- Email: [email protected]
East Asian monsoon variations during Oxygen Isotope Stage 5: evidence from the northwestern margin of the Chinese loess plateau
Quaternary Science Reviews 18 (1999) 1127 } 1135
East Asian monsoon variations during Oxygen Isotope Stage 5: evidence from the northwestern margin of the Chinese loess plateau F.H. Chen , J. Bloemendal *, Z.D. Feng, J.M. Wang, E. Parker, Z.T. Guo Department of Geography, Lanzhou University, Lanzhou, Gansu 730000, China Department of Geography, University of Liverpool, Roxby Building, Liverpool, L69 3BX, UK Department of Earth and Environmental Studies, Montclair State University, Upper Montclair, NJ 07043, USA Beijing Institute of Geology, CAS, Beijing 100029, China
Abstract We present the results of high-resolution multi-proxy climate studies of the S1 palaeosol, corresponding to oxygen isotope stage (OIS) 5, from the northwestern margin of the Chinese Loess Plateau area. Here, S1 is much thicker (ca. 6}8 m) than in the central Loess Plateau areas (ca. 2 m), where most previous studies have been conducted. Hence, much higher-resolution stratigraphic studies are possible, yielding more insight into the temporal variations of the East Asian monsoon during MIS 5. The frequency-dependent magnetic susceptibility, as well as the concentration of secondary carbonate, is used as an indicator of the summer monsoon intensity, and the median particle size as an indicator of the winter monsoon intensity. The results suggest that the northwestern margin of the Chinese Loess Plateau experienced the strongest summer monsoon intensity in sub-stage (OISS) 5e and the weakest in OISS 5a, among the three warmer periods during stage 5. The summer monsoon was weaker in OISS 5b than in OISS 5d. A dusty interval interrupted the second warmer period (5c) and a soil-forming event interrupted the "rst colder period (5d). The results also suggest that the directions of changes in the intensities of summer and winter monsoons may not always have been proportionately opposite. For example, the weakest summer monsoon occurred in OISS 5a during which the winter monsoon was not the strongest. We further conclude that the winter monsoon during the last interglacial was probably driven by global ice volume #uctuations, while the summer monsoon was primarily controlled by the northern hemisphere solar insolation and was probably modi"ed by a feedback mechanism. That is, the climatic bu!ering e!ect of low-latitudinal oceans may have distorted the response of the summer monsoon to insolation variations. Finally, our results do not show the degree of climatic instability comparable to that recorded in the GRIP ice core for the last interglacial (OISS 5e), even though the study area is situated in a region which has been sensitive to climatic changes. 1999 Elsevier Science Ltd. All rights reserved.
1. Introduction Several attempts have been made to reconstruct the behaviour of the east Asian monsoon climate during the last glacial cycle based on stratigraphic investigations of Chinese loess deposits (e.g. An et al., 1991; Guo et al., 1996; Liu et al., 1995; Porter and An, 1995; Sun et al., 1996; Chen et al., 1997). Some of the conclusions of this work are that the east Asian monsoon climate in China is strongly correlated with climatic changes in the high latitudes of the northern hemisphere via the Westerlies (Porter and An, 1995) and the Siberian-Mongolian high
*Corresponding author. Tel: 0044 151 794 2834; fax: 0044 151 7942834; e-mail: [email protected].
(Ding et al., 1995). In addition, the extreme winter monsoon events of the last glacial recorded on the Chinese Loess Plateau were found to be teleconnected with the Heinrich events recorded in North Atlantic deep-sea cores (Porter and An, 1995; Guo et al., 1996), while the extreme summer monsoon events during the last glacial were correlated with the warm episodes recorded in Greenland ice cores (Chen et al., 1997). In contrast, there have been fewer attempts to reconstruct in detail the behaviour of the east Asian monsoon during the period of globally relatively warm climate represented by oxygen isotope (OIS) stage 5, and in particular during sub-stage (OISS) 5e, the last interglacial maximum. Unfortunately, climatic records are poorly resolved in the central Chinese Loess Plateau area, where most palaeoclimatic studies have been conducted.
0277-3791/98/$ - see front matter 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 3 7 9 1 ( 9 8 ) 0 0 0 4 7 - X
1128
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135
Because of the slow sediment accumulation rate in this area, palaeosol S1, corresponding to oxygen isotope stage 5, is only about 2 m thick. In addition, the strong pedogenesis here may have excessively a!ected the palaeoclimatic record (Chen et al., 1997) However, along the northwestern margin of the Loess Plateau the sediment accumulation rate is much higher, producing an S1 palaeosol of ca 6}8 m thickness in the form of a loess-soil pedocomplex. In addition, the intensity of pedogenesis in the more arid northwest is much weaker. Both of these factors enhance the potential of this area for high-resolution palaeoclimatic studies.
2. Pedostratigraphy and chronology The work presented here is a high-resolution stratigraphic study of palaeosol S1, which developed during marine oxygen isotope stage (OIS) 5 (An et al., 1991; Guo et al., 1996; Liu et al., 1995). The three sections investigated are situated in the northwestern margin of the Chinese Loess plateau where the dust deposition rate was 4}7 times higher than in the central Loess Plateau (Yue et al., 1991; Chen and Zhang, 1993). Tuxiangdao (TXD) section is the northwesternmost site with a mean annual precipitation (MAP) of 370 mm; Yuanbo (YB) section is the wettest site with MAP of 500 mm, and Caoxian (CX) section the driest site with a MAP of 340 mm. TXD section is on the fourth terrace of Huangshui River (a branch of the Yellow River) in Xining Basin (Fig. 1) and the last interglacial palaeosol S1 in the section is about 8 m thick (Fig. 2). Palaeosol S1, bracketed by loess layers L1 and L2, actually consists of "ve sub-units: three welldeveloped palaeosols (S1S1, S1S2, S1S3) and two interbedded loess layers (S1L1 and S1L2). Among the three palaeosols, S1S3 is the best developed in the area according to studies of soil micromorphology (Kemp et al., 1996). The loess layers within S1 have similar micromorphological features to that of the Malan Loess (L1) with more secondary CaCO "lling plant root channels. The Yuanbo (YB) section is on the fourth terrace of Daxia River (a branch of the Yellow River) in Linxia Basin and the palaeosol S1 at this section is about 7.4 m thick. Three well-developed palaeosols correlating with S1S1, S1S2 and S1S3 at Tuxiangdao section are identi"ed in this section (Fig. 2). The best-developed palaeosol S1S3 is characterized by a reddish-brown color, needle-shaped CaCO in"llings, abundant grass rootlet channels, and up to 3}4% iron oxide content in the "ne fraction. The other two palaeosols (S1S2 and S1S1) are less well-developed compared to S1S3. The loess layer S1L1 does not show any obvious pedogenic imprints, and loess layer S1L2 contains a weakly developed pedogenic unit at a depth of ca. 38 m. The Caoxian (CX) section is situated on the sixth terrace of the Yellow River in Jingyuan Basin. The terrace is covered with up to 500 m of loess
deposits (Yue et al., 1991). Palaeosol S1 is 6.2 m thick and also consists of three palaeosols and two loess layers. The middle palaeosol S1S2 is interrupted by a weathered loess layer in the middle. The "ve subunits of palaeosol S1, that is three palaeosols separated by two loess layers, have also been found in the Lanzhou loess stratigraphy (Derbyshire et al., 1995; Dai et al., 1995). A series of samples was used for thermoluminescence (TL) dating at the State Key Laboratory for Loess and Quaternary Research (Xian). At the TXD section, three samples were taken for dating: from 2 m above S1S1 in the Malan Loess; at the lower limit of S1S1; and from 20 cm below S1S3. These samples have TL ages of 70$6 ka (standard deviation for the estimated analytical error), 90$6 ka and 149$12 ka, respectively (Fig. 2). At the YB section, a weathered layer close to the bottom of the Malan Loess (L1) has a TL age of 65$5 ka; the bottom of S1S1 is 90$10 ka, and the lower portion of S1S3 is 125$10 ka. In addition, the Blake event, a brief geomagnetic reversal dated at 118128 ka (Nowacozyk et al., 1994), was documented within loess layer S1L2 in the Xining basin (Zhu et al., 1994), and the Lanzhou basin (J.J. Li, unpublished data), and in the Jingyuan section (Yue et al., 1991) of the western Loess Plateau. This reversal episode was also reported to occur within S1L2 of the Huanxian section in the central Loess Plateau (Zheng et al., 1995). Based on the S1 pedostratigraphy in Chinese loess, the position of the Blake event together with the TL dates, it is reasonable to correlate S1S1 with marine oxygen isotope substage (OISS) 5a, S1S2 with MISS 5c, and S1S3 with MISS 5e, and the two loess layers S1L1 and S1L2 with MISS 5b and 5d, respectively (Fig. 2). The revised SPECMAP MIS termination ages (Martinson et al., 1987) were used to constrain the time span of S1 with the assumption that there was no time lag between the global temperature change and variation of the east Asian monsoon. The lower boundary L2/S1 is therefore set at 129.84 ka and the upper boundary S1/L1 at 73.91 ka (Fig. 3). We also used both the sediment particle-size age model of Porter and An (1995) and the magnetic #ux age model of Kukla et al. (1988) to establish a detailed chronology for palaeosol S1. We found that an age discrepancy of up to 4 ka age for S1 exists among the three sections when using the particle-size age model. This di!erence may be caused by the fact that particle size was measured at a lower resolution (8}10 cm) at the CX and YB sections. The magnetic #ux age model works well in practice in the central Loess Plateau (Guo et al., 1996) and seems to produce consistent results in this study, despite the fact that its fundamental assumptions about the origin of the magnetic material have been disputed (e.g Zhou et al., 1990; Maher and Thompson, 1992). Using the high-resolution (2 cm intervals) magnetic #ux modeled ages, together with the presence of the Blake event (within S1L2) and the TL dates, we are con"dent that the three
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135
1129
Fig. 1. The location of the study sites: Tuxiangdao (TXD) section in Xining Basin, Yuanbo (YB) section in Linxia Basin, Caoxian (CX) section in Jingyuan Basin. The Lanzhou (GLS) section in the Lanzhou Basin is included for comparison. The insert shows the extent of loess deposition in central China. Shaded area in main map represents the Tibetan Plateau.
