We studied the magnetostratigraphy and sedimentary facies of a 550-m-long drill core from the Jiudong Basin in the NE Tibetan Plateau. Our aims were to reconstruct the late Cenozoic sedimentary evolution of this foreland basin, and to determine the spatiotemporal pattern of growth of the Qilian Shan. The magnetostratigraphy indicates that the sedimentary sequence was deposited during ca. 7–0 Ma. From ca. 6.7–3.0 Ma, the sediment accumulation rate increased gradually from ∼30 mm/k.y. to 120 mm/k.y., which was associated with the gradual evolution of sedimentary facies from a shallow lake/delta front to braided rivers. The progradation of the depositional system from 7 Ma to 3 Ma probably reflects the growth of the relief of the Qilian Shan caused by tectonic uplift. The occurrence of a continuous braided river environment from 3 Ma to the present suggests that the high relief of the Qilian Shan developed before 3 Ma. An abrupt decrease of the sedimentation rate to ∼46 mm/k.y. during 3.0–1.8 Ma, and the deposition of coarse-grained sediments, indicates the uplift of the basin center. We interpret this to reflect the propagation of the thrust system of the Qilian Shan into the basin along a southward-dipping décollement from ca. 3 Ma. Climatic changes may have influenced the sedimentary sequence by introducing long-distance-transported thin coarse sand/gravel layers which are sandwiched within the sequence, and likely were a response to cooling events or climatic transitions. The widespread occurrence of deformation within the basin region in the NE Tibetan Plateau at ca. 3 Ma indicates that this date marks the basinward growth of the deformation system.
It has been proposed that late Cenozoic increases in the accumulation rate and grain size of sediments in foreland basins was caused either by global cooling (Zhang et al., 2001) or by intensified tectonic activity (Fang et al., 2005b). Late Cenozoic sedimentary deposits are widespread in the basins around the NE Tibetan Plateau, and they are interpreted as reflecting the rapid uplift of mountain ranges since ca. 3.6 Ma (Fang et al., 2005a; Li et al., 2014; Lu et al., 2015). However, this date coincides with the onset of significant global cooling associated with the development of Northern Hemisphere glaciation (Zachos, 2001). In addition to sedimentary records, the evidence of thermochronology (Zheng et al., 2010, 2017) and fault slip rates (Tapponnier et al., 1990; Palumbo et al., 2009; Yuan et al., 2011) also suggests that the NE Tibetan Plateau experienced rapid growth and/or uplift during the late Cenozoic. However, the coincidence of tectonic and climatic events complicates the interpretation of the observed changes in sedimentary deposits and researchers have tended to attribute them to both tectonic processes and climatic change (Chang et al., 2012; Heermance et al., 2013). An interesting attribute of several of the sedimentary records is an increase in grain size which coincides with a decrease in the sediment accumulation rate (Chang et al., 2012; Heermance et al., 2013). This phenomenon is difficult to explain by an increased erosion rate caused by global cooling, which would be expected to cause an increase in both the sediment accumulation rate and the sediment grain size. Furthermore, since most of the exposed strata are located within the folded zone along the mountain front, unconformities and coarse-grained material, associated with the development of folding, complicate the development of an unambiguous magnetostratigraphic framework (Zhao et al., 2017) which is needed to date and hence interpret the sedimentary changes. In short, a continuous sedimentary record with a robust chronological framework is needed to evaluate the relative importance of tectonic and climatic controls on sedimentation within basins adjacent to the uplifting mountain range.
The Hexi Corridor, located at the northeastern margin of the Tibetan Plateau (Fig. 1), is a fragmented and restricted foreland basin adjacent to the North Qilian Shan which has accumulated a thickness of 2.5–5 km of Cenozoic sediments above the basal Cretaceous–Oligocene unconformity (Fang et al., 2005b, 2013; Bovet et al., 2009). Studies of the exposed stratigraphy (Fang et al., 2005b; Chen et al., 2006; Liu et al., 2011; Wang et al., 2016; Zheng et al., 2017) at the basin edges reveal that the Qilian Shan has experienced remarkable northward growth and uplift since the late Miocene, and the growth strata and unconformities within the stratigraphy also record the development of folding along the mountain front and/or uplift events (Fang et al., 2005b; Chen et al., 2006; Liu, et al., 2011). However, unconformities may complicate the development of a magnetostratigraphic framework and make it difficult to determine the timing of deformation of mountain growth (Fang et al., 2005b; Chen et al., 2006; Zhao et al., 2017). Here, we present a continuous late Cenozoic (post-7 Ma) stratigraphy from a drill core from the Jiudong Basin. Studies of the sedimentary facies and magnetostratigraphy provide a detailed lithostratigraphic and chronostratigraphic framework for the basin sediments, which potentially enables us to constrain the linkages between the growth of the Qilian Shan and the evolution of the foreland basin.
The Hexi Corridor is a structurally complex foreland basin (Wang and Coward, 1993; Fang et al., 2005b), located at the northern front of the Qilian Shan, NE Tibetan Plateau (Fig. 1). It has an average elevation of ∼1500 m, in contrast to the Qilian range with elevation of up to 5500 m. The Hexi Corridor Basin is elongated parallel to the Qilian Shan and is divided into four sub-basins: Jiuxi, Jiudong, Zhangye, and Wuwei (Fig. 1), from NW to SE (EGPGYO, 1989; Fang et al., 2005b; Wang et al., 2016). The studied drilling site is located at the center of the Jiudong Basin, which is bounded by the Wenshu Shan to the west, the North Qilian Shan to the south, the Jintanan Shan to the north, and the Yumu Shan to the east (Fig. 2). The Jiudong Basin is hydrologically connected to the Jinta-Ejina Basin by the Heihe River and Beida River (Pan et al., 2016).
