Introduction

Precipitation is critical to the production and livelihood of more than 1.6 billion people in East Asia. It is a major contributor to water resources; however, an excess of rainfall triggers water disasters1,2. In this region, interannual changes in precipitation are closely related to the atmospheric circulations of the Pacific and Indian Oceans3,4. Hence, numerous studies have examined changes in the atmospheric circulations across this region5,6. As a fingerprint, stable oxygen isotope ratios (δ18O) recorded in East Asian speleothems7,8,9, tree-rings10,11,12,13, and deep-sea sediment cores14,15,16 have been used to reconstruct the history of atmospheric circulations. However, the interpretation of these paleoclimate records is clouded by an incomplete understanding of the link between the δ18O signals preserved in these records and the atmospheric circulations. In that regard, establishing the links between atmospheric circulations and annual precipitation δ18O (δ18Op) is essential for interpreting the δ18O records preserved in the paleoclimate archives17,18,19,20,21,22,23,24,25.

There have been a range of studies in southern East Asia that have attributed δ18Op variability to different atmospheric circulations. Some studies have discussed the relationships between annual δ18Op and the East Asian and Indian summer monsoons26,27,28,29,30,31 and proposed the competitive influences of these monsoons on the δ18Op32,33,34,35. However, apart from the monsoons, changes in annual δ18Op in southern East Asia have also been argued to be closely related to the westerlies, with higher δ18Op values associated with a southward-shifting westerly jet during the summer months36. Another group of studies has proposed that the El Niño Southern Oscillation (ENSO) controls the interannual variability of δ18Op in southern East Asia either by modulating the rainout intensity during moisture transport30,35,37,38,39 or by causing changes in convection within the moisture source regions40,41,42,43. However, the influence of ENSO on annual δ18Op in southern East Asia is unstable44. Therefore, factors other than ENSO appear to impact the interannual variability of δ18Op in southern East Asia, and a systematic study is required to explore these additional factors that affect δ18Op variability in this region.

As an important mode of interannual climate variability in the Indian Ocean, the Indian Ocean Dipole (IOD) is generally acknowledged to have a considerable influence on the precipitation across large parts of the tropics and even in the relatively distal southern East Asia45,46,47. However, the role of the IOD on the δ18Op in southern East Asia has received less attention. Importantly, the frequency of extreme positive IOD events significantly increases under global warming, and these events likely have a considerable influence on δ18Op in southern East Asia48,49. However, whether the changes in IOD decouple its relationship with the annual δ18Op in southern East Asia is unknown. These knowledge gaps motivate a further investigation in southern East Asia on the relationship between the IOD and annual δ18Op and the stability of this relationship over time.

In the western Pacific, a permanent large-scale anticyclonic circulation system known as the western Pacific subtropical high (WPSH) develops over the middle and lower troposphere and strongly modulates the climate of southern East Asia50,51,52. Previous studies have focused on the effects of the WPSH on the δ18Op on seasonal timescales in southern East Asia and shown that, during the pre-monsoon season, the adjacent oceanic moisture driven by the WPSH causes relatively high δ18Op values over the region53,54. However, few studies have examined the influence of the WPSH on the δ18Op in southern East Asia on an interannual timescale. Moreover, several studies found that the frequency of strong WPSH events is also increasing under global warming55,56. However, the nature and strength of the relationship between the WPSH and the annual δ18Op in southern East Asia have not been assessed. Therefore, the role of the WPSH on the annual δ18Op in southern East Asia remains to be determined.

This study aims to identify the possible influences of the IOD and WPSH on the δ18Op in southern East Asia on interannual timescales. To achieve this, we first retrieve the observed δ18Op data in Hong Kong over the period 1961–2021 and modeled δ18Op data across southern East Asia over the period 1979–2020. We then clarify the linkages between the IOD and WPSH and the annual δ18Op in southern East Asia, respectively. Finally, we reveal the mechanisms of how the IOD and WPSH modulate annual δ18Op variability across southern East Asia.

