Ultra-High Rates of Loess Sedimentation at Zhengzhou Since Stage 7

中国科技论文在线 http://www.paper.edu.cn

Geomorphology 85 (2007) 131–142 www.elsevier.com/locate/geomorph

Ultra-high rates of loess sedimentation at Zhengzhou since Stage 7: Implication for the Yellow River erosion of the Sanmen Gorge ⁎ Hongbo Zheng a, , Xiangtong Huang a, Junliang Ji a, Rui Liu a, Qingyou Zeng b, Fuchu Jiang c

a State Key Laboratory of Deep-sea Studies, Tongji University, Shanghai, 200092, China b School of Civil Engineering, Tongji University, Shanghai, 200092, China c Institute of Geomechanics, Chinese Academy of Geological Sciences, Beijing, 100081, China Received 19 October 2004; received in revised form 27 March 2005; accepted 29 March 2006 Available online 1 September 2006

Abstract

The Mangshan loess plateau is located 25 km to the west of Zhengzhou on the south bank of the Yellow River. Here the river flows out through the Sanmen Gorge releasing most of its suspended load following a dramatic decrease in gradient. The stratigraphy of the Mangshan loess deposits, consisting of a number of loess and palaeosol sequences, was established following magnetostratigraphic studies and measurements of magnetic susceptibility and grain size distribution. The Bruhnes/Matuyama boundary was found at the depth of about 130 m, indicating that this sequence at Mangshan resembles what is observed elsewhere in the Loess Plateau. The upper part of the Mangshan loess displays extremely high sedimentation rates (∼50 m3 per 1000 years), lower susceptibility values and coarser grain-size distribution, unlike the lower part of the profile and other sections in the Loess Plateau. This striking change indicates that the upper Mangshan loess had a different sediment source, different from the deserts that act as a common source for most of the loess deposits in central China. This sediment source is believed to be the proximal Yellow River floodplain, and the ancient alluvial fan lying at the eastern end of the Sanmen Gorge. The age estimation of the formation of the alluvial fan, based on Mangshan loess, suggests that the Yellow River may have eroded the Sanmen Gorge at approximately MIS 7. © 2006 Elsevier B.V. All rights reserved.

Keywords: Mangshan loess; Sanmen Gorge; Yellow River

1. Introduction processes that shaped the topographic features of present China. These tectonic–climatic processes involved the The timing of the establishment of the eastward growth of the Tibetan Plateau towards west and the drainage systems in China in their present form has been a opening-up of the West Pacific marginal seas towards subject of great interest and vigorous debate (Wang, east, the process starting from early Cenozoic when India 1997). Apparently, the drainage systems could only have collided with Asia (Powell and Conaghan, 1973). developed during, and as part of, the tectonic–climatic Accompanied with the topographic evolution in China, the climatic conditions were also reorganized, evolving from a zonal planetary regime to the present-day ⁎ Corresponding author. monsoonal pattern that prevails over the East Asia and E-mail address: [email protected] (H. Zheng). the marginal seas (Zheng et al., 2004).

转载中国科技论文在线 http://www.paper.edu.cn

132 H. Zheng et al. / Geomorphology 85 (2007) 131–142

As most large rivers in China originate in the Tibetan to be fairly recent (Wang, 1998). Relatively little is known, Plateau and discharge into the west Pacific marginal seas, however, about when and how these rivers came into they act as critical linkages between the two. They have existence. It was assumed that river captures and headward also played significant roles in modifying the oceano- erosion were two major processes that increased the length graphic conditions of the marginal seas, thus exerting and drainage areas of the Yangtze and Yellow Rivers important effects on climatic changes at both regional and following the steepening of the gradient between the west global scales. The study of the evolutionary history of large and the east. Isolated studies at different sections of the rivers in China (and East Asia in general) forms the core of rivers have produced variable age estimations, but often “source to sink” studies, and has attracted a wide scientific with large uncertainties and controversies (Ren et al., community including geomorphologists, palaeoclimatol- 1959; He et al., 1989; Li, 1995; Wang et al., 2001). ogists, and palaeoceanographers (Wang, 1997, 1998). The Yellow River near Lanzhou has developed Compared with many other large rivers in the world, 8 terraces, the best exposures being located in the gullies the main courses of the Yangtze and Yellow Rivers appear to the north of Lanzhou City (Fig. 1). Almost all the

