Gangdese culmination model : Oligocene – Miocene duplexing along the India-Asia suture zone , Lazi region , southern Tibet

The mechanisms for crustal thickening and exhumation along the Yarlung (IndiaAsia) suture in southern Tibet are under debate, because the magnitudes, relative timing, and interaction between the two dominant structures—the Great Counter thrust and Gangdese thrust—are largely unconstrained. In this study, we present new geologic mapping results from the Yarlung suture zone in the Lazi region, located ~350 km west of the city of Lhasa, along with new igneous (5 samples) and detrital (5 samples, 474 ages) U-Pb geochronology data to constrain the crystallization ages of Jurassic– Paleocene Gangdese arc rocks, the provenance of Tethyan Himalayan and Oligocene–Miocene Kailas Formation strata, and the minimum age (ca. 10 Ma) of the Great Counter thrust system. We supplement these data with a compilation of 124 previously published thermochronologic ages from Gangdese batholith, Kailas Formation, and Liuqu Formation rocks, revealing a dominance of 23–15 Ma cooling contemporaneous with slip across the Great Counter thrust system and other potentially linked structures. These data are systematically younger than 98 additional compiled thermo chronologic ages from the northern Lhasa terrane, recording mainly Eocene cooling. Structural and thermochronologic data were combined with regional geological constraints, including International Deep Profiling of Tibet and the Himalaya ( INDEPTH) seismic reflection data, to develop a new structural model for the Oligocene–Miocene evolution of the Tethyan Himalaya, Yarlung suture zone, and southern Lhasa terrane. We propose that a hinterland-dipping duplex beneath the Gangdese mountains, of which the Gangdese thrust is a component, is kinematically linked with a foreland-dipping passive roof duplex along the Yarlung suture zone, the Great Counter thrust system. The spatial and temporal convergence between the proposed duplex structures along the Yarlung suture zone and the South Tibetan detachment system indicate that they may be kinematically linked, though this relationship is not directly addressed in this study. Our interpretation, referred to as the Gangdese culmination model, explains why the Gangdese thrust system is only locally exposed (at relatively deeper structural levels) and provides a structural explanation for early Miocene crustal thickening along the Yarlung suture zone, relief generation along the modern Gangdese Mountains, early Miocene Yarlung River establishment, and creation of the modern internal drainage boundary along the southern Tibetan Plateau. The progression of deformation along the suture zone is consistent with tectonic models that implicate subduction dynamics as the dominant control on crustal deformation.


INTRODUCTION
Documentation of the structural style and timing of crustal thickening that produced the ~5 km average surface elevation of the Tibetan Plateau is key to understanding the response of continental crust to intercontinental collision and recognizing feedbacks among climate, surface processes, and tectonics (e.g., Quade et al., 2003;Harrison et al., 1992;Beaumont et al., 2001;Whipple, 2009).It is also critical to assessing the viability of lithospheric-scale tectonic models (e.g., DeCelles et al., 2011;Laskowski et al., 2017;Webb et al., 2017), which have developed significantly with increasing understanding of the geology and geo physics of Tibet (Yin, 2006;Hu et al., 2016), and alongstrike variations (Replumaz et al., 2010;Leary et al., 2016b;Webb et al., 2017).The significance and timing of Cenozoic fault systems along the ~1300-km-long Yarlung (India-Asia) suture in southern Tibet (Fig. 1), however, remain a subject of debate.Juxtaposition of deeply exhumed magmatic arc rocks of the southern Lhasa terrane against Indian passive-margin strata, as well as thermochronologic data and field mapping, led to the discovery of a northdipping mylonitic shear zone-the Gangdese thrust-that carried magmatic arc rocks southward in its hanging wall (Yin et al., 1994(Yin et al., , 1999)).Primarily documented southeast of the city of Lhasa, the Gangdese thrust was interpreted as a crustal-scale structure that was active by late Oligocene to early Miocene (27-23 Ma) time, based on 40 Ar/ 39 Ar thermochronology data, with a minimum displacement of 46 ± 9 km (Harrison et al., 1992;Yin et al., 1994;Copeland et al., 1995).However, this structure is apparently not exposed along strike to the west of Lhasa, leading others to call into question its significance and along-strike continuity (Aitchison et al., 2003).The dominant structures along the Yarlung suture west of Lhasa are a system of south-dipping reverse faults called the Great Counter thrust (Heim and Gansser, 1939;Yin et al., 1999;Murphy and Yin, 2003), which typically places Indian passive-margin rocks on suture zone mélange, mélange on Cretaceous forearc basin strata, and forearc basin strata on Oligocene-Miocene conglomerate, from south to north.A lack of hanging-wall cutoffs and no clear thermochronological date differences across individual fault splays render constraints on the timing and magnitude of Great Counter thrust activity tenuous, but most studies agree that it was active by late Oligocene-early Miocene time (Quidelleur et al., 1997;Harrison et al., 2000;Yin et al., 1999;Wang et al., 2015), temporally overlapping or closely following activity on the Gangdese thrust.Despite the close GSA Bulletin; July/August 2018;v. 130;no. 7/8;p. 1355-1376;  spatial and temporal relationship between the Great Counter thrust system and the Gangdese thrust (where the Gangdese thrust is exposed), the crosscutting or branching relationships between them are not known.The possibility that the Gangdese thrust is an orogen-scale structure that accommodated significant crustal shortening during Cenozoic time, and the nature of its relationship to the more prominently exposed Great Counter thrust system (Fig. 1) are open questions with major implications for Himalayan-Tibetan tectonics.
To the north of the Yarlung suture zone, there is a belt of calc-alkaline plutonic rocks that are dominantly Cretaceous to Paleogene in age (Schärer et al., 1984), referred to as the Gangdese batholith, and related volcanic and volcani clastic rocks that are dominantly Paleocene to Eocene in age, referred to as the Linzizong Formation (Lee et al., 2009).Collectively, these rocks compose the Gangdese magmatic arc, which developed along the southern Lhasa terrane (Asian) margin during northward subduction of Neo-Tethyan oceanic lithosphere and persisted during India-Asia collision (Kapp et al., 2007).The Gangdese Mountains (Fig. 1) in southern Tibet (also called the Trans-Himalaya) are composed mostly of Gangdese magmatic arc rocks.The Gangdese Mountains define the northern boundary of the Yarlung River watershed, in southern Tibet, and the southern boundary of the internally drained portion of the Tibetan Plateau.
Along the southern flank of the Gangdese Mountains (Fig. 1), an Oligocene-Miocene, conglomerate-rich, continental unit, referred to as the Kailas (Gangrinboche) Formation, is exposed in buttress unconformity atop Gangdese arc rocks (Gansser, 1964;Aitchison et al., 2002;DeCelles et al., 2011DeCelles et al., , 2016;;Leary et al., 2016b).Nonmarine strata of similar composition and structural position are continuous, with some variations in sedimentary facies, for over 1300 km along the Yarlung suture zone (Leary et al., 2016b).Some workers have interpreted the Kailas Formation as the product of contractional deformation, associated with a lithospheric flexure during a late stage of India-Asia collision (Aitchison et al., 2007), flexural foreland basin deposition related to the Great Counter thrust system (Wang et al., 2015), or wedgetop sedimentation related to out-of-sequence Great Counter thrust system activity (Yin et al., 1999).However, recent investigations of the Kailas Formation sedimentology, fossil assemblages, and basin architecture indicate that the Kailas Formation was deposited in an extensional basin bounded by a north-dipping normal fault, perhaps related to Oligocene-Miocene rollback or peeling-back of the subducted Great Indian slab (DeCelles et al., 2011(DeCelles et al., , 2016;;Wang et al., 2013;Leary et al., 2016a).
The Cretaceous-Paleogene Xigaze forearc basin (Einsele et al., 1994;Dürr, 1996;Wang et al., 2012;An et al., 2014;Orme and Laskowski, 2016) is exposed to the south of the Kailas Formation across a splay of the Great Counter thrust system.Xigaze forearc basin strata were deposited atop serpentinite mélange (Orme and Laskowski, 2016)-exposed along its southern margins in the Lazi region-suggesting that the 132-122 Ma Yarlung suture zone ophiolites (Hébert et al., 2012;Chan et al., 2015) were in a suprasubduction-zone position at the onset of forearc basin deposition ca.110 Ma (Huang et al., 2015;Orme and Laskowski, 2016).Another conglomerate unit-the Liuqu Formation-is locally exposed to the south of the Xigaze forearc basin, where it was deposited mainly atop serpentinite-matrix mélange during early Miocene time (Li et al., 2015;Leary et al., 2016a) or late Paleocene time (Ding et al., 2017).The Yarlung suture zone ophiolites and serpentinite-matrix mélanges structurally overlie a belt of subduction-accretion, shale-and sandstone-matrix mélange with (meta-) sedimentary and (meta-) basalt blocks (e.g., Cai et al., 2012).The sedimentary-matrix mélange separates rock units that were located along the southern margin of the Lhasa terrane from the Cambrian to Paleogene, the (meta)sedimentary Tethyan Hima layan sequence, the majority of which was deposited in a passive-margin setting along the northern margin of India (Gaetani and Garzanti, 1991;Liu and Einsele, 1994;Garzanti, 1999).
Thermochronology data from Gangdese batholith rocks exposed along the Yarlung suture zone (Copeland et al., 1987(Copeland et al., , 1995;;Harrison et al., 1992;Dai et al., 2013;Sanchez et al., 2013;Carrapa et al., 2014;Wang et al., 2015;Ge et al., 2016;Li et al., 2016;Laskowski et al., 2017) reveal an orogen-scale Oligocene-Miocene exhumation event, possibly associated with one or more tectonic, climatic, and erosional factors (Carrapa et al., 2014(Carrapa et al., , 2017)).Low-temperature thermochronology data from the Kailas Formation reveal a preponderance of Miocene (19-15 Ma) cooling ages, which have been interpreted to record efficient erosion associated with drainage reorganization and establishment of the Yarlung River during early Miocene time (Carrapa et al., 2014;Lang and Huntington, 2014).Based on these data, the Gangdese magmatic arc appears to have experienced semicontinuous exhumation throughout the period in which it has variably been interpreted to be in the hanging wall of the Gangdese thrust (e.g., Yin et al., 1994), in the footwall of the Great Counter thrust system, and/or in the hanging wall of a north-dipping normal fault that created the accommodation space for burial by up to ~4 km of Oligocene-Miocene nonmarine strata (the Kailas Formation; DeCelles et al., 2011;Wang et al., 2015;Leary et al., 2016b).Further complication arises from the broad range of explanations for relief generation between the presently high-standing Gangdese Mountains (Figs. 1-2) and the low-lying Yarlung suture zone (Fig. 1), including flexure in a foreland basin setting (Wang et al., 2015), crustal thickening driven by the Great Counter thrust (Sanchez et al., 2013) or Gangdese thrust (e.g., Yin et al., 1994), and fluvial incision influenced by the strengthening Asian monsoon coupled with renewed Indian underthrusting (Carrapa et al., 2014).No structural model exists that reconciles the roles of the Gangdese thrust and Great Counter thrust system, provides context for Oligocene-Miocene sedimentary basin development along the suture zone, explains relief generation of the Gangdese Mountains, and provides a mechanism for Yarlung River establishment while maintaining compatibility with thermochronometric data.
In this study, we present a structural model for the Oligocene-Miocene evolution of the southern Lhasa terrane and Yarlung suture zone based on new regional-scale geologic mapping of an ~1500 km 2 area north of the town of Lazi, Tibet (Figs. 1 and 2A), encompassing three detailed (~1:50,000 scale) geologic mapping locales (Figs.2B-2D).In addition, we present five igneous U-Pb ages from the Gangdese arc and younger intrusive rocks and five detrital zircon U-Pb samples from sedimentary rocks in the Lazi region.We interpret that the Great Counter thrust system is a passive roof duplex associated with crustal-scale, hinterland-dipping duplexing-equivalent to the Gangdese thrust.This model explains the occurrence of imbricated, foreland-dipping thrust sheets in the absence of an emergent, hinterland-dipping detachment hori zon.The new structural model is discussed in the context of lithospheric-scale tectonic models that invoke subduction dynamics as the principal driver of crustal deformation, providing an explanation for episodes of internal shortening (duplexing) along the Yarlung suture zone.