The Jiudong Basin contains >3-km-thick Cenozoic sediments, mostly Miocene (locally Oligocene) through Pliocene in age (Li and Yang, 1998; Bovet et al., 2009; Zhuang et al., 2011). Along the basin edge close to the Qilian Shan, exposed strata reveal that Cenozoic sedimentary rocks overlie Paleozoic meta-sedimentary rocks (Fig. 2). In the northern part of the Jiudong Basin, basement rocks crop out within the Jintanan Shan and Heli Shan. They consist of Paleozoic sandstone, siltstone, limestone, and volcanic rocks; and Mesozoic red sandstone, green mudstone, and charcoal-gray mudstone (Fig. 2). From bottom to top, the Cenozoic strata consist of five units: the Paleogene Huoshaogou and Baiyanghe formations, the Neogene Shulehe Formation, and the Quaternary Yumen and Jiuquan formations. The Huoshaogou and Baiyanghe formations consist mainly of fluvio-lacustrine red beds with fine conglomerate to mudstone with occasional playa gypsum beds intercalated. The Shulehe Formation consists of fluvio-lacustrine gray and brown fine conglomerate, sandstone, and siltstone. The Yumen Formation is characterized by conglomerate at the basin edge (Fang et al., 2005b), and by fluvio-lacustrine fine conglomerate to mudstone in the basin center (Wang, 1990). The Jiuquan Formation is dominated by gravel beds.
Apatite (U-Th)/He ages from the hanging wall of the North Qilian Shan fault indicate that the thrust fault was initiated at ca. 10 Ma and has a long-term vertical slip rate of ∼0.5 mm/yr (Zheng et al., 2010). A stratigraphic study also dates the onset of faulting in the Jiuxi Basin to 13.5–10.5 Ma (Wang et al., 2016), and a sedimentary facies change at the mountain front indicates that the uplift of the Qilian Shan was initiated since the late Miocene (Song et al., 2001; Fang et al., 2005b), which can be correlated to a period of sediment recycling (10–8 Ma), based on fission track dating of detrital material (Zheng et al., 2017). The basinward growth of a series of fault-fold systems was initiated mainly during 4–2 Ma, such as the Yumen anticline (Chen et al., 2006), the Yumu Shan (Palumbo et al., 2009; Liu et al., 2010), and the Heli Shan (Zheng et al., 2013a). This evidence reveals the progressive northward growth of the fault-fold systems along the Qilian Shan.
MATERIALS AND METHODS
Drilling and Sampling
Core MH (39°27′47.5″ N, 99°26′00.1″ E, elevation: 1387 m) was drilled in the eastern Jiudong Basin (Fig. 2). A directional sensor was used to confirm the direction of the well every 100 m drilling depth and at the bottom. The core was essentially vertical with apical angles within 1–2° from top to bottom. The drilling depth was 554 m, and the length of the recovered core was 510.79 m, with an average recovery rate of 92.2%. The core sediments are dominated by sand and silt, with occasional interbedded layers of medium-coarse sand, gravel, and clay. The medium to coarse sand and gravels within the interval of 0.0–100.3 m could not be recovered completely because they are unconsolidated and tend to slip out of the core barrel. The sedimentary sequence was described based on color, sedimentary textures, and sedimentary structures, according to Miall (2000). The color of the fresh core surface was determined by comparison with the National Research Council Rock-Color Chart (Goddard et al., 1948). Lithologic classification was based primarily on grain size. For sand-sized sediments, a grain size chart was used for distinguishing the main sub-divisions (very coarse, coarse, medium, fine, very fine) which can provide information on the modal size-range of the sand layers. Silt and clay were distinguished in hand specimens by the presence or absence of a gritty texture, using fingers or the tongue. For gravels, the maximum clast size was typically estimated by taking the average of the ten largest clasts.
2892 oriented samples, at intervals of 0.2–0.5 m, were taken for rock magnetic and paleomagnetic measurements. Each sample was cut into three cubic specimens with dimensions of 2 × 2 × 2 cm. The gravel content of gravel layers was determined by sieving.
Anisotropy of Magnetic Susceptibility
To assess whether the original sedimentary magnetic fabric was disturbed, the anisotropy of magnetic susceptibility (AMS) of 233 discrete samples was measured using a KLY-4s Kappabridge magnetic susceptibility meter. AMS can be represented by the shape and direction of the susceptibility ellipsoid, with the three orthogonal principal axes designated the maximum, intermediate, and minimum susceptibility axes (K1, K2, and K3, respectively) (Hrouda, 1982). The major AMS parameters examined in this study were the lineation (L, defined as L = K1/K2) and foliation (F, defined as F = K2/K3). AMS was measured at the Key Laboratory of Western China’s Environmental Systems (Ministry of Education of China), Lanzhou University.
Rock Magnetic Measurements
To determine the magnetic mineralogy of the sediments, magnetic hysteresis, and thermomagnetic analyses (Dunlop and Özdemir, 1997) were conducted on 35 specimens from different lithologies and depths. Hysteresis loops were measured using a Princeton Measurements Corporation vibrating sample magnetometer (MicroMag 3900); the magnetic field was cycled between +1.0 T and –1.0 T, with a field increment of 5 mT and a time interval of 300–500 ms. The temperature dependence of the magnetization was measured with a variable field translation balance, from room temperature (∼20 °C) to 700 °C and then back to room temperature, with an interval of 4 °C, in an applied field of 110 mT. All measurements were performed at the Institute of Earth Environment, Chinese Academy of Sciences (Xi’an, China).
