Introduction

A long-standing issue on East Asian monsoonal precipitation is how it may vary in a future warmer world, given its profound impact on regional ecosystems, human lives, and socioeconomic development1,2,3,4,5. However, there is no consensus on the projected intensity of the East Asian summer monsoon (EASM) and associated precipitation anomaly pattern6, partly attributed to the limited understanding of monsoon dynamics that may partially arise from the too-short instrumental observations. Therefore, we may look back into past warm intervals, which could provide insights into mechanistic links between global warming and monsoon variation. In particular, the Last Interglacial (LIG, ~130–115 ka BP) was a naturally induced warmest world driven by Earth’s orbit during the past 800 ka, which was witnessed by intense warming over high latitudes (~9 °C over Greenland compared with the preindustrial), reduced glaciers in the polar, and poleward extended boreal forests7,8,9,10,11,12. These climatic characteristics exhibited profound resemblances to the projected future climates, although with different external forcings13,14,15. Thus, the LIG has attracted great interest from both data synthesis and numerical modeling over East Asia11,13,16,17.

Geological evidence indicated a generally warmer and wetter condition over East Asia during the LIG18,19,20,21, in terms of the mean climate state. However, there has been a hot debate on the temporal evolution of East Asian monsoonal precipitation or EASM across the LIG. One group of records (e.g. stalagmites) suggested a stably stronger state of the EASM between ~130 and 120 ka, followed by an abrupt weakening (Fig. 1b, c)22,23. In contrast, another group of records (e.g., loess sequences and lake sediments) indicated a varied intensity of the East Asian monsoonal precipitation across the LIG, e.g., exhibiting either a long-term increasing or a decreasing trend (Fig. 1d–f)19,24,25,26,27,28. These data–data controversies may arise from the discrepancies in the interpretations of proxies (e.g., speleothem δ18O), the low temporal resolution, and spatial inconsistency, which may be resolved via reconstructing precipitation from a single proxy with high-temporal resolution and large latitude span. However, such efforts remain challenging due to great requests for proxies, sites, and techniques. This hence raises a fundamental question: how the East Asian monsoonal precipitation varied across the LIG?

Fig. 1: Evolution of reconstructed East Asian summer monsoon during the LIG based on various proxies.
figure 1

a The residual component of δ18O from the Ocean Drill Program Site 114527. Speleothem δ18O data of the stalagmite from the (b) Dongge Cave and (c) Sanbao Cave22,55. d Spliced East Asian summer monsoon (EASM) index derived from the loess records from Xijin, Chinese Loess records25. e Microcodium δ18O data of the loess records from Huanxian24. f Magnetic susceptibility (MS) of the loess records from Jingbian, Chinese Loess Plateau19. g The fraction of grain <0.63 μm from Miaodao Stratigraphic section28. h Chemical Index of Alteration (CIA) of the lake sediments from the Lake Xingkai26.

Full size image

To help address this issue, transient simulations, rather than time-slice experiments, are useful when focusing on the evolution of climate change across the LIG. However, existing transient simulations targeting the LIG29,30,31,32,33,34,35 may have large uncertainties in depicting East Asian climate, because of the use of orbital acceleration technique32, the simplified physical processes in the intermediate complexity model36, and/or coarse horizontal resolution37. To overcome these limitations, here we performed an transient simulation at high spatial and temporal resolution (i.e., ~1° × 1° without orbital acceleration) from 130 to 120 ka, which was an interval with stably stronger EASM indicated by stalagmites22,23. Our results revealed a divergent evolution of East Asian monsoonal precipitation across latitudes during the LIG, with an increasing (decreasing) trend over south China (northeast Asia) and an indistinctive long-term trend between the two regions. This out-of-phase variation in precipitation was regulated by large-scale atmospheric circulations in summer, which were ultimately driven by the change in May-June-July insolation.

