Volume 46, Issue 15 p. 8741-8752
Research Letter
Free Access

Internal Drainage Has Sustained Low-Relief Tibetan Landscapes Since the Early Miocene

Zhongpeng Han

Zhongpeng Han

State Key Laboratory of Biogeology and Environment Geology, China University of Geosciences, Beijing, China

School of Earth Sciences and Resources, China University of Geosciences, Beijing, China

Search for more papers by this author
Hugh D. Sinclair

Hugh D. Sinclair

School of GeoSciences, The University of Edinburgh, Edinburgh, UK

Search for more papers by this author
Yalin Li

Corresponding Author

Yalin Li

State Key Laboratory of Biogeology and Environment Geology, China University of Geosciences, Beijing, China

School of Earth Sciences and Resources, China University of Geosciences, Beijing, China

Correspondence to: Y. Li,

[email protected]

Search for more papers by this author
Chengshan Wang

Chengshan Wang

State Key Laboratory of Biogeology and Environment Geology, China University of Geosciences, Beijing, China

School of Earth Sciences and Resources, China University of Geosciences, Beijing, China

Search for more papers by this author
Zui Tao

Zui Tao

School of GeoSciences, The University of Edinburgh, Edinburgh, UK

Search for more papers by this author
Xinyu Qian

Xinyu Qian

State Key Laboratory of Biogeology and Environment Geology, China University of Geosciences, Beijing, China

School of Earth Sciences and Resources, China University of Geosciences, Beijing, China

Search for more papers by this author
Zijie Ning

Zijie Ning

State Key Laboratory of Biogeology and Environment Geology, China University of Geosciences, Beijing, China

School of Earth Sciences and Resources, China University of Geosciences, Beijing, China

Search for more papers by this author
Jiawei Zhang

Jiawei Zhang

State Key Laboratory of Biogeology and Environment Geology, China University of Geosciences, Beijing, China

School of Earth Sciences and Resources, China University of Geosciences, Beijing, China

Search for more papers by this author
Yixiong Wen

Yixiong Wen

State Key Laboratory of Biogeology and Environment Geology, China University of Geosciences, Beijing, China

School of Earth Sciences and Resources, China University of Geosciences, Beijing, China

Search for more papers by this author
Jie Lin

Jie Lin

State Key Laboratory of Biogeology and Environment Geology, China University of Geosciences, Beijing, China

School of Earth Sciences and Resources, China University of Geosciences, Beijing, China

Search for more papers by this author
Baosen Zhang

Baosen Zhang

State Key Laboratory of Biogeology and Environment Geology, China University of Geosciences, Beijing, China

School of Earth Sciences and Resources, China University of Geosciences, Beijing, China

Search for more papers by this author
Ming Xu

Ming Xu

State Key Laboratory of Biogeology and Environment Geology, China University of Geosciences, Beijing, China

School of Earth Sciences and Resources, China University of Geosciences, Beijing, China

Search for more papers by this author
Jingen Dai

Jingen Dai

State Key Laboratory of Biogeology and Environment Geology, China University of Geosciences, Beijing, China

School of Earth Sciences and Resources, China University of Geosciences, Beijing, China

Search for more papers by this author
Aorigele Zhou

Aorigele Zhou

State Key Laboratory of Biogeology and Environment Geology, China University of Geosciences, Beijing, China

School of Earth Sciences and Resources, China University of Geosciences, Beijing, China

Search for more papers by this author
Huimin Liang

Huimin Liang

Chengdu Center, China Geological Survey, Chengdu, China

Search for more papers by this author
Shuo Cao

Shuo Cao

State Key Laboratory of Biogeology and Environment Geology, China University of Geosciences, Beijing, China

School of Earth Sciences and Resources, China University of Geosciences, Beijing, China

Search for more papers by this author
First published: 15 July 2019
Citations: 40