Fig. 1. Location map showing the Loess Plateau (LP), the Mu Us Desert (MS) and the North China Plain (NCP). The Mangshan loess section and other loess sections referred in the text are indicated. 中国科技论文在线 http://www.paper.edu.cn

H. Zheng et al. / Geomorphology 85 (2007) 131–142 133 terraces are capped with loess deposits, allowing precise slope of a plateau (locally known as Mangshan loess age estimations. Chronological studies of the terraces by plateau), where the Yellow River has cut through the magnetostratigraphy, optical thermoluminescence dating loess forming a steep cliff. The Mangshan loess plateau and 14C dating suggested that the Yellow River had been is about 18 km in length along the south bank of the river flowing through the region since the Early Pleistocene (Li, and 5 km in width, with its highest point reaching 262 m. 1995). A similar age has also been assigned to the course The Mangshan plateau marks the boundary between the in the upper reaches of the river near Xining. However, no middle and lower reaches of the Yellow River. Down- evidence of an outflowing river system has been found in stream, the river enters the North China Plain with the middle reaches near Yinchuan, where a serious of gentler gradient and an increased number of meanders graben basins along the present river course developed before discharging into the Bohai Sea (Fig. 1). during most of the Plio-Pleistocene (Zhang, 1989). The 171 m thick Mangshan loess represents a num- In this paper, we present results of a stratigraphy ber of well-exposed profiles in the Mangshan Plateau. study of loess deposits along the south bank of the Field observations revealed that Mangshan section Yellow River near Zhengzhou. When combined with contains 11 loess layers (L1–L11) interbedded with 12 chronological estimations of fluvial terraces in the area, palaeosols (S0–S11). The loess–palaeosol sequence re- the present study suggests the date for the cutting of the sembles what is commonly seen elsewhere on the Loess Sanmen Gorge by the Yellow River. Plateau, except that the top two loess–palaeosoil pairs are extremely thick as described below. 2. Mangshan loess deposits The topmost palaeosol commonly known as the Ho- locene S0 palaeosol, is a silty black loam, as seen also in Loess deposits are widespread in central north China other sections on the Loess Plateau (Liu, 1985; Kukla, and are best developed in the so-called Loess Plateau 1987; Zheng et al., 1992)(Fig. 2). The S0 soil has been (Fig. 1), where Plio-Pleistocene loess–palaeosol sequences heavily cultivated, and eroded in many places. The with an average thickness of 150 m can be observed (Liu, thickness of the S0 soil at the section is 1.7 m. The first 1985). These wind-blown silts, when combined with loess layer L1 is 40.8 m in thickness. L1 is of last glacial measurements of magnetic susceptibility, grain-size and age as commonly observed on the Loess Plateau (Liu, other proxies, provide a relatively continuous terrestrial 1985) and composed of two typical loess layers with a record of past global climate changes for the Pleistocene weakly developed soil complex in between (MISe 3 soil). period. Away from the central Loess Plateau, the loess– S1 of the last interglacial age (Liu, 1985) is a soil complex palaeosol sequences generally become less complete, and with a thickness of 18.5 m. It is composed of three soil in most cases thinner, depending on their geographic and layers interbedded with two loess deposits, a feature that is geomorphic locations. In general, loess accumulation rate commonly observed in the northern part of the Loess decreases southeastward across the Loess Plateau, a strong Plateau owing to a relatively high sedimentation rate piece of evidence that indicates the deserts in the north and (Zheng et al., 1995). In the southern part of the Loess northwest as the source regions of loess deposits in central Plateau, such as at Lantian (Zheng et al., 1992), this China (Liu, 1985). Because loess–palaeosol sequences are feature is not visible in the field, but can be identified by laterally correlative and vertically continuous and can be measuring magnetic susceptibility or grain-size. L2 loess correlated with marine isotope records, they are often used is 26 m thick and is a coarse silt layer. S2 is also a soil in establishing litho- and chrono-stratigraphic correlations complex and is composed of three soils layers interbedded (Kukla, 1987). This practice is particularly useful when with two thin loess layers. From L3 down to the bottom of dating river terraces in loess-covered areas. The Yellow the section (S11), the stratigraphy is very similar to that of River terraces near Lanzhou can be taken as an example Luochuan (Liu, 1985) and Lantian (Zheng et al., 1992). (Fig. 1). The Yellow River terraces north of Lanzhou range A comparison between Mangshan and other sections in age from early Pleistocene to Holocene, each being has been constructed (Fig. 2) in order to establish the capped with loess deposits of varying thickness (Li, 1995). loess–palaeosol stratigraphy. All sections indicate the By dating the loess sequences, through soil stratigraphy, general trend of decreasing sedimentation rates from the magnetic polarity or numeric dating (for example, TL), it north and northwest to the south and southeast during and has been possible to assign an upper time limit to terrace before the L3 age. Sections from S0 to S2 also conform to formation, thus determining the tectonic and river history the same trend for most of the profiles except for Mang- of the region. shan, where the layers above S2 are extremely thick. Mangshan loess section is located 25 km to the west Previous studies have assigned the soil horizon be- of Zhengzhou (Fig. 1). The profile is exposed at the north tween 87 m and 97 m as S1, and subsequently the section 中国科技论文在线 http://www.paper.edu.cn

