Transient simulation of orbital-scale precipitation variation in monsoonal East Asia and arid central Asia during the last 150 ka

Authors

  • Xinzhou Li,

    1. State Key Laboratory of Loess and Quaternary Geology, Institute of Earth Environment, Chinese Academy of Sciences, Xi'an, China
    Search for more papers by this author
  • Xiaodong Liu,

    Corresponding author
    1. State Key Laboratory of Loess and Quaternary Geology, Institute of Earth Environment, Chinese Academy of Sciences, Xi'an, China
    2. School of Human Settlements and Civil Engineering, Xi'an Jiaotong University, Xi'an, China
    • Corresponding author: X. Liu, Institute of Earth Environment, Chinese Academy of Sciences, 10 Fenghui S. Rd., Xi'an 710075, China. (liuxd@loess.llqg.ac.cn)

    Search for more papers by this author
  • Linjing Qiu,

    1. State Key Laboratory of Loess and Quaternary Geology, Institute of Earth Environment, Chinese Academy of Sciences, Xi'an, China
    Search for more papers by this author
  • Zhisheng An,

    1. State Key Laboratory of Loess and Quaternary Geology, Institute of Earth Environment, Chinese Academy of Sciences, Xi'an, China
    Search for more papers by this author
  • Zhi-Yong Yin

    1. Department of Marine Science and Environmental Studies, University of San Diego, San Diego, California, USA
    Search for more papers by this author

Abstract

[1] A long-term transient simulation is conducted using the Community Climate System Model version 3 and the orbital acceleration technique to analyze the impact of insolation change caused by the Earth's orbital forcing on precipitation in the monsoonal East Asia (EA) and arid central Asia (CA) over the past 150 ka. Our results show that annual precipitation in both EA and CA has strong signals of the 20 ka precessional cycles and varies in phase with the Northern Hemisphere (NH) summer insolation. Similar characteristics can also be observed from previously published oxygen isotope records of stalagmites near EA and CA. Composite analyses based on seven precessional cycles suggest that the increase (decrease) in the NH summer (winter) insolation enhances EA (CA) summer (winter) precipitation by modulation of the Asian monsoon (westerly) circulation in summer (winter). When the precession-induced NH summer insolation increases, the Asian summer monsoon circulation is enhanced and EA precipitation increases significantly. Meanwhile, the increase in the summer insolation at the precessional scale is accompanied by a decrease in the winter insolation, which causes dramatic cooling of the troposphere in the lower latitudes. Consequently, the CA winter precipitation increases due to the changes in the temperature gradient and the westerly circulation. Therefore, the responses of the Asian monsoon and westerly circulation to summer and winter insolation variations induced by the precessional cycles determine precipitation in the respective rainy seasons and are the primary cause leading to the synchronous variation patterns of annual precipitation in EA and CA at the orbital scale.

1 Introduction

[2] Milankovitch's theory of climate change assumes that the insolation is affected by the Earth's orbital parameters (i.e., eccentricity, axial tilt, and precession of the Earth's orbit) and drives the glacial-interglacial cycles and global climate change from millennial to 100 ka timescales [Milankovitch, 1969]. Many geological records have indicated that the 100, 40, and 20 ka orbital periods corresponding to the eccentricity, axial tilt, and precessional cycles are present throughout the Quaternary climatic evolution in the monsoonal East Asia (EA) east of the Tibetan Plateau (TP) and arid central Asia (CA) west of the TP. For example, studies on the Chinese loess-paleosol sequence, widely accepted as a good proxy of geological climate information [Kukla et al., 1990; Ding et al., 1994; Lu et al., 2004; Sun et al., 2006], and the stalagmite records that provide high-resolution absolute dating [Wang et al., 2001, 2008] show that the winter and summer monsoons in EA have strong signals of the precessional cycles. Meanwhile, the loess [Ding et al., 2002] and stalagmite records from CA [Cheng et al., 2012] suggest that climate variations in the inland Asia have similar signals of orbital cycles as those in EA. Up to date, however, the mechanisms of such synchronous variation patterns have not been fully investigated.

