Are Chinese loess deposits essentially continuous?



[1] We conducted a paleomagnetic examination of the last glacial loess of three representative profiles along an east-west transect in the central Loess Plateau in order to assess the continuity of Chinese loess. The results show that the Xifeng and Luochuan profiles record the Laschamp geomagnetic excursion but with different morphologies. Together with the published results from Weinan, southern plateau, our results suggest that sedimentation of the last glacial loess in the central-southern plateau was continuous at the time scale equivalent to the duration of the Laschamp excursion (∼2 kyr), but probably episodic at finer time scales (<2 kyr). No geomagnetic excursion was found at the Yichuan profile near the Yellow River valley, where loess accumulation may be strongly affected by local environmental changes and thus may have been discontinuous. Both site location and time scale therefore need to be considered when considering continuity of Chinese loess.

1. Introduction

[2] The sequence of alternating loess and paleosols from the Chinese Loess Plateau is one of the most complete terrestrial records of geomagnetic and paleoclimatic signals for the past 2.6 Myr. Since the pioneering work of Heller and Liu [1982], paleomagnetic studies have played a pivotal role in deciphering chronometric signals from the Chinese loess/paleosol sequences (see reviews by Heller and Evans [1995] and Evans and Heller [2001]), thus providing reliable chronological constraints on the sequences. With the help of loess magnetochronology and astronomical tuning, the good correspondences between the Chinese loess-paleosol cycles and the marine oxygen isotope records led several workers to conclude that loess accumulation is basically continuous on the glacial-interglacial timescale [e.g., Ding et al., 2002; Deng et al., 2005]. However, arguments exist about the continuity of the loess deposits at finer time scales. For example, Porter and An [1995] correlated the fluctuations in quartz grain size to the Heinrich events. Using high-resolution multi-proxies, Chen et al. [1997] revealed both Henrich events and Bond cycles from a loess profile at the western Loess Plateau assuming that the loess deposits are essentially continuous. In contrast, using closely spaced optically stimulated luminescence (OSL) dates for samples from the Holocene soil (S0) and the uppermost unit of the last glacial loess (L1) on the loess tablelands from the Beiguoyuan section, central Chinese Loess Plateau, Stevens et al. [2006] documented a depositional hiatus around 10–15 ka. Moreover, several intervals with the highly reduced accumulation rates at greater ages from the Jiuzhoutai loess profile could also correspond most possibly to sedimentation hiatus [Singhvi et al., 1999, 2001]. In addition, Kohfeld and Harrison [2003] systematically investigated the temporal-spatial patterns of the aeolian mass accumulation rates (MARs) from the Chinese Loess Plateau. They suggested that the wrongly assigned age models could be one of the main reasons for uncertainties about the magnitude of the MARs.

[3] To further investigate this problem, we investigated the geomagnetic excursions, which occurred in the last glacial period at three sections along an east-west transect in the central Chinese Loess Plateau. Similar to the geomagnetic polarity reversals, which are the major age controls for Chinese wind-blown sequences, the presence and/or absence of short-term geomagnetic excursions offers an excellent opportunity to assess the continuity of loess sedimentation. In this study, we present new paleomagnetic results and magnetic susceptibility and bulk grain size data from the Xifeng (35.76°N, 107.69°E), Luochuan (35.75°N, 109.42°E) and Yichuan (36.10°N, 110.14°E) sections. The results are further compared with previously published data from the Weinan section (34.4°N, 109.5°E) at the southern margin of the Loess Plateau [Zhu et al., 1999; Pan et al., 2002] (Figure 1).

Figure 1.

Schematic map showing the Loess Plateau and location of the investigated loess sections (solid circles) and other sections mentioned in this study (open circles). The open and solid arrows show summer and winter monsoon directions, respectively.

2. Sampling and Experiments

[4] The sampled stratigraphic interval comprises the Holocene soil (S0), the last glacial loess (L1) and the uppermost part of the last interglacial soil (S1). Total 279 samples were taken from Xifeng at 2.5 ∼ 5 cm intervals, 164 samples from Luochuan at 5 cm intervals, and 216 samples from Yichuan at 2 ∼ 10 cm intervals.

[5] The mass-specific low-field magnetic susceptibility (χ) was measured using a Bartington MS2 meter at a frequency of 470 Hz. The particle size distribution of the samples was measured using a SALD-3001 laser diffraction particle size analyzer. Ultrasonic pretreatment with addition of 20% (NaPO3)6 solution was used to disperse the samples.

[6] Remanence measurements were made using a three-axis cryogenic magnetometer (2G Enterprises, USA) installed in field-free space (<300 nT). Stepwise thermal demagnetization of the natural remanent magnetization (NRM) was performed on all the 659 samples from the Xifeng, Luochuan and Yichuan sections. All samples were heated to 680°C, with 12–20 steps of demagnetization and 10–50°C temperature increments.