Field Methods
We report data collected during field work in 2012 and 2014 from the Yarlung River valley, ~10 km north of the city of Lazi, Tibetan Autonomous Region, China.The city of Lazi, which is also referred to as Lhatse, Quxar, Quxia, or Chusar, is located ~350 km west-southwest of Lhasa City and ~35 km north-northeast of the Mabja Dome (Lee et al., 2004(Lee et al., , 2006)), crowned by Hlako Peak (~6500 m; Fig. 1).Mapping was conducted at ~1:100,000 scale across the study area (Fig. 2A) and at 1:50,000 scale in three areas (Figs.2B-2D) atop topographic maps generated from 3-arc-second Shuttle Radar Topog raphy Mission data with draped Land-Sat orthoimagery.Contacts were interpolated between traverses using both Google Earth and LandSat orthoimagery.Cross sections were drawn from our structural and mapping data through the eastern and western portions of the study area (Figs.2A and 3).

Rock Units and Correlations
The southernmost rocks in the Yarlung suture zone are low-grade metasedimentary rocks, dominated by slate and well-cemented quartz arenite in the Lazi region (Fig. 2A), which we correlate to the Tethyan Himalayan sequence.In the western third of the map area (Fig. 2A), shale-matrix mélange, with blocks of sandstone, limestone, chert, and volcanic rocks, is exposed to the north of the Tethyan Himalayan sequence.We correlate these rocks to the Pomunong mélange (unit JKp), which composed of Late Jurassic to Early Cretaceous rocks, based on radiolarian fossils (Zhu et al., 2005;Cai et al., 2012).Farther north, a unit of shale-and lithic-sandstone-matrix mélange, with blocks of chert, basalt, and gabbro, is exposed between the Pomunong mélange and serpentinite-matrix mélange.We correlate this unit to the Tangga mélange, composed of Late Triassic to Early Cretaceous rocks, based on radiolarian fossils (Ziabrev et al., 2003;Zhu et al., 2005), and is equivalent to the Bainang terrane of Aitchison et al. (2000) and the radiolarian intra-ophiolitic thrust sheet of Tapponnier et al. (1981).The Tangga and Pomunong mélanges likely formed in an accretionary wedge setting during Neo-Tethyan oceanic subduction beneath the southern Lhasa terrane (Cai et al., 2012).
In the southern portion of the study area (Fig. 2A), a sandstone-and chert-clast, pebbleto-cobble conglomerate unit containing interbeds of sandstone and shale was deposited in buttress unconformity atop chert-and matrix-dominated portions of the Tangga mélange, the Pomunong mélange, and between the Tethyan Himalayan sequence and serpentinite-matrix mélange, from west to east (Fig. 2A).We correlate this unit to the Liuqu Formation (Yin et al., 1980), which was deposited in a contractional setting as part of a fluvial and alluvial-fan depositional system (Leary et al., 2016a).The preponderance of geochronological and thermochronological data from the Liuqu Formation suggest that it was deposited during a short interval between 20 and 19 Ma (Li et al., 2015;Leary et al., 2016a).However, Ding et al. (2017) argued that the Liuqu Formation was deposited during latest Paleocene time based on U-Pb geochronology of interbedded tuffs.More work is needed to assess these competing hypotheses.In the Yarlung River valley, the Liuqu Formation was deposited in angular unconformity atop chert blocks in the Tangga mélange, and it is dominated by red chert clasts (Fig. 4D).The maximum preserved thickness of the Liuqu Formation is ~2 km, near the town of Liuxiang (Li et al., 2015), whereas the maximum thickness of the Liuqu Formation in the Lazi region is ~200 m (Leary et al., 2016a).
A belt of serpentinite-and gabbro-block mélange with a serpentinite-dominated matrix is exposed north of the Liuqu Formation, and on both the south and north sides of the Tangga mélange (Fig. 2A).We correlate these rocks to the laterally extensive and variably tectonized south Tibetan ophiolites, which formed between 132 and 122 Ma along strike (Hébert et al., 2012;Chan et al., 2015).However, zircon U-Pb ages from a gabbro block and a fine-grained,   granitic intrusive rock in the serpentinite-matrix mélange, collected within the study area near the town of Jiwa (Fig. 2A), indicate a younger crystallization age of ca. 111 Ma (Orme and Laskowski, 2016).Xigaze forearc basin strata, exposed to the north of Jiwa, were observed to be in depositional contact with the serpentinitematrix mélange (Fig. 2A).The onset of forearc basin deposition was constrained to ca. 110 Ma based on the U-Pb age of a tuffaceous sandstone directly above the basal unconformity, persisting until ca.86 Ma based on detrital zircon maximum depositional ages (Orme and Laskowski, 2016).North of the Xigaze forearc, a narrow (1-2 km north-south width) but relatively thick (~1 km), east-west-trending belt of boulder-to-pebble conglomerate and sandstone of the Kailas (Gangrinboche) Formation (e.g., Aitchison et al., 2002;DeCelles et al., 2011) is exposed in buttress unconformity on Gangdese batholith rocks (Leary et al., 2016b).The thickest accumulations of the Kailas Formation are up to 4 km thick, near the type locality at Mount Kailas (Heim and Gansser, 1939;Gansser, 1964).Along the northern Yarlung River valley, the Kailas Formation is dominated by granite, volcanic, gneiss, limestone, and lithic sandstone cobbles.In the northeast corner of the map area, an ~120 km 2 leucogranitic pluton crosscuts Xigaze forearc, Kailas Formation, and Gangdese batholith rocks.

