1 Introduction

The Asian summer monsoon (ASM) is a complex system that includes the Indian summer monsoon (ISM) and East Asian summer monsoon (EASM) (Wang 2006). ISM precipitation is directly influenced by the intertropical convergence zone, associated with tropical heating and ultimately related to solar radiation (Chao 2000; Fleitmann et al. 2007). However, EASM precipitation is mainly caused by a frontal system, which forms when warm and moist air flows brought by the monsoonal southeasterly winds meet the cold air brought by the westerlies (Ding and Chan 2005; Liu et al. 2014a). Therefore, a better understanding of the interaction between the EASM and mid-latitude westerlies is important to reveal the nature of moisture changes in East Asia. The northeastern Tibetan Plateau (Fig. 1) is ideally located to study the interaction between the EASM and mid-latitude westerlies (Lu et al. 2010; An et al. 2012).

Fig. 1
figure 1

Mean summer water vapour flux vectors (kg m−1 s−1), vertically integrated from the surface to 300 hPa, during the period 1971–2000, based on the NCEP–NCAR Reanalysis (Kalnay et al. 1996). Areas above 3000 m (above sea level) are shaded grey. ‘EASM’, ‘ISM’, and ‘Westerlies’ denote the regions mainly influenced by the East Asian summer monsoon, Indian summer monsoon, and westerlies, respectively. The locations of Dalianhai Lake, Qinghai Lake, Luanhaizi Lake, Tianchi Lake and Gonghai Lake are indicated by the red circle, blue cross, green triangle, magenta dot and blue dot, respectively

Full size image

A large number of proxy records, especially those based on multiple sediment cores from Qinghai Lake, have been studied to reveal the characteristics of the changes in moisture on the northeastern Tibetan Plateau during the Holocene, as well as the history of the ASM. However, significant discrepancies exist among different proxy records; in particular, pollen and ostracod oxygen isotope (δ18O) records indicate different periods for the climatic optimum during the Holocene (Lu et al. 2010; Chen et al. 2016). Pollen records from Qinghai Lake and the surrounding lakes of the northeastern Tibetan Plateau suggest a relatively dry early Holocene, with the wettest period spanning the middle Holocene, and then another dry period in the late Holocene (Shen et al. 2005; Herzschuh et al. 2010; Cheng et al. 2013; Wang et al. 2014a). Pollen-based moisture changes lag the Northern Hemisphere June insolation maxima by ~3 ka. A relatively dry climate during the early Holocene is also indicated by the significantly low lake levels of Qinghai Lake during this period (Liu et al. 2013, 2015b; Wang et al. 2014b). However, ostracod δ18O records from Qinghai Lake indicate a drying trend in this region throughout the Holocene, closely following Northern Hemisphere summer insolation (Liu et al. 2007; Wang et al. 2011; An et al. 2012). This drying trend is further supported by hydrogen isotope records from Qinghai Lake (Thomas et al. 2016) and Genggahai Lake (Rao et al. 2016).

It is commonly believed that insolation-induced ASM recession dominated the changing trend (Colman et al. 2007; An et al. 2012; Thomas et al. 2016), and that North Atlantic cold events contributed to abrupt declines (e.g., 8.2 ka BP) (Ji et al. 2005; Wang et al. 2011; An et al. 2012; Liu et al. 2014b) of precipitation on the northeastern Tibetan Plateau during the Holocene. The northeastern Tibetan Plateau is located in a transitional zone between the EASM and ISM, and it is still not clear which of these monsoon systems (ISM or EASM) controls the precipitation over this area. A number of palynological studies have concluded the EASM to have been the dominant factor influencing precipitation changes during the Holocene (Shen et al. 2005; Liu et al. 2015a; Chen et al. 2016), and the strength of the EASM is believed to have been suppressed by the continental ice sheets during the early Holocene (Chen et al. 2015a, b). Other studies, however, based on stable hydrogen and oxygen isotope records (Liu et al. 2007; Rao et al. 2016; Thomas et al. 2016), suggest that the precipitation over this area has been governed during the Holocene by the ISM.

