The Laschamp geomagnetic excursion (LGE, at ∼41 ka) is a critical age control for hemispherical comparison of paleoclimate records from various kinds of geological archives (e.g., ice core, marine, and lake sediments). The timing of the LGE in Chinese Loess, however, remains poorly constrained due to the lack of a reliable chronology and the complex acquisition processes of natural remanent magnetization. Here, we systematically compare the paleomagnetic results of three optically stimulated luminescence dated loess sequences on the Chinese Loess Plateau. Our results indicate that the timing of the LGE in the northwestern Loess Plateau is slightly older than the absolute radiometric age determination and the timing inferred from Greenland ice core (10Be flux) and marine (i.e., relative paleointensity and authogenic 10Be/9Be stack) records, but younger than the counterparts in the central Loess Plateau. We attribute the different timing of the LGE in the three loess sections to a progressive southeastward increase in the lock-in depth caused by the combined effects of postdepositional processes (e.g., surface mixing, chemical weathering, and lock-in delay) on the remanence acquisition process. We conclude that caution is needed to use the LGE in Chinese Loess as a reliable tie-event for high-resolution chronological correlation to marine and ice core records, only if the potential lock-in effect can be precisely determined.
 The Laschamp geomagnetic excursion (LGE) was first identified in late Pleistocene lava flows of the French Massif Central [Bonhommet and Babkine, 1967]. Early dating of the lava flows suggested a mean age of about 46.2 ± 2.5 ka [Bonhommet and Zahringer, 1969; Hall and York, 1978; Gillot et al., 1979]. Recently, the age of the LGE was determined to be 40.7 ± 0.95 ka using combined 40Ar/39Ar, K-Ar, and 230Th-238U data [Guillou et al., 2004; Singer et al., 2009]. The LGE has been found globally in marine sediments, as evidenced by directional changes and relatively low paleointensity [e.g., Laj et al., 2000, 2004, 2006; Lund et al., 2005; Yamazaki and Kanamatsu, 2007; Channell et al., 2009; Nowaczyk et al., 2012]. Since the cosmogenic production rate is inversely proportional to the intensity of the Earth's magnetic field [e.g., Lal, 1992], the beryllium (Be) records provide proxies of the geomagnetic dipole moment variations, such as the LGE inferred by the 10Be flux and 10Be/9Be ratio in marine sediments [Carcaillet et al., 2004; Leduc et al., 2006; Ménabréaz et al., 2011, 2012] and in Greenland and Antarctic ice cores [Yiou et al., 1997; Raisbeck et al., 2007].
 Synchrony of the LGE in ice cores [10Be flux around 41–42 ka, Raisbeck et al., 2007 and references therein] and globally distributed marine sediments (40–42 ka in relative paleointensity stacks) [e.g., Guyodo and Valet, 1996, 1999; Laj et al., 2000, 2004; Yamazaki and Kanamatsu, 2007; Channell et al., 2009, 2012] (Table 1) suggests that short-term geomagnetic excursions can be employed as tie points for global stratigraphic and paleoclimatic correlation [e.g., Laj et al., 2004; Raisbeck et al., 2007; Channell et al., 2009]. While the LGE is of a sufficiently global nature [e.g., Leonhardt et al., 2009], this excursion is not always recorded in sediments due to the complexity of the postdepositional remanent magnetization (PDRM) mechanisms [Roberts and Winklhofer, 2004]. Therefore, complications remain for the successful application of the LGE for precise synchronization of paleomagnetic and paleoclimatic proxies.
Table 1. The Laschamp Geomagnetic Excursion in Lava Flows, Ice Cores, Marine Sediments, and Chinese Loessa
New Age (ka)
New age is estimated by millennial-scale correlation of loess grain size, and the δ18O records of GRIP and GISP2 ice cores to NGRIP δ18O records record using the GICC05 timescale [Svensson et al., 2008].