Fault Systems
The term "Great Counterthrust" was originally used to describe the south-dipping reverse fault that placed Tethyan Himalayan sequence rocks on the Kailas Formation near Mount Kailas, ~650 km along strike to the west, where intervening sedimentary-and serpentinite-matrix mélange and Xigaze forearc strata are absent (Heim and Gansser, 1939).Here, we expand this nomenclature to include a system of moderately to steeply south-dipping reverse faults that carry Tethyan Himalayan rocks in the structurally highest position over Kailas Formation and Gangdese batholith rocks in the structurally lowest position, with other splays juxtaposing the intermediary units (Fig. 2A).The presence of multiple fault splays of the Great Counter thrust system might reflect a shallower depth of exposure in the Lazi region than in the Mount Kailas region, as Great Counter thrust system splays have previously been interpreted to merge into a single fault at depth (Yin et al., 1994(Yin et al., , 1999;;Laskowski et al., 2017).
Three distinct splays of the Great Counter thrust system were mapped in the Lazi region (Fig. 2A).The southernmost splay juxtaposes the Tethyan Himalayan sequence against serpentinite-matrix mélange, the Liuqu Formation, and Pomunong mélange from east to west.This fault is poorly exposed but was inferred based on juxtaposition of rock units, and it dips to the south based on its relationship with topography.To the north, a second splay juxtaposes hanging-wall serpentinite-matrix mélange against Xigaze forearc strata to the east, transitioning to a zone of anastomosing faults that juxtapose the Liuqu Formation, Pomunong mélange, serpentinite-matrix mélange, and Tangga mélange to the west (Fig. 2A).Xigaze forearc strata in the footwall of this fault zone are steeply dipping and locally overturned (Fig. 2A), and the fault contact between serpentinite-matrix mélange and Xigaze forearc strata in the Yarlung River valley (Fig. 2A) is characterized by an ~15-mthick zone of fault gouge and brecciated cataclasite.Similarly, a nearby fault contact between serpentinite-matrix mélange and a chert block of the Tangga mélange exhibits brecciation and chlorite alteration across an ~20-m-thick fault zone.Some chert blocks are entirely encased in serpentinite-matrix mélange and were likely emplaced as tectonic slivers along the intra-ophiolitic splay fault (Fig. 2A).Preservation of the depositional contact between Xigaze forearc strata and serpentinite-matrix mélange near the town of Jiwa is likely the result of the anastomosing character of this fault zone (Fig. 2D).The faults dip 20°-30° to the south in the eastern map area, 54° to the south near Jiwa (Fig. 2D), 83° to the south ~4 km west of Jiwa, and between 35° and 75° to the south with lo-cally overturned exposures in the anastomosing zone along the Yarlung River valley (Fig. 2B).
Splays of the Great Counter thrust bound both the north and south sides of the Xigaze forearc basin, which is ~15 km wide in the Lazi region (Fig. 2A).Xigaze forearc strata are folded across a broad syncline in-between these zones, producing a roughly fault-parallel average fold axial plane (Fig. 2A).At smaller scale, the Xigaze forearc strata are folded across opento-tight, symmetrical folds with amplitudes on the hundred-meter scale, which likely formed as parasitic folds during north-south contraction.In the east, the ~50° southeast-dipping, northern Great Counter thrust splay (Fig. 2A) juxtaposes hanging-wall Xigaze forearc basin strata against the Kailas Formation.Along the Yarlung River valley to the west, the hanging wall of this splay consists of a fine-grained granitoid that is in intrusive contact with Xigaze forearc basin strata.This contact is sill-like in geometry, and forearc strata structurally above the fine-grained intrusive rock display a strong cleavage oriented ~100/60 SW.The intrusive rock appears to be the product of protracted hypabyssal intrusive activity, which was later subjected to shearing, as foliation-parallel, fine-grained dikes of similar composition to the host rock were observed.
The northernmost splay of the Great Counter thrust system (Fig. 2A) was exhumed from greater depth than the southern splays, because the Kailas Formation in the footwall displays protomylonitic fabrics that were not observed elsewhere.Along the northern Yarlung River valley (Fig. 2C), the hanging wall consists of fine-grained, felsic igneous rocks (Figs.section, brittle fabrics give way to a 20-m-thick zone of top-to-the-north, protomylonitic fabrics that are developed in the granitic clasts and sandstone matrix of the Kailas Formation (Figs. 4A-4B).We interpret these features to indicate that these rocks were tectonically buried in the footwall of the Great Counter thrust to brittle-ductile conditions (250-400 °C) and were later accreted to the hanging wall to accommodate their exhumation.Structurally below the mylonite zone, north-vergent recumbent folds were observed within the Kailas Formation, similarly indicating top-to-the-north shearing (Figs.2C and 4C).
A weak fault-parallel foliation was observed structurally below the Kailas Formation, in Gangdese batholith rocks ~1 km north of the mylonite zone (Fig. 2C).The mylonitic fabrics observed along the Great Counter thrust in the Lazi region allow us to estimate the depth of tectonic burial of Kailas Formation strata and the magnitude of slip along the northern splay of the Great Counter thrust system.The Kailas Formation is ~1.2 km thick in the study area (Leary et al., 2016b), and it achieves a maximum thickness of ~4 km in the Mount Kailas region (Heim and Gansser, 1939;Gansser, 1964).Therefore, the maximum burial that can be attributed to subsidence alone is 4 km.Assuming a geothermal gradient of 25 °C/km, the mylonitic Kailas Formation rocks were likely tectonically buried to depths of at least 10 km to achieve brittle-ductile conditions (250-400 °C) prior to exhumation in the fault zone hanging wall.Considering the measured shear fabric orientation (45°-50° to the south), our interpreted minimum burial depth (10 km), and assuming a steady geothermal gradient and the absence of convective heating by igneous or hydrothermal activity, displacement across this splay of the Great Counter thrust system totaled at least 7-13 km, with 7 km reflecting the maximum 4 km of burial due to subsidence.Mylonitic fabrics that are developed in the proximal footwall of the northernmost Great Counter thrust splay have also been reported near the town of Langxian, located ~500 km along strike to the east in southeastern Tibet (Wang et al., 2015).The occurrence of mylonites at both localities and the >1000 km along-strike continuity of the Great Counter thrust system suggests that it is a crustal-scale structure.
Within the detailed map area near the town of Jiwa (Figs. 2A and 2D), an ~35° north-dipping reverse fault juxtaposes hanging-wall sedimentary-matrix mélange against footwall Liuqu Formation strata.Although this fault is not well exposed east of Jiwa, we interpret that it extends along strike, except where it is cut out by a Great Counter thrust splay with the Tethyan Himalayan sequence and Liuqu Formation in the hanging wall (Fig. 2A).Within the footwall of this fault, bedding in the Liuqu Formation fans up section from locally overturned (steeply northdipping) to moderately south dipping (Fig. 4E), consistent with syntectonic sedimentation.We interpret these features to indicate that this fault is an antithetic splay of the Great Counter thrust system (Fig. 3).
Although fault dips and rock unit juxtapositions are highly variable along different splays of the Great Counter thrust system, interpreted cross sections through the Lazi region map area suggest that the regional structural geom etry is dominated by a set of three Downloaded from https://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/130/7-8/1355/4224640/1355.pdf by guest imbricate, south-dipping thrust sheets (Fig. 3).The steeply dipping faults are interpreted to sole into moderately southwest-dipping (~30°) master faults (Fig. 3), especially in the anastomosing fault zone south of Xigaze forearc basin strata (Fig. 2A).Foliation measurements from the Pomunong and Tangga mélange, bedding measurements in the Liuqu Formation and the Kailas Formation near the northernmost Great Counter thrust splay, and foliation in both Gangdese batholith and Kailas Formation rocks surrounding this fault are consistent with this interpretation (Fig. 3).In contrast, the Linzizong Formation volcanic rocks, exposed on the north side of the Gangdese Mountains, display a regional northward dip (Fig. 1), suggesting that the Gangdese batholith may be exposed in the core of a broad anticline oriented parallel to the Great Counter thrust system.There is no exposed north-dipping fault contact between Gangdese batholith rocks and the Xigaze forearc basin, or between Gangdese batholith rocks and the Tethyan Himalayan sequence, that might correlate to the Gangdese thrust, which has been documented ~400 km along strike to the east (Yin et al., 1994).

GEOCHRONOLOGY AND THERMOCHRONOLOGY U-Pb Geochronology Methods
Five igneous and five detrital zircon U-Pb geochronology samples were collected from map units across the study area to confirm map unit identities, constrain sediment provenance, and determine the age of igneous intrusions.Zircons were extracted from ~2 kg samples by crushing, followed by density and magnetic separation techniques.For detrital samples, a large aliquot of grains (usually >1000) was mounted in epoxy together with crystal standards.For igneous samples, 20-50 selected zircon grains were mounted alongside crystal standards.Mounts were sanded to a depth of ~20 μm to expose zircon cores, polished, imaged with backscattered electron or cathodoluminescence techniques for navigation purposes, and then cleaned prior to analysis.U-Pb geochronology was conducted by laserablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) at the Arizona LaserChron center following the techniques outlined in Gehrels et al. (2006Gehrels et al. ( , 2008) ) and Gehrels and Pecha (2014).
All samples were ablated using a Photon Machines Analyte G2 excimer laser equipped with a HelEx ablation cell, with a spot size ranging from 10 to 30 μm, depending on the size of the target crystal zone.Typically, ablation pits were ~15 μm in depth.One sample (61012AL3) was analyzed using an Element2 high-resolution ICP-MS, which sequences rapidly through U, Th, and Pb isotopes.Signal intensities were measured with a secondary electron multiplier detector that operates in pulse-counting mode for signals less than 50,000 counts per second (cps), in both pulse-counting and analog mode for signals between 50,000 and 5,000,000 cps, and in analog mode above 5,000,000 cps.The remaining samples (n = 9) were analyzed using a Nu high-resolution ICP-MS, which was equipped with a flight tube of sufficient width that U, Th, and Pb isotopes could be measured simultaneously.All Nu measurements were made in static mode using Faraday detectors with 3 × 10 11 ohm resistors for 238 U, 232 Th, and 208 Pb-206 Pb, and discrete dynode ion counters for 204 Pb and 202 Hg.Further details of the Element2 and Nu analytical procedures can be accessed at the Arizona LaserChron Center Web site (Laserchron .org).Detrital zircon U-Pb analytical data are reported in Supplementary Table DR1, and igneous zircon U-Pb analytical data are reported in Supplementary Table DR2. 1ncertainties in these tables are at the 1σ level and include only measurement errors.Analyses that were >20% discordant or >5% reverse discordant were rejected.
Crystallization ages for zircons were isolated from inherited ages in igneous U-Pb samples using the Arizona LaserChron Center (www .laserchron.org)in-house program AgePick, which provides tools for identifying inheritance, Pb loss, and/or overgrowth and recrystallization of metamorphic zircon.Zircons interpreted as inherited in igneous samples are displayed as probability distribution functions alongside weighted mean age determinations incorporating analytical and systematic error (Fig. 5).Detrital zircon data from this study and references from the literature are also displayed as probability distribution functions (Fig. 6), all of which were created using the DZstats program (Saylor and Sundell, 2016).