Results

Relationships between δ18Op and IOD with WPSH

The interannual variations in the δ18Op at Hong Kong during 1961–2021, along with the DMI and WPSHII, are shown in Fig. 1. The δ18Op values show considerable interannual fluctuations, with the values ranging from −6.51‰ to −3.73‰ and a mean of −4.91‰ ± 0.61‰ (Fig. 1b, c). These data show that the trends in annual δ18Op at Hong Kong closely follow the DMI between 1961 and 1998 (Fig. 1b). Indeed, over this timeframe annual δ18Op is significantly and strongly correlated to the DMI, with a correlation coefficient of 0.49 (n = 31, p < 0.01) (Fig. 1b). However, the correlation in the subsequent 1999–2021 period breaks down and is not significant (r = −0.09, n = 23, p = 0.70) (Fig. 1b). That is, the interannual variations in the δ18Op at Hong Kong are remarkably linked to the IOD before 1999, but the linkage disappeared after 1999. Similar comparisons between the annual δ18Op and the WPSHII show opposite changes in the correlations during the two periods to those between the annual δ18Op and the DMI (Fig. 1c). Before 1999 the correlation coefficient between the annual δ18Op and the WPSHII is 0.09 (n = 31, p = 0.63), and it increases to 0.47 thereafter (n = 23, p < 0.05) (Fig. 1c). Therefore, the linkage between the annual δ18Op and the WPSH became significant after 1999, which was completely opposite to the role that the IOD played on the annual δ18Op at Hong Kong (Fig. 1b, c). It is clear that both the IOD and WPSH can affect the changes of annual δ18Op at Hong Kong, but those impacts reversed around 1999.

Fig. 1: Location of Hong Kong and correlations between the DMI/WPSHII and the annual δ18Op at Hong Kong.
Fig. 1: Location of Hong Kong and correlations between the DMI/WPSHII and the annual δ18Op at Hong Kong.
Full size image

a Map of southern East Asia (black rectangle) showing the location of Hong Kong (black dot). The colors show the spatial distributions of the mean annual precipitation amount (P) from 1961 to 2021. The orange box marks the region where the IOD occurs. The orange circle marks the major region where the WPSH occurs. b, c Interannual variations in the δ18Op at Hong Kong compared with DMI (b), and WPSHII (c) during 1961–2021. The black dashed vertical line divides the study period into 1961–1998 and 1999–2021. d–g Spatial distributions of the correlation coefficients between the DMI (d, e), WPSHII (f, g), and annual δ18Op of each grid point during 1979–1998 (d, f) and 1999–2020 (e, g), respectively. Dotted areas highlight correlations exceeding the 95% confidence level. The black dots indicate the location of Hong Kong, and the black rectangles mark the region of southern East Asia.

We further analyzed the spatial correlations between the DMI/WPSHII and the annual δ18Op from the ECHAM6-wiso outputs over a broader southern East Asia (grid points) during 1979–1998 and 1999–2020, respectively (Fig. 1d–g). Similar to the results from the Hong Kong station, the clear shift in the correlations between the DMI/WPSHII and the annual δ18Op is also strongly evident across broader southern East Asia. There are positive correlation coefficients between the DMI and the annual δ18Op in southern East Asia before 1999 which weaken thereafter (Fig. 1d, e), while the positive correlation coefficients between the WPSHII and the annual δ18Op become much stronger after 1999 than those during the 1979–1998 period (Fig. 1f, g). Therefore, the shifting influences of the IOD and WPSH on the annual δ18Op occur across the whole of southern East Asia.

Impact of IOD on annual δ18Op

In light of the significant correlation between the IOD and annual δ18Op in southern East Asia, including Hong Kong station, which mainly appeared before 1999, we first consider how the IOD influenced the interannual variations in δ18Op in southern East Asia during the 1961–1998 period. We define the strongly negative and positive IOD years by the troughs and peaks of the DMI curve, respectively (Supplementary Fig. 1a). In this case, six strongly negative IOD years (1964, 1980, 1985, 1989, 1992, and 1996) and six strongly positive IOD years (1976, 1982, 1987, 1991, 1994, and 1997) were identified over the 1961–1998 period (Note that we have not considered the IOD events during 1966–1972, for there were no observed δ18Op data during this period).