Abstract

The timing of formation of the low-gradient, internally drained landscape of the Tibetan Plateau is fundamental to understanding the evolution of the plateau as a whole. Well-dated sedimentary records of internal drainage of rivers into lakes are used to reveal the timing of this evolution. Here we redate the youngest continental sedimentary successions of central Tibet in the Lunpola Basin and propose a new age range of ca. 35 to 9 Ma, significantly younger than previously thought. We demonstrate long-standing internal drainage in central Tibet since the late Eocene and stable sedimentary environments, source regions, and low topographic relief since at least the early Miocene. We suggest that sediment aggradation of internal drainage and reduction of hillslope gradients by erosion dominate the formation of low-relief landscapes and that the late Cenozoic drainage basins in central Tibet developed in response to flow in the lower crust and/or mantle lithosphere.

Key Points

  • Robust age constraints of the youngest continental stratigraphic unit from the Lunpola Basin in central Tibet are reported
  • Aggradation and erosion of internal drainage dominated the formation of low-relief topography in central Tibet by the early Miocene times
  • Late Cenozoic drainage basins in central Tibet developed in response to flow in the lower crust and/or mantle lithosphere

Plain Language Summary

Internal drainage of rivers into lakes is a characteristic of the high plateaus of the world and, most notably, the Tibetan Plateau. Internal drainage generates local perched base levels for Tibetan rivers, enabling geomorphic isolation from the rapidly incising rivers of the Himalaya and surrounding regions. However, the question of when the low-relief plateau topography was initiated has been largely ignored, and its formation mechanism is controversial. Here we report a detailed investigation in the Lunpola Basin of central Tibet and propose a new depositional age range of ca. 35–9 Ma. We demonstrate that the internal drainage kept eroding the mountain ranges and filling the surrounding lowlands since at least the late Eocene. By no later than the early Miocene, a gentle landscape formed in central Tibet. The late Cenozoic basins in central Tibet developed in response to deep crustal or mantle flow and associated upper crustal deformation.

1 Introduction

The modern-day internally drained hinterland of the Tibetan Plateau has a relatively low-relief topography, in stark contrast to the high-relief externally drained Himalaya and surrounding mountain ranges (Fielding et al., 1994; Figure 1a). As the largest orogenic plateau, the evolution of this striking geomorphic feature is crucial for understanding the Earth's surface and underlying dynamic processes. While there has been extensive discussion of the temporal and spatial evolution of the plateau, the timing and mechanism of development of the low-relief topography remain a challenge (Clark & Royden, 2000; Liu-Zeng et al., 2008). The isolation of elevated river base levels as a result of rivers terminating in lakes within the interior plateau is a requirement for maintaining low-relief topography through sediment aggradation of river channels and reduction in hillslope gradients by weathering and erosion (Liu-Zeng et al., 2008). Detailed dating of the sedimentary successions that record internal drainage is essential for understanding the plateau's evolution. However, few studies have tested the geological evidence for the longevity of internal drainage in central Tibet and, hence, for the low-relief topography.

Details are in the caption following the image
(a) Major geological features and modern internally drained basins (gray lines) of central Tibet. The light blue area is the internal drainage of the Tibetan Plateau. The insert is a topographic cross section along line AA'. (b) Hillshade image of the Lunpola-Silin Co-Nima area with major catchments (black lines), Mesozoic strata, and igneous rocks (adapted from Pan et al., 2011). (c) Geological map of Lunpola Basin (adapted from Chen et al., 2003; Qu et al., 2011; Wang et al., 2012).

Cenozoic continental sediments are preserved in several basins along the Bangong-Nujiang suture of central Tibet, which experienced large-scale upper crustal shortening and thickening during the Late Cretaceous to Paleogene (Kapp et al., 2005). The present deformation pattern of central Tibet is dominated by north-south-oriented extensional structures, with northeast-striking left-slip faults and northwest-striking right-slip faults (Royden et al., 2008). These strike-slip faults exert key control on the elongated geometry of Cenozoic basins such as the Lunpola and Nima Basins, which record the evolution of paleoenvironments and sediment routing in central Tibet (Taylor et al., 2003).