[3] Numerical simulations have been widely used to examine the roles of orbital forcing in the evolution of Asian monsoon and inland arid climate. Earlier simulation studies are mostly concentrated on the responses of Asian monsoon to the changes in insolation caused by orbital forcing from prescribed orbital parameters, either for a particular time slice [Kutzbach, 1981] or for continuous time slice series [Prell and Kutzbach, 1987]. In recent years, advances in computation technology and the application of orbital acceleration technique in climate models [Jackson and Broccoli, 2003; Lorenz and Lohmann, 2004] have made it possible to perform transient simulations to illustrate the continuous responses of climate systems to orbital forcing and the phase relations between orbital forcing and climate responses on long timescales. For example, Kutzbach et al. [2008] conducted a simulation of global monsoon over the past 280 ka and validated the in-phase relations of the summer Asian monsoon and Northern Hemisphere (NH) insolation, using a fully coupled global ocean-atmosphere model with an acceleration scheme by a factor of 100. Previous studies using different acceleration factors (e.g., 10, 20, 50, 100, and 1000 years) [e.g., Jackson and Broccoli, 2003; Lorenz and Lohmann, 2004; Kutzbach et al., 2008] indicated that although the responses of the deep ocean to the orbital acceleration were very slow, the upper ocean and atmosphere showed similar responses to different acceleration speeds. For example, in a study using accelerated forcing in a fully coupled model, Lorenz and Lohmann [2004] found that acceleration factors of 10 and 100 produced similar simulation results. Kutzbach et al. [2008] further suggested that a quasi-equilibrium response to the orbital forcing in the upper ocean can be achieved in the acceleration scheme with a factor of 100. For this current study, we conducted several numerical experiments using the Community Climate System Model version 3 (CCSM3) with different acceleration factors of 10, 20, 50, and 100 and examined the similarities and differences in the simulated atmospheric circulations and air temperatures, as well as different seasonal precipitation patterns (not shown). Our experiments confirmed the conclusions from the aforementioned studies, and therefore, for our long-term transient simulations, it should be acceptable to use a factor of 100 for acceleration.

[4] In order to explain the correlations of climate change between EA and CA exhibited by the geological records [e.g., Ding et al., 2002; Wang et al., 2008; Bar-Matthews et al., 2003; Cheng et al., 2012], this study uses a fully coupled climate model to analyze the spatial and temporal variation patterns of precipitation and to determine the phase relationships between orbital forcing and precipitation at the orbital scale. We focus on the relatively short precessional cycle (20 ka) in the past 150 ka period that covers the last glacial-interglacial (Holocene) cycle.

2 Numerical Model and Transient Experiments

[5] The Community Climate System Model version 3 (CCSM3) [Collins et al., 2006], released by the National Center for Atmospheric Research, was used in this study. CCSM3 is a fully coupled global climate model that consists of four components: atmosphere, land, ocean, and sea ice. It has been widely used for past, present, and future climate simulations [e.g., Meehl et al., 2006a; Otto-Bliesner et al., 2006; Meehl et al., 2006b]. In this study, the horizontal resolution of the atmospheric module is T42 (approximately 2.8° × 2.8°), and the vertical dimension of the atmosphere contains 26 levels. The land module includes dynamic vegetation, drainage confluence, and snow cover/ice changes [Dickinson et al., 2006]. The oceanic component uses a dipole grid and orthogonal curvilinear coordinates with a nominal horizontal resolution of 1° × 1° and a depth (z) coordinate with 40 levels [Smith and Gent, 2002]. The sea ice module consists of a dynamics-thermodynamics model with subgrid-scale parameterization of ice thickness [Briegleb et al., 2002].