134 H. Zheng et al. / Geomorphology 85 (2007) 131–142

Fig. 2. Comparison between Mangshan and other loess sections in the Loess Plateau. Xifeng and Luochuan are located in the middle of the Loess Plateau, while Lantian is located in the south and Mangshan in the southeast.

between 1.7 m and 87 m as L1, and the soil between The upper part of the section at Mangshan (0–97 m) was 42.5 m and 61 m as L1SS1 (Marine Isotope Stage 3) sampled at intervals of 2 cm, while only two samples (Jiang et al., 1998). However, our field observations were collected from each loess and palaeosol unit in the suggest that the soil layer between 42.5 m and 61 m is S1, middle part (L3–S7, 97 m to 146 m). Sampling interval instead of L1SS1, and the soil between 87 m and 97 m is was 20 cm for the lower section (146 m downwards). In S2. To prove this, we reconstruct the stratigraphy for the the field the section was first cleared of weathered sur- whole Mangshan loess–palaeosol sequence by deter- face material, and oriented hand samples were then cut mining the geomagnetic polarity, and by measuring the from the fresh face. Cubic specimens with volume of magnetic susceptibility and grain size distributions. 8cm3 were then prepared in the laboratory. The natural remnant magnetization (NRM) of all specimens was 3. Magnetostratigraphy measured prior to systematic thermal demagnetization. Samples from the upper section were measured in the Sampling strategy for magnetostratigraphic study Palaeomagnetism Laboratory, Institute of Geology and varies through the section, depending on specific needs. Geophysics, Chinese Academy of Sciences. All the rest 中国科技论文在线 http://www.paper.edu.cn