palaeosols within S1 developed during OISS 5a, 5c and 5e, and that the two intervening loess layers were deposited during OISS 5b and 5d (Fig. 2). The resulting chronology is used to generate the multi-proxy-climate time series shown in Figs. 3}5.
3. Climate proxies It is generally agreed that the major carrier of the magnetic susceptibility signal in the Chinese loess-
palaeosol sequences is ultra"ne superparamagnetic minerals produced by in situ pedogenesis (e.g. Zhou et al., 1991; Maher and Thompson, 1992; 1994). However, the magnitude of the magnetic susceptibility signal is in#uenced not only by the concentration of the ultra"ne superparamagnetic minerals, but also by other factors such as the concentration of other types of magnetic minerals, and by diamagnetic and paramagnetic material (Feng, 1996; Feng and Johnson, 1995; Hunt et al., 1995; Thompson and Old"eld, 1986). Therefore, the interpretation of the magnetic susceptibility of loess as a direct
1130
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135
Fig. 2. Pedostratigraphy of palaeosol S1 at TXD, YB and CX sections. Again the Lanzhou section is included for comparison. The black bars beside the lithological columns are the positions of the Blake event. Marine isotope stages (MIS) are used for chronological reference. The black dots are the TL sampling positions and TL dates are expressed in thousands of years (ka).
Fig. 3. The temporal variations in: (1) the percent frequency-dependent susceptibility (sfd%) (three-point moving average), (2) CaCO concentration, and (3) the median grain diameter (MD) of particle size at TXD, YB and CX sections. The heavy dashed lines show the L1/S1 and S1/L2 boundaries.
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135
Fig. 4. Comparison of northern hemisphere July insolation (653N) and the summer monsoon intensity, as indicated by the percent frequencydependent magnetic susceptibility (sfd%) (three-point moving average) at the TXD, YB and CX sections.
Fig. 5. Correlation of the winter monsoon variations as indicated by median grain (MD) at the TXD section with the oxygen isotope records from deep-sea core V19-30 and Greenland GRIP ice core.
indicator of pedogenic intensity is not entirely straightforward. However, the frequency-dependent magnetic susceptibility indicates the proportional contribution of superparamagnetic minerals to the magnetic susceptibility and therefore is a more direct indicator of the degree of pedogenesis (Liu et al., 1990). If the degree of pedogenesis in Chinese loess has indeed recorded the strength of the summer monsoon (An et al., 1991; Maher and Thompson, 1994), the frequency-depen-
1131
dent magnetic susceptibility can be used more con"dently to trace variations in the strength of the summer monsoon The content, morphology and thickness of pedogenic carbonate in a soil pro"le are a function of the soilforming environment and the duration of the interval of carbonate concentration (Gile et al., 1965; 1966; Reheis, 1987). The secondary CaCO in the three palaeosols (S1S1, S1S2, and S1S3) within palaeosol S1 of the three sections is present primarily in the form of small cylinders and nodules, with abundant carbonate-"lled and carbonate-marked grass root channels, suggesting that the carbonate is of pedogenic origin (Gile et al., 1965; Machette, 1985; Feng et al., 1994). The CaCO content in soils under semiarid climates normally re#ects the e!ectiveness of evapotranspiration, which is primarily temperature-dependent, and the depth of CaCO concentration is determined by available soil moisture, which is basically precipitation-dependent, if the duration factor of CaCO concentration is a constant (Aandahl, 1982; Reheis, 1987). The CaCO concentration in the Chinese loess-palaeosol sequence may have recorded the summer monsoon history, which controlled both summer temperature and precipitation. However, it should be noted that the CaCO concentration may not be a sensitive indicator of the summer precipitation and/or temperature, due to uncertainty in the duration factor of CaCO concentration and to the translocation of carbonate within a soil pro"le. The last interglacial palaeosol S1 along the northwestern margin of the Chinese Loess Plateau is basically a cumulic loessial pedocomplex, i.e. it developed when the dust deposition rate was still high enough to inhibit further di!erentiation of soil horizons. The dust particles in this kind of cumulic loessial pedocomplex have experienced very little post-depositional weathering (Ding et al., 1995; Feng et al., 1994). In addition, the particle-size distribution of the modern severe dust storms, which primarily occur in the winter half year, and which is probably representative of the present interglacial (Holocene), is quite similar to that of the last interglacial palaeosol S1 (Dai et al., 1995). This implies that the particle-size distribution of S1 may have not been altered by post-depositional pedogenic processes and that particle size can be used as a reasonably sensitive indicator of the winter monsoon intensity
4. Measurement methods The last interglacial palaeosol S1 from the three sections was sampled at 2 cm intervals (about 100}200 ys) and the low- and high-frequency magnetic susceptibility of the 1759 samples was measured three times using a Bartington Instruments magnetic susceptibility meter and dual frequency sensor. The averaged results were
1132
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135
used to calculate the percent frequency-dependent susceptibility (sfd%), as de"ned by Thompson and Old"eld (1986). The CaCO content was measured using a calcimeter (Bascomb, 1961), at 2 cm intervals for the Tuxiangdao and Caoxian sections, and at 10 cm intervals for the Yuanbo section. For particle size analysis, carbonate in the samples was removed by repeatedly adding 15% HCl and washing the excessive HCl using a centrifuge. The organic matter in the CaCO -free samples was then removed by adding H O and heating for 4 h at 803C. Finally, 2 ml of Cal gon was added to the sample to disperse the particles (Janitzky, 1987). The particle size distribution was measured at 4 cm intervals for the TXD section, at 10 cm intervals for the YB section and at 8 cm intervals for the CX section, using a Coulter Electronics Ltd LS130 laser di!raction particle size analyser. To obtain higher stratigraphic resolution for S1S3, corresponding to MISS 5e, the particle size of this interval was measured at "ner (2 cm) intervals at the TXD section.
5. Results and discussion 5.1. Summer monsoon variations and events The thick (6}8 m) S1 pedocomplex at the northwestern margin of the western Loess Plateau provides the best context for preserving the climatic signature of the last interglacial within the Chinese Loess Plateau. As discussed above, the percent frequency-dependent magnetic susceptibility (sfd%), together with soil morphological features, is here used as an indicator of the summer monsoon intensity. Fig. 3 shows that the three palaeosols of the complex have relatively high sfd% values. The highest sfd% peak in S1S3 supports the morphological inference that palaeosol S1S3 is the best developed amongst the three palaeosols within S1. The sfd% records suggest that S1S2 experienced a slightly greater degree of pedogenesis than S1S1. In the CX section, the sfd% record indicates that the overall relatively strong pedogenesis within S1S2 was interrupted by an episode of weaker pedogenesis represented by a less-weathered unit (at ca. 40 m depth). In this case it is possible that variations in pedogenic intensity (as re#ected by changes in carbonate content) where preserved only in the drier conditions at this site where post-depositional alteration was minimal. The low sfd% values in loess layer S1L1 may result from the fact that this layer was minimally weathered. In contrast, the slightly higher sfd% values in loess layer S1L2 concur with observable weathering imprints. In S1L2 at the YB section, a small peak of sfd% indicates a weakly-developed pedogenic unit (at ca. 38 m). This is in line with the slightly higher modern mean annual precipitation at this site.