Demagnetization of the Natural Remanent Magnetization
Alternating field (AF) demagnetization was conducted on samples and thermal demagnetization was conducted on parallel samples that were well cemented. Stepwise AF demagnetization of the natural remanent magnetization (NRM) was performed on 2403 cubic specimens using a 2G Enterprises Model 760-R cryogenic magnetometer (2G760) installed in a magnetically shielded (<300 nT) space. AF demagnetization steps of 5–10 mT were used, to a maximum AF of 80 mT. 585 discrete specimens were subjected to stepwise thermal demagnetization; they were heated to a maximum temperature of 690 °C at 10–50 °C intervals using a MMTD80 thermal demagnetizer. Remanent magnetization was measured using the 2G Enterprises Model 760-R Cryogenic Magnetometer. Demagnetization results were evaluated using orthogonal projection diagrams (Zijderveld, 1967). Characteristic remanent magnetization (ChRM) directions were calculated using principal component analysis with the least-squares method (Kirschvink, 1980). All the paleomagnetic data were analyzed using PMGSC 4.2 developed by R. Enkin (http://gsc.nrcan.gc.ca/sw/paleo_e.php). Paleomagnetic experiments were conducted at the Key Laboratory of Western China’s Environmental Systems (Ministry of Education of China), Lanzhou University.
Lithology and Sedimentology
The sedimentary sequence of the MH core exhibits an overall coarsening-upward trend (Fig. 3). Based on lithology, sedimentary texture, structure and their vertical associations, the sequence can be divided into four stratigraphic units which represent four different sedimentary environments (Fig. 3). They are described below.
Unit 1(554–465.3 m). This unit is dominated by layers of thick-bedded reddish, yellow-brown fine-grained clay silt, silt, and sandy silt, which are intercalated with thin and medium-bedded (0.2–4 m) gray sand layers. The thick-bedded silt layers contain horizontal, wavy, and rhythmic laminations as well as lenticular bedding; bioturbation features were commonly observed. Flaser and parallel bedding is present in the sand layers. A distinct medium-very coarse sand layer is present at 514.5–509.8 m, which has an erosive lower contact with the silt layer. Intrabasinal clasts of calcrete and mudballs occur above the erosional surface of the sand layer.
The facies assemblage suggests deposition within an intermittent shallow lake, mouth bar/distal bar and distributary channels, as part of a delta front or shallow lake environment (Cabrera et al., 1985; Reading, 1996).
Unit 2(465.3–395.64 m). This unit consists mainly of alternating layers of thin- to medium-bedded (1–3 m) reddish, yellow-brown silt, and brownish, yellowish sand. The lower contacts of the sand layers are erosive. Lag deposits of granules, very coarse sand, calcrete, and mudballs overlie the basal erosional surfaces of the sand layers. Parallel bedding was often observed in the sand layers. The alternating silt layers are usually horizontally laminated and contain several lenticular beddings. Bioturbation features (burrows) were observed in the silt layers. The uppermost part (420–395.64 m) of this unit is dominated by thick-bedded silt (2–7 m) layers containing horizontal, wavy laminations and lenticular bedding.
Horizontally laminated to structureless, mottled, and bioturbated silt layers from 465.3 to 420 m are interpreted as originating in floodplain/interdistributary areas. Sediments deposited within shallow lakes, as expressed by wavy, rhythmic laminations and by lenticular and flaser bedding and thick silt-clay packages, occur from 420 to 395.64 m (Fig. 3) (Reading, 1996). The thin sand beds (<2 m), with ubiquitous scour-fill structures and frequent changes in grain size, suggest deposition by sheet-floods through shallow, simple, unstable channels (Tunbridge, 1981; Hubert and Hyde, 1982; Olsen, 1987, 1989; Kelly and Olsen, 1993), which may have been fluvial distributary channels developed in a delta plain, associated with a shallow lake and delta front environment. We interpret this unit as delta plain or littoral lake facies (Reading et al., 1996).
Unit 3(395.64–292.5 m). This unit consists of alternating layers of medium- to thick-bedded (2.5–15 m) brownish, yellowish sand and silt; in addition, it contains 10 or more sedimentary cycles which grade upwards from sand to silt (Fig. 3). The lower contacts of the sand layers are generally erosive. Intrabasinal clasts of calcrete and mudballs overlie the basal erosional surfaces of some of the sand layers, and cross and parallel bedding was observed in the sand layers. The silt layers are massive and laminated and contain bioturbation features (burrows).
The erosional contact and thickness of the sand layer with cross bedding indicate that the sands were deposited within river channels with a unidirectional current. The silt interlayers are characterized by horizontal laminations, phytoclasts, mottling and bioturbation features, suggesting suspension deposition in a floodplain (Miall, 1996; Reading, 1996). This succession is characterized by 58% channel sands and 42% floodplain silts; the high proportion of floodplain deposits and ubiquitous bioturbation features suggest that the channels were relatively stable (Nichols and Fisher, 2007). We interpret this succession as meandering river facies (Miall, 1996; Reading, 1996).
Unit 4(292.5–0 m). This unit is dominated by thick sand layers, which are yellowish and intercalated with several medium- to thick-bedded (2–10 m) gravel layers and minor intercalated layers of clay and silt. The lower contacts of the sand bodies are generally erosive and the upper contact grades into silt or clay. Multiple vertically stacked sand bodies are usually separated by internal erosion surfaces (Fig. 3). Cross and parallel bedding was frequently observed in the sand layers. The embedded gravel layers contain gravels that are sub-rounded and clast-supported, with clast diameters of up to 30 mm. The intervening silt and clay layers are brown to gray and are largely featureless with a few poorly developed wavy laminations. Four sedimentary cycles which grade upward from sand or gravel to silt or clay were recognized in this unit (Fig. 3).
The dominant thick sand layers with cross bedding indicate deposition within a fluvial channel. Vertical stacking of sands is characteristic of channel bar and channel fill deposits of aggrading, low-sinuosity streams which migrate laterally across an alluvial plain (Gordon and Bridge, 1987). The brown, gray fine-grained interlayers of silt and clay are largely featureless with a few poorly developed wavy laminations, suggesting floodplain deposition. The occurrence of thick clast-supported gravel layers in the depression zones implies that they were transported by high-energy river channel systems. We infer that this unit was deposited within a braided river environment (Gordon and Bridge, 1987; Rust and Jones, 1987; Kumar and Nanda, 1989), like the modern environment.
