Introduction

The Maritime Continent (MC) occupies a pivotal position straddling the Indian and Pacific Oceans, as well as Asia and Australia, boasting the world’s largest archipelagic expanse with intricate geography1. Located near the center of the Indo–Pacific warm pool, the MC region is characterized by strong convection activity, yielding the highest annual mean precipitation in the world2,3. These convective dynamics over the MC constitute the ascending branch of the Indo–Pacific Walker circulation and the local Hadley circulation, thereby exerting a vital influence on global climate patterns4,5,6,7,8. On the other hand, the MC is known as one of the most densely populated regions and boasts the largest economy in Southeast Asia, and the variability in convection/rainfall here is thus crucial for local socio-economic activities. The region is highly vulnerable to extreme weather and climate events such as floods and droughts9. Particularly, droughts associated with rainfall anomalies can potentially lead to forest fires and consequently increase the haze risk10,11. Therefore, understanding rainfall variability in the MC is of great importance.

Rainfall in the MC is affected by various signals from surrounding regions and fluctuates across different timescales12,13,14,15,16. On the synoptic timescale, crucial weather elements in the MC include northerly cold surges17,18,19, tropical synoptic-scale waves20 and tropical synoptic-scale disturbances such as tropical depressions21,22 and the Borneo vortex18,23,24. On the intraseasonal timescale, the most important factor is the Madden–Julian Oscillation (MJO)16,25,26,27,28, which is in association with a large envelope of cloudiness and convection propagating from the Indian Ocean to the Pacific Ocean over approximately 30–60 days. As the MJO moves eastward, the well-organized large-scale enhanced and suppressed convection affect rainfall in the MC alternatively. In addition, the MJO also significantly affects extreme rainfall in the MC16,27. On the interannual to decadal timescales, large-scale circulation patterns–such as cross-equatorial flows28, the monsoon trough or the intertropical convergence zone29, and the monsoon29,30–as well as ocean modes like El Niño–Southern Oscillation, Indian Ocean Dipole, Pacific Decadal Oscillation play crucial roles in controlling MC rainfall12,13,14,15,31,32. There are complicated interactions between these multiscale factors, as documented by previous studies13,32,33,34,35,36,37.

Another significant synoptic weather system in the tropics is the tropical cyclone (TC), which often brings severe weather such as heavy rainfall, damaging winds, and even tornadoes38,39,40,41. Located near the equator, much of the MC is outside the direct path of TCs and is rarely impacted by landfalling TCs. For instance, Indonesia and Timor are generally TC-impact-free regions42. Since 2008, Indonesia has only experienced ten TCs, and only seven low-pressure systems have moved within close proximity (200 km) to Timor Island43.

TCs have another kind of effect on rainfall/convection, namely suppressive effect, in addition to the well-known heavy rainfall. The suppressive effect of TCs has been observed in many basins, including the western North Pacific (WNP)44 and Atlantic45. Recently, Li et al. 46 revealed that TCs in the South China Sea can effectively suppress monsoonal rainfall in southern China by altering the large-scale monsoonal moisture pathway46. In this study, we demonstrate that TCs in the tropical WNP can also significantly suppress rainfall in the MC from a synoptic climatology perspective, which contrasts with the previous cognization that the equatorial MC is rarely affected by TCs. In addition, both MC rainfall and TCs are impacted by intraseasonal and interannual environmental factors, such as the MJO, cross-equatorial flows, and the monsoon trough. Whether the TC–MC rainfall relationship is independent of these external factors remains unclear. Notably, since the convection anomalies associated with the TC–MC rainfall correspondence resemble those associated with the MJO47,48,49, we specifically analyze the impact of TCs during different MJO phases.

Results

Suppressive effect of TCs over the tropical WNP on rainfall in the MC

The tropical WNP hosts the highest frequency of TCs, with two prominent TC centers observed over the east and west Philippines, respectively (Supplementary Fig. 1a, b). There are no clear TC activities in the equatorial MC due to the absence of Coriolis force. Interestingly, MC rainfall shows a significant decrease during WNP TC days (Fig. 1a, c). Significant negative rainfall anomalies are observed across the equatorial MC and the surrounding seas. The probability density distribution of the MC rainfall index (MCRI) on TC days shows a clear decrease in the proportion of rainfall greater than normal (6.72 mm day−1) and an increase in the proportion of rainfall less than normal (Fig. 2b). The former drops from 53% on all days to 35% on TC days, while the latter rises from 47% to 65%. In addition, suppressed rainfall also appears over southern China, which is in association with TC activities in the South China Sea46.