3. Results

[7] Progressive demagnetization successfully isolated the characteristic remanent magnetization (ChRM) components for most of the samples after removing a viscous component of magnetization after the 150–300°C treatment. Demagnetization results were evaluated by orthogonal diagrams [Zijderveld, 1967] (Figure 2) and the principal components direction was computed by a “least-squares fitting” technique [Kirschvink, 1980]. 261 (94%) samples from the Xifeng section, 140 (85%) samples from the Luochuan section, and 202 (94%) samples from the Yichuan section gave reliable characteristic remanent magnetization (ChRM) vector directions. Virtual geomagnetic pole (VGP) latitudes were determined from the ChRM vector directions. These VGP latitudes were subsequently used to define the successions of magnetic polarities in the studied sections (Figures 3a–3c). For comparison, the geomagnetic records obtained from the Weinan section [Zhu et al., 1999] are shown in Figure 3d.

Figure 2.

Orthogonal projections of representative progressive thermal demagnetization of the (a)–(b) Xifeng, (c)–(d) Luochuan, and (e)–(f) Yichuan sections. The solid (open) circles represent the horizontal (vertical) planes. The numbers refer to the temperatures in °C. NRM is the natural remanent magnetization.

Figure 3.

Stratigraphy, low-field magnetic susceptibility (χ), median grain size (Md) and latitude of virtual geomagnetic pole (VGP) of the (a) Xifeng, (b) Luochuan, (c) Yichuan, and (d) Weinan sections, and correlations with the (e) LR04 benthic δ18O stack [Lisiecki and Raymo, 2005]. Md data of the Luochuan section are provided by J. M. Sun. VGP and χ data of the Weinan section are after Zhu et al. [1999] and Pan et al. [2002] and Md data of the section after Song [1994]. Here e1 and e2 correspond to the Mono Lake and Laschamp excursions, respectively. The numbers in Figure 3e indicate the marine oxygen isotope stages.

[8] χ and median grain size (Md) values show variations that correspond to the pedogenic stratigraphy, with higher χ and lower Md values in the Holocene soil S0, the sub-paleosol layer L1SS1 and the last interglacial soil S1, but with lower χ and higher Md values in sub-loess units L1LL1 and L1LL2 (Figure 3).

4. Discussion

[9] The Laschamp excursion (∼40 ka) has been well characterized in previous studies. Lund et al. [2005] investigated the Laschamp excursion from two well separated (150 km apart) sediment cores from the western North Atlantic. The estimated duration of the Laschamp event is about 1.2 and 2 kyr defined by the paleodirection oscillations and the paleointensity anomaly, respectively. Based on the planktonic oxygen isotopic data, the Laschamp event was dated between 39–41 ka for samples from ODP site 919 [Channell, 2006]. Independently, the dominant authigenic 10Be/9Be peak associated with the Laschamp excursion at about 40 ka persisted ∼2 kyr [Leduc et al., 2006]. Therefore, there is a consensus that the duration of the Laschamp excursion is ∼2 kyr.

[10] As shown in Figure 3, except for the Yichuan section, the other three sections consistently record a VGP anomaly in the lower part of the sub-paleosol unit L1SS1, corresponding to marine oxygen isotope stage 3 (MIS3). Zhu et al. [2006] revealed that the VGP anomaly recorded at the Luochuan section is bracketed well by these two Heinrich events (H4 at ∼39 ka and H5 at ∼48 ka), and the linearly interpolated age for the VGP anomaly is about 40–41 ka. Based on the OSL age model of Lu et al. [2007], this VGP anomaly was dated at 42.5 ± 2.5 ka. Therefore, it can be unambiguously correlated to the Laschamp excursion.

[11] The Yichuan section is situated near the Yellow River valley between Shaanxi and Shanxi provinces (see Figure 1). The absence of the Laschamp excursion at Yichuan indicates its highly episodic deposition in the valley where the loess accumulation was probably affected by the local environment. In contrast, the other two sections (Xifeng and Luochuan), located within the flat tablelands in the central Loess Plateau, have the ability to record the Laschamp excursion. This indicates that the loess deposition at the lower part of the sub-paleosol layer L1SS1 of the two profiles, from the viewpoint of paleomagnetism, can be considered to be continuous on the time scale of about 2 kyr. At the Weinan profile in the southern margin of the Loess Plateau (Figure 1), both the Laschamp geomagnetic excursion and the Mono Lake geomagnetic excursion (∼26 ka) were recorded [Zhu et al., 1999; Pan et al., 2002] (see also Figure 3d). Therefore, we suggest that the last glacial loess in the central-southern Loess Plateau is continuous at the time scale of about 2 kyr. Moreover, the loess deposits were transported by the East Asian winter monsoon. The median grain size (Md) and magnetic susceptibility (χ) are widely-used indicators for the intensity of the East Asian winter monsoon [Porter and An, 1995; Ding et al., 2002] and the magnetic enhancement, hence summer monsoon intensity, of Chinese eolian deposits [Heller and Evans, 1995; Liu et al., 2007], respectively. Due to age errors, the interprofile correlations of these proxies are not precise on the millennial timescale. However, on the longer or orbital timescale, these proxies between profiles are well correlated, indicating that the overall environmental setting was relatively uniform.