Here we present our field observations from the Lunpola Basin, apply geochronological, sedimentological, and provenance techniques to establish the stratigraphic framework and reconstruct depositional systems and drainage networks, and then compare these results with the modern drainage of central Tibet. Our newly dated continental sedimentary successions of central Tibet are significantly younger than previously thought, thus providing new insights into the basin evolution and the formation of the low-relief landscape that characterizes the plateau.

2 Geological Background

The Lunpola Basin and adjacent Nima Basin are the two largest Cenozoic basins in central Tibet, and between them is the largest saline lake of Tibet, Silin Co (Figure 1b). The two largest interior rivers of Tibet (Zhajia Zangbo and Bocang Zangbo) originate in the Qiangtang and Lhasa terranes and terminate in the Lunpola-Silin Co-Nima area (Figure 1b). The Bangong-Nujiang suture separates the Qiangtang and Lhasa terranes; the southern Qiangtang terrane is bounded to the north by the Tanggula Shan Range-Qiangtang Central Uplift, and the Gangdese mountains define the northern Lhasa terrane to the south (Figure 1a and Text S1 of the supporting information).

More than 4-km-thick Cenozoic fluvial to lacustrine deposits in the Lunpola Basin unconformably overly the Mesozoic basement, which were divided into the Paleocene to Eocene Niubao Formation and Oligocene to early Miocene Dingqinghu Formation in previous studies (Wei et al., 2017). Sun et al. (2014) proposed a succession of the Dingqinghu Formation ranges from 25.5 to 19.8 Ma based on a single magnetostratigraphic section supported by biostratigraphy and dating of the bentonite layer (He et al., 2012). A mammalian fossil with an age of 18 to 16 Ma was used to define the age of the deep lake deposits of the Dingqinghu Formation in the basin center (Deng et al., 2012), and a tuff layer with an age of 20.6 Ma was found below the mammalian fossil (Mao et al., 2019). However, despite the above constraints, most of the outcrops were assigned a broad age range of Paleocene to Oligocene based on fossil ostracods and palynological assemblages (Xia & Liu, 2008). Thus, the precise depositional ages of these sediments remain uncertain, and stratigraphic correlations with more robust constraints are required.

3 Methods

Sixteen stratigraphic sections of Cenozoic deposits were measured in the Lunpola Basin (Figure 1c). The stratigraphic framework is established by zircon U-Pb ages, facies correlation analyses, fossil-bearing beds, and lateral correlation of outcrops in the field (Figures 2, 3, and S1). Paleocurrent directions were acquired from the entire succession from planar and trough cross-bedding, current ripple lamination, imbricated clasts in conglomerates, and sole marks on beds. Point counting was performed on 92 petrographic thin sections according to the Gazzi-Dickinson method (Ingersoll et al., 1984). Approximately 400 grains were counted in each thin section, and the data were plotted on standard Ternary diagrams with provenance fields established after Dickinson and Suczek (1979). Compositions of conglomerate and modern river gravels were determined by point-counting lithology from measured sections and riverbeds in the Lunpola-Silin Co-Nima area. To ensure that all the counting data are statistically significant and comparable, all countings were performed on the channelized sediments. Fifteen sandstone and tuffite samples were collected from measured sections in the Lunpola Basin for zircon U-Pb dating (Figures 2 and 3 and Text S2); two sandstone samples were collected from measured sections described by DeCelles, Kapp, et al. (2007) in the Nima Basin. Seven modern sand samples were collected from mainstream riverbeds of several large rivers that terminate in the Lunpola-Silin Co-Nima area (Figure 1b and Table S1).