Considering the considerable controversy outlined above regarding the changes in precipitation on the northeastern Tibetan Plateau during the Holocene, it is of great significance to clarify the situation using some other form of evidence. In this respect, paleoclimate modelling and comparison with proxy records is an extremely effective approach that can be used to understand paleoclimatic changes and their forcing mechanisms. In this study, therefore, we simulated the climate changes over the past 9.5 ka and compared the simulation results with proxy records to explore the changes in moisture on the northeastern Tibetan Plateau during the Holocene and their driving factors.

Below, we introduce the methods used in this research (Sect. 2), including proxy records and climate model simulation. Section 3 compares the simulation results with the proxy records. Section 4 analyses the climatic significance of ostracod δ18O in Qinghai Lake, discusses the physical mechanisms responsible for the changes in precipitation on the northeastern Tibetan Plateau during the Holocene, and provides the reason for the mismatch between the simulation and proxy during the late Holocene. The main conclusions are shown in Sect. 5.

2 Methods

2.1 Proxy records

Most of our knowledge on climatic and environmental changes on the northeastern Tibetan Plateau during the Holocene comes from the sediments of Qinghai Lake (Colman et al. 2007; Chen et al. 2016), which is the largest natural lake in China. The δ18O values of ostracod shells and pollen assemblages are widely used as proxies for reconstructing the changes in moisture in this area during the Holocene, as well as the history of the summer monsoon (Shen et al. 2005; An et al. 2012). As a hydrologically closed lake, Qinghai Lake’s hydrology is relatively simple, being ultimately controlled by atmospheric precipitation and local evaporation. Most of the precipitation occurs in summer (June–July–August, JJA), which is influenced by the ASM (Xu et al. 2007; Thomas et al. 2016). The δ18O within ostracod shells inherits the isotopic composition of lake water (Liu et al. 2009), and it has been suggested that Qinghai Lake has a relatively constant hypolimnion temperature of ~6 °C in summer, owing to the thermal stratification of its waters (Colman et al. 2007). Therefore, the δ18O values of the water are independent of temperature and mainly mirror the balance between the amount of precipitation [the δ18O values are negatively correlated with the amount of precipitation—the so-called “amount effect” (Dansgaard 1964)] and evaporation in the closed basin (Wei and Gasse 1999; Johnson and Ingram 2004; Colman et al. 2007). Ultimately, the ostracod δ18O records from Qinghai Lake can be used to reconstruct the effective humidity of the northeastern Tibetan Plateau. In this study, ostracod δ18O records from three sediment cores of Qinghai Lake, with reliable dating control, were employed to compare with the simulation results.

Pollen assemblages directly reflect the vegetation cover associated with certain hydro-thermal conditions (Wang et al. 2014a). Therefore, pollen-based proxies are direct and unambiguous indicators of moisture change (Liu et al. 2015a). Here, a pollen record from a sediment core (QH-2000) of Qinghai Lake (Shen et al. 2005) was selected to compare with the ostracod δ18O records and simulation results. In addition, there are a number of small lakes spread around the Qinghai Lake region, many of which have been studied in reconstructing the climatic and environmental changes over this area. We used pollen records from two of them—Dalianhai Lake (Cheng et al. 2013) and Luanhaizi Lake (Herzschuh et al. 2010; Wang et al. 2014a).

Finally, we also used a Mg/Ca-based SST proxy record from core A7 (Sun et al. 2005) in the northwestern Pacific to analyse the possible physical mechanism underlying the variations in precipitation on the northeastern Tibetan Plateau during the Holocene. Two proxy records from Tianchi Lake and Gonghai Lake on the Chinese Loess Plateau were also used to compare with our simulation results. A summary of the abovementioned proxy records is provided in Table 1. The locations of the proxy records are shown in Fig. 1.

Table 1 Summary of the proxy records used in this study
Full size table

2.2 Climate simulation

The model used in this study is the Kiel Climate Model (KCM; Park et al. 2009), which is a state-of-the-art coupled ocean–sea–ice–atmosphere general circulation model. The KCM comprises the ECHAM5 spectral atmospheric model and the NEMO (Nucleus for European Modelling of the Ocean) ocean–sea-ice general circulation model, coupled using the OASIS3 coupler. The horizontal resolution of ECHAM5 is T31 (3.75° × 3.75°), and in the vertical direction there are 19 levels up to 10 hPa. The horizontal resolution of NEMO is, on average, 1.3°, based on 2° Mercator meshes with grid refinement in the tropical regions, where the meridional grid-point separation reaches 0.5°. NEMO assumes 31 vertical levels. More details on the model can be found in Park et al. (2009).