AMS, Rock Magnetic and Paleomagnetic Results
The magnetic fabric of the samples is generally oblate, with the magnetic foliation (F) larger than the magnetic lineation (L) (Fig. 4A). The minimum susceptibility axes of the AMS ellipsoid for most samples are subvertical and perpendicular to the bedding plane (K3-incl.mean = 80.4°), whereas the maximum axes are close to horizontal and parallel to the bedding plane (K1-incl. mean = 7.2°) (Fig. 4B). K3 are clustered around the center of the projection plane (Fig. 4B). Given that core MH is located in the foreland center and far from the thrust belt, we suggest that the AMS is of sedimentary origin and may be overprinted by tectonic strain (Yu et al., 2014); however, it is still in the earliest stage of deformation (Mattei et al., 1997; Parés et al., 1999; Robion et al., 2007).
Rock Magnetic Results
The temperature-dependent magnetic susceptibility curves (χ–T) (Figs. 5A–5C) for the sand samples show a marked decrease in magnetic susceptibility at ∼580 °C, indicating the ubiquitous occurrence of magnetite. In addition, there is steady increase of χ during heating at temperatures below ∼300 °C (Figs. 5A–5C), which may result from the gradual unblocking of fine-grained (i.e., the superparamagnetic and small single-domain) ferrimagnetic particles, or the release of stress upon heating (Deng et al., 2006; Liu et al., 2005; van Velzen and Zijderveld, 1995). A subsequent fall in χ from 300 °C to 500 °C may be caused by the conversion of ferrimagnetic maghemite to weakly magnetic hematite (Løvlie et al., 2001; Deng et al., 2008), or by changes in crystallinity, grain size, and /or the morphology of the magnetic grains (Dunlop and Özdemir, 1997; Ao et al., 2009). Susceptibility continues to decrease until 680 °C, indicating that hematite is also present in those specimens; however, it is not the dominant magnetic carrier for the sand samples. The cooling curves have lower susceptibility values than the heating curves, which indicate the conversion of ferrimagnetic maghemite to weakly magnetic hematite.
The χ–T curves for the clay and silt samples (Figs. 5D–5F) show a major drop to a minimum at ∼680 °C, indicating that hematite is the main magnetic carrier. In addition, a decrease in χ at ∼120 °C suggests the presence of goethite, while a gentle decrease in χ at 580 °C indicates the presence of magnetite. The absence of a sharp χ increase between 400 °C and 500 °C in the χ–T heating curves (Fig. DR2 in the Data Repository1; Fig. 5) demonstrates that paramagnetic Fe sulfides (Passier et al., 2001; L. Chang et al., 2014b; Fu et al., 2015) are of limited occurrence in the samples from core MH.
The dominance of the magnetic properties of the sand samples by magnetite is further supported by relatively narrow hysteresis loops that are closed below ∼0.5 T (Figs. 5A–5C). By contrast, the clay and silt samples have pronounced wasp-waisted hysteresis loops that are unclosed below 0.8 T (Figs. 5D–5F), which is consistent with the presence of a significant proportion of hematite within mixtures of hematite and magnetite (Roberts et al., 1995).
The rock magnetic results show that the dominant magnetic carrier of the sand samples is magnetite, indicating that AF demagnetization is appropriate. The demagnetization results also show that the sand samples can be completely demagnetized at 80 mT (Figs. 6A–6C), suggesting the presence of a significant magnetite component. However, the clay and silt samples could not be completely demagnetized by AF demagnetization, and the high-temperature component persisted up to 680 °C during thermal demagnetization (Figs. 6D–6F). There were no apparent directional differences between the components demagnetized from 450 °C to 580 °C and from 610 °C to 680 °C (Fig. 6), indicating that the magnetic remanence carried by both magnetite and hematite was acquired during or soon after deposition (Zhu et al., 2005). The similarity of the remanent directions defined by the high-temperature components (450–680 °C), and by the components demagnetized between 30 mT and 80 mT during AF demagnetization (Figs. 6D–6F), indicates that reliable ChRM directions for the silt and clay samples could be obtained by both AF and thermal demagnetization. For all the samples, the low field or low temperature components (carried either by low coercivity magnetite grains and/or present as a viscous overprint) could be removed below 10–30 mT or below 300–450 °C (Fig. 6). Thus, for most samples, the ChRM component could be separated between 30 mT and 80 mT or between 450 °C and 680 °C (Figs. 6A–6C).
Four criteria were used to estimate the ChRM directions: (1) Ambiguous or noisy orthogonal demagnetization diagrams were rejected; (2) ChRM directions were determined by at least four successive demagnetization steps starting from at least 30 mT or 450 °C; (3) maximum angular deviation was <15°; and (4) the absolute value of the inclination was >5°. Using these criteria a total of 1928 specimens (65% of the total of 2988 samples) gave reliable ChRM directions, with 1460 specimens treated by AF demagnetization and 468 specimens by thermal demagnetization. Since there was no control on the azimuth of the core during drilling, the magnetic polarity sequence was constructed solely from the inclination.