Fig. 1: MC rainfall associated with the WNP TCs.
figure 1

a Composite rainfall anomalies (shading; mm day−1) associated with the tropical WNP TCs. Results are based on CPC data. Rainfall anomalies significant at the 95% confidence level are stippled as red. Green box indicates the MC region. b Probability density distributions of the MC rainfall index (MCRI) for TC days (orange line) and all days (black line). Vertical dashed line indicates the mean of MCRI on all days, respectively. Orange and cyan shadings represent the increased and decreased proportions of rainfall, respectively. c is as a, but based on ERA5 data that includes records over the sea.

Full size image
Fig. 2: Horizonal wind and its divergence anomalies in the lower and upper troposphere, respectively, as well as W500, associated with the WNP TCs.
figure 2

a Composite 850-hPa horizontal wind (vectors; m s−1) and 500-hPa vertical velocity (shading; Pa s−1) anomalies associated with TCs. Green box indicates the MC region. b is as a, but for the divergence (10−7 s−1) anomalies of 850-hPa horizontal winds. Shading significant at the 95% confidence level is stippled as red. c and d are as a and b, respectively, but for 150 hPa.

Full size image

TCs not only suppress moderate rainfall but also significantly reduce the occurrence of extreme rainfall and may increase the likelihood of drought over the MC (Fig. 1b). The proportion of extremely strong rainfall decreases from 5% on all days to 2% on TC days and that of extremely weak rainfall increases from 5% to 9%. These results indicate that on the synoptic timescale, TCs in the tropical WNP effectively suppress rainfall in the MC and its surrounding seas.

We confirm the suppressive effect of TCs on rainfall in the MC by examining TC activities associated with both enhanced and suppressed rainfall (Supplementary Fig. 1). Compared to enhanced MC rainfall (Supplementary Fig. 1b), TC activity in the tropical WNP notably intensifies during suppressed rainfall events (Supplementary 1a). The discernible differences are predominantly observed south of 22.5°N (Supplementary Fig. 1c), with less evident impacts north of this latitude. This suggests that TCs in the tropical regions exert a more pronounced influence on MC rainfall than those in the extratropics.

Mechanisms responsible for the suppressed MC rainfall induced by TCs

To clarify the mechanisms behind the suppressed MC rainfall caused by TCs, we analyzed circulation anomalies associated with the tropical WNP TCs. In the lower troposphere, a cyclonic anomaly emerges over the tropical WNP (Fig. 2a). Along its southern periphery, strong and significant southwesterly airflows from the equatorial MC converge toward the center of this cyclonic anomaly. In contrast, wind anomalies over the MC are weak, possibly due to its complex geography. These weak wind anomalies, along with robust southwesterly airflows to the south of TCs, result in significant divergence anomalies over the equatorial MC (Fig. 2b). In the upper troposphere, there are northeasterly anomalies to the south of TC convection (Fig. 2c), which converge over the MC (Fig. 2d). This upper-tropospheric convergence, combined with lower-tropospheric divergence, promotes subsidence and suppresses rainfall in the MC. The equatorial outflows of TCs in the upper troposphere, southeasterly inflows in the lower troposphere, ascending motion at the TC center, and descending motion over the MC suggest a closed vertical circulation cell. In addition, poleward southeasterly outflows north of TC convection in the upper troposphere correspond to anomalous subsidence and suppressed rainfall in South China (Fig. 2c). Compared to this northward branch, the southward outflows toward the MC extend over a broader meridional range, which is possibly due to the weak Coriolis force near the equator.

We further illustrate the role of TC circulation anomalies by examining the water vapor budget (Fig. 3). The vertically integrated moisture flux shows a cyclonic anomaly over the WNP (Fig. 3a), with southwesterly moisture flux diverging over the MC. These moisture divergence anomalies favor suppressed rainfall in the region. Integrations from the surface to 700 hPa and from 700 to 300 hPa indicate that moisture transport is primarily concentrated in the lower troposphere (Fig. 3b and Supplementary Fig. 2). Both zonal and meridional fluxes contribute to moisture divergence over the MC (Supplementary Fig. 3), with area-averaged divergence values of 0.51 and 0.17 × 10−5 kg m−2 s−1, respectively. Analysis based on Eq. (1) shows that the anomalous moisture divergence is dominated by the term \(\bar{q}\nabla {V}^{{\prime} }\) (Fig. 3f), confirming the crucial role of anomalous divergence induced by the lower-tropospheric southwesterly anomalies.