[6] We performed a long-term transient experiment for the past 150 ka using CCSM3 and the orbital acceleration technique [Jackson and Broccoli, 2003; Lorenz and Lohmann, 2004]. We started the simulation using the fixed orbital parameters at 150 ka B.P. for the first 100 years. In the subsequent simulation years, the model was forced by the orbital parameters calculated for the next century using an acceleration factor of 100 (1 model year = 100 calendar years). All other boundary conditions stayed the same during the whole integration. Therefore, the actual length of the simulation was reduced to 1500 model years by accelerating the variations in the orbital cycles with a factor of 100. In this way, we obtained the transient simulation results covering the entire last glacial-interglacial cycle during the past 150 ka (including seven complete precessional cycles).

[7] Composite analysis was used to compare the effects of insolation variation of the precessional cycle on precipitation variation in two regions: EA (100°E–120°E, 35°N–50°N; the box in Figure 1c) and CA (50°E–70°E, 35°N–50°N; the box in Figure 1d). In each of the precession cycles, we took the time with the maximum (minimum) of June insolation at 45°N as the central year, and then we chose 5 model years before and after the central year. Thus, the actual time span of 1100 calendar years served as a period of high (low) insolation corresponding to 11 model years with the orbital acceleration factor of 100. Therefore, there are seven periods of high (low) insolation in the seven precession cycles of the past 150 ka, as indicated by the red (blue) columns in Figure 3. The average of the seven periods of high (low) insolation is used to represent the high-insolation (low-insolation) phase and abbreviated as HI (LI) in the composite analysis. In this way, we averaged the climatic conditions for the HI and LI phases separately and then analyzed the temporal-spatial variations of precipitation and characteristics of atmospheric circulations. We also used power spectral and cross-spectral analyses to examine the variation patterns and phase relationships between the simulated precipitation series, the corresponding geological records, and the NH June insolation. It should be noted that the time series data in this paper were processed using the 1 ka moving averages to filter out the noise and outliers for easy interpretation.

Figure 1.

Comparisons between the (a, b) CMAP-observed and (c, d) CCSM3-simulated climatologically averaged percentages of summer (JJA) (Figures 1a and 1c) and winter (DJF) (Figures 1b and 1d) precipitation to annual total. The red boxes in Figures 1c and 1d represent the locations of the East Asia (EA) and central Asia (CA) regions in this study, respectively.

3 Orbital-Scale Variations of Regional Annual Precipitation in Asia

[8] The temporal-spatial distributions of modern Asian precipitation were well simulated by CCSM3. Figure 1 shows the percentages of average summer (June–August or JJA; Figure 1a) and winter (December–February or DJF; Figure 1b) precipitation to annual precipitation during 1980–2009 based on the Climate Prediction Center Merged Analysis of Precipitation (CMAP) from the National Oceanic and Atmospheric Administration (NOAA) [Xie and Arkin, 1997] and the same seasonal percentages to annual total averaged for the same 30 years simulated by CCSM3 with the present-day orbital parameters (Figures 1c and 1d). Although summer precipitation in northwestern China was underestimated, the simulated precipitation distribution patterns matched well with the observed patterns in the other regions, particularly in the EA and CA regions specified for this study. The EA region is located in the summer rain climate, while CA is in the winter rain climate. Accordingly, Figure 1 shows that summer precipitation is the dominant component of the annual total in EA and winter precipitation is the dominant component in CA. Approximately 66.8% of the observed and 47.8% of the simulated annual total precipitation occur during summer in the monsoonal EA region, while the observed and simulated winter precipitation accounts for 32.3% and 49.2% of the annual totals in the arid CA region, respectively. This means that the rainy season in EA (CA) occurs in summer (winter) and that the rainy season precipitation contributes a significant amount to the respective annual totals in these regions.