Results

Evolution of monsoonal precipitation across the LIG

East Asian summer monsoonal precipitation exhibited divergent evolutionary patterns across latitudes during 130–120 ka. Comparisons showed that summer terrestrial precipitation had a long-term increasing trend over south China (25–35°N), whereas a decreasing trend was observed over northeast Asia (45–55°N) (Fig. 2). In contrast, precipitation over North China (35–45°N; i.e., between south China and northeast Asia) fluctuated near a basic state during 130–120 ka, without a significant long-term trend (Fig. 2). This was induced by the southeastward retreat of the EASM and the resulting opposite variation of precipitation between the west and east of North China (Supplementary Fig. 1). Moreover, the divergent evolution of precipitation was clearly observed in the spatial difference between the late LIG (121–120 ka) and the early LIG (130–129 ka). Precipitation anomalies generally showed a northeast-southwest dipole pattern over the East Asian monsoon domain, with enriched (deficient) amount extending from south China to Japan (pan-northeast Asia), supporting the southeastward retreat of the EASM during the late LIG relative to the early LIG (Fig. 2g). Notably, in addition to the meridional contrast of precipitation, there was a zonal contrast in precipitation change over East Asia, e.g., between the northeastern Tibetan Plateau and the west of North China, which was primarily contributed by the vertical dynamic term (Fig. 2g, Supplementary Fig. 2).

Fig. 2: Variations of summer terrestrial precipitation over East Asia.
figure 2

Evolution of precipitation across different latitude bands during the Last Interglacial (i.e. 130–120 ka) (units: mm d−1) with the latitude bands being (a) 50–55°N, (b) 45–50°N, (c) 40–45°N, (d) 35–40°N, (e) 30–35°N, and (f) 25–30°N. The blue, black, and red lines in (a–f) represent original precipitation evolutions applying a 101-yr running average, their corresponding millennial trends derived from Singular Spectrum Analysis and long-term (130–120 ka) trends, respectively. g Anomaly of precipitation over East Asia in summer between 121–120 ka and 130–129 ka (units: mm d−1). The dotted area in (g) represents the difference between 121–120 ka and 130–129 ka passing the significance test at the 95% level. The blue boxes in (g) indicate the region of south China (SC), North China (NC), and northeast Asia (NA), respectively. The blue, red, and gray markers in (g) denote the reconstructed EASM showed a long-term strengthening trend, a long-term weakening trend, and no significant long-term trend, respectively, during 130–120 ka displayed in Fig. 1.

Full size image

Atmospheric circulations associated with monsoonal precipitation

The differences in summer precipitation between the late and early LIG were closely linked with the variations of atmospheric circulations. There were anomalous northerly winds over East Asia and cyclonic circulations over Japan during the late LIG relative to the early LIG at 850 hPa (the lower troposphere) (Fig. 3a). This indicated a weakened summer monsoonal circulation during the late LIG, which restricted the northward advance of rainy belt and contributed to increased (decreased) precipitation over south China (northeast Asia). Moreover, the enhanced eddy geopotential height over the North Pacific in the mid-troposphere (500 hPa) suggested a southward retreat of the western North Pacific subtropical high (WNPSH) (Fig. 3b). This also provided favorable conditions for the formation of upward motion over south China, hence resulting in enriched precipitation over there (Figs. 2g, 3d). Moreover, the stronger (weaker) zonal winds over the south (north) of 50°N at 200 hPa showed that the East Asian westerly jet (EAWJ) was intensified and moved more equatorward during the late LIG relative to the early LIG (Fig. 3c). The equatorward located EAWJ can steer transient eddies over south China, supporting increased precipitation over there as well as the reduced precipitation over northeast Asia. The configurations of atmospheric circulation anomalies in the troposphere over East Asia gave rise to anomalous upward (downward) motions over south China (northeast Asia) in the non-divergence level (i.e. 500 hPa) during the late LIG relative to the early (Fig. 3d), and ultimately resulted in the northeast-southwest dipole pattern of precipitation over East Asia.