H. Zheng et al. / Geomorphology 85 (2007) 131–142 135 were measured in the Geomagnetism Laboratory, Uni- tions of this geographic location. Therefore, zone III is versity of Liverpool, UK. not a full geomagnetic event but a geomagnetic excur- Close-interval sampling from the upper part of the sion. This geomagnetic excursion has widely been section was used to identify possible geomagnetic events reported in the loess deposits in China. In the section at or excursions that have been reported to have existed Lantian it was named BMPC, meaning a pre-cursor during this time period, i.e. the Blake event around before the B/M reversal (Zheng et al., 1992). 100 ka found in lake sediments (Denham, 1976)andin Zone IV is mostly reversed, except two normal sam- the Chinese loess deposits (Zhu et al., 1994). About 900 ples at 167 m and 169 m, respectively. In general, zone samples spanning the whole S1 (42.5 m to 61 m) were IV belongs to Matuyama reversed polarity, and the measured, but no significant geomagnetic anomaly was normal samples may indicate that the bottom of the detected (Zeng et al., 2001). Zheng and others have section is getting closer to Jaramillo event. In many previously carried out a similar study on the Huanxian other sections in the Loess Plateau, the Jaramillo event loess section in the northern Loess Plateau, and found no lies between S10 and S12, with slight variations from evidence of a major geomagnetic event recorded in S1 one site to another (Heller and Liu, 1982; Liu, 1985; (Zheng et al., 1995). Their results suggested that the Zheng et al., 1992). relative intensity of the geomagnetic field decreased In summary, palaeomagnetic measurements have greatly during Blake, a feature which has also been identified the B/M boundary, which is located in the observed elsewhere in the world. The relative palaeo- lower part of L8. More importantly, magnetostrati- intensity of geomagnetic field during S1 as suggested by graphic study has confirmed the field observation of the the present study shows a similar trend as Huanxian sequence, suggesting that the soil horizon between 87 m section (Zheng et al., 1995). and 97 m is MIS 7 soil (S2). All samples from the middle part of the section (L3–S7) are normally magnetized, with declinations 4. Magnetic susceptibility and grain-size around due north and inclinations about 50°–60°. Therefore, the section between 97 m to 146 m was Magnetic susceptibility and grain-size distribution assigned to a period of the Brunhes normal polarity. are the two most commonly-used proxies in the study of The lower part of the section (from 146 m down- East Asian monsoon history of the Chinese loess depo- wards) is composed of loess layers L8 to L11, and sits (An et al., 1990). In addition, as the down-section palaeosols S8 to S11, and the polarity stratigraphy has variations of these two parameters exhibit strong simi- been divided into 4 zones (zones I to IV) according to larity between sections even hundreds of kilometers their characteristics in terms of field directions or inten- apart, they are also a very useful tool in stratigraphic sities (Fig. 3). The topmost 9 samples (146 m to 148 m) correlation. This is particularly noticeable in the north- in Zone I are assigned to the Brunhes normal polarity. ern and western part of the Loess Plateau, where pedo- Most samples exhibit normal geomagnetic directions, genesis is relatively weak and the field identification of except one that has reversed declination and inclination. soil is frequently very difficult. In this case, the estab- This is a common phenomenon during a field reversal, lishment of stratigraphy of a loess sequence relies in which declination and/or inclination often experience heavily on measuring magnetic susceptibility and grain- a number of normal-reversed cycles. The inclinations of size distribution of the section (Zheng et al., 1995). samples decrease gradually towards the boundary of The sampling strategy for magnetic susceptibility zone I/II, and become fully reversed in zone II, which is and grain-size studies of Mangshan section varies. The also a common character of geomagnetic field before or top 97 m (from S2 upwards) was continuously sampled after a reversal. Therefore, the B/M boundary is as- with each sample spanning a thickness of 2 cm, while a signed to be at 148.1 m, in the lower part of L8. sampling interval of 10 cm remained constant for the L9 is a thick silty loess layer which is regarded as a lower part of the profile. Samples were first dried and marker horizon all over the Loess Plateau (Liu, 1985). In weighed, and then measured by using Bartington MS2 Mangshan section, L9 is 6.5 m thick and has a very magnetic susceptibility meter (both low and high fre- weakly developed soil as suggested by magnetic suscep- quencies). For grain-size measurements, pre-treatment tibility measurement in the lower part, a feature which is of samples followed the method commonly adapted by also commonly observed elsewhere (Zheng et al., 1992). others (Lu and An, 1998), and a laser grain-size analyzer Polarity zone III is characterized by many normal sam- with measuring range of 0.04–2000 μm was used. ples, which may or may not be fully normalized, and in Fig. 4 shows the low frequency magnetic suscepti- many cases do not show typical declinations and inclina- bility (LF) of the Mangshan section while comparing 中国科技论文在线 http://www.paper.edu.cn

136 H. Zheng et al. / Geomorphology 85 (2007) 131–142

Fig. 3. Magnetostratigraphy of the lower part of Mangshan section. Note that B/M is located in loess layer L8, while BMPC is located in L9. Four polarity zones are marked.

with Luochuan (Kukla, 1987). It can be seen that from Fig. 5 shows the low frequency magnetic susceptibil- S3 to S11, the susceptibility curves of the two sections ity and median size (MD) of the Mangshan section from show remarkable similarity. The susceptibility of the 97 m upwards (S2 and above), and their comparison with section above L3 also shows similar fluctuations, except the MD of the same sequence of Lijiayuan section in the that Mangshan section exhibits greater detail. northern Loess Plateau (Ding et al., 2002)(forlocations 中国科技论文在线 http://www.paper.edu.cn