During the strongly negative IOD years, the SSTs display positive anomalies in the Indo-Pacific Warm Pool and negative anomalies in the western tropical Indian Ocean (Fig. 2a). Those patterns in SST anomalies tend to strengthen convection in the Indo-Pacific Warm Pool and weaken convection in the western tropical Indian Ocean. Correspondingly, the opposite patterns in the spatial OLR anomalies between the Indo-Pacific Warm Pool (negative values) and the Western Tropical Indian Ocean (positive values) also demonstrate the strong convection in the Indo-Pacific Warm Pool and the weak convection in the Western Tropical Indian Ocean (Fig. 2c). The differences of convection in the two areas are favorable to enhance the Indian Ocean Walker circulation. The vertical velocity profiles clearly illustrate the changes in the Indian Ocean Walker circulation. As shown in Fig. 2e, the negative anomalies in the vertical velocity occur in the Indo-Pacific Warm Pool (80°E–150°E), including the eastern tropical Indian Ocean, while positive anomalies occur in the western tropical Indian Ocean and adjacent region (45°E–80°E). These results support that the stronger rising motion in the Indo-Pacific Warm Pool and the stronger sinking motion in the western tropical Indian Ocean enhance the Indian Ocean Walker circulation during the strongly negative IOD years.

Fig. 2: Spatial distributions of annual SST, OLR, and vertical velocity anomalies during the strongly negative and positive IOD years.
Fig. 2: Spatial distributions of annual SST, OLR, and vertical velocity anomalies during the strongly negative and positive IOD years.
Full size image

a, b Spatial distributions of annual SST anomalies during the strongly negative (a) and positive (b) IOD years. c, d Spatial distributions of annual OLR anomalies during the strongly negative (c) and positive (d) IOD years. The black dots indicate the location of Hong Kong, and the black rectangles mark the region of southern East Asia. e, f Vertical profiles of meridional mean vertical velocity (ω) anomalies for the latitude range 15°S–15°N during the strongly negative (e) and positive (f) IOD years. Negative and positive anomaly values for vertical velocity indicate rising and sinking motions, respectively. The anomalies are defined as the mean values during the strongly negative and positive IOD years subtracted by the mean values during 1961–1998, respectively.

Influenced by the enhanced Indian Ocean Walker circulation during the strongly negative IOD years, the moisture supplied from the equatorial Indian Ocean (southeastern side of the Indian Peninsula) to southern East Asia increases significantly (oval drawn with a solid pink line in Fig. 3a, c). On the contrary, the contribution of moisture from the northern Bay of Bengal and northern Indochina Peninsula to southern East Asia relatively decreases (rectangle drawn with a dashed blue line in Fig. 3a, c). Therefore, the influence of the moisture from the equatorial Indian Ocean on the δ18Op in southern East Asia is stronger. It should be noted that the moisture supplied from the equatorial Indian Ocean is characterized by lower δ18Ov values at 1000–500 hPa than the northern Bay of Bengal and northern Indochina Peninsula (Supplementary Fig. 2a), which provides a lower background value for the δ18Op in southern East Asia. In other words, the lower δ18Ov values will significantly decrease the δ18Op in southern East Asia. In addition, the moisture supplied from the equatorial Indian Ocean to the study area has a relatively long transport pathway and a larger latitudinal deference. Hence, the δ18Ov along the transport pathway experiences a stronger “rainout effect” and “latitude effect.” As a consequence, the subsequent δ18Op in southern East Asia becomes lower (Fig. 4a). More importantly, when the moisture passes through the strong convective region of the Indo-Pacific Warm Pool, the enhanced convection and rising motion in the region uplifts the moisture from the near-surface to higher altitudes which enhances the recycling of moisture. The recycled moisture is isotopically depleted due to the rainout process57,58 and could be advected toward the study area. Obviously, the enhanced convection further decreases the δ18Op values in southern East Asia (Fig. 4a). It is precisely because of the combined influences of the moisture supply and convection that the range of δ18Op values at Hong Kong during the strongly negative IOD years is as relatively as −4.36‰ to −6.38‰, with a mean of −5.09‰ (Figs. 1b, c and 4a).