Igneous Zircon U-Pb Geochronology Results
Igneous U-Pb samples were collected from the Gangdese batholith (sample 61012AL1) in the northwestern portion of the map area (Figs.2A and 2C), a fine-grained felsic intrusive rock (61012AL7) in the hanging wall of the northernmost Great Counter thrust   splay directly below the sill-like intrusive contact with Xigaze forearc strata (Figs.2A and  2C), and from fine-grained dikes (6912AL1, 6912AL2) and a large, detached boulder of twomica leuco granite (61312AL1) in close proximity to a large (~100 km 2 ) leucogranite pluton (Fig. 2A).Sample 61012AL1, a hornblendeplagioclase-biotite-quartz granodiorite, was located ~1.3 km north of an outcrop exposure of the northernmost splay of the Great Counter thrust, beneath the unconformable contact with the Kailas Formation (Fig. 2C).The outcrop from which this sample was collected displays a weak foliation striking approximately east-west (87°) and dipping 60° to the south, similar to the orientation of shear fabrics along the Great Counter thrust (Fig. 2C).The sample yielded a U-Pb age of 170.1 ± 3.8 Ma (Fig. 5).This was the only sample of the five samples that lacked evidence for complexly zoned zircons with xenocrystic cores.Sample 61012AL7 yielded a poorly constrained crystallization age of 43.7 ± 0.8 Ma based on the weighted mean of the two youngest dates.Four other zircons from this sample yielded dates between 48 and 52 Ma, whereas the remaining 10 analyses yielded Paleozoic and Proterozoic dates (Fig. 5).We interpreted the dates not included in the weighted mean age determination as inherited based on the presence of zircon cores identified using high-resolution back scattered electron and cathodoluminescence imagery; however, it is also possible that the true crystallization age is younger than the poorly defined youngest age population.Samples 6912AL1 and 6912AL2 were collected along an approximately north-southoriented valley south of Zha Xilincun and the Phuntsoling Monastery, west of the leucogranite pluton (Fig. 2A).Both samples were collected from fine-grained dikes that intruded subparallel to east-west-striking bedding of Xigaze forearc sandstone and shale, which dips steeply to the north (strike ~350°, dip 70°-85°).Sample 6912AL1 yielded an age of 9.9 ± 0.3 Ma (Fig. 5).Five 600-300 Ma dates from this sample are interpreted to reflect inheritance (Fig. 5).Sample 6912AL2 yielded an age of 9.7 ± 0.6 Ma (Fig. 5).The 17 older dates from this sample included 300-600 Ma and older, Proterozoic-age populations that are also interpreted to reflect inheritance.Sample 61312AL1, a phaneritic two-mica leucogranite, was collected due east of the dike samples on the east side of the pluton (Fig. 2A) from a boulder in a large terminal moraine below a cirque composed entirely of leucogranite.This sample yielded an age of 9.9 ± 0.3 Ma (Fig. 5).Two 15-14 Ma dates and five others between 800 and 30 Ma are interpreted to reflect inheri-tance.This pluton provides a minimum age of ca. 10 Ma for the Great Counter thrust system, which it crosscuts in the northeastern portion of the study area (Fig. 2A).

Detrital Zircon U-Pb Geochronology Results
Detrital zircon samples were collected from Tethyan Himalayan sequence rocks (three samples) and the Kailas Formation (two samples) to confirm field identifications and refine depositional ages.Two samples were collected from Triassic (62211PK3) and Jurassic (62211PK5) Tethyan Himalayan sequence quartz arenites exposed 15-25 km southwest of the city of Lazi outside of the detailed study area (Fig. 1) to examine previously documented provenance differences between the Triassic interval of the Tethyan Himalayan sequence and the rest of the Tethyan Himalayan sequence (e.g., Cai et al., 2016).Sample 62211PK3 produced a detrital zircon age spectrum characterized by a broad 1230-180 Ma age-probability peak with a few older Paleoproterozoic and Archean ages (Fig. 6).In contrast, sample 62211PK5 produced an age spectrum characterized by a broad distribution of ages between 1250 and 490 Ma, alongside a few older Proterozoic and Archean ages, with a dominant age-probability peak at ca. 509 Ma (Fig. 6).Sample 7712AL2 was collected from a Jurassic quartz arenite exposed on a mountaintop ~1.5 km south of a Great Counter thrust splay (Figs.2A and 2D), and it is characterized by a similar age spectrum to that of 62211PK5, with the addition of four 450-320 Ma ages (Fig. 2D).
Two samples were collected from sheared sandstone beds within the conglomerate-dominated Kailas Formation, exposed in the footwall of the northernmost Great Counter thrust splay (Figs.2A and 2C).Both samples revealed a dominance of 100-40 Ma ages, largely distributed between 100-80 Ma and ca.50 Ma age-probability peaks (Fig. 6).Although these data suggest an Eocene maximum depositional age for the Kailas Formation, another detrital zircon sample from the same locality presented in a separate study provided a wellconstrained Miocene maximum depositional age of 22.8 ± 0.3 Ma (Leary et al., 2016b).In contrast, three other samples from the same study yielded older, Eocene maximum depositional ages.Therefore, five of the six detrital zircon samples from this locality (Figs.2A  and 2C) overestimate the depositional age of the Kailas Formation by ~20 m.y., suggesting that zircon availability, extreme local sourcing, and/or complex zonation (possibly with inherited cores and young rims that might have been missed during analysis via LA-ICP-MS) prevented reliable maximum depositional age determination in this case.
The proximity of thermochronology samples to the Yarlung suture was calculated by measuring the shortest straight-line distance from each sample to the nearest exposure of ophiolitic rocks, and it was plotted against thermo chronological age to reveal orogen-perpendicular (generally north-south) trends (Fig. 7).Thermochrono logi cal data for the Kailas Formation, Gangdese batholith, and northern Lhasa terrane reveal an increase in cooling age with distance  DR3 (see text footnote 1), which also contains references for data sources.ZFT-zircon fission track age.
from the Yarlung suture, punctuated by an apparently sudden transition across the Gangdese Mountains from mainly Oligocene-Miocene ages in the southern Lhasa terrane to mainly Eocene and older ages in the northern Lhasa terrane (Fig. 7; Rohrmann et al., 2012).A plot of age versus distance from the Yarlung suture ophiolitic rocks reveals a dominance of 25-7 Ma ages within ~20 km, and a broader 60-6 Ma range between 20 and 125 km (Fig. 7).The oldest ages in the southern Lhasa terrane are from the zircon (U-Th)/He and ZFT systems, whereas the youngest ages are mainly from the apatite (U-Th)/He and AFT systems (Fig. 7), likely reflecting the higher and lower closure temperatures, respectively.In the Lhasa and Lopu Range regions, biotite 40 Ar/ 39 Ar ages are between 27 and 17 Ma, overlapping in age with zircon (U-Th)/He, AFT, and apatite (U-Th)/He data, and likely indicating rapid cooling from ~300 °C (biotite 40 Ar/ 39 Ar) to ~80 °C (apatite U-Th/He).Probability distribution functions created from Yarlung suture zone data, separated by thermochronological system and geologic unit (Fig. 8), reveal a dominance of late Oligocenemiddle Miocene cooling, except for apatite (U-Th)/He data from the Gangdese batholith and Liuqu Formation, which record late Miocene cooling (Figs. 7 and 8).Age-probability peaks for Gangdese batholith and Kailas Formation samples display a slight dependence on closure temperature (decreasing age with decreasing closure temperature).Aside from a few older ages from the Gangdese batholith in the Lhasa region, which may reflect igneous activity or conductive cooling after emplacement, the majority of cooling through 40 Ar/ 39 Ar biotite to AFT closure temperatures (~350-60 °C) took place between ca. 26 and 12 Ma (Figs. 7 and 8).Therefore, we conclude that if the Gangdese thrust is a major structure, it was likely active during this period, which also overlaps in time with the permitted age range of the Great Counter thrust.Age spectra from the Gangdese batholith are dominated by age-probability peaks at ca. 22 Ma and ca.17 Ma, accompanied by a ca. 25 Ma peak in the biotite 40 Ar/ 39 Ar data and a younger ca. 9 Ma peak in apatite (U-Th)/He data, which may reflect differences in closure temperature (Fig. 8).Age-probability peaks for the Kailas Formation are between 20 and 12 Ma, i.e., slightly younger than those of the Gangdese batholith, consistent with a southward progression of exhumation.Thermo chrono logic data from the Liuqu Formation indicate that it cooled through apatite (U-Th)/He closure between 10 and 4 Ma, likely recording Yarlung River incision (e.g., Carrapa  et al., 2014) and possibly influenced by Miocene (ca.16 Ma) to recent orogen-parallel extension (Fig. 8; e.g., Sundell et al., 2013).