The transient simulation was initialized using output from an equilibrium simulation of the early Holocene forced by Earth’s orbital parameters [eccentricity, obliquity and precession (Berger and Loutre 1991)] for 9.5 ka BP, and then forced by varying the orbital parameters according to the respective period from 9.5 to 0 ka BP with a tenfold acceleration scheme (Lorenz and Lohmann 2004). Hence, the climate trends and feedbacks of the last 9500 years (9.5–0 ka BP) imposed by the external orbitally driven insolation changes were represented in the experiment by 950 model years in 10-year intervals. Other forcing factors apart from insolation forcing—for example, greenhouse gas concentrations, continental ice sheets and meltwater fluxes—were taken at pre-industrial (AD 1850) levels. In this regard, this approach is effectively a sensitivity test of climate models to variations in insolation. More details on the model can be found in Jin et al. (2014).

The KCM climatology for present-day conditions has been validated against observations in Park et al. (2009) and Jin et al. (2014). Temperature trends from time slice simulations over the Holocene have been compared with proxy data in Schneider et al. (2010). The KCM reasonably well simulates the ASM precipitation during the Holocene (Jin et al. 2014; Dallmeyer et al. 2015). The KCM’s rainfall is also consistent with that simulated by five other climate models in transient mode (Dallmeyer et al. 2015), which focused on trends in global zonal-mean June-July-August (JJA) precipitation and surface air temperature during the last 6 ka.

3 Results

On the northeastern Tibetan Plateau, the moisture (precipitation) changes during the Holocene revealed by the pollen records (Fig. 2a–c) were different from those revealed by the ostracod δ18O records (Fig. 2e–g). The pollen records suggested a wetting trend from 9.5 to ~6.2 ka BP and a persistent drying trend thereafter, with the wettest interval in the middle Holocene. This wettest interval lagged the Northern Hemisphere summer insolation maximum (Fig. 2h) by ~3.5 ka. However, the ostracod δ18O values shifted from low to high during the Holocene, implying a drying trend over this period according to the “amount effect”, in which the amount of precipitation is inversely correlated with precipitation δ18O values (Dansgaard 1964). The moisture history derived from the ostracod δ18O records resembles that from stalagmite δ18O records, widespread across monsoonal China (Liu et al. 2015a; Cheng et al. 2016). Moreover, the ostracod δ18O-based moisture (precipitation) variations closely followed the monsoon wind index and summer (June) insolation in the Northern Hemisphere during the Holocene (Fig. 2e–h).

Fig. 2
figure 2

Moisture changes during the Holocene on the northeastern Tibetan Plateau, from proxy records and climate simulation: a tree pollen percentages of Dalianhai Lake (Cheng et al. 2013); b tree pollen percentages from Core QH-2000 of Qinghai Lake (Shen et al. 2005); c annual precipitation reconstructed based on pollen assemblages from Luanhaizi Lake (Wang et al. 2014a); d summer rainfall averaged over the region (35°–45°N, 95°–105°E), as simulated by the KCM (99-point Gaussian filter); The green line indicates the mean summer precipitation during 1971–2000 calculated based on the meteorological data of 49 stations over the region (35°–45°N, 95°–105°E) (http://data.cma.cn). e ostracod δ18O records from Core QH-2000 of Qinghai Lake (Liu et al. 2007); f ostracod δ18O records from Core QH-2005 of Qinghai Lake (Wang et al. 2011); g ostracod δ18O records from Core 1Fs of Qinghai Lake (An et al. 2012); h EASM index [averaged summer meridional wind at 850 hPa in East China (30°–45°N, 105°–120°E)], as simulated by the KCM (99-point Gaussian filter). June insolation at 45°N is plotted for comparison (Laskar et al. 2004)