Most of the polarity reversals are characterized by large changes in declination (∼180°, Fig. 7). The NRM within normal magnetozones is not higher than that within reversed magnetozones (Fig. DR1; see footnote 1), implying that post-depositional remagnetization is unlikely. Moreover, we compiled additional magnetic parameters (MDF, medium destructive field), Bc (coercivity), and Mrs/Ms (ratio of saturation remanent magnetization to saturation magnetization) (Fig. DR1) and found no significant differences between normal and reverse polarity samples. The mean positive and negative inclinations were calculated using the Arason-Levi method based on maximum likelihood estimation (Arason and Levi, 2010). The AF demagnetized specimens (Table 1) have a mean normal inclination of 52.25° and a mean reversed inclination of –53.45°; and the directions isolated by thermal demagnetization have a mean normal direction of I = 50.83° and a reversed direction of –51.05°. The difference between the normal and reversed inclinations is <2°, suggesting that they are antipodal and they pass the reversal test (McFadden and McElhinny, 1990). The mean inclination values indicate slight inclination shallowing compared to the inclination of the axial dipole field at the study site [∼59.5° during 3.1–0 Ma and ∼60.9° during 3.1–11.9 Ma, according to the apparent polar wander path of Eurasia (Besse and Courtillot, 2002)]. This agrees with earlier reports of low inclinations during the Pliocene and Pleistocene in the NE Tibetan Plateau (Cogné et al., 1999; Dupont-Nivet et al., 2002; Yan et al., 2005; Zhang et al., 2012). The occurrence of inclination shallowing is also evidence for the detrital origin of the calculated ChRM components (Dupont-Nivet et al., 2002; Zhang et al., 2012, 2014). The minor inclination shallowing does not affect the establishment of a geomagnetic polarity stratigraphy. Furthermore, declinations of the corrected ChRMs are consistent with the inclinations and have passed the reversal test (Fig. DR1 and Fig. DR4; see footnote 1). Overall, therefore, the measured inclinations are deemed reliable for reconstructing a geomagnetic polarity sequence.
During the establishment of the polarity sequence, at least three successive specimens of the same polarity were used to define a magneto-zone. A total of 36 magnetozones were determined (Fig. 7), with 18 normal polarities (N1–N18) and 18 reversed polarities (R1–R18).
Age of the Sedimentary Sequence
The drilling site is located in the depocenter of the Jiudong Basin and is experiencing sediment deposition at the present day; therefore, we assume an age of zero for the top of the sedimentary sequence. The magnetic polarity sequence of the MH core matches well with the Geomagnetic Polarity Time Scale (Gradstein et al., 2012) and records a continuous magnetic polarity sequence from C3An.2n to C1n (Fig. 7). The topmost predominantly normal polarity zone N1 should thus correlate to Chron C1n (Brunhes normal polarity epoch), and the sequence comprising the reversed zones R1–R9 correlates to Chrons C1r–C2r (Matuyama reverse polarity epoch). The long normal polarity zone N10–N12, with two reversed polarity zones (R10, R11), correlates to Chron C2An (Gauss normal polarity chron). The five reversed polarity zones R12–R16 are readily correlated with the five reversed polarity intervals within the Gilbert reversed polarity epoch. Consequently, the four short normal polarity intervals N13–N16 correlate to the four normal polarity intervals of Chron C3n (Chrons C3n.1n-C3n.4n). The sequence N17–N18 is readily correlated to Chrons C3An.1n–C3An.2n within the Epoch-4. Based on the magnetostratigraphy (Fig. 7), the basal age of the MH core is estimated as ca. 7.0 Ma, and the age of sedimentary units 1–4 are estimated to be 7.0–5.2 Ma, 5.2–4.1 Ma, 4.1–3.0 Ma, and 3.0–0 Ma, respectively.
Sediment Accumulation Rate
In determining the sediment accumulation rate (SAR), we only used the major polarity boundaries of chrons and subchrons and ignored possible events and excursions. The SAR plot (Fig. 8) shows an increasing trend from ∼30 mm/k.y. to ∼180 mm/k.y. since 7 Ma. Within this overall increasing trend, there is substantial decrease in the SAR from 107 mm/k.y. to 46 mm/k.y. at ca. 3 Ma, and a lower SAR persists for ∼1.2 m.y. (3.0–1.8 Ma), spanning five magnetozones (Fig. 8). Clearly, the age of a change in SAR will not necessarily coincide with a magnetic polarity transition and the listed ages of the SAR changes are approximations. Despite the inability to resolve SAR variations on a finer scale than provided by the polarity reversal boundaries, we consider that both the long-term trend of the SAR, and the 1.2 m.y. interval of lower SAR values are reliable.
Reversed Polarities within Chron C1n
At least seven reversed polarity intervals are identified within Chron C1n (Fig. 7). These reversed directions may reflect genuine geomagnetic reversals, or they may result from post-depositional disturbances or diagenetic remagnetization (Hu et al., 1998; Fu et al., 2008). Diagenetic remagnetization can be excluded for the following reasons: (1) the χ–T curves (Fig. DR2) show that the ferrimagnetic greigite and pyrrhotite are not significantly present; (2) the samples from these reversed intervals have similar MDF and NRM values to neighboring samples of normal polarity and with the same lithology (Fig. DR1); (3) ΔGRM/ΔNRM (ΔGRM—the gyromagnetic remanence intensity acquired between final and minimum values of the remanence intensity during alternating-field demagnetization (AfD), ΔNRM—the difference of initial and minimum values of the remanence intensity during AfD) (Fig. DR1) also indicates that greigite is absent from the reversed polarity intervals.
Among the seven reversed intervals, three intervals (r1, r4, and r7) are characterized by small changes in declination (ΔDecl. <<180°, Fig. 7), indicating that they are probably not geomagnetic excursions. Ge et al. (2012) suggested that disturbed sedimentary intervals can be recognized by anomalous values of K3-Inc.. In our stratigraphy, the three short-lived directional anomalies (r1, r4, and r7) have relatively low K3-Inc. values (<60°), indicating that they are likely to reflect post-depositional disturbance (Wang et al., 2004; Duan et al., 2016). By contrast, r2 (20.1–21.2 m), r3 (27.8–28.9 m), r5 (56.9–57.8 m), and r6 (114.5–115.8 m) are characterized by large changes in declination (ΔDecl. close to 180°) and steep K3-Inc. (>60°), indicating that they probably represent intervals of reversed polarity within the Brunhes normal polarity epoch. Based on the SAR within N1, the median ages of the four intervals are estimated to be 116 ka for r2 (coeval with the Blake excursion of 110–120 ka), 160 ka for r3 (close to the Iceland Basin excursion of 188 ka), 323 ka for r5 (close to the Calabrian Ridge 1 excursion of 318 ka), and 646–653 ka for r6 (close to the Stage 17 of 670 ka).