Whereas the CaCO concentration may not be a sensi tive indicator of summer precipitation and/or temperature, it may provide additional information about the summer monsoonal history. At the moister site (YB section), soil CaCO apparently was heavily depleted within S1S1 and S1S3. However, CaCO content fails to distin guish S1S2 from the two loess layers S1L1 and S1L2 at this site. At the two drier sites (TXD and CX), the CaCO variations seem to predate the sfd% variations, i.e. the CaCO peaks appear stratigraphically below the sfd% peaks. This consistent stratigraphic lag between sfd% and CaCO content is logical because the carbonate is always leached downward and re-deposited in the lower portion of contemporary soils and even beyond under semiarid climates (Aandahl, 1982; Reheis, 1987). The summer monsoon intensity, as indicated by the CaCO content, seems to show a generally decreasing trend from S1S3 (MISS 5e) to S1S1 (MISS 5a), which is in agreement with the conclusion drawn from the Xfd% data (Fig. 4). Thus, the interval corresponding to the formation of S1S3 experienced the strongest summer monsoon intensity, and that for S1S1 the weakest intensity, among the three palaeosol units. The second warm/moist period (S1S2) was interrupted by a dusty (colder and drier) interval. The "rst dust depositional period (S1L2) was more favorable for weathering than the second dust depositional period (S1L1), and a soil-forming (warmer and moister) phase was recorded in the "rst loess-depositing period (S1L2). 5.2. Winter monsoon variations and events In semi-arid climates the particle-size distribution of the S1 palaeosol may not have been altered by postdepositional pedogenic processes and therefore can be interpreted as a reasonably sensitive indicator of the winter monsoon intensity. The variations in median particle size of S1 (MD in Fig. 3) suggest that the winter monsoon intensity, in terms of the maximum strength of the winds, changed abruptly during the L2/S1 (MIS 6/5) and S1/L1 (OIS 5/4) transitional periods. The soil-forming periods (S1S1, S1S2, and S1S3) and dust-depositional periods (S1L1 and S1L2) are also readily distinguished by the median particle size. Among the three palaeosols, S1S3 has the "nest median particle size and S1S2 the coarsest, indicating that winter monsoon intensity was weakest in OIS 5e (S1S3) and strongest in OIS 5d (S1S2). The median particle size also indicates that the winter monsoon was stronger during OIS 5d (S1L2) than during the OIS 5b (S1L1). It should be mentioned that the weakly developed pedogenic unit within S1L2 at the YB section and the less-weathered unit within the S1S2 at CX section are not clearly distinguishable in the medianparticle-size data. By comparing the variations of the summer monsoon intensity, as indicated by sfd%, with the variations of the
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135
winter monsoon, as indicated by the median particle size, several conclusions can be drawn. First, the winter monsoon during the two strong winter monsoon periods (S1L1 and S1L2) of the last interglacial (S1) was considerably weaker than that during the two glacials represented by loess units L1 and L2. In contrast, the summer monsoon during the two strong winter monsoon periods (S1L1 and S1L2) was also as weak as that during the two glacials. Second, all of the three sections show that the strength of the winter monsoon was gradually weakening while the strength of the summer monsoon increased relatively abruptly during the L2/S1 (MIS 6/5) transitional period. During the S1/L1 (MIS 5/4) transitional period, the summer monsoon may have weakened quickly and the strengthening of the winter monsoon was considerably delayed at the YB and CX sections. However, the TXD section (the northwesternmost site) shows that the dominance of the winter monsoon occurred abruptly during the S1/L1 transitional period, implying that the winter monsoon became dominant at the northwestern limit of the area in#uenced by the summer monsoon as soon as the climate entered the glacial (L1) mode. Third, this study also shows that the negative correlation between the intensities of the summer and the winter monsoons has not always been linear. For example, the S1S1 palaeosol experienced the weakest summer monsoon and the S1S2 the strongest winter monsoon among the three palaeosols within S1. Both the winter and summer monsoons were stronger in S1L2 than in S1L1. 5.3. Global comparisons Fig. 4 compares northern hemisphere insolation (653N) with time series of the sfd% records for the three sections. During MIS 5, the variations in the summer monsoon intensity followed the variations in the northern summer solar insolation. For the three warmer periods, both the summer monsoon intensity and the solar insolation were greatest in OISS 5e, and lowest in OISS 5a. However, the two minima of the summer monsoon intensity corresponding to OISS 5b and 5d do not accurately re#ect the northern hemisphere solar insolation; that is, the summer monsoon intensity was higher in 5d than in the 5b, while the insolation was lower in 5d than in 5b (Fig. 4). We speculate that the maximum energy storage in low-latitudinal oceans during 5e bu!ered the minimum insolation conditions in the interior during 5d, and it was for this reason that the summer monsoon intensity was unexpectedly higher during 5d than that during 5b. We also observe that the 5c maximum of the summer monsoon intensity extended into the 5b minimum of insolation, and consequently the 5b summer monsoon lagged behind the corresponding minimum of the insolation by more than 6000 yr. A possible explanation is that during the two high-insolation periods (5e
1133
and 5c), low-latitudinal oceans stored enough energy and acted as a bu!er to delay the weakening of the summer monsoon in response to decreasing insolation. Finally, the weakest summer monsoon in 5b might be attributable to both the 5b insolation minimum and to the trend of generally decreasing insolation for the whole of MIS 5. The comparison between the summer monsoon and the insolation also shows that the summer monsoon dominance of the last interglacial was established when the insolation reached its peak in 5e and that the dominance was ended when the insolation was approaching its lowest in 5a. Our results are also of interest in the context of speculations about the occurrence (or otherwise) of rapid climatic #uctuations in Greenland, based on ice core data (GRIP Members, 1993; and cf Mayewski and Bender, 1995). Based on modern analog (0.7 per mille per 3C: Broecker, 1995), the temperature at Summit, Greenland, oscillated at a magnitude of 6}73C during the early part (OISS 5e and 5d) and in a range of 2}43C during the later part (OISS 5c, 5b, and 5a) of OIS 5 on decadal to millennial time scales (Fig. 5) (GRIP ice core: GRIP Members, 1993). However, the reconstructed winter monsoon intensity from the TXD section which has the highest time resolution (samples at about 150-year intervals) does not show such dramatic oscillations. The contrast of the winter monsoon intensity between the glacials (L1 and L2) and during OIS 5 (S1) is striking. Even the contrast between warmer periods (5e, 5c, and 5a) and the colder periods (5d and 5b) are unambiguously distinguishable. However, the oscillations superimposed on the "rst-order variations (glacial/interglacial) and the second-order variations (stadial/interstadial) are not clearly identi"able. In contrast, the variations in the reconstructed winter monsoon intensity are quite similar to the variations in the global ice volume as recorded by the dO record from Paci"c Ocean core V19-30 (Fig. 5). It has been proposed that the variation in the winter monsoon of the Chinese Loess Plateau my be teleconnected with variations in the climate of the North Atlantic region (Porter and An, 1995; Ding et al., 1995). It seems logical to suggest that the variations in the climate of the North Atlantic region high latitudes were dynamically related to the changes in the global ice volume. If so, the teleconnection between "rst-order winter monsoon variations in China and the climate in high latitudes might have been driven by the global ice volume.
6. Conclusions This study shows that the northwestern margin of the Chinese Loess Plateau experienced the strongest summer monsoon during OISS 5e and the weakest during OISS 5a, among the three warmer periods during the OIS 5. The second warmer period (5c) was interrupted by
1134
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135
a dusty (colder and drier) interval. The summer monsoon was weaker in OISS 5b than in OISS 5d during which a soil-forming (warmer and moister) event occurred. In contrast to previous suggestions that the intensities of the summer and winter monsoons always changed in opposite directions (An et al., 1991; Liu, et al., 1995), we found that the directions might not have been always proportionately opposite. For example, we infer that the weakest summer monsoon occurred in OISS 5a and the strongest winter monsoon in OISS 5c. Also, both the winter and summer monsoons were stronger in OISS 5d than in OISS 5b. We also conclude that the winter monsoon intensity in China during the last interglacial was probably driven by global ice volume #uctuations, and that the summer monsoon intensity was primarily controlled by northern solar insolation and probably modi"ed by a feedback mechanism. We suggest that the feedback mechanism was the bu!ering e!ect caused by heat storage of lowlatitude oceans, which had the e!ect of reducing the sensitivity of the summer monsoon intensity to solar insolation. As a result, the summer monsoon intensity and the solar insolation were not completely in phase during the last interglacial. Finally, the three sections investigated did not record any climate instability comparable to that recorded in the GRIP ice core during the last interglacial (OISS 5e), even though the study area is situated in the northwestern margin of the Chinese Loess Plateau area in#uenced by the summer monsoon, and which has been shown to be very sensitive to climate change (Chen and Zhang, 1993; Li et al., 1988; Yie et al., 1990). Acknowledgments We want to express our thanks to Mr. Yutian Zhang and Jixiu Cao for "eld and laboratory help; to Dr. Max Zhao for helping to improve the English; and to Steve Porter and an anonymous referee for helpful comments. TL dates were measured at the Xi'an State Key Laboratory for Loess and Quaternary Geology. This research was supported by NSFC grant (No. 49571065) and by the Royal Society K. C. Wong Fellowship. References Aandahl, A. R. (1982). Soils of the Great Plains. Lincoln, Nebraska: University of Nebraska Press. An, Z. S., Kukla, G. J., Porter, S. C., & Xiao, J. L. (1991). Magnetic susceptibility evidence of monsoon variation on the loess Plateau of central China during the last 130,000 years. Quaternary Research, 36, 29}36. Bascomb, C. L. (1961). A Calcimeter for routine use on soil samples. Chemistry and Industry, 45, 1826}1827. Broecker, W. S. (1995). ¹he Glacial =orld according to =ally. LamontDoherty Earth Observatory of Columbia University, Eldigio Press.