Fig. 3: Integrated moisture flux and its divergence, as well as the related components.
figure 3

a Composite water vapor flux integrated from the surface to 300 hPa (vectors; kg m−1 s−1) and its divergence (shading; kg m−2 s−1) anomalies associated with TCs. Green box indicates the MC region. b is as a, but integrated from the surface to 700 hPa. Only vectors significant at the 95% confidence level are shown and shading significant at the 95% confidence level is stippled as red. c–h Terms in the right-hand-side of Eq. (1) integrated from the surface to 700 hPa.

Full size image

One might be concerned about reversed causality, where MC rainfall could potentially suppress WNP TCs. To rule out this possibility, we investigated the evolution of anomalies associated with TC events (Supplementary Figs. 4–6). It is found that suppressed MC rainfall lags TC appearance by approximately 3 days, suggesting that TCs are responsible for suppressing MC rainfall rather than the other way around. In addition, although westerly anomalies over the MC are present on the first day of TC appearance, the lower-tropospheric divergence becomes strong and significant 3 days later (Supplementary Fig. 6), coinciding with significant rainfall anomalies. This suggests that the divergence anomalies responsible for the suppressed MC rainfall are primarily a result of TC circulation rather than environmental conditions.

Combined effect of tropical WNP TCs and the MJO on MC rainfall

We observed striking similarities in rainfall patterns between TCs and the MJO (Fig. 1c and Supplementary Fig. 7a). One might question whether the suppressed rainfall associated with TCs result from the impact of the MJO. This section aims to demonstrate that the suppressive effect of TCs on MC rainfall is somewhat independent of the MJO, and subsequently explore their combined impact on MC rainfall.

The MJO significantly affects rainfall in the MC and its surrounding seas (Supplementary Fig. 7a). MJO phases 2–4 enhance rainfall, while phases 6–8 suppress it (Fig. 4). When TCs coincide with the MJO, MJO-related rainfall in the MC decreases in all of the MJO phases (Fig. 4), indicating that the suppressive effect of TCs is independent of the MJO. Notably, TCs occur more frequently during MJO phases 6 and 7 (Supplementary Table 1), aligning with previous findings that these phases are conducive to TC formation47,48,49. Therefore, the suppressed rainfall during MJO phases 6–8 partly stems from TC activities.

Fig. 4: MCRI associated with MJO and MJO+TCs.
figure 4

MCRI anomaly (mm day−1) associated with each of the eight active MJO phases and the inactive MJO and the concurrent occurrence of TCs.

Full size image

To confirm the sole effect of TCs on MC rainfall, we conducted analyses focusing on TC days with inactive MJO (Supplementary Figs. 8 and 9). The results closely mirror those observed across all TC days (Figs. 1 and 2). Notably, rainfall anomaly tends to be zero on inactive MJO days but shows significant suppression in the presence of TCs (Fig. 4). The MCRI is −0.01 mm day−1 on inactive MJO days, contrasting with −0.45 mm day−1 during TC days with inactive MJO–a value very similar to that observed on all TC days (−0.53 mm day−1).

We merge MJO phases 2–4 and 6–8, respectively, to increase the sample size in each category (Fig. 5a, b). These two categories have similar case counts (3542 vs. 3410) and correspond to enhanced and suppressed rainfall in the MC and its surrounding seas, respectively (Fig. 5a, b). Rainfall anomalies associated with these two categories tend to be symmetrically opposite-signed, with the MCRI anomaly being 1.01 and −1.01 mm day−1, respectively. Concurrent TC occurrences significantly reduce enhanced rainfall associated with MJO phases 2–4 (Fig. 5c), with the MCRI anomaly dropping to 0.59 mm day−1. Conversely, suppressed rainfall associated with MJO phases 6–8 (Fig. 5b) is further reduced during concurrent TC events (Fig. 5d), with the MCRI anomaly decreasing to −1.37 mm day−1.

Fig. 5: UV850 and W500 associated with MJO and MJO+TCs, as well as their difference.
figure 5

Composite rainfall (shading; mm day−1) and 850-hPa horizontal wind (vectors; m s−1) anomalies associated with a MJO phases 2–4 and b 6–8, respectively. Green box indicates the MC region. c and d are as a and b, respectively, but with concurrent TCs. e is the difference between c and a and f is the difference between d and b. Rainfall anomalies over lands based on the CPC data are magnified in the bottom right-hand corner. Only vectors significant at the 95% confidence level are shown and shading significant at the 95% confidence level is stippled as red.