 Similar to the marine and ice core records, the LGE in Chinese Loess has been studied using geomagnetic [Zhu et al., 1999, 2000, 2006a, 2006b, 2007] and cosmogenic nuclide (10Be) approaches [Zhou et al., 2007, 2010]. However, the timing of the LGE in Chinese Loess has not well constrained, ranging between 37.4 and 47.5 ka derived from various age models on six loess sections from the central Loess Plateau (Figure 1 and Table 1). Moreover, due to the complex postdepositional remanent magnetization (PDRM) processes of Chinese Loess and variable sedimentary settings, the LGE was not identified in three loess sections (Mangshan, Yichuan, and Baicaoyuan, Figure 1) near the Yellow River [Zeng et al., 2002; Zhu et al., 2007; Deng, 2008]. Thus, detailed geomagnetic and chronological investigations on Chinese Loess are essential to quantify the timing and lock-in effect of the LGE in Chinese Loess.
 In this study, we present new paleomagnetic results revealing a distinct geomagnetic excursion in an optically stimulated luminescence (OSL) dated loess sequence from the northwestern Chinese Loess Plateau. Unlike previous geomagnetic studies mainly focusing on the mechanism of the loess remanence acquisition processes [e.g., Zhu et al., 1999; Spassov et al., 2003; Zhao and Roberts, 2010], our main objective is to estimate the timing and lock-in effect of the LGE in Chinese Loess by comparison of the paleomagnetic results of three OSL-dated loess sections (Gulang, Luochuan, and Weinan). The results are further compared with the 10Be flux record in a Greenland ice core and the relative paleointensity and 10Be/9Be stack from deep-sea sediment cores to address the difference PDRM lock-in processes in loess and marine sediment. Finally, based upon the spatial changes in the timing and lock-in depth of the LGE at three loess sections, we evaluate whether or not the LGE in Chinese Loess can be regarded as a reliable tie point for correlation with marine and ice core records.
2. Materials and Methods
 The loess sequence at Gulang (37.49°N, 102.88°E, 2400 m above sea level) is situated in the northwestern Chinese Loess Plateau (Figure 1). High sedimentation rates (19–63 cm/kyr at Gulang) and weak pedogenesis make this loess sequence sensitive to rapid monsoon changes [Sun et al., 2012]. Unlike previous paleomagnetic studies based on cubic specimens collected from fresh outcrops, we took continuous 50 cm U-channel samples from a 20 m pit excavated at Gulang for paleomagnetic analyses. Sampling strategy and quality of the loess U-channel samples are described in Liang et al. . In this study, we conducted a paleomagnetic examination of these loess U-channel samples and magnetic measurements on several representative paleosol and loess samples.
 Stepwise alternating field (AF) demagnetization at 2 cm resolution on the U-channels was performed from 5 to 100 mT with increments of 5–20 mT, using a Model 810 Automatic Sample Handler with a 42 mm diameter aperture. The remanence measurements were made using a 2G Enterprises Model 755 cryogenic magnetometer installed in a magnetically shielded space (<150 nT). All magnetic measurements were conducted at the Institute of Earth Environment, Chinese Academy of Sciences. Demagnetization results were evaluated by orthogonal diagrams and the characteristic remanent magnetization was determined after 30 mT AF demagnetization using a least-squares method and with a maximum angle deviation usually less than 10° (Figure 2). Virtual geomagnetic poles (VGP) were calculated from the vector directions of the characteristic remanent magnetization. The VGP latitudes were subsequently used to define the short-term excursion in Gulang section and to further compare with previously published VGP results of the Luochuan and Weinan sections [Zhu et al., 1999, 2006a, 2007].
 Temperature-dependent susceptibility (χ-T) curves for three representative samples were measured using a MFKI-FA Kappabridge equipped with a CS-3 high-temperature furnace (Agico Ltd., Brno) in an argon atmosphere, with a field of 200 A/m and a frequency of 976 Hz. The heating temperature increased from room temperature to 700°C, with a heating rate of 11°C/min, while the cooling curve is obtained from 700°C to room temperature with the same field, frequency, and cooling rate. A record of the variation of remanent magnetization with temperature was obtained. Three representative samples were cooled from room temperature (300 K) to 10 K in a zero field and then saturation isothermal remanent magnetization (SIRM) was imparted with a 2.5 T field at 10 K. After the magnetic field was switched off, SIRM was measured from 10 to 300 K at 5 K intervals using a Quantum Designs Magnetic Property Measurement System to further determine the main magnetic carriers.