Chen, F. H., & Zhang, W. X. (1993). ¹he loess stratigraphy and Quater nary glacial problems in the areas of Gansu province and Qinghai province. Beijing: Science Press, (in Chinese). Chen, F. H., Bloemendal, J., Wang, J. M., Li, J. J., Old"eld, F., & Ma, H. Z. (1997). High-resolution multi-proxy climate records from Chinese loess: evidence for rapid climatic changes over the last 75 kyr. Palaeogeography, Palaeoecology, Palaeoclimatology, 130, 323}335. Dai, X. R., Shi, Y. X., & Xue, B. (1995). Granulometric characteristics and signi"cance of the deposits from a recent extraordinary heavy dust storm in Lanzhou, Gansu, North China. Journal of ¸anzhou ;niversity (Natural Sciences), 31, 168}174 (in Chinese) Derbyshire, E., Keen, D., Kemp, R. A., Rolph, T. C., Shaw, J., & Meng, X. M., (1995). Loess-palaeosol sequences as recorders of palaeoclimatic variations during the last glacial-interglacial cycle: some problems of correlation in north-central China. In E. Derbyshire (ed.), =ind blown sediments in the quaternary. London: Wiley. (pp. 7}18). Ding, Z. L., Liu, T. S., Rutter, N. W., Yu, Z. W., Guo, Z. T., & Zhu, R. X. (1995). Ice-volume forcing of east-Asian winter monsoon variations in the past 800,000 years. Quaternary Research, 44, 149}159. Feng, Z.-D., Johnson, W. C., Sprowl, D. R., Lu, Y.-C., & Ward, P. A. (1994). Climatic signals from loess-soil sequences in the central Great Plains, USA. Palaeogeography, Palaeoecology, Palaeoclimatology, 110, 345}358. Feng, Z.-D. (1996). Climatic implications of magnetic susceptibility and Be-10 #ux in Chinese loess-soil sequences. Catena, 27, 143}147. Feng, Z.-D., & Johnson, W. C. (1995). Factors a!ecting the magnetic susceptibility of a loess-soil sequence, Barton County, Kansas, USA. Catena, 24, 25}37. Gile, J. H., Peterson, F. F., & Grossman, R. B. (1965). The K horizon: a master soil horizon of carbonate accumulation. Soil Science, 99, 74}82. Gile, J. H., Peterson, F. F., & Grossman, R. B. (1966). Morphological and genetic sequences of carbonate accumulation in desert soils. Soil Science, 101, 347}360. Guo, Z., Liu, T., Guiot, J., Wu, N., Lu, H., Han, J., Liu, J., & Gu, Z. (1996). High frequency pulses of east Asian monsoon climate in the last two glacials: link with the North Atlantic. Climate Dynamics, 12, 701}709. GRIP Members (1993). Climate instability during the last interglacial period recorded in the GRIP ice core. Nature, 364, 203}207. Janitzky, P. (1987). Particle size analysis. In M. Singer, & P. Janitzky (Ed.), Field and laboratory procedures used in soil chronosequence study. ;.S. Geological Survey Bulletin 1648, (pp. 11}16). Hunt, C. P., Banerjee, S. K., Han., J., Solheid, P. A., Oches, E., Sun, W., & Liu, T. (1995). Rock magnetic proxies of climatic change in the loess-palaeosol sequences of the western loess plateau of China. Geophysical Journal International, 123, 232}244. Kemp, R. A., Derbyshire, E., Chen, F. H., & Ma, H. Z. (1996). Padosedimentary development and paleoenvironmental signi"cance of the S1 palaeosol on the northeastern margin of the Qinghai-Xizang (Tibetan) Plateau. Journal of Quaternary Science, 11, 95}106. Kukla, G., Heller, F., Liu, X. M., Chun, X. T., & An, Z. S. (1988). Pleistocene climates in China dated by magnetic susceptibility. Geology, 16, 811}814. Li, J. J., Feng, Z. D., & Tang, L. Y. (1988). Late Quaternary monsoon pattern on the Loess Plateau of China. Earth Surface Processes and ¸andforms, 13, 125}135. Liu, X. M., Liu, T. S., Heller, F., & Xu, T. C. (1990). Frequencydependent susceptibility of loess and Quaternary palaeosol. Quaternary Science, (1), 42}50. (in Chinese) Liu, T. S., Guo, Z. T., Liu, J. Q., Ding, Z. L., Gu, Z. Y., & Wu, N. Q. (1995). variations of eastern Asian monsoon over the last 140,000 Years. Bulletin of Society Geol. France, 166, 221}229.
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135 Machette, M. N. (1985). Calcic soils of the southwest U. S. In D. L. Weide (Ed.), Soils and quaternary of the southwest ;. S. Geological Society of America Special Paper 203, (pp. 1}20). Maher, B. A., & Thompson, R. (1992). Paleoclimatic signi"cance of the mineral magnetic record of the Chinese loess and palaeosols. Quaternary Research, 37, 155}170. Maher, B. A., & Thompson, R. (1994). Pedogenesis and paleoclimate } interpretation of the magnetic susceptibility record of Chinese loess-palaeosol sequences * comment. Geology, 22, 857}858. Martinson, D. G., Pisias, N. G., & Hays, J. D. (1987). Age dating and the orbital theory of the ice age: development of a high-resolution 0 to 30,000 years chronostratigraphy. Quaternary Research, 27, 1}29. Mayewski, P. A., & Bender, M. (1995). The GISP2 ice core recordpaleoclimate highlights. Reviews of Geophysics Supplement, p.1287}1296. US National Report to the IUGG 1991}1994. Nowacozyk, N. R., Frederichs, T. W., Eisenhauer, A., & Gard, G. (1994). Magnetostratigraphic data from Late Quaternary sediments from the Yermak Plateau, Arctic Ocean: evidence for four geomagnetic polarity events within the last 170Ka of the Brunhes Chron. Geophysical Journal International, 117, 453}471. Porter, S. C., & An, Z. S. (1995). Correlation between climate events in the North-Atlantic and China during the last glaciation. Nature, 375, 305}308.
1135
Reheis, M. C. (1987). Climatic implication of alternating clay and carbonate formation in semiarid soils of south-central Montana. Quaternary Research, 27, 270}282. Sun, D. H., Zhou, J., Jiang, F. C., & Porter, S. C. (1996). Preliminarystudy on the summer monsoon climate of Chinese loess plateau during the last interglacial period. Chinese Science Bulletin, 41, 321}325. Thompson, R., & Old"eld, F. (1986). Environmental magnetism. London: Allen and Unwin. Yie, D. Z. (1990). Study on global changes of China in advance. Beijing: Meteorological Press Beijing. (in Chinese) Yue, L. P., Lei, X. Y., & Qu, H. J. (1991). Magnetostratigraphy study on the Jingyuan loess section, Gansu, China. Quaternary Science, 4, 349}353. (in Chinese). Zheng, H. B., Rolph, T., Shaw, J., & An, Z. S. (1995). A detailed palaeomagnetic record for the last interglacial period. Earth and Planetary Science ¸etters, 133, 339}351. Zhou, L. P., Old"eld, F., Wintle, A. G., Robinson, S. G., & Wang, J. T. (1990). Partly pedogenic origion of magnetic variations in Chinese Loess. Nature, 346, 737}739. Zhu, R. X., Zhou, L. P., Laj, C., Mazaud, A., & Ding, Z. L. (1994). The Blake geomagnetic polarity episode recorded in Chinese loess. Geophysical Research ¸etters, 21, 697}700.