H. Zheng et al. / Geomorphology 85 (2007) 131–142 137

Fig. 4. Comparison of magnetic susceptibility of Mangshan section with that of Luochuan section. see Fig. 1). In common with other sequences from the et al., 2002). Not only do the major loess and palaeosol Loess Plateau, both magnetic susceptibility and grain size layers have similar grain-size distribution curves, but also fluctuate with loess–palaeosol variations: loess layers are the weakly developed palaeosols such as the two weak coarser and lower in magnetic susceptibility values, while peaks in L2 (Fig. 5). The comparison of grain-size palaeosols contain more clay fractions and are more distribution of Mangshan section with the well-estab- enriched with magnetic grains owing to pedogenesis. The lished Lijiayuan section also suggests that the section grain-size distribution of Mangshan section also exhibits from 97 m upwards in Mangshan belongs to S2 to S0 remarkable similarity with the Lijiayuan section (Ding units. Most importantly, the soil layer between 87 m and 中国科技论文在线 http://www.paper.edu.cn

138 H. Zheng et al. / Geomorphology 85 (2007) 131–142

Fig. 5. Comparison of median diameter of Mangshan section with that of Lijiayuan and Xinzhuangyuan sections.

97 m is S2, not S1 as previously suggested by Jiang et al. below S2, and increases two times above it (Fig. 6). The (1998). N100 μm size fraction displays the same trend, with an almost zero value below S2, but becoming a significant 5. Cause of the difference of the sections below and component above it, indicating an increase of coarse above S2 particles. In fact given its southerly location, Mangshan contains far too many coarse particles. Not only is the section from S2 upwards at Mangshan The average sedimentation rate of S0–S2 at Mangshan extremely thick, when compared with other sections in is calculated to be about 50 cm per a thousand years, much the Loess Plateau, but it may also have a different source higher than that of the section below S2. Compared with region as suggested by magnetic susceptibility and grain- other sections in the Loess Plateau, the sedimentation rate size analyses. The size fraction of N63 μm averages 5% (as judged from the thickness) of the lower section of 中国科技论文在线 http://www.paper.edu.cn

H. Zheng et al. / Geomorphology 85 (2007) 131–142 139

Fig. 6. Magnetic susceptibility and grain-size distribution of Mangshan section.

Mangshan is slightly lower than the average value, which plateau. In contrast, during interglacial periods when is what should be expected given the southerly locality of summer monsoon was dominant, climates were rela- the Mangshan section. However, the sedimentation rate of tively humid and warm, dust accumulation was reduced the upper section is much higher than the average value of and soils formed. In general, dust accumulation rates are other sections. In addition, the grain-size distribution is higher during loess-forming periods than during soil significantly coarser, and the magnetic susceptibility val- formation. Spatially speaking, sedimentation rates de- ues much lower. All of these point out that loess depo- crease southward as the depositional sites get farther sition at Mangshan since S2 has been under a different away from the source area (Liu, 1985). This is the case regime, which produced and transported coarser dust in for most loess sequences such as Luochuan (Liu, 1985), greater quantities to the site. Xifeng (Kukla, 1987) and Lantian (Zheng et al., 1992), Climatic conditions were generally cold and dry even for the lower part of the Mangshan sequence. We during glacial and sub-glacial periods when the East interpret that the increase in sedimentation rate and grain Asian winter monsoon was strong and summer monsoon size after S2 has been caused by addition of a new source. was relatively weak (An et al., 1990). As a result, more Judging from the grain size distribution, we suggest that dust would be produced by the northerly monsoon winds this new source (apart form the deserts) would be the from the deserts, which favoured loess formation on the proximal flood plain and alluvial fan of the Yellow River. 中国科技论文在线 http://www.paper.edu.cn