Fig. 3: Changes in atmospheric circulations responsible for variations of precipitation over East Asia in summer during the late LIG relative to the early LIG.
figure 3

Anomaly of (a) wind field at 850 hPa (units: m s−1), (b) eddy geopotential height at 500 hPa by removing the zonal mean (units: gpm), (c) zonal wind at 200 hPa (units: m s−1), and (d) vertical velocity at 500 hPa (units: 0.01 Pa s−1) between 121–120 ka and 130–129 ka. Orange solid and dotted lines in (b) represent the zero line of zonal wind at 500 hPa to indicate the location of the western North Pacific subtropical high during 130–129 ka and 121–120 ka, respectively. Orange solid and dotted lines in (c) represent the jet axis of the East Asian westerly jet during 130–129 ka and 121–120 ka, respectively. The dotted area in (a–d) represents the difference between 121–120 ka and 130–129 ka passing the significance test at the 95% level.

Full size image

The alternations in atmospheric circulations as well as the precipitation anomalies over East Asia are in turn tied to the adjustment of the atmospheric thermal structure and the associated climatic changes. The cooling over the East Asian inland was more intense than over south China and the subtropical Pacific during the late relative to the early LIG (Fig. 4a, Supplementary Fig. 3). This was associated with increased surface pressure over Asian inland and decreased one over south China and the subtropical Pacific, which weakened the north-south pressure gradient and resulted in the northerly winds over East Asia (Fig. 4a). Moreover, the reduced meridional gradient of sea surface temperature (SST) over the North Pacific, manifested by anomalous warming over the north of Taiwan and cooling over the south. This was disadvantage for the northward location of the WNPSH (Fig. 4d), similar to the casual relationship observed in modern climate38. The anomalous anticyclone (cyclone) over south China and cyclone (northeast Asia) at 200 hPa resulted in the southward displacement of the EAWJ (Fig. 4b). The configuration of anticyclonic circulation at 200 hPa and cyclonic circulation at 850 hPa over south China suggested a baroclinic structure (Fig. 4a, b). Based on the Matsuno–Gill response, the anticyclone/cyclone may be partially induced by increased precipitation over its southeastern flank near the east of the Philippines arising from the increased SST over there (Fig. 4c, d). These results indicated the important roles of changes in the SSTs during the late LIG relative to the early in regulating the precipitation-associated atmospheric circulations over East Asia. Additionally, in the upper troposphere, the extratropical Rossby waves originated from North Africa and then propagated northeastward to Central Asia (Fig. 4b). Then the wave train branched two pathways, with the major one propagating eastward and turning southward to south China and the other propagating southeastward directly (Fig. 4b). This suggested that the horizontal propagation of extratropical Rossby waves was likely to play a role in the formation of the anomalous atmospheric circulations over East Asia. Overall, variations of SSTs over the subtropical/North Pacific and the propagation of extratropical Rossby waves jointly regulated anomalous atmospheric circulations over East Asia during the late LIG relative to the early.

Fig. 4: Large scale climatic changes during the late LIG relative to the early LIG.
figure 4

Anomaly of (a) surface pressure (shade, units: hPa) and wind field (vector, units: m s−1) at 850 hPa, (b) eddy geopotential height by removing the zonal mean (shade, units: gpm) and wave activity flux (vector, units: m2 s−2) at 200 hPa, (c) precipitation (units: mm d−1), and (d) eddy sea surface temperature by removing the zonal mean (units: °C) between 121–120 ka and 130–129 ka. The dotted area represents the difference between 121–120 ka and 130–129 ka passing the significance test at the 95% level.