Fig. 3: Atmospheric circulation patterns during the strongly negative and positive IOD years.
Fig. 3: Atmospheric circulation patterns during the strongly negative and positive IOD years.
Full size image

Vertically integrated moisture flux fields (a, b) and moisture flux field anomalies (c, d) at 1000–500 hPa during the strongly negative (a, c) and positive (b, d) IOD years. The anomalies are defined as the mean values during the strongly negative and positive IOD years subtracted by the mean values during 1961–1998, respectively. The black dots indicate the location of Hong Kong, and the black rectangles mark the region of southern East Asia. The solid line circles or rectangles with pink mark the areas of increased moisture flux, and the dashed line circles or rectangles with pink mark the areas of decreased moisture flux.

Fig. 4: Schematic diagrams of the impacts of the IOD and WPSH on annual δ18Op in southern East Asia before and after 1999, respectively.
Fig. 4: Schematic diagrams of the impacts of the IOD and WPSH on annual δ18Op in southern East Asia before and after 1999, respectively.
Full size image

a, b The impacts of the IOD on annual δ18Op in southern East Asia during the strongly negative (a) and positive (b) IOD years before 1999. c, d The impacts of the WPSH on annual δ18Op in southern East Asia during the extremely strong (c) and relatively weak (d) WPSH years after 1999. The δv indicates water vapor δ18O, and δp indicates δ18Op. The plus (minus) signs indicate increased (decreased) δ and SST anomalies values, and the numbers of the plus (minus) signs indicate the magnitudes of those increases (decreases).

During the strongly positive IOD years, however, the SST, OLR, and vertical velocity anomalies show the opposite patterns to those during the strongly negative IOD years (Fig. 2b, d, f). The results indicate that, under the influence of the strongly positive IOD events, the positive SST anomalies enhance convection and form the rising motion in the western tropical Indian Ocean, while the negative SST anomalies in the Indo-Pacific Warm Pool weaken convection and cause a sinking motion (Figs. 2b, d, f and 4b). This process leads to anomalous easterlies over the lower troposphere of the tropical Indian Ocean but anomalous westerlies over the upper troposphere of the tropical Indian Ocean (Fig. 4b). Clearly, the direction of the circulation caused by the strongly positive IOD events greatly weakens the Indian Ocean Walker circulation (Fig. 4b). In this case, the moisture contribution from the equatorial Indian Ocean to the study area is significantly reduced (ovals drawn with a dashed pink line in Fig. 3b, d). Meanwhile, more moisture from the northern Bay of Bengal and northern Indochina Peninsula is transported to southern East Asia (rectangles drawn with a solid pink line in Fig. 3b, d). The moisture supplied from the northern Bay of Bengal and northern Indochina Peninsula contains relatively high δ18Ov (Supplementary Fig. 2b), which contributes to the relatively high background values for the δ18Op in southern East Asia. In addition, such moisture is transported to the study area via a relatively short transport pathway and experiences a relatively small latitude difference, leading to weaker isotopic depletion along the moisture transport pathway (Fig. 4b). Furthermore, the weaker convection and even a sinking motion occur within the upstream Indo-Pacific Warm Pool, which result in less recycling of moisture. This process also limits isotopic depletion in the study area during the strongly positive IOD years (Fig. 4b). As a result, δ18Op values at Hong Kong during the strongly positive IOD years ranging from −3.73‰ to −4.46‰, with the mean value of −4.13‰, which is 0.96‰ higher than that during the strongly negative IOD years (Figs. 1b, c and 4b).

Our results show that the influence of the IOD on the annual δ18Op in southern East Asia greatly reduced after 1999. Here, we discuss the causes of this major shift from correlation to non-correlation in the linkage between DMI and annual δ18Op in our study area around the year 1999. Over the 1961–1998 period, there were fluctuations in the DMI around zero, with some positive and negative phases (Supplementary Fig. 1a). These fluctuations indicate that the IOD was in a symmetrical phase during this period. The pattern of the IOD led to the Indian Ocean Walker circulation showing interannual differences between positive and negative IOD years, which contributed to the significant interannual variations of the δ18Op in southern East Asia. However, the DMI over the 1999–2021 period is characterized by an apparent increasing trend, and almost all DMI values are positive (Supplementary Fig. 1a). These results indicate that the IOD represents positive phases after 1999, and the IOD changed from a symmetric to asymmetric pattern. This transition of the IOD reduces the interannual variability of the Indian Ocean Walker circulation after 1999 and likewise the influence of the IOD on the annual δ18Op in southern East Asia. As a result, the annual δ18Op variations in southern East Asia are decoupled from the changes in IOD during the period 1999–2021 (Fig. 1b, e).