140 H. Zheng et al. / Geomorphology 85 (2007) 131–142

An earlier study has reported scattered relicts of ancient and palaeoclimatologists over almost a century (Yang, alluvial fans distributed widely in the region to the west 1936; Wang et al., 1999, 2001). Recent magnetostrati- of Mangshan, although their ages have remained undated graphic study of the Sanmen Group revealed that the (Shi et al., 1985). The alluvial fans have been interpreted lake started to infill with sediments in the early Pliocene, to be the result of the Yellow River flooding through the and disappeared in the Pleistocene (Wang et al., 2001). Sanmen Gorge. Our stratigraphic study of the Mangshan The drying up of the lake is believed to have been the section implies that the age of the formation of alluvial result of the Yellow River cutting through the Sanmen fans corresponds to S2. Gorge area, an area which consists of pre-Cenozoic rocks that used to be the lake margin (Wang et al., 2001). 6. Sanmen palaeo-lake and the Yellow River As the river ran out of the gorge, it dumped most of its channelization load to form alluvial fans on the east side of the gorge. The fans, together with flood plains acted as a proximal The Sanmen palaeo-lake is one of the many grabens source for the nearby loess deposits at Mangshan. which were developed along the course of the Yellow Judging from the ultra-high sedimentation rate of the River during the latest part of the Cenozoic (Zhang, section above S2 at Mangshan, the alluvial fan would 1989)(Fig. 7). These graben basins, with various incep- have been formed at the beginning of MIS 7 and the tion ages, were once internal-flowing basins, and pro- gorge cutting of the Yellow River would have occurred bably isolated from each other for most of the Pliocene at the same time. and also for the early part of Pleistocene (Zhang, 1989). After eroding the gorge, the Yellow River would cut The Sanmen Lake stretched from Baoji in the west to and remove the palaeo-Sanmen lake beds. The regional Sanmen Gorge in the east, covering most of the drainage base level was lowered, which may have led to terrace area of the present Weihe River and part of the Yellow forming. Near the town of Pinglu (Fig. 7) on the north River (Fig. 7). The stratigraphy of Sanmen Lake depo- bank of the Yellow River, two terraces (T1 and T2) can sits has been a subject of research for palaeontologists be observed. T1 is about 13 m high, and is constructed of

Fig. 7. Plio-Pleistocene graben basins along the course of the Yellow River (after Zhang, 1989). Sanmen palaeo-lake is marked as Sanmen PL, Sanmen Gorge as Sanmen G. 中国科技论文在线 http://www.paper.edu.cn