Detrital Zircon Provenance Analysis
Detrital zircon samples from the Yarlung suture zone in the vicinity of the Lazi region track sediment provenance prior to and during India-Asia collision.Samples 62211PK5 and 7712AL2 (Figs. 1 and 2A), which we correlated with the Tethyan Himalayan sequence, are dominated by Pan-African (500-600 Ma) detrital zircons that are typical of both the Lhasa terrane, which rifted from Gondwana, and the Tethyan Himalayan sequence (Fig. 6).However, the absence of younger detrital zircons that might have been derived from the Late Triassic-Paleogene Gangdese magmatic arc is indicative of derivation from Indian sources alone, as is typical of the majority of the Tethyan Himalayan sequence (DeCelles et al., 2000;Gehrels et al., 2011).Similarities between age spectra from these two samples and a composite reference curve for the Tethyan Himalayan sequence (Fig. 6; Gehrels et al., 2011) support our field correlations.Sample 62211PK3, located northeast of sample 62211PK5 and southwest of sample 7712AL2 (Fig. 1), is characterized by Pan-African and older ages alongside younger, Paleozoic-Early Jurassic (~ca.180 Ma, n = 2) ages.The detrital zircon age spectrum for this sample is similar to that of other Triassic-Early Jurassic Tethyan Hima layan rocks in southern Tibet (Fig. 6; Aikman et al., 2008;Cai et al., 2012;Webb et al., 2013;Li et al., 2015;Cai et al., 2016).Since the Lhasa terrane had rifted from Gondwana prior to deposition of the Triassic-Early Jurassic Tethyan Himalayan sequence strata (e.g., Li et al., 2016), and there is no known Indian source of Permian-Early Jurassic detrital zircons, we interpret that zircons in sample 7712AL2 were derived from crustal fragments along the northwestern margin of Australia (modern-day West Papua), transported westward onto the northern margin of India, and incorporated into the Tethyan Himalayan sequence, in accordance with previous interpretations of age-equivalent Tethyan Himalayan rocks in southern Tibet (Cai et al., 2016).
Two detrital zircon samples from conglomeratic strata of the Kailas Formation (61012AL2, 61012AL3) yielded markedly different age spectra than those of the Tethyan Himalayan sequence (Fig. 6).Detrital zircons in these samples are mainly distributed between two populations with age-probability peaks at ca. 90 Ma and ca.50 Ma.A few additional Middle Jurassic and Permian detrital zircons were present in sample 61012AL2 (Fig. 6).Comparison with a probability distribution function representative of Gangdese arc activity generated from a compilation of igneous samples (Orme et al., 2015) indicates that samples 61012AL2 and 61012AL3 were likely derived almost exclusively from Gangdese magmatic arc rocks (Fig. 6).Similar detrital zircon age spectra characterize the Kailas Formation at other locations along the Yarlung suture zone (DeCelles et al., 2011(DeCelles et al., , 2016;;Leary et al., 2016b) and are indicative of local derivation, as no Pan-African or older zircons from either the Lhasa terrane or India were identified (Fig. 6).Interestingly, the granodiorite immediately below the Kailas Formation, close to the sample locality, is Middle Jurassic in age (sample 61012AL1; Figs.2A and 3), while only a small population of zircons with similar ages from sample 61012AL2 was identified.Therefore, the Middle Jurassic Gangdese batholith rocks were likely exposed over a much smaller area compared to those of Late Cretaceous-Paleogene age at the time of deposition.The youngest detrital zircons in both samples overlap in age with the fine-grained granitic igneous rocks (sample 61012AL7) located in the hanging wall of the northernmost Great Counter thrust splay to the south (Fig. 2A).Therefore, it is possible that these rocks were exposed to the south of the Kailas Basin during deposition.

Timing Constraints for Major Structures in Southern Tibet
A synthesis of crosscutting relationships and geochronological results from the Lazi region (Fig. 1) with regional tectonic data provides some new information on the timing of slip across major structures in southern Tibet.From north to south, these include the north-dipping Gangdese thrust (or equivalent north-dipping structures beneath the Gangdese Mountains; Fig. 1), the south-dipping Great Counter thrust system, the inferred north-dipping listric normal fault that provided accommodation space for deposition of the Kailas Formation (Kailas fault; e.g., DeCelles et al., 2011), and the South Tibetan detachment system (Fig. 1).The Gangdese thrust was originally interpreted to have been active between 27 and 23 Ma, based on 40 Ar/ 39 Ar thermochronology data (Harrison et al., 1992;Yin et al., 1994;Copeland et al., 1995).However, these constraints overlap with the timing of deposition of the lower Kailas Formation, which was deposited between 25 and 23 Ma (Lazi Region detrital zircon dataalong-strike variation up to ~3 m.y.; Leary et al., 2016b).If the extensional nature of the Kailas Basin is correct (DeCelles et al., 2011(DeCelles et al., , 2016;;Leary et al., 2016b), then the Gangdese thrust was unlikely to have initiated until after 23 Ma in the Lazi region.If the Kailas Basin is contractional (e.g., Yin et al., 1999;Wang et al., 2015), then the Gangdese thrust might have been active as early as 27 Ma.The Great Counter thrust system cuts the Kailas Formation (Figs. 1, 2, and 3), and provenance shifts and clast composition in the upper Kailas Formation indicate deposition synchronous with contraction.Therefore, we interpret that the Great Counter thrust system initiated ca.23 Ma, possibly at the same time as the Gangdese thrust (or equivalent structures).
The younger limit of Great Counter thrust and Gangdese thrust activity is constrained by the timing of the transition from north-south contraction to east-west extension in southern Tibet, as north-south-oriented normal faults crosscut the Great Counter thrust system in multiple locations (Fig. 7).A compilation of extension onset and acceleration ages (Sundell et al., 2013) indicates that the majority of extensional structures initiated between 15 and 11 Ma.Some structures may have initiated earlier, such as the ≥16 Ma Lopukangri fault, which crosscuts the Great Counter thrust system ~250 km to the west (Laskowski et al., 2017).An absolute minimum age for the Great Counter thrust system is provided by the crosscutting relationship with the ca. 10 Ma pluton in the northeast corner of the study area (Figs. 1,2A,and 3B).The timing of slip on the South Tibetan detachment system is largely interpreted from a compilation of tectonic data provided by Webb et al. (2017), who indicated a 26-16 Ma range at the longitude of the study area, with significant along-strike variation.If the South Tibetan detachment system is kinematically linked with the Great Counter thrust system, then it may not have been activated until ca.23 Ma, if the constraints from the extensional interpretation of the Kailas Basin apply.