Impact of WPSH on annual δ18Op

To investigate the changing influence of the WPSH on the annual δ18Op in southern East Asia before and after 1999, we analyzed the 500 hPa geopotential height anomalies during the two periods of 1961–1998 and 1999–2021, respectively (Supplementary Fig. 3). During 1961–1998, the 500 hPa geopotential height displays negative anomalies in the tropical and mid-latitude Pacific and forms an anomalously low-value center in the northern Pacific at about 37°N (Supplementary Fig. 3a). Correspondingly, the normalized WPSHII also displays frequent negative anomalies during this period (Supplementary Fig. 1b). These results indicate that the intensity of the WPSH was very weak in the 1961–1998 period. Therefore, during this period, the WPSH had negligible influence on the interannual variations of the δ18Op at Hong Kong and across southern East Asia. That is, the changes of the annual δ18Op in southern East Asia are independent of the changes of the WPSHII over the 1961–1998 period (Fig. 1c, f).

In contrast, the 500 hPa geopotential height during the 1999–2021 period displays positive anomalies in the tropical and mid-latitude Pacific and forms an anomalously high-value center in the northern Pacific (Supplementary Fig. 3b). In addition, the normalized WPSHII experiences an increasing trend and contains more frequent positive anomalies with very few negative anomalies during the 1999–2021 period (Supplementary Fig. 1b). These results indicate that the WPSH intensifies considerably during this period. Our results are consistent with previous studies finding an intensification of the WPSH during this period59. This intensification of the WPSH in the 1999–2021 period greatly influenced the interannual changes of the δ18Op across southern East Asia. In this case, the correlation between the interannual fluctuations of δ18Op across southern East Asia and the WPSHII between 1999 and 2021 becomes significant (Fig. 1c, g).

We then focus on the extremely strong (2010, 2016, and 2020) and relatively weak (2000 and 2012) WPSH years during the 1999–2021 period (the years were selected by the WPSHII in Supplementary Fig. 1b) to explore how the WPSH affects the annual δ18Op in southern East Asia. During the extremely strong WPSH years, the 500 hPa geopotential height field anomalies show a significant positive value center in the northern Pacific at about 30°N (Fig. 5a). The extremely strong WPSH expands westward and largely blocks the eastward transport of moisture from the Indian Ocean. This results in decreased moisture supply from the Indian Ocean (rectangles drawn with dashed pink lines in Fig. 5c, e). In comparison, this extremely strong WPSH drives a strong anticyclone circulation to form in the northwestern Pacific (110°E–180°, 0°–30°N) (Fig. 5c, e). The easterly winds on the south side of the anticyclone drove moisture from the northwestern Pacific westward, and the southerly winds on the west side of the anticyclone transported the moisture northwards to Hong Kong (ovals drawn with a solid pink line in Fig. 5c, e). Consequently, the moisture sourced from the northwestern Pacific and delivered to southern East Asia considerably increased in the extremely strong WPSH years (ovals drawn with a solid pink line in Fig. 5c, e). It is worth noting that the δ18Ov in the northwestern Pacific is higher than in the Indian Ocean (Supplementary Fig. 4a), indicating the relatively high background δ18Ov. This can result in higher δ18Op values in southern East Asia. In addition, as more isotopically enriched moisture traveled a shorter distance and a relatively small latitude range from the northwestern Pacific to southern East Asia in the extremely strong WPSH years, the isotopic depletion was comparatively weaker and the δ18Op in southern East Asia is higher (Fig. 4c). Furthermore, the OLR values in the extremely strong WPSH years show positive anomalies in the Indo-Pacific Warm Pool (Fig. 6a), which imply a significantly weakened convection in this region. The weak convection along the transport pathway also contributes to the relatively high δ18Op values during the extremely strong WPSH years (Fig. 4c). As a result, the range of δ18Op values at Hong Kong extends from −4.27‰ to −5.11‰, with a mean of −4.76‰. (Figs. 1b, c and 4c).