Full size image

The combined effects of TCs and the MJO on MC rainfall can be explained by lower-tropospheric circulation anomalies. During MJO phases 2–4 (Fig. 5a), westerly anomalies prevail over the equatorial Indian Ocean and easterly anomalies dominate the equatorial western Pacific. These anomalies converge over the MC, promoting ascent motion and enhancing rainfall. However, the appearance of TCs weakens the easterly anomalies over the equatorial western Pacific and enhances those over the outer-equatorial western Pacific. This alteration weakens the lower-tropospheric convergence over the MC, reducing ascent motion and rainfall. Conversely, during MJO phases 6–8 (Fig. 5b), easterly anomalies prevail over the equatorial Indian Ocean and westerly anomalies dominate the equatorial western Pacific. These anomalies diverge over the MC, inducing descent motion and suppressing rainfall. When TCs appear during these phases (Fig. 5d), the westerly anomalies in the equatorial western Pacific strengthen, promoting southwesterly offshore winds. This strengthens lower-tropospheric divergence over the MC, intensifying descent motion and further suppressing rainfall.

The difference between the TC–MJO combined anomalies and the MJO-related anomalies (Fig. 5e, f) resembles those associated with TCs (Fig. 2a), for both MJO phases 2–4 and 6–8, suggesting that these differences result from TCs activities. It is notable that although there is a great similarity in anomalous rainfall pattern between TCs and the MJO (Figs. 1c and 5b), the circulation anomalies are completely different. Specifically, TCs exhibit southwesterly anomalies to the north of the MC (Fig. 2a), while the MJO shows equatorial easterly and westerly anomalies (Fig. 5b).

Furthermore, the concurrent occurrence of TCs and the MJO may decrease the likelihood of extremely strong rainfall associated with phases 2–4 and increase the likelihood of extremely weak rainfall associated with phases 6–8. The probability density distributions of MCRI for different categories (Supplementary Fig. 10) show that TCs and MJO phases 6–8 correspond to a similar proportion of both extremely strong and weak rainfall. Extremely weak rainfall accounts for 9% on TC days and 10% on MJO phases 6–8 days (Supplementary Table 2). This percentage increases to 16% on MJO phases 6–8 days with simultaneous occurrence of TCs. On the other hand, extremely strong rainfall accounts for 9% on MJO phases 2–4 days, and this percentage decreases to 5% when there are concurrent TC activities.

In addition to the MJO, other environmental factors, such as cross-equatorial flows, the monsoon trough, and the WNP monsoon, may also influence both MC rainfall and the frequency of TC occurrences (Supplementary Figs. 11–14 and Supplementary Table 3). We examined the impact of TCs in these contexts and found that TCs suppress rainfall anomalies associated with these factors, regardless of the sign of the denoted indices (Supplementary Fig. 15). Moreover, the suppressive effect of TCs on MC rainfall exists under weak conditions of these factors (Supplementary Fig. 16), suggesting that the suppressive effect of TCs on MC rainfall is notably stable.

Discussion

Based on the climatological composite analysis, this study reveals that TCs over the tropical WNP significantly suppress rainfall in the MC. This finding challenges the common belief that the MC is rarely affected by TCs and introduces a new factor influencing its weather patterns from a synoptic climatology perspective.

The suppressive effect of TCs on MC rainfall exists in all of the phases of the MJO, a dominant factor influencing MC rainfall on an intraseasonal timescale. During MJO phases 6–8, which typically suppress MC rainfall, the concurrent occurrence of TCs can further increase the likelihood of extremely weak rainfall. This scenario can exacerbate drought conditions, fire events, and the resultant air quality issues in MC countries. Notably, MJO phases 6 and 7 are conducive to TC formation, making this configuration more frequent. Conversely, during MJO phases 2–4, which enhance MC rainfall, TCs coinciding with these phases diminish the enhanced rainfall, potentially hindering the eastward propagation of the MJO. Moreover, the suppressive effect of TCs across all phases of the MJO provides strong evidence that TCs suppress MC rainfall. That is, the connection between WNP TCs and MC rainfall is not barely a result of MJO’s impacts on them. Accordingly, this study also supports the negative relationship between WNP TCs and MC rainfall in summer (June–August) on the interannual timescale50.

On interannual and decadal timescales, TC frequency is modulated by sea surface temperature anomalies51,52,53,54. Whether these anomalies can affect the variability of MC rainfall through TC activities deserves further investigation. Additionally, the lag appearance of suppressed MC rainfall to TCs is noteworthy, suggesting that the suppressive effect of TCs may be related to their intensity. This topic also warrants further exploration.