 Due to significant developments in laboratory protocols [e.g., Murray and Wintle, 2000, 2003], the reliability of OSL dating has been significantly improved and can now be used to address loess accumulation processes [Stevens et al., 2006; Lu et al., 2007; Sun et al., 2010; Kang et al., 2013] and abrupt monsoon changes [Stevens et al., 2008; Sun et al., 2012]. These high-resolution OSL dates also permit a reliable age estimation of the LGE in loess sections at Luochuan [Lu et al., 2007], Gulang [Sun et al., 2012], and Weinan [Kang et al., 2013]. To eliminate the age uncertainties in the ice core and marine records published by previous studies, e.g., relative paleointensity stack (NAPIS-75) on the GISP timescale [Meese et al., 1997; Laj et al., 2000] and 10Be flux in the GRIP record on the SS09sea age model [North Greenland Ice Core Project (NGRIP) Members, 2004; Raisbeck et al., 2007], the GRIP and GISP2 δ18O records were directly matched with the NGRIP results on a recently developed Greenland Ice Core Chronology (hereafter referred to as the GICC05 timescale) [Andersen et al., 2006; Svensson et al., 2008]. Fine adjustment of the LGE timing is less than 500 years based on the GICC05 chronology (Table 1).
3. Results and Discussion
 Magnetic mineralogy can be determined using the χ-T plots and low-temperature demagnetization spectra for representative paleosol and loess samples. The heating curves for loess and paleosol specimens resemble those from loess sequences in the western Chinese Loess Plateau (Figure 3a). The susceptibility exhibits a more prominent drop from 260 to 450°C for paleosol than loess samples, due to the transformation of pedogenic maghemite particles to hematite [Liu et al., 2005a]. For the loess and paleosol samples, a susceptibility peak above 510°C is due to the neoformation and/or presence of single domain ferrimagnetic particles during heating [Dunlop and Özdemir, 1997; Liu et al., 2005a]. The χ-T curves are characterized by a major decrease in susceptibility at about 585°C, implying the presence of magnetite. The cooling curves were highly enhanced with maximum susceptibility at ∼400°C, corresponding to the unblocking temperatures of the newly formed single domain magnetite particles during heating [Liu et al., 2005a]. Zero field cooling results of both loess and paleosol samples reveal the existence of the Verwey transition at 110–120 K (Figure 3b), indicating that coarse-grained magnetite particles are the main carriers of the natural remanent magnetization (NRM) [Liu et al., 2003; Deng, 2008].
 The demagnetization results for representative specimens reveal that the NRM varies between 10−4 and 10−5 A/m (Figure 4). For the paleosol samples, a stable component of magnetization is isolated after AF demagnetization at 10–15 mT. In contrast, loess samples show a remanence component that trends toward the origin except for the soft component removed at an AF of 30 mT or lower. However, a considerable amount of remanence remains after demagnetization at an AF of 100 mT, probably attributable to the presence of coarse-grained magnetite and hematite [Liu et al., 2003, 2005b]. A representative sample from the Laschamp excursion is characterized by a stable component of magnetization passing through the origin of the orthogonal projection after AF demagnetization at 30 mT (Figure 4C).
 The inclination of Gulang Loess sequence indicates a large oscillation around 13.5–13.9 m (Figure 2b), characterized by a gradual decrease from 55.7° to −27.3° at the depth of 13.9–13.7 m, an abrupt increase at 13.7 m and small fluctuations (38–59°) afterward. The recorded VGP paths of the Laschamp excursion are different at the three sites (Figures 5 and 6). First, the VGP latitudes demonstrate negative shifts at Gulang and Weinan sites, but only a subtle change (not extending southward beyond 45°N) at Luochuan. Second, the LGE is characterized by a short interval and abrupt changes at Gulang and Luochuan sections. However, it spans around a 0.8 m depth interval with significant fluctuation of the VGP latitudes at the Weinan section. Third, the VGP paths at Gulang and Weinan exhibit a clockwise loop reaching to southern hemisphere high latitude, similar to the paths inferred from marine records [Laj et al., 2006; Leonhardtetal., 2009]. However, the VGP paths of the Luochuan show a very distinctive pattern, with a very shallow loop toward the northeast Pacific, likely representing a PDRM-smoothed record of the LGE.