Fig. 5: Geopotential height anomalies and atmospheric circulation patterns in the extremely strong and relatively weak WPSH years.
Fig. 5: Geopotential height anomalies and atmospheric circulation patterns in the extremely strong and relatively weak WPSH years.
Full size image

a, b Annual 500 hPa geopotential height anomalies in the extremely strong (a) and relatively weak (b) WPSH years. c–f Vertically integrated moisture flux fields (c, d) and moisture flux field anomalies (e, f) at 1000–500 hPa in the extremely strong (c, e) and relatively weak (d, f) WPSH years. The anomalies are defined as the values in the extremely strong and relatively weak WPSH years subtracted by the mean values during 1999–2021, respectively. The black dots indicate the location of Hong Kong, and the black rectangles mark the region of southern East Asia. The solid (dashed) line circles or rectangles with pink in (c–f) mark the areas of increased (decreased) moisture flux.

Fig. 6: Spatial distributions of annual OLR anomalies during the extremely strong and relatively weak WPSH years.
Fig. 6: Spatial distributions of annual OLR anomalies during the extremely strong and relatively weak WPSH years.
Full size image

a The extremely strong WPSH years. b The relatively weak WPSH years. The black dots indicate the location of Hong Kong, and the black rectangles mark the region of southern East Asia.

In contrast, in the relatively weak WPSH years, the 500 hPa geopotential height displays negative anomalies (Fig. 5b). The WPSH weakens and retreats eastward, which has less power to drive moisture from the northwest Pacific to southern East Asia but allows moisture from the Indian Ocean to be transported almost unimpeded to southern East Asia. Consequently, the moisture supplied from the northwestern Pacific to southern East Asia is reduced (ovals drawn with a dashed pink line in Fig. 5d, f) with a larger amount of moisture derived from the southern Bay of Bengal and Indian Ocean (rectangle drawn with solid pink lines in Fig. 5d, f). Correspondingly, the lower δ18Ov values from the southern Bay of Bengal and Indian Ocean (Supplementary Fig. 4b) contribute to the lower background values for the δ18Op in southern East Asia. In addition, the moisture from the southern Bay of Bengal and the Indian Ocean is transported to southern East Asia experiences a long transport pathway and an obvious latitude change, which cause a stronger isotopic depletion along the moisture transport pathway. These processes result in the lower δ18Op in southern East Asia (Fig. 4d). More importantly, in this case, convection obviously enhances in the Indo-Pacific warm Pool indicated by the negative anomalies of OLR (Fig. 6b). Therefore, when the moisture from the remote southern Bay of Bengal and Indian Ocean in the lower latitude is transported northward and passes through the Indo-Pacific Warm Pool, the significantly enhanced convection lifts moisture from near-surface to high altitude. The process further decreases the δ18Ov and is responsible for the lower δ18Op in southern East Asia when the moisture is transported to this area in the relatively weak WPSH years (Fig. 4d). That is, the increasing moisture supply from the southern Bay of Bengal and the Indian Ocean, coupled with the enhanced convection, results in the lower values of annual δ18Op in southern East Asia (annual δ18Op values ranged from −5.27‰ to −6.03‰, with an average value being −5.65‰ at Hong Kong, which is 0.89‰ lower than during the extremely strong WPSH years) (Fig. 1b, c and 4d).

Discussion

This study investigated the relationships between the IOD/WPSH and the annual δ18Op at Hong Kong and even in the broader southern East Asia during 1961–2021. We found that both the IOD and WPSH influenced the interannual changes of δ18Op in this region, albeit over different periods, and that a major shift occurred around 1999. That is, the IOD played an important role on annual δ18Op in southern East Asia during 1961–1998 while the influence of the WPSH becomes important in the following 1999–2021 period.