Details are in the caption following the image
(a) Concordia plots (left column) and probability density diagrams (middle column) of <65-Ma zircons from tuffite samples and (right column) YC1σ (2+) ages. MSWD: Mean Standard weighted Deviation. (b) Selected cathodoluminescence images of dated zircons from tuffite samples. (c) Probability density diagrams of <100-Ma detrital zircons from four sandstone samples that show YPP ages.
Details are in the caption following the image
(a) Sedimentary logs of the measured sections. All the sections are correlated by zircon ages (YC1σ (2+) and YPP ages in red), published age constraints (annotation), fossil-bearing beds, facies analyses, and lateral correlation of outcrops. (b) Pie charts show conglomerate clast composition data from the Lunpola (this work) and Nima Basins (DeCelles, Kapp, et al., 2007) and modern riverbeds (this work). n = counted grains. (c) Histograms show the distribution of gravel grain sizes from conglomerate (from measured sections of the Lunpola Basin, see (a) and Figure S1 for the counting locations) and modern riverbeds (see Figure 1b for counting sites). (d) Ternary diagrams are showing how modal sandstone petrographic data from this work compare to coeval sediments in Nima Basin. (e) Comparisons of kernel density estimation spectra of zircon U-Pb ages in this work and published data from modern rivers, Lunpola Basin, Nima Basin, Qiangtang terrane, and Lhasa terrane. Scale change at 250 Ma. n = number of analyses; bracketed red number is the number of grains with ages <250 Ma. Data sources are (Chapman & Kapp, 2017; DeCelles, Kapp, et al., 2007; Fan et al., 2016; Gehrels et al., 2011, and references therein; Guynn et al., 2012; Huang et al., 2017; Lai et al., 2017; Leier, DeCelles, et al. 2007; Leier, Kapp, et al., 2007; Li et al., 2014; Li, Ding, et al., 2017; Li, Guilmette, et al., 2017;  Ma et al., 2017; Ma, Hu, et al., 2018; Meng et al., 2018; Pullen et al., 2008, 2011; Sun et al., 2015; Zhang et al., 2017; Zhang et al., 2011; Zhu et al., 2011). Qt = total quartz; F = total feldspar; L = total lithic grains; CB = continental block; RO = recycled orogen; MA = magmatic arc.

4 Results and Interpretations

4.1 Geochronology

Six tuffites were sampled from different sections in the Lunpola Basin (Figures 2 and 3). Crystal and vitric pyroclasts dominate the microstratified tuffites with interstitial materials of tephra, silt, and clay (Figure S2a). All tuffite samples yield Eocene to Miocene zircon U-Pb ages with more than 50% of measurements yielding ages <30 Ma (Figure 2a). These young zircons are euhedral, displaying well-developed prismatic crystal forms and clear oscillatory zoning (Figures 2b and S2b). We interpret the euhedral zircons as first-cycle zircons that originated from the adjacent coeval volcanic eruptions; hence, the age of the youngest population must be near the actual depositional ages of the sampled succession. In this study, YC1σ (2+) ages (weighted mean ages of two or more grain ages overlapping in age at 1σ; Dickinson & Gehrels, 2009) are adopted to constrain the maximum depositional ages of tuffites, which are 9.22 ± 0.13 Ma (15H1-254U), 16.77 ± 0.34 Ma (14H8-370U), 17.59 ± 0.32 Ma (13H4U), 21.10 ± 0.42 Ma (14H4-128U), 22.57 ± 0.48 Ma (14H3-48U), and 35.31 ± 0.93 Ma (ZP15U1; Figure 2a and Table S2). Notably, four tuffite-bearing successions were previously assigned as Paleocene to Eocene in age (Ma, Wang, Wang, Li, et al., 2015), namely, Sections 15H1, 14H8, 14H3, and 14H4 (Figure 1c). The YPP ages (youngest graphical detrital zircon age peak on age-probability plot, Dickinson & Gehrels, 2009) are used to constrain the maximum depositional ages of four sandstone samples that yield the youngest age clusters with Cenozoic ages, namely, 22.8 Ma (ZP03), 38.5 Ma (13H14), and 23.9 Ma and 26.5 Ma (15H2; Figure 2c), and Sections ZP03 and 15H2 were also previously regarded as Paleocene to Eocene in age (Ma, Wang, Wang, Li, et al., 2015).