East Asian monsoon variations during Oxygen Isotope Stage 5: evidence from the northwestern margin of the Chinese loess plateau F.H. Chen , J. Bloemendal *, Z.D. Feng, J.M. Wang, E. Parker, Z.T. Guo Department of Geography, Lanzhou University, Lanzhou, Gansu 730000, China Department of Geography, University of Liverpool, Roxby Building, Liverpool, L69 3BX, UK Department of Earth and Environmental Studies, Montclair State University, Upper Montclair, NJ 07043, USA Beijing Institute of Geology, CAS, Beijing 100029, China
Abstract We present the results of high-resolution multi-proxy climate studies of the S1 palaeosol, corresponding to oxygen isotope stage (OIS) 5, from the northwestern margin of the Chinese Loess Plateau area. Here, S1 is much thicker (ca. 6}8 m) than in the central Loess Plateau areas (ca. 2 m), where most previous studies have been conducted. Hence, much higher-resolution stratigraphic studies are possible, yielding more insight into the temporal variations of the East Asian monsoon during MIS 5. The frequency-dependent magnetic susceptibility, as well as the concentration of secondary carbonate, is used as an indicator of the summer monsoon intensity, and the median particle size as an indicator of the winter monsoon intensity. The results suggest that the northwestern margin of the Chinese Loess Plateau experienced the strongest summer monsoon intensity in sub-stage (OISS) 5e and the weakest in OISS 5a, among the three warmer periods during stage 5. The summer monsoon was weaker in OISS 5b than in OISS 5d. A dusty interval interrupted the second warmer period (5c) and a soil-forming event interrupted the "rst colder period (5d). The results also suggest that the directions of changes in the intensities of summer and winter monsoons may not always have been proportionately opposite. For example, the weakest summer monsoon occurred in OISS 5a during which the winter monsoon was not the strongest. We further conclude that the winter monsoon during the last interglacial was probably driven by global ice volume #uctuations, while the summer monsoon was primarily controlled by the northern hemisphere solar insolation and was probably modi"ed by a feedback mechanism. That is, the climatic bu!ering e!ect of low-latitudinal oceans may have distorted the response of the summer monsoon to insolation variations. Finally, our results do not show the degree of climatic instability comparable to that recorded in the GRIP ice core for the last interglacial (OISS 5e), even though the study area is situated in a region which has been sensitive to climatic changes. 1999 Elsevier Science Ltd. All rights reserved.
1. Introduction Several attempts have been made to reconstruct the behaviour of the east Asian monsoon climate during the last glacial cycle based on stratigraphic investigations of Chinese loess deposits (e.g. An et al., 1991; Guo et al., 1996; Liu et al., 1995; Porter and An, 1995; Sun et al., 1996; Chen et al., 1997). Some of the conclusions of this work are that the east Asian monsoon climate in China is strongly correlated with climatic changes in the high latitudes of the northern hemisphere via the Westerlies (Porter and An, 1995) and the Siberian-Mongolian high
*Corresponding author. Tel: 0044 151 794 2834; fax: 0044 151 7942834; e-mail: [email protected].
(Ding et al., 1995). In addition, the extreme winter monsoon events of the last glacial recorded on the Chinese Loess Plateau were found to be teleconnected with the Heinrich events recorded in North Atlantic deep-sea cores (Porter and An, 1995; Guo et al., 1996), while the extreme summer monsoon events during the last glacial were correlated with the warm episodes recorded in Greenland ice cores (Chen et al., 1997). In contrast, there have been fewer attempts to reconstruct in detail the behaviour of the east Asian monsoon during the period of globally relatively warm climate represented by oxygen isotope (OIS) stage 5, and in particular during sub-stage (OISS) 5e, the last interglacial maximum. Unfortunately, climatic records are poorly resolved in the central Chinese Loess Plateau area, where most palaeoclimatic studies have been conducted.
0277-3791/98/$ - see front matter 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 3 7 9 1 ( 9 8 ) 0 0 0 4 7 - X
1128
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135
Because of the slow sediment accumulation rate in this area, palaeosol S1, corresponding to oxygen isotope stage 5, is only about 2 m thick. In addition, the strong pedogenesis here may have excessively a!ected the palaeoclimatic record (Chen et al., 1997) However, along the northwestern margin of the Loess Plateau the sediment accumulation rate is much higher, producing an S1 palaeosol of ca 6}8 m thickness in the form of a loess-soil pedocomplex. In addition, the intensity of pedogenesis in the more arid northwest is much weaker. Both of these factors enhance the potential of this area for high-resolution palaeoclimatic studies.
2. Pedostratigraphy and chronology The work presented here is a high-resolution stratigraphic study of palaeosol S1, which developed during marine oxygen isotope stage (OIS) 5 (An et al., 1991; Guo et al., 1996; Liu et al., 1995). The three sections investigated are situated in the northwestern margin of the Chinese Loess plateau where the dust deposition rate was 4}7 times higher than in the central Loess Plateau (Yue et al., 1991; Chen and Zhang, 1993). Tuxiangdao (TXD) section is the northwesternmost site with a mean annual precipitation (MAP) of 370 mm; Yuanbo (YB) section is the wettest site with MAP of 500 mm, and Caoxian (CX) section the driest site with a MAP of 340 mm. TXD section is on the fourth terrace of Huangshui River (a branch of the Yellow River) in Xining Basin (Fig. 1) and the last interglacial palaeosol S1 in the section is about 8 m thick (Fig. 2). Palaeosol S1, bracketed by loess layers L1 and L2, actually consists of "ve sub-units: three welldeveloped palaeosols (S1S1, S1S2, S1S3) and two interbedded loess layers (S1L1 and S1L2). Among the three palaeosols, S1S3 is the best developed in the area according to studies of soil micromorphology (Kemp et al., 1996). The loess layers within S1 have similar micromorphological features to that of the Malan Loess (L1) with more secondary CaCO "lling plant root channels. The Yuanbo (YB) section is on the fourth terrace of Daxia River (a branch of the Yellow River) in Linxia Basin and the palaeosol S1 at this section is about 7.4 m thick. Three well-developed palaeosols correlating with S1S1, S1S2 and S1S3 at Tuxiangdao section are identi"ed in this section (Fig. 2). The best-developed palaeosol S1S3 is characterized by a reddish-brown color, needle-shaped CaCO in"llings, abundant grass rootlet channels, and up to 3}4% iron oxide content in the "ne fraction. The other two palaeosols (S1S2 and S1S1) are less well-developed compared to S1S3. The loess layer S1L1 does not show any obvious pedogenic imprints, and loess layer S1L2 contains a weakly developed pedogenic unit at a depth of ca. 38 m. The Caoxian (CX) section is situated on the sixth terrace of the Yellow River in Jingyuan Basin. The terrace is covered with up to 500 m of loess
deposits (Yue et al., 1991). Palaeosol S1 is 6.2 m thick and also consists of three palaeosols and two loess layers. The middle palaeosol S1S2 is interrupted by a weathered loess layer in the middle. The "ve subunits of palaeosol S1, that is three palaeosols separated by two loess layers, have also been found in the Lanzhou loess stratigraphy (Derbyshire et al., 1995; Dai et al., 1995). A series of samples was used for thermoluminescence (TL) dating at the State Key Laboratory for Loess and Quaternary Research (Xian). At the TXD section, three samples were taken for dating: from 2 m above S1S1 in the Malan Loess; at the lower limit of S1S1; and from 20 cm below S1S3. These samples have TL ages of 70$6 ka (standard deviation for the estimated analytical error), 90$6 ka and 149$12 ka, respectively (Fig. 2). At the YB section, a weathered layer close to the bottom of the Malan Loess (L1) has a TL age of 65$5 ka; the bottom of S1S1 is 90$10 ka, and the lower portion of S1S3 is 125$10 ka. In addition, the Blake event, a brief geomagnetic reversal dated at 118128 ka (Nowacozyk et al., 1994), was documented within loess layer S1L2 in the Xining basin (Zhu et al., 1994), and the Lanzhou basin (J.J. Li, unpublished data), and in the Jingyuan section (Yue et al., 1991) of the western Loess Plateau. This reversal episode was also reported to occur within S1L2 of the Huanxian section in the central Loess Plateau (Zheng et al., 1995). Based on the S1 pedostratigraphy in Chinese loess, the position of the Blake event together with the TL dates, it is reasonable to correlate S1S1 with marine oxygen isotope substage (OISS) 5a, S1S2 with MISS 5c, and S1S3 with MISS 5e, and the two loess layers S1L1 and S1L2 with MISS 5b and 5d, respectively (Fig. 2). The revised SPECMAP MIS termination ages (Martinson et al., 1987) were used to constrain the time span of S1 with the assumption that there was no time lag between the global temperature change and variation of the east Asian monsoon. The lower boundary L2/S1 is therefore set at 129.84 ka and the upper boundary S1/L1 at 73.91 ka (Fig. 3). We also used both the sediment particle-size age model of Porter and An (1995) and the magnetic #ux age model of Kukla et al. (1988) to establish a detailed chronology for palaeosol S1. We found that an age discrepancy of up to 4 ka age for S1 exists among the three sections when using the particle-size age model. This di!erence may be caused by the fact that particle size was measured at a lower resolution (8}10 cm) at the CX and YB sections. The magnetic #ux age model works well in practice in the central Loess Plateau (Guo et al., 1996) and seems to produce consistent results in this study, despite the fact that its fundamental assumptions about the origin of the magnetic material have been disputed (e.g Zhou et al., 1990; Maher and Thompson, 1992). Using the high-resolution (2 cm intervals) magnetic #ux modeled ages, together with the presence of the Blake event (within S1L2) and the TL dates, we are con"dent that the three
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135
1129
Fig. 1. The location of the study sites: Tuxiangdao (TXD) section in Xining Basin, Yuanbo (YB) section in Linxia Basin, Caoxian (CX) section in Jingyuan Basin. The Lanzhou (GLS) section in the Lanzhou Basin is included for comparison. The insert shows the extent of loess deposition in central China. Shaded area in main map represents the Tibetan Plateau.