A total of 14 excursions have been confirmed during the Brunhes Chron (Singer, 2014), while only four short-lived intervals are identified in the MH core. The absence of other excursions may be the result of the lower recovery rate in the upper part of the core (average recovery rate of 85.0% within 0.0–100.3 m) and/or the presence of short-duration sedimentary hiatuses.
Uncertainties in the Interpretation of the Magnetostratigraphy
The variations in SAR, which are based on the average rate within individual polarity intervals, include the abrupt changes represented by N1 and N4 which correspond with the base of C1n and C1r.1n. They may result from possible biases in the correlation of the magnetostratigraphy. In addition to the preferred correlation presented in the Results section, alternative possibilities need to be considered. Zone N1 is undoubtedly correlated to the Brunhes epoch. In the Matuyama epoch, an alternative possibility is that N4 is correlated to C2n, and consequently N6-N8 correlate to the Gauss epoch, N9-N12 correlate to the four normal zones within the Gilbert epoch, and N13-N14 correlate to C3An.1n-C3An.2n within Epoch-4. However, this alternative correlation produces much more variable SARs (Fig. DR3; see footnote 1) and results in anomalously high SARs within C1n, C2An.1r, and C3n.2n, and anomalously low SARs within C1r.1r, C2An.1n, C2An.1n–C3n.1n, and C3n.2r, compared to the neighboring intervals and to our preferred correlation (Fig. DR3). Thus, this correlation can be excluded, and our preferred magnetostratigraphic correlation is deemed the most reliable according to the criteria of best fit accumulation rate (Talling and Burbank, 1993).
Sedimentary Evolution of the Jiudong Basin
Since 7 Ma, the sedimentary facies document the gradual coarsening-upward evolution of the sedimentary system, from a shallow lake environment to a braided river environment (Fig. 3). The timing of the shift from shallow lake/delta front to delta plain, at 5.2 Ma, is roughly synchronous with the transition from the Shulehe Formation to the Yumen Formation (Fig. 9), which has been determined by magnetostratigraphic and cosmogenic burial dating (Fang et al., 2005b; Zhao et al., 2017) in the Laojunmiao section in the southern edge of the basin (Fig. 2). In the Laojunmiao section in the Jiuxi Basin, the upper part (7–5 Ma, Zhao et al., 2017) of the Shulehe Formation is interpreted as a fan delta-braided river system, which changed to an alluvial fan system represented by the Yumen Formation (Fang et al., 2005b; Li et al., 2014) (Fig. 9). The transitions of sedimentary facies at 5.2 Ma and 4.1 Ma are not readily correlated with large climatic oscillations and/or global cooling evident in the deep-sea oxygen isotope records (Zachos, 2001) (Fig. 8). This indicates that the stepwise upward-coarsening of the sediments was not driven predominantly by climate change. The correlation between the sediments of the basin edge, close to the Qilian Shan, and in the basin center (Fig. 9) indicates rather that the evolution of the sedimentary environment in the basin was closely related to the growth of the relief of the Qilian Shan.
The sedimentary facies changed from delta to meandering river at ca. 4.1 Ma, indicating another interval of progradation of the sedimentary system which was caused by the continuous relief growth of the Qilian Shan. This sedimentary facies transition is also documented in the Laojunmiao section (Fig. 9) (Fang et al., 2005b; Zhao et al., 2017) and the Niugetao section (Chen et al., 2006). Around 3 Ma, the meandering river regime changed to a braided river regime, which persisted to the present. At the mountain front, the Laojunmiao and Yumushan sections record continuous conglomerate deposition since ca. 3.6 Ma (Fang et al., 2005b; Liu et al., 2010) (Fig. 9). This evidence probably indicates that the high relief of the mountain range along the Qilian Shan developed at 3.6–3.0 Ma. Subsequently, the relief of the Qilian Shan range has not changed significantly, indicating that the evolution of the relief has attained a steady-state in which the uplift rate is balanced by the erosion rate (Hetzel, 2013; Hu et al., 2010).
The sustained growth of the relief of the Qilian Shan before 3 Ma is also documented by the remarkable increase in the sedimentation rate in the MH core (Fig. 8), in the Laojunmiao section (Fang et al., 2005b), and in the Wenshushan section (Zhao et al., 2001). An apatite (U-Th)/He age/elevation transect along the Qilian Shan range (Zheng et al., 2010) records an interval of rapid exhumation since 9.5 Ma, and apatite fission track ages from the Jiuxi Basin reveal an overturn pattern at ca. 10.5 Ma (Zheng et al., 2017). Both lines of evidence indicate that the onset of the North Qilian fault and the growth of the Qilian Shan occurred since ca. 10 Ma. The increase of the mountain relief caused by the uplift of the Qilian Shan caused the increase in the grain size and the sediment supply to the Jiudong Basin, hence the upward-coarsening trend within the sedimentary sequence, which records the progradation of the sedimentary system in the Jiudong Basin from ca. 10 Ma to 3 Ma (Fig. 9).
Basinward Growth of the Qilian Shan
The most prominent feature of the stratigraphy of core MH is the out-of-phase change in sediment accumulation rate and grain size between 3.0 and 1.8 Ma (Fig. 8). The SAR decreases substantially at 3.0 Ma, when the sedimentary environment changed from a meandering river system (unit 3) to a braided river system (unit 4), as indicated by the coarsening of the grain size (from silts and sands to sands and gravels). SARs within terrestrial basins are controlled primarily by the rate of sediment supply and the accommodation space (Heermance et al., 2007). The accommodation space is created either by tectonically induced subsidence beneath an uplifted block (Jordan, 1981), or by uplift that traps sedimentation within a piggyback basin (DeCelles and Giles, 1996). The lower SAR during 3.0–1.8 Ma can be explained by three possibilities: (1) significantly reduced sediment supply during this time interval, (2) spatial changes in the course of the braided river system, or (3) a reduction in the accommodation space resulting from a decreased subsidence rate (basement movements).