[9] Figures 2a–2e show the time series of the June insolation at 45°N, the simulated annual precipitation of EA and CA, and the stalagmite oxygen isotope records with accurate dating and high resolution near EA and CA. The EA stalagmite records came from the Sanbo and Hulu Caves in China [Wang et al., 2008], whereas the CA stalagmite records came from the Soreq and Peqiin Caves in Israel [Ayalon et al., 2002; Bar-Matthews et al., 2003]. In general, the arid CA lacks proper geological climate records for comparison. We chose the Israeli stalagmite records because of their relatively close proximity to the CA region and specifically for the reason that they were from the winter rain area. In order to compare the cyclic patterns and phases of the time series data, we performed power spectral analyses on the simulated precipitation and stalagmite records (Figures 2f–2j). The time series and power spectral analyses both indicate that annual precipitation in these regions not only had the evident 20 ka cycles but also had the apparent in-phase relations with the NH midlatitude summer insolation during the past 150 ka. The stalagmite records displayed the 20 ka cycles with some minor variations from the simulated series, although other cycles of lower frequencies were visible in these records as well.

Figure 2.

Time series of (a) June insolation at 45°N and simulated annual precipitation in the (b) monsoonal East Asia (EA) and (d) arid central Asia (CA) regions, as well as the (c) Chinese stalagmite records near EA [Wang et al., 2008] and (e) Israeli stalagmite records near CA [Bar-Matthews et al., 2003] during the past 150 ka. Results of power spectral analysis (f–j) are also presented corresponding to the series in Figures 2a–2e. The dashed lines represent the precessional band at 20 ka.

4 Responses of Summer and Winter Precipitation and Atmospheric Circulation to Insolation Forcing

[10] Since the rainy season precipitation contributes heavily to the annual totals, we focus on the respective rainy seasons in EA and CA to explore the mechanisms behind the apparent in-phase relationships at the orbital scale as illustrated in Figure 2. Previous studies have demonstrated that summer precipitation in EA is closely linked with the Asian summer monsoon [Ding and Chan, 2005], which has an in-phase relation with the NH summer insolation variation [Kutzbach et al., 2008; Liu and Shi, 2009]. Figure 3 shows the potential relationships between the June insolation, JJA precipitation in EA, and East Asian summer monsoon index (EASMI) during the past 150 ka. The EASMI is calculated based on the definition by Wang [2000] using the following equation, with consideration of the EA extent defined for this study:

display math

where U′ and V′ are the zonal and meridional JJA wind departures from the zonal average at 850 hPa in the area 105°E–120°E and 25°N–45°N, respectively. Figure 3 indicates that when the NH summer insolation is high (low) corresponding to the HI (LI) phase of the precessional cycles, the EASMI is remarkably enhanced (weakened), which causes increased (reduced) precipitation in the EA region with a northward (southward) shift of the monsoon rain belt. These results suggest that the EA monsoon intensity has an in-phase variation pattern with the NH summer insolation at the precessional scale, which is also consistent with the stalagmite records reported by Wang et al. [2001, 2008]. In summary, insolation appears to be the key factor that determines the orbital-scale summer precipitation variation in EA through its influences on the East Asian monsoon intensity (the actual mechanism will be further explained in the following).

Figure 3.

Time series of (a) the June insolation at 45°N (same as in Figure 2a), (b) the simulated summer (JJA) precipitation in EA, (c) the East Asian summer monsoon index (EASMI) according to Wang [2000], (d) the December insolation at 45°N, (e) the winter (DJF) precipitation in CA, and (f) the winter westerly index (WWI) according to Rossby et al. [1939]. The red and blue columns represent the high-insolation (HI) and low-insolation (LI) phases of the precessional cycles, respectively.

[11] The orbital-scale precipitation variation in the arid CA, due to its winter rain regime, is mainly regulated by the winter insolation. Figure 3 shows the 150 ka series of the December insolation (Figure 3d), the DJF precipitation (Figure 3e), and the winter westerly index (WWI) (Figure 3f). The WWI is defined as the global average of the sea level pressure differences between 35°N and 55°N, according to the zonal index proposed by Rossby et al. [1939]. In the following, “high index” and “low index” are used to represent the strong and weak midlatitude westerlies, respectively. The simulation results reveal that both annual precipitation (Figure 2d) and DJF precipitation in CA (Figure 3e) exhibit cyclic patterns of 20 ka intervals similar to that of EA, while the WWI (Figure 3f) displays an inverse variation pattern compared to winter precipitation. With the decrease (increase) in the NH winter insolation, the midlatitude westerly circulation weakens (intensifies) in general and shifts southward (northward) upstream of CA, which is linked to the enhanced (reduced) upstream westerly winds south of 40°N, eventually leading to the increases (decreases) in DJF precipitation in CA (the actual mechanism will be further explained in the following). Consequently, annual precipitation in CA varies in phase with annual precipitation in EA, but with the underlying mechanisms poorly understood so far, especially at the orbital scale.