 The above differences in the shapes and VGP paths of the LGE indicate that sedimentary environments have significant influences on the behaviors of short-term geomagnetic events recorded in marine and terrestrial sediments [Roberts and Winklhofer, 2004; Zhu et al., 2007; Jin and Liu, 2010]. Particularly in Chinese Loess, the possibility of multiple PDRM processes and unstable directional change during relatively weak paleointensity periods complicate the robust detection of short-term geomagnetic excursions and reversals [Spassov et al., 2003; Jin and Liu, 2010]. Unlike the global nature of the LGE [e.g., Leonhardt et al., 2009], the Mono Lake excursion is probably not sufficiently global. This excursion is evident in the Weinan and Datong Loess sections [Zhu et al., 1999, 2006b] and the volcanic rocks of New Zealand [Cassata et al., 2008], but is less clear in the other loess sections and in the 10Be/9Be results from the Western Pacific [Menabreaz et al., 2011]. The absence of the Mono Lake excursion in Chinese Loess is likely due to its relative short duration and variable depositional environments involving sedimentation rate, pedogenesis, and postdepositional disturbance [Zhu et al., 2006b, 2007].
 It is noteworthy that the stratigraphic positions of the Laschamp excursion in three loess sections are different, varying from the middle of L1LL2 at Gulang, the bottom of L1LL2 at Luochuan to the most L1SS2 at Weinan [Zhu et al., 1999, 2007] (Figure 5). Such a systematic southeastward deepening of the stratigraphic position of the LGE in last glacial loess indicates that the lock-in feature of the LGE in Chinese Loess is strongly linked to postdepositional processes such as surface mixing and particularly pedogenic alterations related to the summer monsoon precipitation. Based on the recently published OSL chronology [Lu et al., 2007; Sun et al., 2012; Kang et al., 2013], the timing of the Laschamp excursion was estimated to be about 42–43 ka at Gulang, significantly younger than that at Luochuan (about 47–51 ka) and Weinan (about 54–56.7 ka) (Figure 5 and Table 1). Different timing of the LGE in Chinese Loess attributes to variable lock-in depths and sedimentation rates over the Loess Plateau, which significantly hampers the application of the LGE as a tie-point for precise correlation of loess-based proxies with ice core and marine records.
 After synchronizing the chronologies of ice core and marine records to the GICC05 timescale, the age of the LGE in Chinese Loess (42–42.5 ka) is slightly older than the timing inferred from GRIP 10Be flux (40.8–41.8) and oceanic stack records (40.5–41.5 ka) of the relative paleointensity and 10Be/9Be ratios (Table 1 and Figure 7). Judging from the high-resolution and large-amplitude grain size oscillations, the LGE at Gulang section is located between Dansgaard-Oeschger (DO) events 10 and 11, apparently older than those around DO event 10 in ice core and marine records. Such a discrepancy in the stratigraphic position (∼40 cm, estimated from the sedimentation rate of ∼40 cm/kyr around the LGE) indicates that the age difference is not attributable to the chronological uncertainties, but rather to different postdepositional processes between loess and marine sediments, including surficial mixing and lock-in effects.