Specifically, before 1999, the Indian Ocean Walker circulation strengthened during the strongly negative IOD years, which drove more moisture from the equatorial Indian Ocean to our study area. The isotopically depleted moisture supply resulted from the low background δ18Ov values, the long transport pathway, and the large latitude difference, combined with the stronger convection, contributed to the lower δ18Op values in southern East Asia. However, the relatively high δ18Op values occurred in southern East Asia during the strongly positive IOD years, which were caused by the obviously increased moisture supply from the northern Bay of Bengal and the northern Indochina Peninsula and weaker convection in the Indo-Pacific Warm Pool. Meanwhile, the WPSH was very weak during this period and had little influence on the annual δ18Op in our study area.

After 1999, the symmetric pattern of the IOD changed to an asymmetric pattern, which effectively decoupled the linkage of the IOD and the annual δ18Op in southern East Asia. Hence, the annual δ18Op in southern East Asia “lost” the signal of the IOD. In contrast, the enhanced WPSH during this period significantly affected the interannual changes of δ18Op across southern East Asia. In other words, the annual δ18Op “retained” the WPSH signal. Therefore, the correlation between the annual δ18Op and the WPSH becomes significant, with the higher (lower) δ18Op values corresponding to the stronger (weaker) WPSH.

Our results highlight the important roles of the IOD and WPSH that influence the interannual variation of δ18Op across southern East Asia. We suggest that researchers need to consider the influences of the IOD and WPSH when interpreting paleoclimate archives. In particular, we found that a marked shift in the influences of the IOD and WPSH on the annual δ18Op in southern East Asia occurred around 1999. Hence, the influences of the IOD and WPSH on the annual δ18Op are likely to vary over different historical periods, and it may be necessary to interpret the paleo-δ18O records within specific time segments.

In this study, we found that the transition of the IOD and the enhancement of the WPSH in 1999 resulted in shifting influences of the IOD and WPSH on the annual δ18Op across southern East Asia. Similarly, the effects of the WPSH on Asian monsoon precipitation and its teleconnection between India and northern China also changed around 199950,51. What caused the changes in the IOD and WPSH around 1999 remains unclear and should be the subject of future research. It is noted that, on an interdecadal scale, the Pacific Decadal Oscillation (PDO) turned more negative also around 199960. Therefore, it is worth exploring the possible relationships between the PDO and IOD/WPSH and verifying the trigger that caused the changes in the IOD and WPSH.

Methods

Study sites and stable isotope data

Hong Kong (114.17°E, 22.32°N, 66 m a.s.l.) resides within a typical monsoon climate region of southern East Asia (Fig. 1a). Mean air temperature between 1961 and 2021 at the site is relatively high at 23.1 °C. The annual air temperatures over this timeframe display small interannual differences ranging from 21.6 °C to 24.1 °C. The corresponding annual mean precipitation amount is about 2300 mm. In contrast to air temperature, precipitation amount has much higher interannual variability ranging from less than 1000 mm to over 3000 mm.

In this study, we first obtained the observed monthly δ18Op data at Hong Kong during 1961–2021 (data from 1966 to 1972 are missing) from the Global Network of Isotopes in Precipitation (GNIP). To analyze the interannual variability of δ18Op, we then calculated the monthly δ18Op values to annual values. The δ18Op series from Hong Kong represents the longest observed record from East Asia and provides the foundation for revealing the links between annual δ18Op and atmospheric circulations.

In addition, we used the modeled δ18Op and water vapor δ18O (δ18Ov) outputs from ECHAM6-wiso for the period 1979–202061,62. ECHAM6 is the sixth generation of the general atmospheric circulation model ECHAM63, and ECHAM6-wiso is its isotopic version. The 3D fields of temperature, vorticity, and divergence, as well as the surface pressure field of the ECHAM6-wiso simulation were nudged toward the ERA5 reanalysis dataset. The monthly mean sea surface temperature and sea-ice fields from the ERA5 reanalysis have been applied as ocean surface boundary conditions, too64. The ECHAM6-wiso has a relatively high spatial resolution (0.9° × 0.9° and 95 vertical levels) and a 6-h temporal resolution62. Details about the implementation of stable water isotopes in ECHAM6 have been described in Cauquoin et al.61. The modeled δ18Op data from ECHAM6-wiso at Hong Kong, are in good agreement with the observed δ18Op data from GNIP, with a significant and positive correlation of 0.77 (n = 473, p < 0.01) (Supplementary Fig. 5). Moreover, previous studies have also evaluated the ECHAM6-wiso data using various observational data from other regions54,62, and found that ECHAM6-wiso results are in good agreement with the observations. Therefore, the ECHAM6-wiso model outputs can be used to confidently reproduce the variations in δ18Op across southern East Asia.