palaeosols within S1 developed during OISS 5a, 5c and 5e, and that the two intervening loess layers were deposited during OISS 5b and 5d (Fig. 2). The resulting chronology is used to generate the multi-proxy-climate time series shown in Figs. 3}5.
3. Climate proxies It is generally agreed that the major carrier of the magnetic susceptibility signal in the Chinese loess-
palaeosol sequences is ultra"ne superparamagnetic minerals produced by in situ pedogenesis (e.g. Zhou et al., 1991; Maher and Thompson, 1992; 1994). However, the magnitude of the magnetic susceptibility signal is in#uenced not only by the concentration of the ultra"ne superparamagnetic minerals, but also by other factors such as the concentration of other types of magnetic minerals, and by diamagnetic and paramagnetic material (Feng, 1996; Feng and Johnson, 1995; Hunt et al., 1995; Thompson and Old"eld, 1986). Therefore, the interpretation of the magnetic susceptibility of loess as a direct
1130
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135
Fig. 2. Pedostratigraphy of palaeosol S1 at TXD, YB and CX sections. Again the Lanzhou section is included for comparison. The black bars beside the lithological columns are the positions of the Blake event. Marine isotope stages (MIS) are used for chronological reference. The black dots are the TL sampling positions and TL dates are expressed in thousands of years (ka).
Fig. 3. The temporal variations in: (1) the percent frequency-dependent susceptibility (sfd%) (three-point moving average), (2) CaCO concentration, and (3) the median grain diameter (MD) of particle size at TXD, YB and CX sections. The heavy dashed lines show the L1/S1 and S1/L2 boundaries.
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135
Fig. 4. Comparison of northern hemisphere July insolation (653N) and the summer monsoon intensity, as indicated by the percent frequencydependent magnetic susceptibility (sfd%) (three-point moving average) at the TXD, YB and CX sections.
Fig. 5. Correlation of the winter monsoon variations as indicated by median grain (MD) at the TXD section with the oxygen isotope records from deep-sea core V19-30 and Greenland GRIP ice core.
indicator of pedogenic intensity is not entirely straightforward. However, the frequency-dependent magnetic susceptibility indicates the proportional contribution of superparamagnetic minerals to the magnetic susceptibility and therefore is a more direct indicator of the degree of pedogenesis (Liu et al., 1990). If the degree of pedogenesis in Chinese loess has indeed recorded the strength of the summer monsoon (An et al., 1991; Maher and Thompson, 1994), the frequency-depen-
1131
dent magnetic susceptibility can be used more con"dently to trace variations in the strength of the summer monsoon The content, morphology and thickness of pedogenic carbonate in a soil pro"le are a function of the soilforming environment and the duration of the interval of carbonate concentration (Gile et al., 1965; 1966; Reheis, 1987). The secondary CaCO in the three palaeosols (S1S1, S1S2, and S1S3) within palaeosol S1 of the three sections is present primarily in the form of small cylinders and nodules, with abundant carbonate-"lled and carbonate-marked grass root channels, suggesting that the carbonate is of pedogenic origin (Gile et al., 1965; Machette, 1985; Feng et al., 1994). The CaCO content in soils under semiarid climates normally re#ects the e!ectiveness of evapotranspiration, which is primarily temperature-dependent, and the depth of CaCO concentration is determined by available soil moisture, which is basically precipitation-dependent, if the duration factor of CaCO concentration is a constant (Aandahl, 1982; Reheis, 1987). The CaCO concentration in the Chinese loess-palaeosol sequence may have recorded the summer monsoon history, which controlled both summer temperature and precipitation. However, it should be noted that the CaCO concentration may not be a sensitive indicator of the summer precipitation and/or temperature, due to uncertainty in the duration factor of CaCO concentration and to the translocation of carbonate within a soil pro"le. The last interglacial palaeosol S1 along the northwestern margin of the Chinese Loess Plateau is basically a cumulic loessial pedocomplex, i.e. it developed when the dust deposition rate was still high enough to inhibit further di!erentiation of soil horizons. The dust particles in this kind of cumulic loessial pedocomplex have experienced very little post-depositional weathering (Ding et al., 1995; Feng et al., 1994). In addition, the particle-size distribution of the modern severe dust storms, which primarily occur in the winter half year, and which is probably representative of the present interglacial (Holocene), is quite similar to that of the last interglacial palaeosol S1 (Dai et al., 1995). This implies that the particle-size distribution of S1 may have not been altered by post-depositional pedogenic processes and that particle size can be used as a reasonably sensitive indicator of the winter monsoon intensity
4. Measurement methods The last interglacial palaeosol S1 from the three sections was sampled at 2 cm intervals (about 100}200 ys) and the low- and high-frequency magnetic susceptibility of the 1759 samples was measured three times using a Bartington Instruments magnetic susceptibility meter and dual frequency sensor. The averaged results were
1132
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135
used to calculate the percent frequency-dependent susceptibility (sfd%), as de"ned by Thompson and Old"eld (1986). The CaCO content was measured using a calcimeter (Bascomb, 1961), at 2 cm intervals for the Tuxiangdao and Caoxian sections, and at 10 cm intervals for the Yuanbo section. For particle size analysis, carbonate in the samples was removed by repeatedly adding 15% HCl and washing the excessive HCl using a centrifuge. The organic matter in the CaCO -free samples was then removed by adding H O and heating for 4 h at 803C. Finally, 2 ml of Cal gon was added to the sample to disperse the particles (Janitzky, 1987). The particle size distribution was measured at 4 cm intervals for the TXD section, at 10 cm intervals for the YB section and at 8 cm intervals for the CX section, using a Coulter Electronics Ltd LS130 laser di!raction particle size analyser. To obtain higher stratigraphic resolution for S1S3, corresponding to MISS 5e, the particle size of this interval was measured at "ner (2 cm) intervals at the TXD section.
5. Results and discussion 5.1. Summer monsoon variations and events The thick (6}8 m) S1 pedocomplex at the northwestern margin of the western Loess Plateau provides the best context for preserving the climatic signature of the last interglacial within the Chinese Loess Plateau. As discussed above, the percent frequency-dependent magnetic susceptibility (sfd%), together with soil morphological features, is here used as an indicator of the summer monsoon intensity. Fig. 3 shows that the three palaeosols of the complex have relatively high sfd% values. The highest sfd% peak in S1S3 supports the morphological inference that palaeosol S1S3 is the best developed amongst the three palaeosols within S1. The sfd% records suggest that S1S2 experienced a slightly greater degree of pedogenesis than S1S1. In the CX section, the sfd% record indicates that the overall relatively strong pedogenesis within S1S2 was interrupted by an episode of weaker pedogenesis represented by a less-weathered unit (at ca. 40 m depth). In this case it is possible that variations in pedogenic intensity (as re#ected by changes in carbonate content) where preserved only in the drier conditions at this site where post-depositional alteration was minimal. The low sfd% values in loess layer S1L1 may result from the fact that this layer was minimally weathered. In contrast, the slightly higher sfd% values in loess layer S1L2 concur with observable weathering imprints. In S1L2 at the YB section, a small peak of sfd% indicates a weakly-developed pedogenic unit (at ca. 38 m). This is in line with the slightly higher modern mean annual precipitation at this site.
