atlantic Niño/Niña on the WNP TC activity during the early WNP TC season (JJA), as the Atlantic Niño/Niña does not cause a statistically significant change in WNP TC activity during the late WNP TC season (SON). Atlantic Niño/Niña typically peaks in JJA and weakens in SON40,42,44,45, thus leading to the most pronounced impact on WNP TC activity during the early WNP TC season. The interbasin teleconnection of Atlantic Niño/Niña also varies greatly between the early and late WNP TC seasons. For example, the anomalous upper-level atmospheric circulation pattern associated with Atlantic Niño/Niña during JJA is distinctly different from that observed during SON (Fig. 3a and Supplementary Fig. S8a). Specifically, the anomalous upper-level divergence in the tropical Atlantic driven by Atlantic Niño during SON is comparable to that seen during JJA. However, the anomalous upper-level convergence in the tropical Pacific near the Western Pacific warm pool region seen during JJA vanishes during SON. This results in incoherent changes in precipitation over the central tropical Pacific and Western Pacific warm pool regions in SON, differing notably from the significantly suppressed precipitation pattern observed in the Western Pacific warm pool region during JJA (Fig. 3b and Supplementary Fig. S8b). These differences in the tropical Pacific atmospheric anomalies may account for the negligible impact of Atlantic Niño/Niña on WNP TC activity during SON. On the other hand, the tropical Pacific atmospheric circulation and precipitation patterns associated with ENSO are quite similar between JJA and SON seasons (Fig. 3c, d and Supplementary Fig. S8c, d). Nevertheless, the influence of ENSO on WNP TC activity is still much stronger during SON, largely because ENSO is in its developing phase in SON and in its onset and developing phase in JJA, thus is much stronger in SON than in JJA25,68.

It is also worthwhile to note that there are other known factors modulating WNP TC activity, but not discussed in this study. For example, previous studies have shown that certain conditions in the Arctic (e.g., Arctic sea ice69 and Arctic Oscillation70) and the regional Hadley circulations in the Pacific71 can modulate WNP TC frequency, genesis location, and tracks. Therefore, it is of interest to further investigate the roles of Atlantic Niño/Niña and ENSO in modulating WNP TC activity under various conditions of these other factors.

Lastly, previous studies have shown that the seasonal prediction skill of Atlantic Niño/Niña is significantly lower than that of ENSO72,73,74,75,76. However, according to those studies, skillful Atlantic Niño/Niña forecasts can still be achieved at lead times of up to 3–4 months. This suggests that Atlantic Niño/Niña predicted based on boreal spring initial conditions may serve as guidance or as a potential predictor to enhance the current seasonal prediction skill for WNP TC activity during the early TC season, especially in ENSO-neutral years.

Methods

Observational and reanalysis datasets

We used WNP TC best-track data from the International Best Track Archive for Climate Stewardship (IBTrACS)77 version 4 to derive TC genesis, TC track density, and TC landfall. To detect TC genesis, we considered the first positions of TCs with wind speeds greater than 34 knots; thus, tropical depressions were excluded. In addition, TC landfall was defined as when the TC center passed within 60 miles of land, which is the same method used by IBTrACS77. To test the robustness of the results derived from IBTrACS. We also use two additional TC best-track datasets, one from the Regional Specialized Meteorological Center Tokyo at the Japan Meteorological Agency (JMA) and the other from the Joint Typhoon Warning Center. Monthly atmospheric variables (i.e., vertical wind shear, omega, relative humidity, relative vorticity, and LI) were obtained from the National Centers for Environmental Prediction—National Center for Atmospheric Research Reanalysis version 178. Monthly precipitation was obtained from NOAA’s Gridded Precipitation Reconstruction dataset79. All atmospheric variables were linearly detrended to avoid the potential influence of anthropogenic climate change. Extended Reconstructed Sea Surface Temperature version 5 (ERSSTv5)80 was used to derive Atlantic Niño/Niña and Niño3.4 indices. Given the poor quality of TC best-track data in the pre-satellite era (before 1979), we limit our analysis to 44 years (1979–2022) of data from the early summer WNP TC season (JJA), which is the peak season of Atlantic Niño/Niña.

Definition of Atlantic Niño/Niña and El Niño/La Niña

Atlantic Niño and Niña are identified based on the threshold that the magnitude of the area-averaged SSTAs over the ATL3 region (3S°–3°N, 20°W–0°) exceeds a half standard deviation (σ = 0.30 K) during June–August (JJA). The same method was used to identify El Niño and La Niña events (σ = 0.48 K), but using area-averaged SSTAs over the Niño3.4 region (5S°–5°N, 170°E–120 W°). In total, 14 Atlantic Niño, 14 Atlantic Niña, 11 El Niño, and 15 La Niña events are identified (Supplementary Table S1). In order to objectively separate the impacts of Atlantic Niño/Niña from ENSO, we computed partial regressions of TC genesis, track density, and environmental variables onto the ATL3 and Niño3.4 indices. Partial regression coefficients of ATL3 represent the change in TC activity, or other environmental variables, per unit change in ATL3 while Niño3.4 is held constant. Similarly, partial regression coefficients of Niño3.4 represent the change in the TC activity, or fields per unit change in Niño3.4 while ATL3 is held constant47.

Community earth system model-large ensemble simulation

To test the results derived from the observational and reanalysis datasets, we analyzed 1100 model-years from the Community Earth System Model-Large Ensemble Simulation (CESM-LENS) with a constant preindustrial CO2 level81. Atlantic Niño/Niña and ENSO indices from CESM-LENS were obtained in the same manner used in the observational analysis.

AGCM sensitivity experiments

To further test the robustness of the results derived from reanalysis and CESM-LENS, we carry out three sets of AGCM experiments (i.e., CTRL, +ATL3, and –NINO34 experiments) using CESM version 282. The atmospheric model has 27 vertical levels with a 1.98° × 2.58° horizontal resolution. The CTRL experiment is prescribed with monthly climatological SST for 1980–2020 based on ERSSTv5. The other two sensitivity experiments are prescribed with a combination of climatological SST and SSTAs composites of Atlantic Niño (+ATL3 experiment) and La Niña (-NINO34 experiment) years. All experiments are integrated for 30 years, and the last 20 years are analyzed to avoid potential spin-up issues.