4.2 Stratigraphy

Deposits of Sections 13H5, 13H4, 13H2, and 12H1 are dominated by hundred-meter thick sequences of structureless dark gray to green argillaceous rocks (Figure 3a), including thin-bedded siltstones, mudstones, marlstones, and paper-thin oil shales with fossil fish (Figures 3a, S3a, S3b, S4a, and S4b). Sections ZP15, 14H4, 15H1, and ZP03 are dominated by 10- to 200-m-thick sequences of mottled, yellow to reddish, laterally continuous, upward-fining, and coarsening packages of conglomerates and sandstones (Figures 3a and S4c). Gravels from pebble to boulder matrix-supported conglomerates are angular to subangular, disorganized, and unstratified (Figure S3c). Clast-supported pebble conglomerates are well sorted with imbricated clasts. Planar and trough cross-bedded coarse- to fine-grained pebbly sandstones are intercalated with rippled siltstones (Figure S3d). Massive and laminated, rippled, and bioturbated argillaceous rocks with soft-sediment deformation structures dominate the upper part of these sections (Figures S3e and S3f). Alternating argillaceous and coarse clastic rocks dominate the other eight sections (Figure 3a). Both upward-fining and upward-coarsening successions comprise interbedded conglomerates, sandstones, siltstones, marlstones, and mudstones (Figures S4d and S4e). Conglomeratic beds are clast-supported with well-sorted, subrounded to rounded, imbricated gravels and occasionally sharp erosional surfaces underlain by red nodule-bearing paleosol (Figures S4f and S4g). Planar and trough cross-beddings and ripples are common in the sandstones (Figures S3g and S3h). Argillaceous rocks consist of massive and laminated siltstones, marlstones, mudstones, and shales with ripples, rootlets, burrows, mud cracks, rain drops, slumps, and water escape structures (Figures S3i and S3j). Laterally traceable laminated dolomitic limestones are occasionally present (Figure S4h).

The hundred-meter thick sequences of dark argillaceous rocks intercalated with fine-grained sandstones are interpreted as lacustrine deposits (Ma, Wang, Wang, Li, & Hu, 2015; Figures 3a and S1). Dark gray, oil-rich shales with interbedded marlstones and siltstones suggest a semideep or deep lake environment (Wang et al., 2011b). Well-sorted, clast-supported conglomerates with wave rippled siltstones and bioturbated mudstones indicate lake margin settings (Wei et al., 2017). The sequences of conglomerates and coarse- to fine-grained sandstones suggest pulses of coarse deltaic sedimentation (Wei et al., 2017). Subaqueous deltaic environments are evidenced by interbedded siltstones, marlstones, and mudstones, which contain rootlets, burrows, and soft-sediment deformation (DeCelles, Kapp, et al., 2007; Wei et al., 2017). Braided river deposits are recorded by channelized, cross-bedded sandstones and erosive conglomerates with interbedded cross-bedded sandstones and cross-laminated siltstones and mudstones (Wei et al., 2017). These facies are intercalated with nodular red beds with desiccation cracks recording paleosols (DeCelles, Kapp, et al., 2007). Matrix-supported conglomerates suggest debris flows possibly sourced from a nearby alluvial fan (DeCelles, Kapp, et al., 2007). We interpret the depositional environments represented by these sections as being dominated by braided river channels and associated floodplains feeding fluvial-deltaic and lacustrine settings with proximal alluvial fans near the source areas. By combining these interpretations with the above age constraints, we correlate these sections and reestablish a stratigraphic framework from the late Eocene to late Miocene (ca. 35–9 Ma) and demonstrate that large and stable fluvio-lacustrine systems were present in the Lunpola area probably throughout most of the Miocene (ca. 23–9 Ma; Figure 3a).