[12] Cross-spectral analysis is an effective tool to test the relationship between two sequences within different frequency bands [Hannan, 1970]. The results of cross-spectral analyses show that both the simulated rainy season precipitation series and the geological records display the 20 ka precessional cycles and vary in phase with the June insolation at 45°N (Figure 4). The 20 ka cross-spectral cycles in the simulated rainy season precipitation series of EA and CA and the Chinese stalagmite records are statistically significant at the 90% confidence level, while the 20 ka cycles in the Israeli stalagmite records are very close to the 90% confidence level. Cross-spectral analyses also reveal that the JJA precipitation of EA leads the June insolation of NH by 20° (about 1200 years), while the DJF precipitation of CA leads the June insolation of NH by 22° (about 1350 years). Considering the period of the precessional cycles (20 ka), the phase differences of 1200–1350 years are relatively small and consequently can be considered as in phase.

Figure 4.

Cross-spectral analyses of the CCSM3-simulated precipitation and stalagmite records with June insolation at 45°N: (a) EA summer precipitation (solid line) and the Chinese stalagmite records (dashed line) and (b) CA winter precipitation (solid line) and the Israeli stalagmite records (dashed line). The dash-dotted lines represent the 90% confidence level.

[13] Based on the above results, we can conclude that the NH summer and winter insolation variations caused by the precessional forcing are associated with the EA summer monsoon and winter westerly circulation in midlatitudes, ultimately leading to the precipitation variations in CA and EA at the orbital scale. In the following, we will analyze the responses of the atmospheric circulations to the orbital forcing to provide explanations to the mechanisms that cause such linkages. Figure 5a shows the latitudinal distribution of the JJA insolation in the HI and LI phases of the precessional cycle, and the insolation differences between the HI and LI phases as HI-LI values. It can be seen that the insolation during the HI phase is higher than that of LI in summer, and the large differences appear in the low to middle latitudes (Figure 5a). In the meantime, the insolation of HI is lower than that of LI in winter (Figure 5b), and although the differences are also greater in the low latitudes, they minimize rapidly for the high-latitude regions. Different latitudinal distributions of insolation cause changes in the latitudinal temperature gradients, leading to variations in the atmospheric circulation patterns. Comparing the HI phase with the LI phase of the precessional cycle, Figure 5c shows the summer surface air temperature anomalies of HI relative to LI. In response to the higher NH summer insolation in the HI phase, the land surface temperature rises significantly while the sea surface temperature remains essentially unchanged, leading to the enhanced land-sea thermal contrast that further causes enhanced ocean-land air pressure gradient. Therefore, in the HI phase, the East Asian summer monsoon circulation (Figure 5e) is intensified and the monsoon rain belt extends farther to the north, which causes increases in summer precipitation in EA (Figure 3b).

Figure 5.

Differences in insolation, temperatures, and wind fields between the composites of the high-insolation (HI) and low-insolation (LI) phases of the seven precessional cycles (as HI-LI): (a) NH (0°N–70°N) latitudinal distributions of summer (JJA) insolation during the HI and LI phases (red solid and dashed lines) and the HI-LI differences (black solid line), (b) same as Figure 5a but for winter (DJF), (c) differences in summer surface air temperatures (unit: °C) for the region 40°E–180°E, (d) differences in mean winter tropospheric temperatures (unit: °C) for the region 40°W–100°E, (e) differences in the summer 850 hPa wind (unit: m/s) field in 40°E–180°E, and (f) differences in the winter 850 hPa wind (unit: m/s) field in 40°W–100°E. The shaded areas in Figures 5c and 5d represent regions exceeding the 99% confidence level of Student's t test. The red boxes in Figures 5e and 5f represent the EA and CA regions, respectively, and the black bold lines are the 2000 m contour that shows the approximate boundary of the Tibetan Plateau.