H. Zheng et al. / Geomorphology 85 (2007) 131–142 141

loess deposits with weakly developed soils. The bottom the loess–paleosol sequence in central China. Quaternary Inter- – of T1 is not exposed and precise age estimation will national 7/8, 91 95. Denham, C.R., 1976. Blake polarity episode in two cores from the depend on further dating. Our field observations suggest Greater Antilles outer ridge. Earth and Planetary Science Letters that T1 is most likely of last glacial age. T2 is typically 29, 422–434. dual-structured: fluvial gravel and sand in the lower part Ding, Z.L., Derbyshire, E., Yang, S.L., Yu, Z.W., Xiong, S.F., Liu, T.S., and loess deposits in the upper part. The loess deposits on 2002. Stacked 2.6-Ma grain size record from the Chinese loess δ18 T2 contain two layers of well developed soil. Field based on five sections and correlation with the deep-sea O record. Paleoceanography 17 (3), 1033–1053. observation indicated that the two soil layers are S2 and He, H., He, K., Zhu, X., Zhu, Z., 1989. A discussion on the problem of S1, thus suggesting that T2 was formed prior to S2. river capturing of the Jinshajiang River in northwest Yunnan. The Sanmen Group (palaeo-Sanmen lake beds), to Geoscience 3 (3), 319–330. which T2 is attached, is about 70 m thick, and is also Heller, F., Liu, T., 1982. Magnetostratigraphical dating of loess depo- – capped by loess deposits. The Sanmen Group consists of sits in China. Nature 300, 431 433. Jiang, F., Wu, X., Sun, D., Xiao, H., Wang, S., An, Z., Tian, G., Liu, interbedded grey, green and yellowish fine to medium K., Yin, W., Xue, B., 1998. On Mangshan loess stratigraphy in sands, silt, and clay layers. The loess deposits contain China central plains. Journal of Geomechanics 4 (4), 90–97 (in two well developed soil horizons, and they are most Chinese with English abstract). likely S2 and S1. Field observation of the loess strati- Kukla, G., 1987. Loess stratigraphy in central China. Quaternary – graphy suggested that the Sanmen Lake dried up before Science Reviews 6, 191 219. Li, J., 1995. Uplift of Qinghai–Xizang (Tibet) Plateau and Global S2 was formed, the cause being attributed to the Yellow Change. Lanzhou Univ. Press, Lanzhou. 207 pp. River eroding its gorge. Liu, T., 1985. Loess and the Environment. China Ocean Press, Beijing. 147 pp. 7. Conclusion Lu, H., An, Z., 1998. Pretreated methods on loess–palaeosol samples granulometry. Chinese Science Bulletin 43, 237–240. Ren, M., Bao, H., Han, T., 1959. Geomorphology of the Jinshajiang The evolutionary history of the Yellow River is of River and river-capturing. Acta Geographica Sinica 25 (20), great interest to a wide community of scholars in geo- 135–155. morphology, tectonics, sedimentology and palaeoclima- Powell, C., Conaghan, P.J., 1973. Plate tectonics and the Himalayas. tology. Available information seems to suggest that the Earth and Planetary Science Letters 20, 1–12. Yellow River has been rather “fragmental”, indicating Shi, C., Dong, Y., Han, S., 1985. Problems on the formation of the Huanghe (Yellow) River delta. Geological Review 31 (6), 539–547 that different sections of the river have their own evo- (in Chinese with English abstract). lutionary histories, depending on regional settings. The Wang, P.X., 1997. Late Cenozoic environmental evolution in China: present study only dealt with a particular section of the marine factors and records. In: Jablonski, N.G. (Ed.), The Changing river. Nevertheless, the channeling of the river through Face of East Asia During the Tertiary and Quaternary. The Univer- – the Sanmen Gorge is a significant event in the river's sity of Hong Kong Press, pp. 263 274. Wang, P.X., 1998. Deformation of Asia and global cooling: searching history in that it has become an out-flowing drainage links between climate and tectonics. Quaternary Sciences 3, 213–221 system ever since. The results presented in this paper are (in Chinese with English abstract). rather preliminary. Further geochemical study of the Wang, S., Jiang, F., Wu, X., Wang, S., Tian, G., Xue, B., Zheng, H., Mangshan loess is needed to document the source re- 1999. A study of the age of Sanmen Group in Sanmenxia area. – gions. Precise dating is also critical for age estimations of Journal of Geomechanics 5 (4), 105 113 (in Chinese with English abstract). the river terraces and the palaeo-fan deposits, in order to Wang, S., Wu, X., Zhang, K., Jiang, F., Xue, B., Tong, G., Tian, G., quantify the timing of the river cutting through the gorge. 2001. Sedimentary records of environmental evolution in the Sanmen Lake Basin and the Yellow River running through the Acknowledgements Sanmenxia Gorge eastward into the sea. Science in China 31 (9), 585–608. Yang, Z., 1936. The chronostratigraphy of the Sanmen Group. Geological The authors would like to thank NSFC and Ministry Review 1 (3), 323–330 (in Chinese with English abstract). of Education of China for financial support through Zeng, Q., Zheng, H., Zhu, R., Jiang, F., Qiang, X., 2001. The absence grants 40025107, 90211019, G2000078501 and Cheung of the Laschamp excursion in the Mangshan loess section and its Kong Program. Discussions with Prof. Pinxian Wang cause of formation. Marine Geology & Quaternary Geology 22 (1), – and Dr Zhongli Ding are greatly appreciated. 89 95 (in Chinese with English abstract). Zhang, K., 1989. The development of drainage systems in the middle reaches of the Yellow River. Quaternary Sciences in China 8 (1), References 185–193 (in Chinese with English abstract). Zheng, H., An, Z., Shaw, J., 1992. New contribution to Chinese Plio- An, Z.S., Liu, T.S., Lu, Y.C., Porter, S.C., Kukla, G., Wu, X.H., Hua, Pleistocene magnetostratigraphy. Physics of the Earth and Plane- Y.M., 1990. The long term paleomonsoon variation recorded by tary Interiors 70, 146–153. 中国科技论文在线 http://www.paper.edu.cn

142 H. Zheng et al. / Geomorphology 85 (2007) 131–142

Zheng, H., Rolph, T., Shaw, J., An, Z., 1995. A detailed palaeo- soon as viewed from land and sea. Global and Planetary Change magnetic record for the last interglacial period. Earth and Planetary 41, 147–155. Science Letters 133, 339–351. Zhu, R.X., Zhou, L.P., Laj, C., Mazaud, A., Ding, Z.L., 1994. The Zheng, H., Powell, C., Rea, D., Wang, J., Wang, P., 2004. Late Blake geomagnetic polarity episode recorded in Chinese loess. Miocene and mid-Pliocene enhancement of the East Asian Mon- Geophysical Research Letters 21, 679–700.