Methods

Observational and reanalysis data

This study uses unified gauge-based global daily precipitation data provided by the Climate Prediction Center (CPC)55 and daily mean reanalysis data from the fifth generation European Centre for Medium–Range Weather Forecast (ECMWF) reanalysis (ERA5)56. The CPC rainfall data includes records over land and the ERA5 data also encompasses records over the sea. Both datasets have a horizontal resolution of 0.5° × 0.5°. TC data, recorded at 6-h intervals (0000, 0600, 1200, and 1800 UTC), are sourced from the Joint Typhoon Warning Center. This dataset includes all TCs, ranging from tropical depressions to typhoons. The analysis period spans from 1979 to 2020.

MJO RMM index

We use the Real-time Multivariate MJO index (RMM) created by Wheeler and Hendon (2004)57 to quantify the characteristics of the MJO. This MJO index is derived by projecting daily observed data onto the first two leading Multivariate Empirical Orthogonal Functions modes of 200- and 850-hPa zonal wind and OLR variability in the tropics. The process begins by meridionally averaging the variables within the band between 15°S to 15°N. Anomalies at each longitude are then calculated by removing the mean and the first three harmonics of the annual cycle. Following this, the El Niño–Southern Oscillation signal is eliminated from the anomalies through linear regression. Finally, for each day, the mean of the preceding 120 days is subtracted, as outlined in Wheeler and Hendon (2004). This index consists of two components, RMM1 and RMM2, which represent the standardized principal component time series derived from the first two EOFs. The MJO is categorized into eight phases based on the angle spanned by RMM1 and RMM2, and the MJO amplitude is quantified by the length of the vector formed by these two components. An active MJO is identified when the amplitude exceeds a threshold of 1.0. In this study, the anomalies associated with the MJO specifically pertain to periods of active MJO.

Definitions

The MC rainfall index (MCRI) is defined as the land rainfall averaged over the area 8°S–8°N, 95°–140°E (Fig. 1a). Following the definition of extreme rainfall at a single station, rainfall is considered extremely strong when it exceeds the 95th percentile of MCRI and extremely weak when it falls below the 5th percentile of MCRI. TC occurrences are counted in each 5° × 5° latitude–longitude grid. We focus on TC activities in the tropical WNP, specifically the area south of 22.5°N. This area is of particular interest due to the strong connection between TC occurrence frequency and MC rainfall (Supplementary Fig. 1c). A total of 5827 days with TC activities in this area were recorded during 1979–2020. The results associated with TC activities refer to the average of these TC days. Given that both the MJO and TC activities exhibit seasonal variations, for instance, TCs are active in July–September and MJO is strong during December–February, we have examined our major results month by month and found that TCs play a similar suppressive role throughout the year. Therefore, we present the results based on the entire year in this study.

Water vapor budget

The water vapor flux can be expressed as \({\bf{Q}}=q{\bf{V}}\), and the divergence of moisture flux can be decomposed as follows:

$$\nabla {{\bf{Q}}}^{\text{'}}={q}^{\text{'}}\nabla \overline{{\bf{V}}}+\overline{{\bf{V}}}\nabla {q}^{\text{'}}+{{\bf{V}}}^{\text{'}}\nabla \overline{q}+\overline{q}\nabla {{\bf{V}}}^{\text{'}}+{q}^{\text{'}}\nabla {{\bf{V}}}^{\text{'}}+{{\bf{V}}}^{\text{'}}\nabla {q}^{\text{'}}$$
(1)

The vertically integrated water vapor flux is given by:

$$\frac{1}{g}{\int }_{{p}_{s}}^{300}(q{\bf{V}})dp$$
(2)

where g, ps, q and V are the acceleration of gravity, surface pressure, specific humidity, and horizontal wind vector, respectively. Overbars denote climatological means and primes denote perturbations associated with TCs.

Statistical analyses and significance

This study primarily employs composite analysis. Daily anomalies are derived by subtracting the climatological values for a specific day from the raw data. A two-tailed Student’s t-test is used to assess the significance of each variable at every grid point58. The effective sample size, represented as N*, is computed at each grid point using the formula N* = N×(1 − r1)/(1 + r1), where N is the original length of the time series and r1 is the lag-1 autocorrelation coefficient59. A composite vector is deemed significant if either its zonal or meridional components passes the test.