Whereas the CaCO concentration may not be a sensi tive indicator of summer precipitation and/or temperature, it may provide additional information about the summer monsoonal history. At the moister site (YB section), soil CaCO apparently was heavily depleted within S1S1 and S1S3. However, CaCO content fails to distin guish S1S2 from the two loess layers S1L1 and S1L2 at this site. At the two drier sites (TXD and CX), the CaCO variations seem to predate the sfd% variations, i.e. the CaCO peaks appear stratigraphically below the sfd% peaks. This consistent stratigraphic lag between sfd% and CaCO content is logical because the carbonate is always leached downward and re-deposited in the lower portion of contemporary soils and even beyond under semiarid climates (Aandahl, 1982; Reheis, 1987). The summer monsoon intensity, as indicated by the CaCO content, seems to show a generally decreasing trend from S1S3 (MISS 5e) to S1S1 (MISS 5a), which is in agreement with the conclusion drawn from the Xfd% data (Fig. 4). Thus, the interval corresponding to the formation of S1S3 experienced the strongest summer monsoon intensity, and that for S1S1 the weakest intensity, among the three palaeosol units. The second warm/moist period (S1S2) was interrupted by a dusty (colder and drier) interval. The "rst dust depositional period (S1L2) was more favorable for weathering than the second dust depositional period (S1L1), and a soil-forming (warmer and moister) phase was recorded in the "rst loess-depositing period (S1L2). 5.2. Winter monsoon variations and events In semi-arid climates the particle-size distribution of the S1 palaeosol may not have been altered by postdepositional pedogenic processes and therefore can be interpreted as a reasonably sensitive indicator of the winter monsoon intensity. The variations in median particle size of S1 (MD in Fig. 3) suggest that the winter monsoon intensity, in terms of the maximum strength of the winds, changed abruptly during the L2/S1 (MIS 6/5) and S1/L1 (OIS 5/4) transitional periods. The soil-forming periods (S1S1, S1S2, and S1S3) and dust-depositional periods (S1L1 and S1L2) are also readily distinguished by the median particle size. Among the three palaeosols, S1S3 has the "nest median particle size and S1S2 the coarsest, indicating that winter monsoon intensity was weakest in OIS 5e (S1S3) and strongest in OIS 5d (S1S2). The median particle size also indicates that the winter monsoon was stronger during OIS 5d (S1L2) than during the OIS 5b (S1L1). It should be mentioned that the weakly developed pedogenic unit within S1L2 at the YB section and the less-weathered unit within the S1S2 at CX section are not clearly distinguishable in the medianparticle-size data. By comparing the variations of the summer monsoon intensity, as indicated by sfd%, with the variations of the
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135
winter monsoon, as indicated by the median particle size, several conclusions can be drawn. First, the winter monsoon during the two strong winter monsoon periods (S1L1 and S1L2) of the last interglacial (S1) was considerably weaker than that during the two glacials represented by loess units L1 and L2. In contrast, the summer monsoon during the two strong winter monsoon periods (S1L1 and S1L2) was also as weak as that during the two glacials. Second, all of the three sections show that the strength of the winter monsoon was gradually weakening while the strength of the summer monsoon increased relatively abruptly during the L2/S1 (MIS 6/5) transitional period. During the S1/L1 (MIS 5/4) transitional period, the summer monsoon may have weakened quickly and the strengthening of the winter monsoon was considerably delayed at the YB and CX sections. However, the TXD section (the northwesternmost site) shows that the dominance of the winter monsoon occurred abruptly during the S1/L1 transitional period, implying that the winter monsoon became dominant at the northwestern limit of the area in#uenced by the summer monsoon as soon as the climate entered the glacial (L1) mode. Third, this study also shows that the negative correlation between the intensities of the summer and the winter monsoons has not always been linear. For example, the S1S1 palaeosol experienced the weakest summer monsoon and the S1S2 the strongest winter monsoon among the three palaeosols within S1. Both the winter and summer monsoons were stronger in S1L2 than in S1L1. 5.3. Global comparisons Fig. 4 compares northern hemisphere insolation (653N) with time series of the sfd% records for the three sections. During MIS 5, the variations in the summer monsoon intensity followed the variations in the northern summer solar insolation. For the three warmer periods, both the summer monsoon intensity and the solar insolation were greatest in OISS 5e, and lowest in OISS 5a. However, the two minima of the summer monsoon intensity corresponding to OISS 5b and 5d do not accurately re#ect the northern hemisphere solar insolation; that is, the summer monsoon intensity was higher in 5d than in the 5b, while the insolation was lower in 5d than in 5b (Fig. 4). We speculate that the maximum energy storage in low-latitudinal oceans during 5e bu!ered the minimum insolation conditions in the interior during 5d, and it was for this reason that the summer monsoon intensity was unexpectedly higher during 5d than that during 5b. We also observe that the 5c maximum of the summer monsoon intensity extended into the 5b minimum of insolation, and consequently the 5b summer monsoon lagged behind the corresponding minimum of the insolation by more than 6000 yr. A possible explanation is that during the two high-insolation periods (5e
1133
and 5c), low-latitudinal oceans stored enough energy and acted as a bu!er to delay the weakening of the summer monsoon in response to decreasing insolation. Finally, the weakest summer monsoon in 5b might be attributable to both the 5b insolation minimum and to the trend of generally decreasing insolation for the whole of MIS 5. The comparison between the summer monsoon and the insolation also shows that the summer monsoon dominance of the last interglacial was established when the insolation reached its peak in 5e and that the dominance was ended when the insolation was approaching its lowest in 5a. Our results are also of interest in the context of speculations about the occurrence (or otherwise) of rapid climatic #uctuations in Greenland, based on ice core data (GRIP Members, 1993; and cf Mayewski and Bender, 1995). Based on modern analog (0.7 per mille per 3C: Broecker, 1995), the temperature at Summit, Greenland, oscillated at a magnitude of 6}73C during the early part (OISS 5e and 5d) and in a range of 2}43C during the later part (OISS 5c, 5b, and 5a) of OIS 5 on decadal to millennial time scales (Fig. 5) (GRIP ice core: GRIP Members, 1993). However, the reconstructed winter monsoon intensity from the TXD section which has the highest time resolution (samples at about 150-year intervals) does not show such dramatic oscillations. The contrast of the winter monsoon intensity between the glacials (L1 and L2) and during OIS 5 (S1) is striking. Even the contrast between warmer periods (5e, 5c, and 5a) and the colder periods (5d and 5b) are unambiguously distinguishable. However, the oscillations superimposed on the "rst-order variations (glacial/interglacial) and the second-order variations (stadial/interstadial) are not clearly identi"able. In contrast, the variations in the reconstructed winter monsoon intensity are quite similar to the variations in the global ice volume as recorded by the dO record from Paci"c Ocean core V19-30 (Fig. 5). It has been proposed that the variation in the winter monsoon of the Chinese Loess Plateau my be teleconnected with variations in the climate of the North Atlantic region (Porter and An, 1995; Ding et al., 1995). It seems logical to suggest that the variations in the climate of the North Atlantic region high latitudes were dynamically related to the changes in the global ice volume. If so, the teleconnection between "rst-order winter monsoon variations in China and the climate in high latitudes might have been driven by the global ice volume.
6. Conclusions This study shows that the northwestern margin of the Chinese Loess Plateau experienced the strongest summer monsoon during OISS 5e and the weakest during OISS 5a, among the three warmer periods during the OIS 5. The second warmer period (5c) was interrupted by
1134
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135
a dusty (colder and drier) interval. The summer monsoon was weaker in OISS 5b than in OISS 5d during which a soil-forming (warmer and moister) event occurred. In contrast to previous suggestions that the intensities of the summer and winter monsoons always changed in opposite directions (An et al., 1991; Liu, et al., 1995), we found that the directions might not have been always proportionately opposite. For example, we infer that the weakest summer monsoon occurred in OISS 5a and the strongest winter monsoon in OISS 5c. Also, both the winter and summer monsoons were stronger in OISS 5d than in OISS 5b. We also conclude that the winter monsoon intensity in China during the last interglacial was probably driven by global ice volume #uctuations, and that the summer monsoon intensity was primarily controlled by northern solar insolation and probably modi"ed by a feedback mechanism. We suggest that the feedback mechanism was the bu!ering e!ect caused by heat storage of lowlatitude oceans, which had the e!ect of reducing the sensitivity of the summer monsoon intensity to solar insolation. As a result, the summer monsoon intensity and the solar insolation were not completely in phase during the last interglacial. Finally, the three sections investigated did not record any climate instability comparable to that recorded in the GRIP ice core during the last interglacial (OISS 5e), even though the study area is situated in the northwestern margin of the Chinese Loess Plateau area in#uenced by the summer monsoon, and which has been shown to be very sensitive to climate change (Chen and Zhang, 1993; Li et al., 1988; Yie et al., 1990). Acknowledgments We want to express our thanks to Mr. Yutian Zhang and Jixiu Cao for "eld and laboratory help; to Dr. Max Zhao for helping to improve the English; and to Steve Porter and an anonymous referee for helpful comments. TL dates were measured at the Xi'an State Key Laboratory for Loess and Quaternary Geology. This research was supported by NSFC grant (No. 49571065) and by the Royal Society K. C. Wong Fellowship. References Aandahl, A. R. (1982). Soils of the Great Plains. Lincoln, Nebraska: University of Nebraska Press. An, Z. S., Kukla, G. J., Porter, S. C., & Xiao, J. L. (1991). Magnetic susceptibility evidence of monsoon variation on the loess Plateau of central China during the last 130,000 years. Quaternary Research, 36, 29}36. Bascomb, C. L. (1961). A Calcimeter for routine use on soil samples. Chemistry and Industry, 45, 1826}1827. Broecker, W. S. (1995). ¹he Glacial =orld according to =ally. LamontDoherty Earth Observatory of Columbia University, Eldigio Press.