Full size image

Moisture budget analysis for monsoonal precipitation

Moreover, we utilized the moisture budget analysis to further estimate the contributions of dynamic and thermodynamic processes in precipitation variations over East Asia across the LIG (Fig. 5). The horizontal thermodynamic term and the vertical dynamic term played the most important role in determining increased precipitation over south China, with the positive contribution from the horizontal dynamic term. In contrast, all the evaporation, horizontal and vertical thermodynamic/dynamic terms were essential in resulting in decreased precipitation over northeast Asia. This highlighted that the variations of atmospheric circulations and reduced specific humidity (Supplementary Fig. 4) were both important in regulating precipitation over there during the late LIG relative to the early. In addition, over the transition zone (North China), the decreased precipitation over the western part was dominated by evaporation, vertical dynamic, vertical thermodynamic, and non-linear terms, whereas the increased precipitation over the east was controlled by the horizontal thermodynamic and vertical dynamic terms (Supplementary Fig. 5). Notably, the residual term was relatively larger in the moisture budget analysis during the late LIG relative to the early, which may arise from the non-negligible difference of the moisture storage in the atmosphere due to the mean climate state discrepancies, and was also found in previous modeling studies39,40.

Fig. 5
figure 5

Moisture budget analysis over South/East China (25–40°N, 115–130°E) and northeast Asia (40–55°N, 110–130°E) between 121–120 ka and 130–129 ka (units: mm d−1).

Full size image

Discussions

Here, we reveal a divergent evolution of East Asian monsoonal precipitation across latitudes during the LIG, with an increasing (decreasing) trend over south China (northeast Asia) and an indistinctive long-term trend between the two regions (Fig. 2). We speculate that the out-of-phase variations of precipitation over south China and northeast Asia may follow the variation of orbital insolation in May-June-July during the LIG (Fig. 6). Specifically, insolation in May-June-July over East Asia increased from 130 to 128 ka and then decreased profoundly (Supplementary Fig. 6). Decreased May-June-July solar insolation determined the cooling trend in summer over East Asia during the LIG (Supplementary Fig. 7), due to the delayed effect of solar insolation on climate change. The decreased temperature over East Asia was not uniform, with more remarkable cooling over the land at mid-high latitudes than over the ocean at low latitudes, induced by the land-sea thermal contrast (Supplementary Fig. 7). This led to weakened monsoonal circulations and the southeastward retreat of the EASM (Supplementary Fig. 8), which further resulted in the long-term decreasing trend over northeast Asia and an increasing precipitation trend over south China during the LIG. Overall, our results highlighted out-of-phase changes in simulated East Asian monsoonal precipitation across East Asia in response to the precession.

Fig. 6: Schematic diagram of the role of May-June-July solar insolation in driving variations of atmospheric circulations and associated precipitation pattern during the late LIG relative to the early LIG in summer.
figure 6

The shade pattern and circles at the bottom denote surface temperature and precipitation anomalies between the late LIG and early LIG in summer, respectively. The shade in the middle represents the zonal wind/the location of the westerly jet during the early LIG (red) and late LIG (blue) in summer. The red and blue contour in the middle represents the location of the western North Pacific subtropical high (WNPSH) during the early LIG and late LIG in summer, respectively. The vector in the middle indicates the anomaly of meridional wind during the late LIG relative to the early LIG in summer. The curve at the top indicates the evolution of May-June-July solar insolation at 25–55°N between 130 ka and 120 ka.

Full size image

We notice that the responses of East Asian monsoonal precipitation are complicated in a warmer world, which are dependent on the external forcings inducing the warmth. In the transition from the early LIG (warm) to late LIG (relatively cold), we reveal the out-of-phase variations of East Asian monsoonal precipitation over south China and northeast Asia across the LIG, in tandem with the weakened EASM. This was also observed during the Holocene (~11–0 ka), with May-June-July insolation also characterized by a general decreasing trend41,42. In contrast, in an increased greenhouse gas dominated warmer world, an increasing trend in East Asian monsoonal precipitation has been projected4, whereas a dipole anomaly pattern was observed during the warm mid-Pliocene (~3.0 Ma characterized by elevated CO2 concentration (~400 ppm))43. This highlights the complexity of variations in the EASM and associated precipitation in a warmer-than-present world, in tandem with mechanistic links between global warming and monsoon change. It should be noted that although the external forcings and associated precipitation responses are different between the LIG and the future, the profound warming over the mid-high latitudes is similar7,13, in tandem with the strengthened monsoonal circulations relative to the present day5,40. Therefore, understanding changes in the EASM during the LIG may be indicative of its behavior in a future warmer world to some extent.