Observations that Guided Structural Model Development
Mapping, structural geology, geochronology, and thermochronology data presented in this study, alongside constraints from previous studies, guided development of our structural model for Oligocene-Miocene Yarlung suture zone evolution.At the regional scale (Fig. 1), Gangdese batholith rocks are exposed in a belt north of the Xigaze forearc that coincides with the highest-elevation portion of the ~1600-kmlong Gangdese Mountains, with summit elevations that commonly exceed 5500 m north of Lazi (Fig. 1).In contrast, the Xigaze forearc basin and Yarlung suture zone mélanges to the south are exposed in a relatively low-lying region, with peaks between 4500 and 5000 m surrounding the ~3900 m Yarlung River valley.Together, thermochronologic data (Fig. 7) and map relationships suggest that the Gangdese batholith experienced higher exhumation magnitudes in the southern Gangdese Mountains than to the north, where the volcanic carapace is preserved (Fig. 1).The thermochronologic data (Figs.7 and 8) suggest that the Gangdese batholith experienced some cooling during 25-23 Ma deposition of the Kailas Formation (Leary et al., 2016b).The Gangdese batholith appears to be exposed in the core of an east-west-trending antiform, as north-dipping Linzizong volcanic rocks (the Paleocene-Eocene Gangdese magmatic arc volcanic carapace) and the variably (~35°-60°) south-dipping Kailas Formation are exposed along the northern and southern flanks of the Gangdese Mountains, respectively (Fig. 1).This geometry can also be explained by a ramp along the north-dipping Gangdese thrust (Yin et al., 1994).The majority of thermo chrono logic ages older than ca. 25 Ma in our data set are from the samples farthest north in the Gangdese Mountains, particularly in the region surrounding Lhasa, and in the northern Lhasa terrane (Fig. 7).
Structural model development was guided by multichannel seismic reflection data obtained during the International Deep Profiling of Tibet and the Himalaya (INDEPTH) experiment (Brown et al., 1996;Nelson et al., 1996;Alsdorf et al., 1998;Hauck et al., 1998), collected along a transect from the High Himalaya into the Lhasa terrane (Fig. 9).We chose to favor these data over the more recent Hi-CLIMB receiver functions experiment (e.g., Nabelek et al., 2009)  due to the comparatively high mid-to uppercrustal resolution of the INDEPTH experiment.We integrated the orientation, depth, and previously published interpretations of major reflectors by projecting a composite profile generated from migrated rift-parallel seismic lines (Fig. 9; Alsdorf et al., 1998) ~250 km to the west into a roughly parallel, regional-scale cross section through the Lazi region (Fig. 9).Several prominent reflectors are visible in the INDEPTH data (Fig. 9); their interpretation is critical to understanding the subsurface geology of the Yarlung suture zone.The northernmost reflector is the Yamdrok-Damxung reflection band (YDR in Fig. 9; Brown et al., 1996), consisting of a band of reflectors with varying north and south dips between 12 km and 18 km depth that was interpreted either as a midcrustal partial melt layer (Nelson et al., 1996) or as a preexisting structural or lithologic boundary that was subsequently deformed, likely after the shutdown of Gangdese arc magmatism (Alsdorf et al., 1998).Results from the subsequent Hi-CLIMB receiver function experiment indicate that the Yamdrok-Damxung reflection (Fig. 9) is likely constrained to rift valleys (Fig. 9) and is largely discontinuous (Nabelek et al., 2009;Hetényi et al., 2011).Therefore, we did not consider the Yamdrok-Damxung reflection in development of the structural model.Three north-dipping reflectors between ~20 and 30 km depth are present beneath the southern extent of the Yamdrok-Damxung reflection (reflection "4" in Fig. 9), displaying decreasing northward apparent dip with increasing depth.The uppermost of these reflections projects into the Yamdrok-Damxung reflection.These were interpreted as a hinterland-dipping duplex above a footwall ramp along a north-dipping fault that projects to the surface within the Yarlung suture zone, possibly equivalent to the Gangdese thrust (Alsdorf et al., 1998;Makovsky et al., 1999).Deep imbrication structures are also apparent in Hi-CLIMB data at depths of 50-60 km, immediately above the Moho, extending northward of the Yarlung suture zone for ~30 km (Nabelek et al., 2009).Due to the great depth of the anomalies and comparatively low resolution at mid-to upper-crustal depths, it is unclear whether these structures can be linked to "reflection 4" (Fig. 9), or any faults that breach the surface near the Yarlung suture zone.To the south of the Yamdrok-Damxung reflection, a near-horizontal reflection band, referred to as the Yarlung-Zangbo reflector (YZR in Fig. 9), is visible at ~25 km depth (Fig. 9), possibly representing a low-angle structural discontinuity below the surface exposure of the Yarlung suture (Alsdorf et al., 1998), a hydrothermal or magmatic boundary (Alsdorf et al., 1998), or an ophio litic slab separating Indian-and Asianaffinity rocks (Makovsky et al., 1999).Farther south, series of south-dipping reflections (referred to as the "back-thrust system") are visible in the INDEPTH data that project beneath the North Himalayan domes (BTS in Fig. 9), including the Mabja dome south of Lazi (Fig. 1).The back-thrust system (BTS) reflection has previously been interpreted as the downdip projection of the Great Counter thrust system, requiring that the Great Counter thrust system cut Greater Hima layan rocks that are currently exposed in the North Himalayan domes (Fig. 1; Hauck et al., 1998).
The Kailas and Liuqu Formations provide a rich record of sediment transport, suture zone basin evolution, and thermal evolution during Oligocene-Miocene time.The basal unconformity between the Kailas Formation and Gangdese arc plutonic rocks, which are Middle Jurassic to Eocene in age in the Lazi region (Fig. 2A), implies that the Gangdese batholith underwent significant exhumation prior to the onset of Kailas Formation deposition, which occurred between 25 and 23 Ma in the Lazi region (Leary et al., 2016b).Growth strata in the Kailas Formation (Wang et al., 2015) indicate progressive southward steepening of the basal unconformity during deposition, while detrital zircon data, conglomerate clast counts, sandstone modal petrog raphy, and paleocurrent indicators indicate initial Gangdese arc provenance from the north in the lower Kailas Formation, followed by provenance from the south in the upper Kailas Formation (DeCelles et al., 2011(DeCelles et al., , 2016;;Leary et al., 2016b).Leary et al. (2016a) interpreted growth strata in the Liuqu Formation within the footwall of a Great Counter thrust system splay (Fig. 4E; Leary et al., 2016a) to indicate syntectonic sedimentation.
Thermochronologic data from Gangdese batholith rocks just to the east of the Lazi region study area (Fig. 7) indicate that the Gangdese arc was being exhumed immediately prior to, during, and shortly after Kailas Formation deposition, with maximum age probability between 20 and 15 Ma (Figs. 5 and 7).The most proximal Gangdese batholith thermochronologic data, collected along an approximately north-south transect ~5 km to the northeast of the Lazi region geologic map (Figs.2A and 8), include three AFT ages between 25 and 23 Ma alongside one older age at ca. 28 Ma and two ages at ca. 9 Ma (Ge et al., 2016) and seven zircon (U-Th)/He ages between 23 and 17 Ma (Dai et al., 2013;Ge et al., 2016).Also of note is the short duration between Kailas Formation deposition (25-23 Ma) and subsequent exhumation, which began by 17 ± 1 Ma, based on zircon (U-Th)/He and AFT data from the Kailas For-mation along-strike to the west (Fig. 8; Carrapa et al., 2014).
Any tectonic model of Yarlung suture zone evolution during Oligocene-Miocene time must explain the geological, geophysical, and thermochronological constraints summarized herein and, in particular, provide a mechanism for Gangdese arc exhumation both during and after deposition of the Kailas Formation, synchronous with deposition of the upper Kailas Formation.If a north-dipping fault like the Gangdese thrust were not active at the same time as the Great Counter thrust system, then we would expect that that Gangdese batholith rocks would have experienced burial beneath the Kailas Formation and additional tectonic burial in the Great Counter thrust footwall during this period.Furthermore, between 25 and 23 Ma, the Kailas Basin was in a position of low relative elevation and characterized by a warm and wet climate (DeCelles et al., 2011(DeCelles et al., , 2016)), whereas today, the Kailas Formation is exposed at high elevation (4800-6700 m), draped along the southern margin of the Gangdese Mountains, with an arid-to semiarid climate and mean annual temperatures of approximately -5 °C at 5000 m elevation (Quade et al., 2011).A possible alternative explanation involves erosion of the Great Counter thrust system hanging wall(s) at a rate greater than or equal to the rate of rock uplift, preventing tectonic burial of the Gangdese arc rocks in the southern Gangdese Mountains (Fig. 1).We interpret that this scenario is unlikely given that it would require the Yarlung River to remain relatively stationary atop a zone of enhanced rock uplift throughout the period in which the Great Counter thrust system was active (constrained to 23-16 Ma).This scenario is also difficult to evaluate, as evidence for the existence of the Yarlung River during early Miocene time is limited to detrital zircon geochronology data from fluvial deposits in northeast India (e.g., Lang and Huntington, 2014).
Despite the obscurity of north-dipping structures such as the Gangdese thrust, and the poorly constrained magnitude of the Great Counter thrust due to a lack of hanging-wall cutoffs, the Yarlung suture zone in southern Tibet appears to have experienced large-magnitude shortening during Oligocene-Miocene time.Indeed, the >1000 km along-strike continuity of the Great Counter thrust system indicates that it accommodated significant shortening, while differential exhumation of Gangdese batholith rocks relative to Xigaze forearc (Fig. 1) or Tethyan Himalayan strata (primarily east of Lhasa) is consistent with the existence of a fault equivalent to the Gangdese thrust (Searle et al., 1987;Yin et al., 1994).It is possible that the Gangdese thrust reactivated the original, north-dipping megathrust-essentially the India-Asia suture projected downdip-that accommodated convergence between India and Asia.