Atmospheric circulation indices

The IOD is a seesaw phenomenon of sea surface temperature (SST) anomalies that occurs between the eastern and western tropical Indian Oceans. The IOD intensity is indicated by the dipole mode index (DMI), which is calculated by the anomalous SST gradient between the western equatorial Indian Ocean (50°E–70°E and 10°S–10°N) and the southeastern equatorial Indian Ocean (90°E–110°E and 10°S–0°N). When DMI > 0, the phenomenon is known as the positive IOD phase, and when DMI < 0, it falls within the negative IOD phase. The DMI data are provided by the NOAA Physical Sciences Laboratory and are calculated based on the SST data from the United Kingdom Met Office Hadley Center Global Sea Ice and Sea Surface Temperature Dataset version 1 (HadISST1).

The WPSH located in the Pacific Ocean, is one of the most important atmospheric circulation systems, with the intensity changes of the WPSH indicated by the WPSH Intensity Index (WPSHII)65. The WPSHII is defined as the difference between the grid point’s height exceeding 5880 and 5870 gpm multiplied by the sum of the total area encircled by the 5880 gpm isoline at 500 hPa within the region 10°N–60°N, 110°E–180°E65. The potential height data based on the NCEP/NCAR reanalysis datasets are used to calculate the WPSHII. The calculation formula65 of the WPSHII is as follows:

$$\begin{array}{l}{{WPSHII}}={dx}\times {dy}\times \mathop{\sum}\limits_{i}\mathop{\sum}\limits_{j}\left({n}_{{ij}}\times \left({H}_{{ij}}-5870\right)\times \cos {\varphi }_{j}\right)\\ \quad\quad\quad\;{n}_{{ij}}=\left\{\begin{array}{c}1,\,{H}_{{ij}}\ge 5880\\ 0,\,{H}_{{ij}}\, <\, 5880\end{array}\right.\end{array}$$

In the formula, dx is the distance value of the latitudinal grid point, and dy is the distance value of the meridional grid point; i is the ordinal number for the zonal grid point, i = 1, 2, …, nx, and nx is the sum of grid points in the monitoring range, increasing from west to east; j = 1, 2, …, ny, and ny is the sum of grid points in the monitoring range, increasing from south to north. Hij is the potential height value of the grid point on the 500 hPa height field, and φj is the latitude value of the grid point.

The WPSHII data were obtained from the Climate Diagnostics and Prediction Division, National Climate Center, China Meteorological Administration. The relatively high values of the WPSHII indicate a strong WPSH, while relatively low values of WPSHII indicate a weak WPSH.

Meteorological data

In this study, we used outgoing longwave radiation (OLR) to indicate the intensity of convection. Note that the OLR data were calculated by the mean top net longwave radiation flux data. The vertical velocity data were also used to explore the vertical air motion anomalies. The moisture flux was calculated using a combination of the meteorological fields, including both eastward and northward wind components and the specific humidity on various pressure levels, and surface pressure on a single level. The total precipitation data were used to show the distributions of mean annual precipitation amount. We also used the geopotential height to evaluate the changes in WPSH. The meteorological data (1961–2021) mentioned above were obtained from the ERA5 reanalysis dataset provided by the European Center for Medium-Range Weather Forecasts (ECWMF), with a spatial resolution of 0.25° × 0.25°. In addition, we used the SST data to reflect the interannual variations of SST anomalies. The SST data during the period 1961–2021 were obtained from the HadISST1, with a spatial resolution of 1° × 1°.