4.3 Provenance

Paleocurrent directions measured from cross-bedding, current ripple lamination, and imbricated gravels are highly variable but indicate a component of eastward to southward paleoflow with a small component of northward-directed flow (Figures 1c and 3a). Conglomerate compositions of measured sections are enriched in gray bioclastic and micritic limestones, red and green cherts, quartz, grayish-green and reddish sandstones, basalt, and andesite (Figure 3b). The modern rivers have comparable clast compositions and grain sizes in the main channels (Figures 3b and 3c).

Jurassic to Lower Cretaceous marine strata in the southern Qiangtang and northern Lhasa terranes account for the presence of abundant bioclastic and micritic limestone clasts (Ma et al., 2017). Mélanges within the Bangong-Nujiang suture contributed the red and green cherts and basalt clasts (Ma, Hu, et al., 2018). The reddish sandstone clasts were probably derived from the Upper Cretaceous red bed unit, which is predominant in the southern Qiangtang terrane (Meng et al., 2018). The andesite clasts were likely derived from the Lower Cretaceous volcanic unit, which is exposed to the north and south of the basin (Zhang et al., 2004). Sandstone petrology plots indicate that the sediments are mainly of a “recycled orogeny” source, with a fraction of the samples showing a “magmatic arc” source (Figures 3d and S5; Dickinson & Suczek, 1979), which refers to the strongly deformed Mesozoic strata and igneous rocks of central Tibet.

In total, 1,116 zircon U-Pb concordant ages were obtained from the Lunpola Basin, 575 concordant ages from the modern river sands, and 197 measurements from the Nima Basin; the age spectra were plotted by integrating results presented by DeCelles, Kapp, et al. (2007; Figure 3e). The Cenozoic age population of 62–10 Ma in the Lunpola Basin is mainly from the newly dated tuffites. Fifteen Cenozoic ages present in the Nima Basin range from 64 to 34 Ma (Figure 3e). Middle Eocene to late Miocene (45–9 Ma) postcollisional volcanism in the Qiangtang terrane and the western Lhasa terrane could account for the presence of middle to late Cenozoic ages (Liu et al., 2017). The pre-Cenozoic ages of these two basins exhibit distinct clusters of 125–100 Ma, 620–250 Ma, 1,160–720 Ma, 1,950–1,700 Ma, and 2,630–2,360 Ma, which are also evident in the pre-Cenozoic detrital zircon samples from the Qiangtang and Lhasa terranes (Figure 3e). The age distributions of river sands are highly similar to the ancient drainage basins (Figure 3e). Widespread Cretaceous igneous rocks in central Tibet are responsible for the prominent clusters of 125- to 100-Ma ages (Figure 1b).

5 Discussion and Conclusions

The Cenozoic stratigraphic framework of the Lunpola Basin was previously established mainly as a result of hydrocarbon exploration in the region (Xu, 1980), and >5-km-thick continental deposits can be observed in seismic profiles (Wu et al., 2016). Our newly established stratigraphic framework constrains up to ~2-km-thick deposits within the basin (Figure 3a), which represent the uppermost part of the Cenozoic deposits. Therefore, most of the lower Cenozoic strata are still buried, so the new chronological constraints for the upper successions are significant for reinterpreting previous geophysical and borehole data and understanding the sedimentary response to surface and deep processes.