Full size image

The KCM-simulated summer precipitation on the northeastern Tibetan Plateau, averaged over the region (35°–45°N, 95°–105°E), increased from 181 mm at 9.5 ka BP to 191 mm at 6.2 ka BP, and then decreased to 165 mm at 0 ka BP (Fig. 2d). The amount of summer precipitation at 0 ka BP based on the KCM simulation is quite close to the observation value during 1971–2010 (Fig. 2d, green line), supporting the excellent ability of the KCM to simulate precipitation on the northeastern Tibetan Plateau. The model results agreed well with the pollen records, with the climatic optimum taking place over the middle Holocene (Fig. 2a–d). Model–proxy comparison therefore suggested the climatic optimum on the northeastern Tibetan Plateau lagged the peak summer insolation by ~3.5 ka. It is therefore suggested that ostracod δ18O proxy records are not reasonable for measuring the amount of precipitation on the northeastern Tibetan Plateau, arguing the conclusion drawn by others (Lister et al. 1991; Liu et al. 2007; Wang et al. 2011; An et al. 2012).

4 Discussion

4.1 Climatic significance of ostracod δ18O in Qinghai Lake

According to the “amount effect” (Dansgaard 1964), the ostracod δ18O values of core sediments in Qinghai Lake, which receive the signal from precipitation δ18O, are traditionally regarded as a proxy that can reflect the amount of precipitation over this region (Lister et al. 1991; Liu et al. 2007; Wang et al. 2011; An et al. 2012). However, observed precipitation δ18O values were found to be higher in the wet summer and lower in the dry winter during the period of 1986–2003 at Zhangye meteorological station, which is 200 km away from Qinghai Lake (Yao et al. 2013). This implies that changes in precipitation δ18O values on the northeastern Tibetan Plateau are probably not controlled by the “amount effect”. Therefore, the ostracod δ18O in Qinghai Lake is probably not a proxy of precipitation on the northeastern Tibetan Plateau.

In addition to the “amount effect”, it has also been suggested that changes in precipitation δ18O values are influenced by the moisture source of precipitation (LeGrande and Schmidt 2006; Breitenbach et al. 2010). We analysed the transport of water vapour for summer rainfall events at Xining meteorological station, 90 km away from Qinghai Lake, for the period 1954–2011 (provided by the China Meteorological Data Service Center, available online at http://data.cma.cn). The rainfall events were divided into three levels: rainfall greater than or equal to 25 mm/day (R25, n = 58 days); rainfall greater than or equal to 10 mm/day but less than 25 mm/day (R10–25, n = 349 days); and rainfall less than 10 mm/day (R10, n = 2089 days).

An anticyclonic anomaly was generated over Northeast Asia during the R25 events compared to the R10–25 events, bringing anomalous moisture from the northwestern Pacific to the northeastern Tibetan Plateau (Fig. 3a). These moisture sources also contributed considerably to the R10–25 events compared to the R10 events (Fig. 3b). That is, enhanced precipitation in Xining was associated with a stronger western Pacific subtropical high (WPSH) transporting the water vapour from the northwestern Pacific. However, the contributions of the westerly moisture transport from the North Atlantic and the southerly moisture transport from the Indian Ocean to the changes in precipitation at Xining meteorological station were not significant. Therefore, precipitation changes on the northeastern Tibetan Plateau are much more likely to be controlled by the EASM in the present-day climate.

Fig. 3
figure 3

Composite differences of the water vapour vertically integrated from the surface to 300 hPa, a between R25 events (rainfall days with greater than or equal to 25 mm/day) and R10–25 events (rainfall days with greater than or equal to 10 mm/day but less than 25 mm/day), and b between R10–25 events (rainfall days with greater than or equal to 10 mm/day but less than 25 mm/day) and R10 events (rainfall days with less than 10 mm/day), at Xining weather station, in the summers of 1954–2011. Blue vectors indicate changes that are significant at the 90% confidence level. The circle indicates the location of Xining weather station. The colour shading shows the annual mean sea surface oxygen isotope ratio (δ18O) based on the global gridded data set of the oxygen isotopic composition in seawater (LeGrande and Schmidt 2006), which were calculated using c the observed δ18O values from the upper 5 m of the water column since about 1950 (Schmidt et al. 1999)

Full size image

A similar anticyclonic anomaly was also induced over Northeast Asia at 700 hPa during the early (middle) Holocene compared to the middle (late) Holocene, in response to orbital forcing in the KCM transient simulation (Fig. 4b, c). This anticyclonic anomaly increased the southeasterly transport of moisture from the northwestern Pacific to the northeastern Tibetan Plateau (Fig. 4a–c). The southerly winds over the east of the Tibetan Plateau did not strengthen during the early (middle) Holocene compared to the middle (late) Holocene, in spite of an insolation-induced strong ISM (Fig. 4a–c). Therefore, the water vapour contributing to the precipitation changes on the northeastern Tibetan Plateau during the Holocene was probably transported by the EASM circulation from the northwestern Pacific, rather than by the ISM circulation from the Indian Ocean, as suggested by previous studies (Li and Xu 2016; Li et al. 2016; Rao et al. 2016; Thomas et al. 2016).