Structural Model
Passive roof duplexes, which are characterized by imbricated, foreland-dipping thrust sheets in the absence of an emergent, hinterland-dipping fault, have been recognized in the field and in seismic data since the 1980s (Banks and Warburton, 1986).Physical experiments predict that passive roof duplexes are more common in zones of efficient surface erosion (Mora et al., 2014, and references therein), consistent with critical taper wedge mechanics of thin-skinned thrust belts (e.g., Davis et al., 1983).Since their first recognition, it has been postulated that efficient sediment storage in the adjacent monocline along the deformation front is critical to the formation of passive roof duplexes (Mora et al., 2014).In the Lazi region, evidence that might support the existence of a passive roof duplex includes: (1) the presence of imbricate, foreland-dipping (south-dipping) thrust sheets, (2) the absence of an emergent, hinterland-dipping fault, (3) early Miocene syntectonic sedimentation in a contractional setting in the upper Kailas Formation and the Liuqu Formation and (4) growth structures in the Liuqu Formation (Fig. 4E; Leary et al., 2016a), which indicate increasing southward dip of the basal unconformity throughout deposition, and (5) thermochronology data that indicate significant exhumation of the Gangdese batholith and Yarlung suture zone during Oligocene-Miocene time.
Our preferred model (Fig. 10) invokes an initial, late Oligocene (ca.26-23 Ma) phase of extension along a top-to-the-north normal fault, similar in orientation to the India-Asia suture projected downdip, to accommodate lower Kailas Formation deposition in an extensional basin.The rationale for late Oligocene extension is based on the work of DeCelles et al. (2011DeCelles et al. ( , 2016) ) and Leary et al. (2016b), who documented characteristic extensional basin architecture, growth strata, and paleoenvironmental indicators.These authors interpreted that rollback (DeCelles et al., 2011) or peeling-back (Leary et al., 2016b) of the subducted Greater Indian slab led to a phase of upper-crustal extension along the Yarlung suture zone during late Oligocene time.This phase of deformation is included in the structural model for the sake of completeness, though the nature of the subsequently formed contractional structures does not depend on prior development of an extensional basin along the Yarlung suture zone.
During early Miocene time, we interpret that the inferred normal fault that possibly accommodated Kailas Basin deposition was reactivated as a top-to-the-north thrust system that roots into a hinterland-dipping duplex beneath the Gangdese batholith (Fig. 10).South-directed thrusting fed slip into a system of top-to-thenorth thrusts (the Great Counter thrust system), which together comprise a foreland-dipping, passive roof duplex.In this model, splays of the hinterland-dipping duplex, not exposed but possibly evidenced by INDEPTH seismic reflection data (Fig. 9), can be considered equivalent to the Gangdese thrust of Yin et al. (1994).The lack of surface exposure of the Gangdese thrust in southern-central Tibet likely reflects the shallower depth of exhumation relative to southeastern Tibet, where Gangdese batholith rocks are juxtaposed against Tethyan Hima layan sequence rocks and the Yarlung suture zone assemblages are absent.
The structural model (Fig. 10) initiates with deposition of the Kailas Formation between 26 and 23 Ma in a fault-bounded, extensional basin associated with the inferred, north-dipping Kailas normal fault (DeCelles et al., 2011(DeCelles et al., , 2016;;Leary et al., 2016b).We favor this interpretation because it is explains the fanning southwarddipping growth strata in both the lower and upper Kailas Formation (Wang et al., 2015),  9), constraint "BTS" refers to the south-dipping reflectors interpreted as the downdip projection of the Great Counter thrust (Fig. 9), and constraint "4" is the imbricate north-dipping reflectors beneath the Yamdrok-Damxung reflection (Fig. 9), previously interpreted as a hinterland-dipping duplex (Alsdorf et al., 1998).Constraint "a" is the estimated thickness of the Greater Himalayan sequence beneath the Tethyan Himalaya based on exposures in the footwall of the Gurla Mandata metamorphic core complex, ~600 km to the west of the study area (Murphy, 2007).LHS-Lesser Himalayan sequence.the narrow-but-deep, "lacustrine sandwich" basin architecture that is typical of the Kailas Formation (DeCelles et al., 2016), and organic geochemical data consistent with warm-water lacustrine deposition, which are interpreted to reflect deposition at lower elevation than at present (DeCelles et al., 2011(DeCelles et al., , 2016)).The position of the Kailas Basin adjacent to the Gangdese magmatic arc, in the hinterland of the Himalayan thrust belt, implies that the Yarlung suture was likely at high elevation during the Oligocene, following >20 m.y. of collisional orogenesis (e.g., Najman et al., 2010;Hu et al., 2016).Therefore, the Kailas Basin might have transitioned to lower elevation between 26 and 23 Ma as the result of localized crustal extension (DeCelles et al., 2011).The 25-23 Ma AFT ages from Gangdese arc rocks proximal to the Lazi region study area (Fig. 7; Ge et al., 2016) might reflect erosional exhumation of Gangdese arc rocks during deposition of the Kailas Formation.
During deposition of the upper Kailas Formation, a transition to southerly provenance coincided with, or shortly preceded, a return to contraction along the Yarlung suture.Regional exhumation of the Gangdese batholith and Kailas Formation occurred mainly between 27 and 15 Ma, based on prominent 40 Ar/ 39 Ar biotite, ZFT and AFT, and zircon and apatite (U-Th)/He age-probability peaks (Figs. 7 and 8).Biotite Ar-Ar data (Copeland et al., 1987) provide the strongest evidence for cooling of Gangdese arc rocks between 26 and 23 Ma (Fig. 8).It is unclear whether these data can be explained by Kailas Basin extension (e.g., DeCelles et al., 2011), as the sample locations are in the hanging wall of the interpreted north-dipping normal fault (Fig. 7).Alternatively, these data could be explained by an earlier switch from extension to contraction east of the field area, as evidenced by recent summaries of the depositional history of the Kailas Basin (Leary et al., 2016b) and synthesis of regional tectonic data in the Himalaya (Webb et al., 2017).We interpret that subsequent 23-15 Ma exhumation was primarily driven by growth of the hinterland-dipping duplex beneath the Gangdese Mountains, resulting in structurally higher fault-bend folding to tilt the Kailas Formation southward, drive erosional exhumation in the southern Gangdese Mountains, and generate the regional northward dip of Linzizong Formation, in the northern Gangdese Mountains (Fig. 10).Duplexing would be expected in the hinterland of the Himalayan thrust belt, beneath the Yarlung suture zone, to regain crustal thickness and build taper following the extension episode associated with deposition of the Kailas Formation.Similarly, duplexing would also be expected based on the lithospheric tec-tonic model of Webb et al. (2017), in which rollback and breakoff of the Greater Indian slab following Kailas Formation deposition resulted in a dynamic steepening of the basal décollement, driving subcritical taper conditions after ca.20 Ma.
Our structural model provides an explanation for Miocene cooling ages in the Kailas Formation and Gangdese batholith in southern Tibet (Fig. 7), which in a simple structural model should have been experiencing tectonic burial rather than exhumation during Great Counter thrust shortening.As deformation progressed, three splays of the Great Counter thrust system initiated, possibly propagating from north to south based on the depositional ages of the upper Kailas Formation and the Liuqu Formation and trends in the Gangdese batholith and Kailas Formation thermochronologic data (Figs.7 and 8).
Orogen-parallel extension mainly initiated during Miocene time (16-12 Ma) along the suture zone in southern Tibet, providing a minimum age constraint for the Great Counter thrust system (Sanchez et al., 2013;Sundell et al., 2013;Laskowski et al., 2017).In addition, the ca. 10 Ma pluton exposed in the east part of the study area (Figs.2A and 5) provides an unequivocal minimum age for the northern splay of the Great Counter thrust system.Detrital zircon geochronology on samples from foreland basin deposits in India (Lang and Huntington, 2014), the termination of north-south sediment transport associated with the Kailas and Liuqu Formations (Leary et al., 2016a), and previous interpretations of Yarlung suture zone thermochronology (Carrapa et al., 2014) suggest that the Yarlung-Siang-Brahmaputra River system also initiated during early Miocene time.It is possible that the Yarlung River was established through a combination of topographic inversion (i.e., uplift of the Himalayas to higher elevation than the Yarlung suture zone)-thought to have taken place during Miocene time based on paleoelevation records and structural interpretations (Murphy et al., 2009)-and relief generation along the Gangdese Mountains related to duplexing.Middle to late Miocene AFT and apa tite (U-Th)/He ages from the Gangdese batholith, Kailas Formation, and Liuqu Formation (Figs. 5 and 9) likely reflect efficient erosional exhumation along the Yarlung River valley following establishment of this continentscale drainage system (Carrapa et al., 2014).The ca. 10 Ma leucogranite pluton that cuts across the Yarlung suture zone in the eastern portion of the study area (Fig. 2A) was possibly generated by isothermal decompression driven by orogen-parallel extension, or possibly by decompression related to slab breakoff (Webb et al., 2017).Its exhumation might be related to diffuse normal faulting-related, orogen-parallel extension, or efficient fluvial incision along the Yarlung River valley.

Lithospheric-Scale Tectonics and the Gangdese Culmination Model
Documentation of the timing and nature of Yarlung suture zone fault systems, and their relationship to contemporaneous sedimentary basins and igneous activity, enables investigation into the ways in which the suture zone geology might relate to important structures at lithospheric scale.The suture zone geology (Figs. 2 and 3) was integrated with the geology of the Himalaya (specifically, the South Tibetan detachment system) using constraints from geophysical data (Fig. 9).Aside from implementation of the Gangdese culmination model, developed herein, the key interpretation in Figure 10 is linkage of the South Tibetan detachment system and the basal décollement of the Great Counter thrust system.Although this relationship was not observed in the field, their similar timing (Fig. 8), spatial convergence (Figs. 1 and 10), and consistent top-to-the-north kinematics suggest that linkage was possible, if not probable (Yin et al., 1994(Yin et al., , 1999(Yin et al., , 2010;;Yin, 2006;Webb et al., 2007Webb et al., , 2017;;He et al., 2016).The South Tibetan detachment system forms the roof fault of the North Himalayan domes (e.g., Larson et al., 2010), exposed just to the south of the Yarlung suture zone (Fig. 1), consistent with the interpretation in Figure 10.
The geology of the Lazi region (this study), and possibly the Yarlung suture zone as a whole (e.g., Zhang et al., 2011;Laskowski et al., 2017), is characterized by alternating episodes of extension and contraction.The timing of tectonic mode switches appears to be consistent with changes in the behavior of the subducting Great Indian slab, characterized by contraction during episodes of shallow underthrusting, extension during rollback, and duplexing during slab breakoff and return to northward underthrusting (e.g., DeCelles et al., 2011;Laskowski et al., 2016Laskowski et al., , 2017;;Webb et al., 2017).The geology of the Yarlung suture zone in the Lazi region integrates well with these "subduction control" models.It is plausible that the Kailas Basin formed due to north-south extension associated with rollback of the Greater Indian slab between 25 and 23 Ma, with timing variation of up to 3 m.y.along strike (Leary et al., 2016b).Subsequently, the slab broke off, initially resulting in dynamic steepening of the Himalayan basal décollement (and subcritical taper conditions; Webb et al., 2017), a plausible driver of the switch from north-south extension (the Kailas Basin) to north-south contraction in the form of duplexing along the Yarlung suture zone.We interpret that this switch occurred at ca. 23 Ma based on the geology of the Lazi region, but significant along-strike timing variations are possible (e.g., Leary et al., 2016b;Webb et al., 2017).The tectonic mode switched again around 16 Ma, this time from north-south contraction (the Gangdese culmination) to east-west extension (Fig. 7).It is possible that this switch was related to renewed Greater Indian underthrusting, which might have produced critical gravitational potential energy in the upper crust via simple-shear crustal thickening (e.g., Sundell et al., 2013;Styron et al., 2015).