 Previous paleomagnetic studies have investigated extensively the mechanisms of remanence acquisition in loess. Heller and Liu  first attributed the characteristic remanence to a chemical remanent magnetization acquired during in situ growth of haematite. In contrast, most subsequent paleomagnetic investigations considered a PDRM carried by magnetite to be the dominant component of the characteristic remanence for Chinese Loess [e.g., Zhu et al., 1994; Zhou and Shackleton, 1999]. However, Spassov et al.  suggested a detrital-pedogenic lock-in model that comprises a combination of detrital and chemical remanences in Chinese Loess. Recently, laboratory redeposition experiments suggested that Chinese Loess is magnetized by a combination of DRM and PDRM mechanisms, and that water content (rather than dewatering in marine sediments) plays the most dominant control in the PDRM acquisition [Zhao and Roberts, 2010; Wang and Løvlie, 2010]. Therefore, for the Chinese Loess sequences, high sedimentation rates are usually associated with low paleoprecipitation, which will results in a lower lock-in depth of the NRM, and vice versa.
 Unlike terrestrial loess, marine sediments become magnetized through a PDRM process as a result of progressive sediment consolidation and dewatering [e.g., Kent, 1973]. The PDRM processes not only cause a delayed acquisition of remanent magnetization in sediments below the sediment/water interface (namely PDRM lock-in depth) [Verosub, 1977] but also result in smoothing of the original geomagnetic signals (i.e., excursions and reversals) [RobertsandWinklhofer, 2004]. The PDRM lock-in depth can be constrained using paired 10Be flux and paleomagnetic records (e.g., ∼15 cm for the Matuyama-Brunhes boundary) [Suganuma et al., 2010, 2011]. Notably, the amplitude of this offset has long been debated for both Chinese Loess [e.g., Zhou and Shackleton, 1999; Liu et al., 2008] and marine sediments [e.g., RobertsandWinklhofer, 2004; Tauxeetal., 2006], mainly due to the complexity of PDRM processes induced by variable sedimentary characteristics (physical, chemical, and biological properties) and environments (sedimentation rate and water availability).
 Compared with previous paleomagnetic studies, our study provides better constraints on the timing and lock-in depth of the LGE in the three studied loess sections. Sun et al.  estimated the surface mixing layer of profiles from the western Loess Plateau to be less than 20 cm and around 5 cm in the central Loess Plateau. Therefore, the lock-in depth seems to be of the same order (∼20 cm) as the depth of the surface mixing layer at the western Loess Plateau, but it increases significantly in the southeastern Loess Plateau. The southeastward deepening of the lock-in depth is strongly related to the mean annual precipitation changes, which increase from <300 mm at Gulang to 650 mm at Weinan. The downward displacement of the LGE provides new insights on the influence of monsoonal precipitation on stable remanence acquisition of Chinese Loess [proposed by Zhao and Roberts, 2010; Wang and Løvlie, 2010], which probably occurred to about 40 cm depth below the surface in the western Loess Plateau during the last glaciation. Therefore, an appropriate adjustment of the timing of the LGE should be made before applying it to the high-resolution correlation of loess-based proxies with marine and ice core records.
 Paleomagnetic investigation of loess U-channel samples reveals a distinct LGE in the Gulang Loess sequence from the northwestern Loess Plateau. Comparison of the LGE in three OSL-dated loess sections indicates that both the timing and lock-in depth increase from the northwestern to the southeastern Loess Plateau, probably attributable to the southeastward strengthening of pedogenic alterations and decreasing sedimentation rate. The limited age discrepancy for the LGE between the Gulang Loess, ice core, and marine records suggests that short geomagnetic excursions in high sedimentation, weakly weathered loess sequences can be utilized as tie points for chronological and paleoclimatic correlations after appropriate adjustments of the PDRM lock-in effect. However, in the case of strongly weathered loess profiles with relatively low sedimentation rates from the central Loess Plateau, the use of paleomagnetic results for stratigraphic and paleoclimatic correlations should be done cautiously, due to the significant downward displacement of the lock-in depth of the NRM [Zhou and Shackleton, 1999] and the unstable directional change during relatively weak paleointensity periods [Jin and Liu, 2010].
 We thank Simo Spassov, Nicolas Thouveny, and an anonymous reviewer for their valuable comments. Pengxiang Hu and Hui Zhao are acknowledged for their laboratory assistance. This work was supported by grants from the Chinese Academy of Sciences (KZCX-EW-114) and the Natural Science Foundation of China (41072272 and 41272208). Qingsong Liu also thanks supports from NSFC (41025013 and 40974036).