The representative zircon U-Pb age signals of the Qiangtang Central Uplift (240–200 Ma), Amdo microterrane (190–160 Ma), and southwestern Qiangtang terrane (160–140 Ma) are not detected in neither ancient nor modern drainage (Chapman & Kapp, 2017), indicating that the river networks of the Lunpola-Silin Co-Nima area have not drained these regions since the late Eocene. Typical age signals of the Linzizong volcanism (65–45 Ma; Yin & Harrison, 2000) in the Gangdese mountains are not observed in the ancient drainage, whereas strong signals are present in the contemporaneous sediments accumulated in southern Tibet (DeCelles et al., 2011, 2014, 2016). Organic geochemistry studies indicate that the Miocene lakes in the Lunpola and Nima Basins were hypersaline (Wang et al., 2011a, 2011b), similar to the modern Silin Co Lake. Ma et al. (2015) interpreted the Miocene Lunpola Lake as a balance-fill lake (Carroll & Bohacs, 1999); combined with the paleocurrent directions, we conclude that the Lunpola Basin has already been hydrologically closed since then. Integrating these data with the paleogeographic reconstruction of the Nima Basin presented by DeCelles, Kapp, et al. (2007), we demonstrate that during the late Eocene to late Miocene, the internally drained Lunpola-Silin Co-Nima area was confined by the Tanggula Shan-Qiangtang Central Uplift to the north and the Gangdese mountains to the south, which also bound the proto-Tibetan Plateau (Wang et al., 2008). The rivers terminated in deltaic deposits on the edge of large saline lakes much as they do today (Figure 4). Facies and provenance analyses reveal the Miocene large and stable fluvio-lacustrine systems with similar source regions (Figures 1b and 4a) and grain-size distributions to the modern system, which indicates that the ancient and modern drainages probably have similar hydrodynamic conditions. The sediment dispersal patterns and depositional environments are sensitive to the regional topographic relief and climate changes (Blum & Tornqvist, 2000; Dalrymple, 2010; Miall, 1977), and our results reveal that they are likely to have been persistent in central Tibet since the early Miocene. Besides, central Tibet has been suggested to have been in an arid and cold climate since the Oligocene (DeCelles, Quade, et al., 2007) and low erosion rates since the Eocene (Rohrmann et al., 2012). Thus, we propose that the low topographic relief has not changed significantly at least since the large and stable fluvio-lacustrine systems formed in the early Miocene.

Details are in the caption following the image
(a) Reconstruction of the paleodrainage. The dark gray area is pre-Mesozoic geological units that are not discussed in this work. (b) The schematic diagram for the drainage basin evolution and underlying deep processes. The schematic Moho morphology is adopted from Gao et al. (2013) and Wu et al. (2016). Thrust faults and conjugate strike-slip faults are after Wang et al. (2008) and Yin and Taylor (2011).

Seismic observations and numerical modeling suggested that lower crustal flow and mantle processes were the main controlling mechanisms that caused flattening of the plateau surface (Bird, 1991; McKenzie & Jackson, 2002; Shen et al., 2001). However, the development of internal drainage was essential to developing the low-gradient landscape by aggradation and sediment redistribution (Liu-Zeng et al., 2008; Meyer et al., 1998). While central Tibet was internally drained, western Tibet was externally drained, and fluvial incision led to the present strong regional relief (Gourbet et al., 2016). The deceleration of erosion rates in central Tibet since the mid-Eocene (Hetzel et al., 2011) most likely implies a reduction in river incision, followed by sedimentation, which implies the initiation of internal drainage. The inability of rivers to flow out of the surrounding highlands and their connection to a constant base level is required to maintain a long-standing internal drainage network (Sobel et al., 2003). As topographic growth of the Himalaya resulted in enhanced aridification in central Tibet, the ability of rivers to incise through surrounding mountains was reduced (Dupont-Nivet et al., 2007; Ramstein et al., 1997; Sobel et al., 2003), thus facilitating the development of internal drainage. Under such hydraulic and climatic conditions, many endorheic basins were formed in central Tibet with anoxic and saline water bodies (Han et al., 2014; Horton, 2012; Ma, Wang, Wang, Li, & Hu, 2015; Ma, Li, et al., 2018). Ponding of the fluvial base level resulted in the lowering of hillslope gradients through weathering, erosion, and increased sediment aggradation; these processes combined to reduce overall gradients and thus define a plateau morphology. We conclude that central Tibet has been internally drained since at least the late Eocene; this geomorphic isolation enabled millions of years aggradation of fluvial systems and consequent diffusive lowering of hillslopes, resulting in the development of the relatively low-relief landscape no later than the early Miocene (Figure 4b).