[14] In contrast to the above analysis on summer precipitation in EA, we examined winter tropospheric temperatures and circulation patterns for winter precipitation in CA. Figure 5d shows the differences of winter tropospheric (surface to 200 hPa) temperature between HI and LI for the realm relevant to CA. Due to the lower winter insolation in the HI phase, especially in the low-latitude areas (Figure 5b), temperatures in the middle- to low-latitude regions are also lower dramatically (similar to the findings in Kutzbach et al. [2013]), while temperatures in the high-latitude regions vary with only minor changes (Figure 5d). This would cause a reduced temperature gradient in the middle to high latitudes, which is the primary cause for the reduced intensity of the westerlies north of 40°N during the HI phase. In the meantime, we observe that the westerly winds upstream of CA but south of 40°N are significantly enhanced (Figure 5f). As suggested by Namias [1950], during the periods of high index of WWI, the westerlies shift into the high-latitude region with the development of a strong circumpolar vortex, and the cold air activities are mostly contained in the polar region. This setting, according to Figure 3, tends to occur during the LI phase of the precessional cycle. On the other hand, during the low index of WWI associated with the HI phase, cold air outbreaks would occur more frequently with a weakened but expanded circumpolar vortex. This would lead to the weakened westerlies north of 40°N with more frequent southward extrusions of cold air activities and southward shift of the westerlies to the region south of 40°N, which intensifies the abnormal trajectory of westerly flows in the lower troposphere upstream of CA (e.g., at the 850 hPa level), transporting moisture from the Atlantic to central Asia via the Mediterranean and Southern Europe (Figure 5f), leading to the increased precipitation in CA. This mechanism for long-distance moisture transport is also facilitated by the fact that the winter westerly winds in these latitudes extend from the surface to the upper troposphere. Therefore, the combination of increased frequencies of cold air outbreaks from the north due to reduced intensity of zonal flows in the high latitudes and increased availability of moisture may have contributed to the increases in winter precipitation in CA during the HI phase. This is why there is an in-phase variation pattern in annual precipitation between CA and EA, but caused by different processes and in different seasons, although all are linked to the insolation cycles.

5 Conclusions and Discussion

[15] In this study, the potential effects of insolation variations caused by orbital parameters, especially the precessional cycle, on precipitation in the monsoonal EA and the arid CA regions during the past 150 ka are investigated using a coupled general circulation model CCSM3 with an orbital acceleration factor of 100. We have discovered in-phase relationships between the NH midlatitude summer insolation and annual precipitation in both EA and CA with the 20 ka precessional cycles. The statistically significant precessional cycles and the in-phase relationships with the NH insolation can also be observed in the oxygen isotope records of stalagmites from China near EA and from Israel near CA. Composite analyses suggest that the inverse-phased NH midlatitude summer insolation and winter insolation modulated by the precessional forcing are responsible for the corresponding variations in summer precipitation in EA and winter precipitation in CA, respectively. The intensification of summer insolation mainly occurs in the middle to low latitudes in the HI phase of the precessional cycle, which causes a greater warming over the Asian landmass than over the surrounding ocean, leading to the enhanced East Asian monsoon circulation and northward shift of the monsoon rain belt due to an increased land-ocean thermal contrast. As a result, the precipitation increases significantly in the monsoonal EA region. In the meanwhile, during the HI phase of the precessional cycle, there is a decrease in the NH winter insolation, which causes more dramatic cooling of the troposphere in the low-latitude regions comparing to the LI phase. The changes in the NH tropospheric temperature gradients cause weakened westerlies north of 40°N but enhanced westerly flows south of this latitude upstream of CA, transporting moisture from the Atlantic to CA via Southern Europe and the Mediterranean and thereby increasing winter precipitation in CA together with increased frequencies of cold air outbreaks. In conclusion, the responses of the monsoon and westerly circulations to insolation variations at the orbital scale determine the variation patterns of the rainy season precipitation in EA and CA. It should be pointed out that the large-scale topography such as the Tibetan Plateau may not only be closely related to the formation and evolution of the Asian monsoon and inland arid climate but also affect the orbital-scale precipitation changes in CA and EA by modulating the atmospheric circulation, which will be a subject of future research.