Fig. 4
figure 4

a Simulated climatological mean summer wind fields (m/s) at 700 hPa. b Composite differences in summer wind fields at 700 hPa between the periods of 9.5–8.5 and 7–6 ka BP. c Composite differences in summer wind fields at 700 hPa between the periods of 7–6 and 1–0 ka BP. d–f As in a–c, but at 500 hPa. Blue vectors indicate changes that are significant at the 95% confidence level. ‘A’ and ‘C’ indicate anticyclone and cyclone, respectively. The circle, cross and triangle indicate the locations of Dalianhai Lake, Qinghai Lake and Luanhaizi Lake, respectively. The green arrows indicate the wind directions. The pressure levels below surface pressure are omitted

Full size image

Observed surface δ18O values were much lower in the mid-latitude North Pacific than in the tropical Pacific and Indian oceans (Fig. 3). Moisture evaporated from low δ18O water can decrease the precipitation δ18O values significantly (Breitenbach et al. 2010). Thus, it is expected that moisture from the mid-latitude North Pacific (Yellow Sea) causes lower values of precipitation δ18O. Furthermore, the anticyclonic anomaly over Northeast Asia caused by insolation-induced enhanced EASM during the early Holocene (Fig. 4b, c) was much stronger than that in the present-day climate (Fig. 3a, b). Therefore, the northeastern Tibetan Plateau could receive water vapour from the high-latitude North Pacific (Okhotsk Sea) during the early Holocene (Fig. 4b, c). Observed surface δ18O values were much lower (~ 2‰) in the Okhotsk Sea than in the Yellow Sea, with the greatest difference up to 7‰ (Fig. 3). In addition, the distance covered by the moisture transport from the high-latitude North Pacific is much longer than that from the mid-latitude North Pacific (Fig. 4). The long-distance moisture transport is associated with negative δ18O values (Tan 2014), which further reduced the δ18O values of the water vapour over the northeastern Tibetan Plateau during the early Holocene. These two factors can cover the most variability in the ostracod δ18O (~4‰) from Qinghai Lake during the Holocene. The surface δ18O values documented in planktonic foraminifera shells show no significant changes in the North Pacific Ocean throughout the Holocene (Ohkushi et al. 2003; Sun et al. 2005; Kubota et al. 2010). The weakened WPSH, which reduces the transport of water vapour from the high-latitude North Pacific, has probably been an important reason for the enrichment of 18O in Qinghai Lake during the Holocene. The weakened WPSH is a vital component of the EASM system (Ding and Chan 2005). Therefore, the ostracod δ18O values probably reflect the strength of the EASM circulation.

Tan (2016) suggested that the WPSH is an important bridge connecting Chinese stalagmite δ18O to the circulation of the tropical Pacific (Tan 2014). Our results indicate that the WPSH might also bridge Chinese stalagmite (ostracod) δ18O with the mid-and high-latitude North Pacific. The asynchronous variations of ostracod δ18O and precipitation suggest that the precipitation changes on the northeastern Tibetan Plateau during the Holocene have not been completely controlled by the transport of water vapour related to the insolation-forced circulation of the EASM (Fig. 2h).

4.2 Physical mechanisms responsible for the changes in precipitation on the northeastern Tibetan Plateau during the Holocene

The simulated changes in summer precipitation on the northeastern Tibetan Plateau during the Holocene did not fully follow the summer insolation, with a delay of ~3.5 ka (Fig. 2). Previous studies have attributed this lag to the effects of slow ice-sheet melting and the discharge of meltwater into the North Atlantic during the early Holocene (Chen et al. 2015a, b; Li and Xu 2016). However, our simulation experiment considered orbital forcing only and did not take into account these effects. Therefore, internal feedbacks have likely played an important role during the Holocene in modulating the insolation forcing of summer precipitation changes on the northeastern Tibetan Plateau.