The well-established stratigraphic framework provides a spatial-temporal constraint for basin evolution and its possible response to deeper geological processes (DeCelles et al., 2015). As a balance-fill lake (Ma, Wang, Wang, Li, & Hu, 2015), the ancient Lunpola Lake subsided sufficiently to accommodate water and sediment fill (Carroll & Bohacs, 1999). The Lunpola Basin was suggested to have been a flexural basin bounded by steep, basement-involved thrust faults (Horton, 2012; Leeder et al., 1988) or a rift basin in a transpressional setting (Schneider et al., 2003). The significant N-S thrusting in central Tibet, which was regarded as the primary control mechanism of the basin formation (Horton, 2012; Wei et al., 2017), ceased before ~23 Ma (Kapp et al., 2007) and hence predates the main phase of basin formation. Thus, the post-Oligocene basin evolution requires another suitable mechanism.

The deformation pattern of central Tibet substantially changed from compression to extension (Li et al., 2015). Subsequently, central Tibet has been dominated by N-S extensional structures and V-shaped conjugate strike-slip faults, which have been suggested that facilitate the presence of topographic lowlands along the Bangong-Nujiang suture zone (Taylor et al., 2003). The E-W extension in central Tibet was proposed to have initiated in the early to middle Miocene (18–13 Ma; Li et al., 2015) or even to have occurred concurrently with the N-S shortening after the Eocene (ca. 47 Ma; Wang et al., 2010). The modern rivers flow dominantly westward over the site of the ancient lake into Silin Co (Figure 1b), and the E-W extension and strike-slip faulting likely account for the diversion of the river network and the relocation of the lake center. A deep seismic reflection profile reveals that the Moho depth is approximately 5–6 km deeper beneath the Bangong-Nujiang suture than beneath the southern Qiangtang and northern Lhasa terranes (Gao et al., 2013), which has been interpreted to result from ramp faulting in deep crust (Wu et al., 2016), and this thickness difference is slightly greater than those of the Cenozoic deposits in the Lunpola-Silin Co-Nima area. A paired general-shear model was suggested by Yin and Taylor (2011) as the mechanism for the formation of V-shaped conjugate strike-slip faults, which are attributed to the asthenosphere flow and eastward spreading of the thickened Tibetan lithosphere. DeCelles et al. (2015) and Wang et al. (2015) suggested that delamination or dripping of the lithosphere could have exerted a significant influence on basin subsidence in the central Andes, which also features high and broad internally drained hinterland and thickened crust, and this probably also occurred in the late Cenozoic Tibetan hinterland (Kapp & DeCelles, 2019). Taken all the abovementioned events into account, we suggest that deep processes across the Bangong-Nujiang suture, such as lower crustal flow and/or even processes in the mantle lithosphere and asthenosphere, are likely coupled with the surface topography, basin subsidence, and sedimentation (Gao et al., 2013; Wu et al., 2016; Yin & Taylor, 2011). Deep geological processes may not have been required to form the low-relief landscape, but it is reasonable to suggest that subsidence of these lake-filled basins along the Bangong-Nujiang suture responded to deep crustal or mantle flow. This speculation is supported by the latest 3-D simulations presented by Bischoff and Flesch (2018), which suggest that the flow of a weak lower crust could have led to surface subsidence in central Tibet.

Acknowledgments

The data supporting the conclusions can be obtained in the supporting information. This work was financially supported by the Ministry of Science and Technology of the People's Republic of China (National Key R&D Plan, 2017YFC0601405), Strategic Priority Research Program of Chinese Academy of Sciences (XDA20070303), National Natural Science Foundation of China (41802124), the Second Tibetan Plateau Scientific Expedition and Research Program (2019QZKK0204), and Key Laboratory of Sedimentary Basin and Oil and Gas Resources, Ministry of Natural Resources of the PRC (cdcgs2018006).