[12] It is notable that the morphology of the Laschamp excursion differs greatly between the studied three profiles. The Weinan section, with high χ values and probably suffered a higher degree of pedogenic alteration than the other sites, which therefore records both the sophisticated structure of the Laschamp excursion and the short-lived Mono Lake excursion. This indicates that at this site loess accumulation was more continuous but that stronger pedogenesis may have altered the morphology of the paleomagnetic records. However, for the Xifeng and Luochuan sections with a reduced and comparable degree of pedogenesis, weak pedogenic alteration is surely not the major factor accounting for the diversified features of the paleomagnetic records. Liu et al. [2005] have clearly characterized the grain size distribution of pedogenic maghemite particles. The grain size distribution of these fine-grained particles peaks at the single-domain and superparamagnetic threshold. Thus, they are very magnetically viscous and are the major magnetic carrier for chemical and/or viscous remanent magnetization, which can be easily cleaned by thermal demagnetization. Thus, the inconsistent patterns of the Laschamp excursion among the profiles are possibly due to the discontinuous depositions on a time scale less than 2 kyr. This could also be the main reason that the Mono Lake excursion with a duration (e.g., 1.2 kyr [Wagner et al., 2000]) less than the Laschamp excursion is not present at either Xifeng or Luochuan sections.

[13] It is not surprising that Stevens et al. [2006] found a depositional hiatus around 10–15 ka at the Beiguoyuan section. Our findings further suggest that sedimentary hiatuses with duration less than 2 kyr can be common even in the loess unit with relatively higher sedimentation rate in the central Chinese Loess Plateau. Specifically, areas near valleys or on hillslopes, such as the Yichuan section, might experience the sheetwash-induced erosion at some times. Another possible area is the transition zone close to the margins of deserts (dust sources), which are probably subjected to more serious discontinuities during intervals of desert expansion. For example, sand layers are identified in the Caijiagou and Shimao loess sections adjacent to the Mu Us desert [Sun and Liu, 2000] and in the Shagou loess section near the Tengger desert [Wu et al., 2006] (see Figure 1). Sedimentation is probably continuous in the central-southern Loess Plateau at the time scale larger than 2 kyr. Therefore, when retrieving the paleoclimatic records at the millennial time scale, great cares should be taken to consider both the site location and the possibility of discontinuities of the loess deposition.

[14] Paleomagnetic records (both paleomagnetic reversals and excursions) represent only a short period of several thousand years. This could prevent us confidently generalizing our conclusion to the whole L1 sequence. Nevertheless, this study provides useful constraints on the sedimentation continuity at least for the bottom of MIS3, where OSL ages have a relatively large error of several thousand years.

5. Conclusions

[15] Paleomagnetic investigation of loess sediments of the last glaciation revealed the presence of the short-duration Laschamp and/or Mono Lake geomagnetic excursions at the Xifeng, Luochuan and Weinan sections, central-southern Loess Plateau, and the absence of these excursions at the Yichuan section, close to the Yellow River valley. The integration of geomagnetic and paleoclimatic records suggests that pedogenesis-induced smoothing effects on the primary natural remanent magnetization (NRM) are weak. Because the duration of these two geomagnetic excursions is no more than 2 kyr, the presence of the excursions suggests that loess sedimentation at the central and southern Loess Plateau at least during the last glaciation is continuous at the time scale of about 2 kyr. Discontinuities may exist at these profiles at the time scale <2 kyr, which may have generated the different morphology of the Laschamp excursion records. Significant discontinuity of sedimentation may have occurred in profiles close to large valleys and deserts. Therefore, it is vital to consider both profile location and time scale when discussing the short-term paleoclimatic and paleomagnetic records from the Chinese loess deposits.


[16] Financial assistance was provided by the National Natural Science Foundation of China grants 40674032 and 40221402 and the Chinese Academy of Sciences grant KZCX-3-SW-150. Q. S. Liu was supported by the European Commission through a Marie-Curie Fellowship (IIF), proposal number 7555.