To explore the role of internal feedbacks to the precipitation changes on the northeastern Tibetan Plateau during the Holocene, we removed insolation forcing by deleting the linear trends of climate variabilities that approximate the general monotonic trend in summer insolation during the Holocene (Fig. 2h). The simulated summer precipitation on the northeastern Tibetan Plateau was negatively correlated with 700-hPa meridional winds over western Central Asia and the northwestern Pacific, and positively correlated with 700-hPa meridional winds over East Asia, on millennial time scales during the Holocene (Fig. 5a). In addition, the simulated summer precipitation was negatively associated with 700-hPa zonal winds over the Middle East, the east of Baikal Lake, East Asia, and the northwestern Pacific, on millennial time scales during the Holocene (Fig. 5b). This correlation pattern resembles a wave-like anomaly pattern at middle latitudes, shown by the composite difference in the summer wind fields at 700 hPa between the middle and late Holocene (Fig. 4c). This wave-like pattern is conducive to the southward displacement of cold air and the westward transport of warm water vapour from the northwestern Pacific. When the cold air meets the warm and moist flow on the northeastern Tibetan Plateau, a frontal system will form, generating anomalous precipitation over the area. The southward displacement of cold air was especially obvious at 500 hPa (Fig. 4f). However, no significant ridge and trough was induced over the west of the northeastern Tibetan Plateau during the early Holocene compared with the middle Holocene (Fig. 4b, e), in response to orbital forcing. This circulation pattern reduced frontal rainfall on the northeastern Tibetan Plateau during the early Holocene, in spite of enhanced westward transport of warm water vapour from the northwestern Pacific to this region (Fig. 4b). Since frontal rainfall is the major type of precipitation in summer in this area, the amount of precipitation was lower during the early Holocene than the middle Holocene (Fig. 2d).

Fig. 5
figure 5

Map showing correlations between the summer precipitation averaged over the region (35°–45°N, 95°–105°E) and a meridional winds, and b zonal winds at 700 hPa in summer on millennial time scales (99-point filtered and detrended). Areas in colour indicate that correlations are significant at the 95% confidence level, as estimated by Monte-Carlo experiments. The circle, cross and triangle indicate the locations of Dalianhai Lake, Qinghai Lake and Luanhaizi Lake, respectively. The arrows indicate the directions of the wind anomalies related to the summer precipitation. The pressure levels below surface pressure as a result of topography are omitted

Full size image

The amplitude of the ridge–trough system was negatively related with the strength of the mid-latitude westerlies (Fig. 4c, f). The mid-latitude westerlies, which were negatively correlated with summer precipitation on the northeastern Tibetan Plateau (Fig. 6a), were determined by the meridional thermal difference between the middle-latitude and subpolar regions, ultimately influenced by the meridional insolation gradient. However, the simulated mid-latitude westerlies at 500 hPa showed a slight weakening trend from 9.5 to 4.5 ka BP, which inversely followed the insolation difference between 30°N and 60°N (Fig. 6b). This divergence was probably affected by the Arctic Oscillation (AO), which was positively correlated with the strength of the westerlies at 500 hPa (Fig. 6b); a negative phase of the AO is conductive to the development of a ridge–trough system at middle latitudes (Blunden et al. 2011). The strength of the AO shifted from positive phase during the early Holocene to negative phase during the middle Holocene, and then returned to positive phase during the late Holocene—a finding that agrees well with a proxy record identified in Arctic sea-ice drift (Darby et al. 2012). The positive phase of the AO during the early Holocene was able to overwhelm the insolation forcing, causing a slight decreasing trend in the strength of the mid-latitude westerlies at 500 hPa during this period (Fig. 6b).