Chen, F. H., & Zhang, W. X. (1993). ¹he loess stratigraphy and Quater nary glacial problems in the areas of Gansu province and Qinghai province. Beijing: Science Press, (in Chinese). Chen, F. H., Bloemendal, J., Wang, J. M., Li, J. J., Old"eld, F., & Ma, H. Z. (1997). High-resolution multi-proxy climate records from Chinese loess: evidence for rapid climatic changes over the last 75 kyr. Palaeogeography, Palaeoecology, Palaeoclimatology, 130, 323}335. Dai, X. R., Shi, Y. X., & Xue, B. (1995). Granulometric characteristics and signi"cance of the deposits from a recent extraordinary heavy dust storm in Lanzhou, Gansu, North China. Journal of ¸anzhou ;niversity (Natural Sciences), 31, 168}174 (in Chinese) Derbyshire, E., Keen, D., Kemp, R. A., Rolph, T. C., Shaw, J., & Meng, X. M., (1995). Loess-palaeosol sequences as recorders of palaeoclimatic variations during the last glacial-interglacial cycle: some problems of correlation in north-central China. In E. Derbyshire (ed.), =ind blown sediments in the quaternary. London: Wiley. (pp. 7}18). Ding, Z. L., Liu, T. S., Rutter, N. W., Yu, Z. W., Guo, Z. T., & Zhu, R. X. (1995). Ice-volume forcing of east-Asian winter monsoon variations in the past 800,000 years. Quaternary Research, 44, 149}159. Feng, Z.-D., Johnson, W. C., Sprowl, D. R., Lu, Y.-C., & Ward, P. A. (1994). Climatic signals from loess-soil sequences in the central Great Plains, USA. Palaeogeography, Palaeoecology, Palaeoclimatology, 110, 345}358. Feng, Z.-D. (1996). Climatic implications of magnetic susceptibility and Be-10 #ux in Chinese loess-soil sequences. Catena, 27, 143}147. Feng, Z.-D., & Johnson, W. C. (1995). Factors a!ecting the magnetic susceptibility of a loess-soil sequence, Barton County, Kansas, USA. Catena, 24, 25}37. Gile, J. H., Peterson, F. F., & Grossman, R. B. (1965). The K horizon: a master soil horizon of carbonate accumulation. Soil Science, 99, 74}82. Gile, J. H., Peterson, F. F., & Grossman, R. B. (1966). Morphological and genetic sequences of carbonate accumulation in desert soils. Soil Science, 101, 347}360. Guo, Z., Liu, T., Guiot, J., Wu, N., Lu, H., Han, J., Liu, J., & Gu, Z. (1996). High frequency pulses of east Asian monsoon climate in the last two glacials: link with the North Atlantic. Climate Dynamics, 12, 701}709. GRIP Members (1993). Climate instability during the last interglacial period recorded in the GRIP ice core. Nature, 364, 203}207. Janitzky, P. (1987). Particle size analysis. In M. Singer, & P. Janitzky (Ed.), Field and laboratory procedures used in soil chronosequence study. ;.S. Geological Survey Bulletin 1648, (pp. 11}16). Hunt, C. P., Banerjee, S. K., Han., J., Solheid, P. A., Oches, E., Sun, W., & Liu, T. (1995). Rock magnetic proxies of climatic change in the loess-palaeosol sequences of the western loess plateau of China. Geophysical Journal International, 123, 232}244. Kemp, R. A., Derbyshire, E., Chen, F. H., & Ma, H. Z. (1996). Padosedimentary development and paleoenvironmental signi"cance of the S1 palaeosol on the northeastern margin of the Qinghai-Xizang (Tibetan) Plateau. Journal of Quaternary Science, 11, 95}106. Kukla, G., Heller, F., Liu, X. M., Chun, X. T., & An, Z. S. (1988). Pleistocene climates in China dated by magnetic susceptibility. Geology, 16, 811}814. Li, J. J., Feng, Z. D., & Tang, L. Y. (1988). Late Quaternary monsoon pattern on the Loess Plateau of China. Earth Surface Processes and ¸andforms, 13, 125}135. Liu, X. M., Liu, T. S., Heller, F., & Xu, T. C. (1990). Frequencydependent susceptibility of loess and Quaternary palaeosol. Quaternary Science, (1), 42}50. (in Chinese) Liu, T. S., Guo, Z. T., Liu, J. Q., Ding, Z. L., Gu, Z. Y., & Wu, N. Q. (1995). variations of eastern Asian monsoon over the last 140,000 Years. Bulletin of Society Geol. France, 166, 221}229.
F.H. Chen et al. / Quaternary Science Reviews 18 (1999) 1127}1135 Machette, M. N. (1985). Calcic soils of the southwest U. S. In D. L. Weide (Ed.), Soils and quaternary of the southwest ;. S. Geological Society of America Special Paper 203, (pp. 1}20). Maher, B. A., & Thompson, R. (1992). Paleoclimatic signi"cance of the mineral magnetic record of the Chinese loess and palaeosols. Quaternary Research, 37, 155}170. Maher, B. A., & Thompson, R. (1994). Pedogenesis and paleoclimate } interpretation of the magnetic susceptibility record of Chinese loess-palaeosol sequences * comment. Geology, 22, 857}858. Martinson, D. G., Pisias, N. G., & Hays, J. D. (1987). Age dating and the orbital theory of the ice age: development of a high-resolution 0 to 30,000 years chronostratigraphy. Quaternary Research, 27, 1}29. Mayewski, P. A., & Bender, M. (1995). The GISP2 ice core recordpaleoclimate highlights. Reviews of Geophysics Supplement, p.1287}1296. US National Report to the IUGG 1991}1994. Nowacozyk, N. R., Frederichs, T. W., Eisenhauer, A., & Gard, G. (1994). Magnetostratigraphic data from Late Quaternary sediments from the Yermak Plateau, Arctic Ocean: evidence for four geomagnetic polarity events within the last 170Ka of the Brunhes Chron. Geophysical Journal International, 117, 453}471. Porter, S. C., & An, Z. S. (1995). Correlation between climate events in the North-Atlantic and China during the last glaciation. Nature, 375, 305}308.
1135
Reheis, M. C. (1987). Climatic implication of alternating clay and carbonate formation in semiarid soils of south-central Montana. Quaternary Research, 27, 270}282. Sun, D. H., Zhou, J., Jiang, F. C., & Porter, S. C. (1996). Preliminarystudy on the summer monsoon climate of Chinese loess plateau during the last interglacial period. Chinese Science Bulletin, 41, 321}325. Thompson, R., & Old"eld, F. (1986). Environmental magnetism. London: Allen and Unwin. Yie, D. Z. (1990). Study on global changes of China in advance. Beijing: Meteorological Press Beijing. (in Chinese) Yue, L. P., Lei, X. Y., & Qu, H. J. (1991). Magnetostratigraphy study on the Jingyuan loess section, Gansu, China. Quaternary Science, 4, 349}353. (in Chinese). Zheng, H. B., Rolph, T., Shaw, J., & An, Z. S. (1995). A detailed palaeomagnetic record for the last interglacial period. Earth and Planetary Science ¸etters, 133, 339}351. Zhou, L. P., Old"eld, F., Wintle, A. G., Robinson, S. G., & Wang, J. T. (1990). Partly pedogenic origion of magnetic variations in Chinese Loess. Nature, 346, 737}739. Zhu, R. X., Zhou, L. P., Laj, C., Mazaud, A., & Ding, Z. L. (1994). The Blake geomagnetic polarity episode recorded in Chinese loess. Geophysical Research ¸etters, 21, 697}700.