In the late Cenozoic Wenshushan section (Zhao et al., 2001), along the southern margin of the Jiudong Basin, the SAR remained constant at ∼0.1 mm/yr from 4.5 Ma to 0.9 Ma, with no reduction during 3.0–1.8 Ma; and in the Yumushan section (Liu et al., 2011), the SAR remained almost constant at ∼0.1 mm/yr from 3.6 Ma to 1.8 Ma. This evidence suggests that the sediment supply from the Qilian Shan did not decrease during 3.0–1.8 Ma. Moreover, the occurrence of clastic progradation at ca. 3 Ma also indicates no reduction of the sediment supply. The MH site is located in a floodplain region rather than in a piedmont fan region (Fig. 2), and the sediments at the site were probably supplied by the Maying River, Bailang River, and other small fluvial systems in between (Fig. 2). Thus, the changes in the courses of the Maying and Bailang rivers would probably have caused an alternation of sediment types rather than a change in the sediment accumulation rate.
According to the basin-infilling model (Heller and Paola, 1992; Paola et al., 1992), an out-of-phase relationship between sedimentation rate and grain size which lasts for more than 1 m.y. is probably caused by a reduction in the accommodation space in the Jiudong Basin at the location of the MH core. A reduction in the accommodation space in flexure basins may be caused by uplift within the basin interior as the deformation front propagates into the region (Heermance et al., 2007; H. Chang et al., 2014a; Wang et al., 2016). Furthermore, comparison of the sedimentary records of the MH core and the Wenshushan and Yumushan sections reveals that their SARs were similar (0.09–0.12 mm/yr) before 3.0 Ma (back to ca. 4.5 Ma for the MH core and the Wenshushan section, and back to ca. 3.6 Ma for the Yumushan section). This similarity suggests a spatially uniform rate of aggradation in the Jiudong Basin; while during 3.0–1.8 Ma, the lower rate of ∼0.05 mm/yr in the MH core (compared to the rate of ∼0.1 mm/yr in the southern part of the basin) suggests the occurrence of uplift in the basin center (basement movements).
Several faults bordering the mountain ranges around the Jiudong and the Jiuxi basins were initiated in the late Pliocene. At the southern margin of the Jiuxi Basin, the magnetostratigraphy (from 13 Ma to 0 Ma) and results of cosmogenic burial dating (Zhao et al., 2017) of sites along the Laojunmiao anticline indicate that the first angular unconformity developed during 3.1–2.6 Ma and growth strata started to develop at ca. 3 Ma (Chen et al., 2006). An age range of 3.8–2.8 Ma is also estimated for the onset of the thrust fault along the Hei Shan (Liu, 2017) in the northern Jiuxi Basin. At the southeastern margin of the Jiudong Basin, estimation of the fault slip rate by cosmogenic exposure dating gave a relatively lower rate of 0.5–0.8 mm/yr for the thrusting of the Yumushan fault, and a new age range of 2.8–4.6 Ma was proposed for the start of the growth of the Yumu Shan (Palumbo et al., 2009). Furthermore, in the western Yumu Shan, the magnetostratigraphy of the Cenozoic sediments has constrained the angular conformity to the age range of 2.9–2.6 Ma (Liu et al., 2011), which probably corresponds to the initiation of the folding of the Yumu Shan range. The coincidence between the timing of fault growth around the basin and the out-of-phase deposition inside the basin at ca. 3 Ma suggests they were related.
From the foregoing, we conclude that the thrust system along the northern margin of the Qilian Shan began to propagate into the basin along a southward-dipping décollement at ca. 3 Ma (Fig. 10). The Jiuxi and Jiudong basins were proposed as a foreland basin that was active during the Jurassic and Early Cretaceous (Wang and Coward, 1993) which accumulated several kilometers of sediments above the pre-Mesozoic basement. In the late Cenozoic, this re-activated foreland basin (Wang and Coward, 1993) continued to accumulate sediments due to the thrusting of the North Qilian Shan fault (Zheng et al., 2010). The propagation of the thrust fault from the North Qilian Shan fault to the north probably occurred along a décollement in the Precambrian basement, exposing both the Precambrian basement and Cenozoic sediments along the Jintanan Shan. A seismic reflection profile (Gao et al., 1999) also shows a southward-dipping reflection layer at a shallow depth (5–7 km). Slippage along the southward-dipping décollement caused the growth of the Laojunmiao anticline, the Hei Shan and the Yumu Shan, and caused the uplift in the basinal area, inducing the foreland basin to change to a piggyback basin. To the north, the southward-dipping Jintanan Shan fault (Zheng et al., 2013a) probably defines the northern front of the thrust system. Alternatively, the uplift of the basin may have resulted from faulting along a deep-seated thrust. From an analysis of the seismic reflection data, Gao et al. (1999) proposed a southward-dipping thrust fault that borders the Alashan terrane and the Qilian Shan terrane. A study of the fault behavior in the east of the Jiudong Basin indicated that the South Heli Shan fault is a northward-dipping thrust fault and the exposed crystalline rock belongs to the Alashan terrane (Zheng et al., 2013a) (Figs. 1 and 2). Thus the uplift of the basin may have been caused by the activation of the surrounding crustal-scale faulting.
Irrespective of the type of tectonic model applicable to the evolution of the Jiudong Basin, the fluctuations in the SAR suggest that the deformation started at ca. 3 Ma. After 1.8 Ma, the SAR started to increase again, which probably indicates the beginning of the development of the mountain relief (the Jinatanan Shan and the Heli Shan) in the north (Fig. 10). As mentioned previously, the accommodation space was created by uplift (DeCelles and Giles, 1996), resulting in sedimentation being confined within a piggyback basin in the Jiudong Basin rather than being distributed within a wide region across the Jintanan Shan and the Heli Shan. The date of 1.8 Ma is roughly coincident with the onset of the growth of mountains to the north of the basin. Calculation of the age of initiation of mountain growth (Zheng et al., 2013b) gave an estimate of 1.5–1.6 Ma for the Jintanan Shan; however, the significantly younger age of ca. 0.23 Ma has also been proposed (Hetzel et al., 2004). The initiation of the uplift of the Heli Shan is estimated at 4–1 Ma based on the fault slip rate (Zheng et al., 2013a).