Fig. 6
figure 6

a Comparisons between summer rainfall (99-point Gaussian filter, black line) on the northeastern Tibetan Plateau and the strength of the westerlies (99-point Gaussian filter, red line) in the region (30°–60°N, 0°–80°E) at 500 hPa, based on the KCM simulation. b Comparison between the Arctic Oscillation index (AOI) (99-point Gaussian filter, black line) and the strength of the westerlies (99-point Gaussian filter, red line) in the region (30°–60°N, 0°–80°E) at 500 hPa, based on the KCM simulation. The AOI is defined as the difference in the normalized monthly zonal-mean sea level pressure between 35°N and 65°N (Li and Wang 2003). The difference in June insolation between 30°N and 60°N (magenta line) is plotted for comparison

Full size image

The variation of summer precipitation on the northeastern Tibetan Plateau during the Holocene was a result of the interaction between the mid-latitude westerlies and the EASM circulation. The relatively weaker precipitation during the early Holocene, as compared with the middle Holocene, was caused by stronger mid-latitude westerlies resulting from a positive phase of the AO. Therefore, the lagged response of the peak precipitation on the northeastern Tibetan Plateau to the summer insolation maximum during the Holocene was probably caused by the AO.

In addition, the precipitation changes on the northeastern Tibetan Plateau according to the pollen records were inversely correlated with the Mg/Ca-based sea-surface temperatures (SSTs) in the northwestern Pacific during the Holocene (Fig. 7). The simulated summer SSTs in the northwestern Pacific (15°–30°N, 120°–150°E) match well with the Mg/Ca record, showing a cooling trend from the early to middle Holocene and a consistent warming trend thereafter (Fig. 8a). The simulation results also revealed a significantly negative correlation (r = −0.79, P < 0.05) between the northwestern Pacific SSTs and summer precipitation on millennial time scales (Fig. 8b). Their relationship is independent from the orbital forcing, as the correlation (r = −0.68) is still significant at the 95% confidence level after removing the millennial-scale trend. We further applied the generalized difference method to remove the influence of autocorrelation on the p-values. The correlation coefficient between the differential sequences is 0.51 (P < 0.05), indicating an internal relationship between them.

Fig. 7
figure 7

Comparison between the moisture changes on the northeastern Tibetan Plateau (a–c) and d northwestern Pacific SSTs: a tree pollen percentages of Dalianhai Lake (Cheng et al. 2013); b tree pollen percentages from Core QH-2000 of Qinghai Lake (Shen et al. 2005); c annual precipitation reconstructed based on pollen assemblages from Luanhaizi Lake (Wang et al. 2014a); d northwestern Pacific SSTs derived from the Mg/Ca of core A7 (Sun et al. 2005)

Full size image
Fig. 8
figure 8

a Comparison of northwestern Pacific SSTs between the proxy record (black line) and KCM simulation result (red line). The KCM simulation result is the mean summer SSTs averaged over the region (15°–30°N, 120°–150°E) (99-point Gaussian filter). b Comparison between northwestern Pacific SSTs (99-point Gaussian filter, red line) and summer rainfall (99-point Gaussian filter, red line) over the northeastern Tibetan Plateau, based on the KCM simulation

Full size image

Anomalously warm SSTs over the northwestern Pacific tended to increase the integrated tropospheric air temperature (TT) over the western Pacific and Indian oceans (Fig. 9a). In addition, the TT over East Asia was negatively correlated with the northwestern Pacific SSTs (Fig. 9a). That is, anomalously warm northwestern Pacific SSTs would have decreased the meridional thermal contrast between East Asia and the tropical western Pacific. This meridional thermal contrast is suggested to be the cause of the EASM (Zhou and Zou 2010), which induces easterly wind anomalies over East Asia transporting water vapour from the northwestern Pacific to the northeastern Tibetan Plateau (Fig. 9b). Therefore, warm SST anomalies in the northwestern Pacific decreased this water vapour transport, and the resultant precipitation (Figs. 7, 8b). In this way, summer precipitation changes during the Holocene have been modulated by the SSTs in the northwestern Pacific.

Fig. 9
figure 9

Map showing the correlations between northwestern Pacific SSTs and a tropospheric temperature (1000–200 hPa), and b zonal wind at 700 hPa, in summer on millennial time scales (99-point filtered and detrended). Areas in colour indicate that correlations are significant at the 95% confidence level, as estimated by Monte-Carlo experiments. The star, circle, cross and triangle indicate the locations of A7, Dalianhai Lake, Qinghai Lake and Luanhaizi Lake, respectively. The pressure levels below surface pressure as a result of topography are omitted

Full size image

4.3 Possible reason for the mismatch between the simulation and proxy during the late Holocene

During the late Holocene, especially the last millennium, pollen records indicated a remarkably stable climate on the northeastern Tibetan Plateau (Fig. 2a–c). In the contrast, the KCM simulated a quick decrease in the summer precipitation on the northeastern Tibetan Plateau during this period (Fig. 2d). Archaeological research found that there was an increasing intensity of human activities since 3.6 ka BP over this area (Chen et al. 2015a, b). This area is an important sector of the ancient Silk Road. Strong human activities would have remarkably altered the natural vegetation, causing a broken connection between the climate and vegetation evolution.