[16] Controlled by the insolation variations at the orbital scale, the East Asian monsoon and midlatitude westerlies have significant impacts on regional climate changes in Asia. Our long-term transient experiments illustrate that the Asian monsoon and midlatitude westerlies exhibit prominent signals of the precessional cycles and cause the in-phase variations in precipitation between these two regions from the last glacial period to the Holocene. Although our study only considers the impact of insolation changes, specifically the precessional cycles, the simulated results are consistent with certain geological records [Ding et al., 2002; Wang et al., 2008; Bar-Matthews et al., 2003]. This implies that even though there are still many other factors that may influence orbital-scale climate changes, such as ice sheet evolution, vegetation changes, ocean circulations, and greenhouse gas concentration, insolation plays a dominant role in the climate system at this timescale. Undoubtedly, concentration of the atmospheric greenhouse gases has changed significantly and is an important forcing or feedback factor affecting climate change in the geological history especially in the last precessional cycle [Yin and Berger, 2012]. In future work, we will further consider the combined effects of the Earth's orbital forcing and other forcing factors including atmospheric CO2.

[17] The loess and stalagmite records have demonstrated strong and positive correlations of climate changes between the monsoonal EA and the arid CA regions [e.g., Ding et al., 2002; Wang et al., 2008; Bar-Matthews et al., 2003; Cheng et al., 2012], but many relevant issues are still worthy of further study. For example, Cheng et al. [2012] discovered signals of the precessional cycle in precipitation in the arid Asian inland reconstructed from the Kesang Cave stalagmite records over the past 500 ka, but they proposed that the variation pattern was likely to be associated with the moisture transport by the Indian summer monsoon because their study site was located in the eastern part of the Asian arid region and belongs to the summer-rain-dominant climate. In contrast, our analyses indicate that the precipitation variation in CA of the winter-rain-dominant climate is mainly determined by the westerly circulation in the rainy winter season. Additionally, a composite analysis of sediments from 11 lakes suggested that the moisture condition in the arid central Asia has an inverse phase relationship with that in the Asian monsoon region during the early and late Holocene periods [Chen et al., 2008], which covers half a precessional cycle, rather than the in-phase relation illustrated in the current study and also in others. Similar conclusions were also proposed by Feng et al. [2011] and Jin et al. [2012]. Although, at the present, it is difficult to fully explain why there are discrepancies in regard to the phase relations of moisture conditions between the arid CA and monsoonal EA in different studies, we believe that such discrepancies are at least partly due to the definition of the arid central Asia region. The extent of CA in our study is located in 50°E–70°E and 35°N–50°N, while the lake sediment samples used in Chen et al. [2008], for example, come from a much larger area of 43.20°E–117.38°E and 38.40°N–51.00°N [see Chen et al., 2008, Table 1]. Recently, An et al. [2012] analyzed sediment cores from Lake Qinghai in the transition zone between the Asian monsoon region and the inland arid region to explore the interplay between the westerlies and the Asian summer monsoon during the past 32 ka and again found an inverse phase relationship between them. They also illustrated that the climate in this region was dominated by the westerlies during the glacial period, whereas the Asian summer monsoon dominated during the Holocene. Through this discussion, it is clear to us that within the Asian inland region, there are complex interactions between multiple circulation systems in different geological periods, and much more work is needed to achieve a full understanding of the climatic effects of the Asian monsoon and westerly circulation and their interactions.

Acknowledgments

[18] This work was supported by the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB03020601), the National Basic Research Program of China (2010CB833406), and the National Natural Science Foundation of China (41290255 and 41075067).

Ancillary