CONCLUSIONS
Structures and rocks units exposed in the Lazi region record a pronounced episode of late Oligocene-early Miocene contractional deformation that drove exhumation of the Yarlung suture zone and southern Lhasa terrane.Our mapping reveals the presence of southdipping imbricate splays of the Great Counter thrust system that juxtapose Tethyan Himalayan sequence strata, sedimentary-and serpentinitematrix suture zone mélange, Xigaze forearc basin strata, and the Kailas Formation, from south to north.Detrital zircon provenance analysis of the Tethyan Himalayan sequence reveals a difference between Triassic strata-derived at least partially from West Papua-and the rest of the Tethyan Himalayan sequence-derived from more local, Indian sources.Dominance of Cretaceous-Eocene detrital zircons in the Kailas Formation indicates local provenance from the Gangdese arc, and the likely absence of Oligocene-Miocene igneous rocks within the source area.Compilation of thermochronological data from a >1000-km-long swath along the southern Lhasa terrane and Yarlung suture zone reveals a dominance of late Oligocene to early Miocene (ca.23-17 Ma) exhumation, concomitant with slip across the Great Counter thrust system.The paradox of exhumation of the Kailas Formation and Gangdese batholith at the same time or shortly after they were experiencing tectonic burial to brittle-ductile conditions in the footwall of the Great Counter thrust system supports the existence of a north-dipping fault system that carried Gangdese batholith and Kailas Formation rocks in its hanging wall.Using constraints from INDEPTH seismic reflection data, we propose a new structural model for the Oligocene-Miocene Yarlung suture zone in which the Great Counter thrust system is a passive roof duplex that is kinematically linked with structures equivalent to the Gangdese thrust.The proposed existence of two detachment horizons structur-ally above (the Great Counter thrust) and below (the Gangdese thrust) a duplex beneath the Gangdese Mountains explains the lack of Gangdese thrust exposure west of the city of Lhasa in central-southern Tibet, and it provides a mechanism for suture zone exhumation concomitant with slip across the Great Counter thrust system and deposition of the Liuqu Formation.If this model is correct, then the Oligocene-Miocene contractional deformation along the Yarlung suture zone might have generated relief between the low-lying Yarlung suture zone and the high-standing Gangdese Mountains, which largely define the southern boundary of internal drainage on the Tibetan Plateau and the northern boundary of the incipient Yarlung-Siang-Brahma putra River watershed in southern Tibet.

Figure 2 (
Figure 2 (on this and following three pages).(A) Geologic map of the Lazi region study area, with geochronology sample locations and results, structural data for the Xigaze forearc (Kx) on an equal-area, lower-hemisphere stereonet, and geographical features referenced in the text.Detailed geologic mapping locales are indicated (detail maps 1-3).(B-D) Detailed geologic mapping locales of (B) the northern Yarlung River valley detailed mapping area (detail map 1), (C) the southern Yarlung River valley detailed mapping area (detail map 2), and (D) the Jiwa detailed mapping area (detail map 3).

Figure 4 .
Figure 3. Cross sections through the Lazi region map area along profiles A-A′ and B-B′ (Fig. 2A) with no vertical exaggeration (1:1 vertical to horizontal scale).Map units and colors are keyed to Figure 2. Structural data from nearby measurements were projected into the crosssection plane.

Figure 5 .Figure 6 .
Figure 5. Igneous U-Pb results shown as both weighted mean age determinations with 1σ error (left), plotted as a function of U/Th ratio, and probability distribution functions (PDFs) of U-Pb ages interpreted to reflect inheritance (right).Zircon dates excluded from the weighted mean age determination are indicated by the light weighted symbols.PDFs were generated using DZStats (Saylor and Sundell, 2016).MSWD-mean square of weighted deviates.

Figure 7 .
Figure 7. Top: Compiled thermochronological data for Gangdese batholith and Kailas Formation rocks in the southern Lhasa terrane, and an Early Cretaceous pluton in the northern Lhasa terrane, with thermochronological system indicated by the symbol shape and age indicated by color.Zircon (U-Th)/He and apatite fission track (AFT) ages for the Kailas Formation are indicated by the black dots and are labeled with thermochronologic age (zircon [U-Th]/He in bold).Major geologic structures, including the Gangdese thrust (GT), Great Counter thrust system (GCT), South Tibetan detachment, and active N-S rifts (in red) are plotted for reference (modified after Orme et al., 2015).The base map is a digital elevation model from the Global Multi-Resolution Topography Synthesis (Ryan et al., 2009) annotated with geographical features discussed in the text (top).Bottom: Gangdese batholith and Lhasa terrane data are plotted as a function of distance from the Yarlung suture zone ophiolites and ophiolitic mélange.Compiled thermochronological data are available in Supplementary TableDR3(see text footnote 1), which also contains references for data sources.ZFT-zircon fission track age.

Figure 8 .
Figure 8. Summary of tectonic events discussed in the text, including magmatism, sedimentation, deformation, and metamorphism.In addition, probability distribution functions (PDFs) of compiled thermochronological data are shown, separated by system and rock unit (Supplementary Table DR3 [see text footnote 1]), from the Yarlung suture zone across southern Tibet.Timing of South Tibetan detachment system (STDS) is interpreted from data summarized in Webb et al. (2017).Sample locations and ages are provided in Figure 7. PDFs were generated using DZStats (Saylor and Sundell, 2016).KF-Kailas Formation; GT-Gangdese thrust; GCT-Great Counter thrust; ZFT-zircon fission track age; AFT-apatite fission track; ZHe-zircon (U-Th)/He; AHe-apatite (U-Th)/He.
Figure 9. (A) International Deep Profiling of Tibet and the Himalaya (INDEPTH) common-midpoint projected seismic reflection profile from Alsdorf et al. (1998), including labels for prominent seismic reflectors.These include the Moho, the Main Himalayan thrust (MHT), the back-thrust system (BTS), the Yarlung-Zangbo reflector (YZR), the Yamdrok-Damxung reflection (YDR), and the Gangdese deep reflector (GDR).The simplified regional cross section presented in Figure 10 is shown atop these data to illustrate where our interpretations are supported by projected seismic reflection data.(B) Generalized tectonic map of southern Tibet, showing the location of the regional-scale cross-section C-C′ through the Lazi region map area (labeled with the box), and the location of INDEPTH seismic reflection profiles used to create the profile in A. Active structures (adapted from Taylor and Yin, 2009) are shown to highlight the coincidence of the INDEPTH profiles and rift hanging walls.

Figure
Figure 10.Restored (top) and modern (bottom) cross sections from the Himalayas, through the Lazi region study area, and into the southern Lhasa terrane (A-A′; Fig. 1) based on the geologic map in Figure 1, International Deep Profiling of Tibet and the Himalaya ( INDEPTH) seismic reflection data (Fig. 9), and our mapping results (Figs. 2A-2D).No vertical exaggeration.The syntectonic Liuqu Formation was deposited 20-19 Ma (Li et al., 2015; Leary et al., 2016a), so it does not appear in the restored cross section.Color-coding and unit names are keyed to Figures 1 and 2A.Structural data are indicated by the ball and tick mark symbols, whereas the circled numbers indicate constraints from previous studies.Constraint "YZR" is the subhorizontal orientation of structural fabric referred to by previous workers as the Yarlung-Zangbo reflector (Fig.9), constraint "BTS" refers to the south-dipping reflectors interpreted as the downdip projection of the Great Counter thrust (Fig.9), and constraint "4" is the imbricate north-dipping reflectors beneath the Yamdrok-Damxung reflection (Fig.9), previously interpreted as a hinterland-dipping duplex(Alsdorf et al., 1998).Constraint "a" is the estimated thickness of the Greater Himalayan sequence beneath the Tethyan Himalaya based on exposures in the footwall of the Gurla Mandata metamorphic core complex, ~600 km to the west of the study area(Murphy, 2007).LHS-Lesser Himalayan sequence.