Given that the reconstructed precipitation over East Asia varied across sites and proxies, it was a matter of course that our simulated results were consistent with some proxies (e.g.25,26,44), but still contradicted with others (e.g.24,28). Therefore, a future study that reconstructs precipitation from a single proxy with high-temporal resolution and a large latitude span together with a series of transient simulations needs to be carried out to better decipher the evolution of East Asian monsoonal precipitation during the LIG. Nevertheless, we provide a modeled scenario on the evolution of East Asian monsoonal precipitation across the LIG. This may have important implications for reconciling the discrepancies in the reconstructed precipitation across proxies/sites and shed light on precipitation behavior in a warmer-than-present world.

Methods

Model and experimental setups

We use the Community Earth System Model (CESM) version 1.2.2 to perform a Transient Global Climate Simulation covering the LIG (i.e., 130 to 120 ka) (hereafter referred to TGCS-LIG). The CESM is a fully coupled global climate model that has interactive flux among the atmosphere, ocean, land, and sea ice components45. In this study, we perform the TGCS-LIG at f09g16 resolution. The Community Atmosphere Model version 4 (CAM4) has a horizontal resolution of ~0.9° × 1.25°with 26 hybrid coordinate vertical levels, and the horizontal resolution of the Community Land Model version 4 (CLM4) is identical to the CAM4. The Parallel Ocean Program version 2 (POP2) with 60 vertical levels and the Community Ice CodE version 4 (CICE) are at ~1° horizontal resolution. The CESM1.2 well reproduces the modern spatial pattern of East Asian monsoonal precipitation compared with the observations, with the spatial correlation coefficient and root-mean-square error of 0.98 and ~0.45 mm d−1, respectively (Supplementary Fig. 9). We perform the TGCS-LIG experiment using 384 cores on a Linux cluster supercomputer, with a throughput of ∼18 model years per wall-clock day. For the 10,000 model years, the TGCS-LIG experiment costs ~5.5 million core-hours and produces ~166 TB of outputs.

To perform the TGCS-LIG experiment, we first carry out an equilibrium run at 130 ka (LIG_130 ka), which serves as the initial condition for the TGCS-LIG experiment. In the LIG_130ka experiment, orbital parameters are set to the levels of 130 ka based on Berger46 and greenhouse gas concentrations use the mean value between 130.5–129.5 ka according to Köhler et al.47 (Supplementary Table 1). Ice sheets and vegetation in the LIG_130ka are kept the same as the preindustrial conditions. The LIG_130ka experiment is branched off from a 1300-year long-time picontrol experiment48, in which the last 100-yr annual net radiation at the top of the atmosphere shows no long-term trend (~0.0002 W m−2 yr−1). The LIG_130ka experiment is then integrated for additional 450 years to allow the surface climate to reach quasi-equilibrium (i.e., net radiation at the top of the atmosphere averaged over the last 100 years is ~0.0001 W m−2 yr−1).

Next, the TGCS-LIG experiment is initialized from the LIG_130ka experiment and integrated for another 10,000 years (i.e., from 130 to 120 ka). Regarding the external forcings, orbital parameters are adopted from Berger46, and greenhouse gas concentrations are obtained from Köhler et al.47, following the guidelines of the PMIP449. Briefly, July solar insolation at 25°N increases since 130 ka, reaches the peak at ~125 ka, and then decreases afterward (Supplementary Fig. 10a). Atmospheric CO2 increases rapidly since 129 ka and then drops, with slight fluctuations since 128 ka (Supplementary Fig. 10b). The evolution of atmospheric CH4 is similar to that of atmospheric CO2 (Supplementary Fig. 10c), whereas atmosphere N2O generally shows an increasing trend during 130–120 ka (Supplementary Fig. 10d). Besides, all the other boundary conditions are kept the same as the preindustrial (e.g., ice sheets and vegetation). Although the roles of ice sheets and vegetation were neglected in the TGCS-LIG experiment, they may have limited impact on the temporal evolution of the East Asian monsoonal precipitation, but influence the changing magnitude, as indicated in previous studies50,51.