Basinward propagation of fault systems at around 3 Ma has also been reported for other regions in the NE Tibetan Plateau. A higher SAR at 3.6–2.6 Ma occurred in the center of the Qaidam Basin (Zhang et al., 2014), which is coeval with the lower SAR, unconformities, and coarse-grained deposition close to the basin edge (Fang et al., 2007; Zhang et al., 2013). Furthermore, the growth of folding in the basin center started at ca. 3 Ma in the Qaidam Basin (Heermance et al., 2013), and the period of most intense shortening started at ca. 2.5 Ma (Wei et al., 2016). From this evidence, we propose that the Qaidam Basin experienced basinward propagation of the fault system at ca. 3 Ma. This tectonic event was also recorded in the Guide Basin by coarse-grained sediment deposition and a lower SAR at 3–2 Ma (Fang et al., 2005b); in the Linxia Basin by deposition of the Jishi conglomerate and basin deformation at 3.6–2.6 Ma (Fang et al., 2003); and in the Lake Qinghai by an increase in sediment grain size and a lower SAR from 3 Ma to 2 Ma (Fu et al., 2013). In the NE Tibetan Plateau, the coincidence of the widely distributed deformation in the basin region at ca. 3 Ma (or during 3.6–1.8 Ma) indicates that the proposed tectonic event at ca. 3 Ma (the Qing-Zang Movement) (Li et al., 1996) was a time of basinward growth of the fold and fault system rather than the intense uplift of the principal mountain ranges.
Several gravel or pebbly sand layers (Fig. 8) are present in the MH core, which have larger grain sizes compared with the surrounding beds. Based on the depth of the deposits and the mean sedimentation rate, the ages of these seven sedimentary events (E1–E7) are estimated at ca. 6.0 Ma, ca. 3.3 Ma, ca. 2.5 Ma, ca. 0.76 Ma, ca. 0.63 Ma, ca. 0.26 Ma, and ca. 0.18 Ma, respectively. The deposition of these coarse-grained layers occurred during intervals of climatic cooling and/or climatic transition (Fig. 8). E1 is synchronous with latest Miocene glacial development at ca. 6 Ma, and E2 with Pliocene glaciation at ca. 3.3 Ma (Larsen et al., 1994; Amidon et al., 2017); E3 (ca. 2.5 Ma) coincides with the establishment of Northern Hemisphere ice sheets as well as with the Pliocene–Pleistocene climatic transition (Sosdian and Rosenthal, 2009); E4 and E5 are synchronous with the occurrence of maximum glaciation (Wangkun Glaciation) on the Tibetan Plateau during marine isotope stage (MIS) 18–16 (Zhou et al., 2006); and E6 and E7 are synchronous with the penultimate glaciation during MIS 8–6 (Zhou et al., 2006).
These relatively thin coarse-grained intervals represent the transport of coarse grains over distances far from the deformation front. Progradation of this long-distance-transported fluvial gravel into actively subsiding sedimentary basins is unusual in the stratigraphic record (Heller et al., 2003). These thin intervals require flows that are deep enough, and/or slopes that are steep enough, to achieve and maintain a critical shear stress over large areas that can mobilize and transport coarse sediment. Given the thin, abrupt nature of these deposits, and the synchrony with the timing of cold glacial periods and climatic transitions, it might be concluded that climatic cooling and shifts from a previous relatively constant climate to an oscillating climate would be accompanied by a significant increase in erosion rates and a reduction in vegetation coverage in the North Qilian Shan. Indeed, this is indicated by previous research in the regions surrounding the Tibetan Plateau and the Tian Shan (Zhang et al., 2001; Charreau et al., 2011; Li et al., 2014). Increased physical erosion rates and vegetation degradation in the sediment source area might be expected to increase the influx of coarse debris into the basin. Previous research has shown that glacial “cold” periods were synchronous with phases of aridity in the NE Tibetan Plateau (Li et al., 2014), implying that the increased occurrence of short-term high-intensity storms promoted the increased mobilization of coarse debris (Molnar, 2001). Thus, we propose that coarse debris was deposited in the central-distal foreland basin during the timing of glacial “cold” periods and during climatic transitions.
A new stratigraphic investigation of the MH drill core from the Jiudong Basin provides a robust magnetostratigraphy for the sedimentary evolution of the foreland basin and builds a reliable and continuous sedimentary sequence from 7 Ma to 0 Ma around the NE Tibetan Plateau. Sedimentary facies analysis, together with the magnetostratigraphy, reveals a coarsening-upward sequence which represents an environmental transition from delta front/shallow lake to braided rivers since 7 Ma. Several thin layers of coarse sediment within the sequence represent the long-distance transport of coarse material which was a response to global climate cooling and climatic transitions. The basinward encroachment of depositional systems in the Jiudong Basin occurred at 5.2 Ma, 4.1 Ma, and 3.0 Ma, and is attributed to the growth of the relief of the Qilian Shan, which attained a steady-state at ca. 3 Ma. Tectonic deformation was propagated to the foreland basin after 3 Ma, which induced the uplift of the basin region.
We are grateful to Yaoyang Lu, Xilin Cao, Yunxia Jia, Lianyong Guo, and Ying Chen for their assistance with the hard work in the field and laboratory. Thoughtful reviews from Massimo Mattei, Emilio L. Pueyo, Jaume Dinarès-Turell, and another reviewer improved the science of the manuscript. We thank Dr. Jan Bloemendal for English improvement. This research is financially supported by the National Natural Science Foundation of China (NSFC grant nos. 41730637, 41571003, and 41471009) and the Key Project of the Major Research Plan of the NSFC (grant no. 91125008).