By comparison, EASM precipitation changes revealed from two remote alpine lakes, Tianchi Lake (Fig. 10a) and Gonghai Lake (Chen et al. 2015a, b) (Fig. 10b), on the Chinese Loess Plateau experienced a sharp decrease during the last millennium, matching well with the KCM simulation results (Fig. 10c). These two lakes are remote, and remain largely untouched during the historical period. Specially, Gonghai Lake is located in Guancen Mountain where was a royal forest, and therefore was highly protected during the historical period. In addition, diatom records suggested that Gonghai Lake had not been significantly disturbed by human activities during the historical period (Liu et al. 2017). Therefore, EASM history documented in Gonghai Lake mainly reflects actual climate change. However, pollen records imprinted by significant human influence fail to indicate moisture change, leading to a mismatch between the proxy and simulation during the late Holocene (Fig. 2).

Fig. 10
figure 10

Comparison of the KCM-simulated summer precipitation on the northeastern Tibetan Plateau (c) and pollen-based EASM precipitation (a) from Tianchi Lake (Chen et al. 2015a, b), and b from Gonghai Lake (Chen et al. 2015a, b) on the Chinese Loess Plateau

Full size image

5 Conclusions

The temporal evolution of precipitation changes on the northeastern Tibetan Plateau during the Holocene and its forcing mechanisms are by no means properly understood. To help address the knowledge gap, this study investigated this topic through combining proxy records with the results from a transient simulation forced by orbital variations in a coupled ocean–sea–ice–atmosphere general circulation model.

The ostracod δ18O records from Qinghai Lake suggested that the northeastern Tibetan Plateau has experienced a persistent drying trend during the Holocene, following the trend of declining summer insolation in the Northern Hemisphere. However, the pollen records from Qinghai Lake and surrounding lakes suggested that the early Holocene was drier than the middle Holocene, showing a lagged response to solar variability. The simulation results match the pollen records well. The model–proxy comparison therefore indicates that the ostracod δ18O is not a proper proxy record of the amount of precipitation on the northeastern Tibetan Plateau. Further study revealed that the ostracod (precipitation) δ18O in Qinghai Lake might reflect the water vapour from the high-latitude North Pacific, in which the δ18O values are much lower than they are in other sea areas, e.g., the low-and mid-Pacific and Indian oceans. Reduced water vapour from the high-latitude North Pacific, resulting from insolation-induced weakening monsoon circulation, was probably an important factor contributing to the increasing values of ostracod δ18O in Qinghai Lake during the Holocene. In addition, the decreased distance of the moisture transport further promoted the enrichment of δ18O in Lake Qinghai during the Holocene.

The changes in summer precipitation on the northeastern Tibetan Plateau during the Holocene have been controlled by the interaction between the EASM circulation and mid-latitude westerlies. A stronger EASM circulation, related to the insolation-induced sea–land thermal contrast, enhances the transport of water vapour from the northwestern Pacific to the northeastern Tibetan Plateau. Weaker mid-latitude westerlies generate troughs and ridges, leading to the invasion of cold air into the northeastern Tibetan Plateau. The mid-latitude westerlies, meanwhile, have been jointly controlled during the Holocene by the meridional thermal contrast triggered by the insolation gradient and the AO. A positive phase of the AO during the early Holocene overwhelmed the effect of the weak meridional insolation gradient, leading to relatively stronger mid-latitude westerlies in this period compared to the middle Holocene. These relatively stronger mid-latitude westerlies ultimately reduced the amounts of precipitation during the early Holocene, in spite of being affected by the most powerful EASM throughout the Holocene. Therefore, the AO has played an important role during the Holocene in the lagged response of summer precipitation to insolation forcing. In addition, the SSTs in the northwestern Pacific have further modulated summer precipitation through their effect on the circulation of the EASM.