Extratropical and tropical Rossby waves

We use the wave-activity flux (W) to indicate the horizontal propagation direction of Rossby waves52, which is defined as:

$${\bf{W}}=\frac{1}{2|\overline{U}|}\left[\begin{array}{c}\overline{u}({\psi }_{x}^{{\prime} 2}-\psi ^{\prime} \psi ^{{\prime} }_{xx})+\overline{v}({\psi }_{x}^{{\prime} }\psi ^{{\prime} }_{y}-\psi ^{\prime} {\psi }_{xy}^{{\prime} })\\ \overline{u}({\psi ^{\prime} }_{x}{\psi ^{\prime} }_{y}-\psi ^{\prime} {\psi ^{\prime} }_{xy})+\overline{v}({\psi ^{\prime} }_{y}^{2}-\psi ^{\prime} {\psi ^{\prime} }_{yy})\end{array}\right]$$
(1)

where u is the zonal wind velocities, and v is the meridional wind velocities. ψ stands for the stream function. Here, the overbar indicates the mean climate state during 130–120 ka, and the prime indicates the differences between the late LIG (121–120 ka) and the early LIG (130–129 ka).

Additionally, the barotropic structure of the anomalous circulations (i.e. a cyclone in the lower troposphere accompanied by a cyclone in the upper troposphere; and vice versa for the anticyclone) may primarily result from the propagation of extratropical Rossby waves. In contrast, the baroclinic structure of the anomalous circulations may arise from anomalous diabatic heating/cooling over the tropics, indicated by anomalous cyclone (anticyclone) in the lower (upper) troposphere over the northwestern flank of the heating source in the Northern Hemisphere based on the Matsuno–Gill response53,54; and vice versa for the diabatic cooling.

Moisture budget analysis

We use the moisture budget formulation to analyze the relative contribution of dynamic and thermodynamic terms in regulating summer precipitation over East Asia during 130–120 ka. The moisture budget relationship is defined as:

$$P^{\prime} =-\langle \bar{\omega }{\partial }_{p}q^{\prime} \rangle -\langle \omega ^{\prime} {\partial }_{p}\bar{q}\rangle -\langle \bar{{\bf{V}}}\cdot \nabla q^{\prime} \rangle -\langle {\bf{V}}^{{{\prime} }}\cdot \nabla \bar{q}\rangle -\langle \omega ^{\prime} {\partial }_{p}q^{\prime} \rangle -\langle {\bf{V}}^{{{\prime} }}\cdot \nabla q^{\prime} \rangle +E^{\prime} +\delta$$
(2)

where P stands for the precipitation, and q represents the specific humidity. V is the horizontal velocity vector, and E is on behalf of the evaporation. δ indicates the residual term. The overbar indicates the climatology between 130 and 120 ka, and the prime indicates the differences between the late LIG (121–120 ka) and the early LIG (130–129 ka). Term \(-\langle \bar{\omega }{\partial }_{p}q^{\prime} \rangle\), \(-\langle \omega ^{\prime} {\partial }_{p}\bar{q}\rangle\), \(-\langle \bar{{\bf{V}}}\cdot \nabla q^{\prime} \rangle\) and \(-\langle {\bf{V}}^{{{\prime} }}\cdot \nabla \bar{q}\rangle\) indicates the vertical thermodynamic term, vertical dynamic term, horizontal thermodynamic term, and horizontal dynamic term, respectively. Term \(-\langle \omega ^{\prime} {\partial }_{p}q^{\prime} \rangle\) and \(-\langle {\bf{V}}^{{{\prime} }}\cdot \nabla q^{\prime} \rangle\) stand for non-linear terms. Variations of humidity dominate the thermodynamic processes, whereas atmospheric circulation changes (i.e. velocity anomalies) control the dynamic processes.