Contrasting models of upper crustal shortening versus lower crustal flow have been proposed to explain the formation of thickened crust in the Longmen Shan (LMS), eastern Tibetan Plateau (TP) margin. These models require different structural kinematics along the LMS, whose structural geometry is defined by three parallel NW-dipping fault zones. From foreland (southeast) to hinterland (northwest), they are the Guanxian-Anxian Fault, Yingxiu-Beichuan Fault (YBF), and Wenchuan-Maowen Fault (WMF). Newly derived and previously published low-temperature thermochronology data from the LMS were synthesized to constrain the spatial exhumation and test previous models. The results show that (1) exhumation increases abruptly across the range-bounding YBF, suggesting the fault being the main thrust boundary between the LMS and the Sichuan Basin to the east; (2) Younger Late Cenozoic cooling ages are found on the hinterland WMF, where a dichotomy of ages on the hanging wall versus footwall suggests Late Cenozoic thrust activity; and (3) toward the hinterland to the west, exhumation rates decrease twofold over a distance of ~30–40 km. This exhumation pattern indicates a westward decrease of tectonic uplift, providing the regional topography approached a steady state, whereby exhumation is in balance with tectonic uplift. The observed exhumation estimates support an upper crustal configuration where thrusts in the LMS merge gradually into a gentle detachment seated at a depth of ~20–30 km. Results of this study support a revised upper crustal thrusting model.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The devastating M 7.9 Wenchuan Earthquake, 12 May 2008, in the Longmen Shan (LMS), eastern Tibetan Plateau (TP) margin, has triggered numerous studies on the earthquake physics [e.g., Liu-Zeng et al., 2009; Xu et al., 2009; Zhang et al., 2010a], geodynamics and Cenozoic tectonic uplift of the LMS [e.g., Burchfiel et al., 2008; Royden et al., 2008; Hubbard and Shaw, 2009; Zhang et al., 2010b], and the role of large earthquakes in forming the plateau margin [e.g., Parker et al., 2011]. The abrupt increase in crustal thickness, from ~35 km in the Sichuan Basin (SB), to the east of the LMS, to ~50–60 km beneath the eastern TP and LMS [e.g., Zhang et al., 2009], ranks the area as one of the world's premier natural laboratories for studying intracontinental crust dynamics (Figure 1) [Tapponnier et al., 1982, 2001; Royden et al., 1997, 2008]. It is generally regarded that the nucleation of earthquakes, thickened crust, and the high topography in the LMS were responses to ongoing outward growth of the TP. However, debate continues as to whether the dominant mechanism of deformation in forming the active plateau margin was by brittle upper crustal thrusting [Tapponnier et al., 2001; Hubbard and Shaw, 2009; Wang et al., 2011] or ductile lower crustal flow [Royden et al., 1997, 2008; Meng et al., 2006; Burchfiel et al., 2008; Zhang, 2013] (Figure 2).
 Uncertainty also exists as to the structural kinematics of the LMS. Structural geometry of the LMS is defined by three parallel NW-dipping fault zones, which are, from foreland (southeast) to hinterland (northwest), the Guanxian-Anxian Fault (GAF), the Yingxiu-Beichuan Fault (YBF), and the Wenchuan-Maowen Fault (WMF) (Figures 1, 3b, and 3c). Consensus has been reached regarding the Late Cenozoic thrust and dextral-thrust nature of the GAF and YBF [Xu et al., 1992; Burchfiel et al., 1995; Chen et al., 1995; Densmore et al., 2007]. However, as to the WMF, while dextral movement is unequivocal [Burchfiel et al., 1995; Densmore et al., 2007; Xu et al., 2008], debate still exists as to the nature of its Late Cenozoic dip slip between top-to-SE brittle thrusting [Xu et al., 1992; Liu et al., 1994; Hubbard and Shaw, 2009; Xu et al., 2009] and top-to-NW ductile normal faulting [Burchfiel et al., 1995; Meng et al., 2006].
 The geodynamic and structural controversies introduced above are, in fact, highly related (Figure 2). The model, highlighting brittle upper crustal shortening, apparently requires all of the three faults to be thrusts (Figure 2a). However, the ductile lower crustal flow model requires an upper crustal normal fault to accommodate the subvertical extrusion of lower crustal rocks (Figure 2b). This model argues that the flow of ductile lower crustal material away from the central TP, impeded by the strong crust of the Yangtze Craton beneath the SB, would pump up topographic relief in the LMS by inflating its lower crust [Royden et al., 1997, 2008; Clark et al., 2005a; Burchfiel et al., 2008]. In this model, the upper crust responds passively to the extrusion of deeper materials, and thus, normal faulting along the backside of the “channel” is predicted, as suggested by numerical modeling experiments allowing horizontal displacement [e.g., Beaumont et al., 2001; 2004]. It has been speculated that in the LMS, the WMF, along which and only ductile normal faulting features developed (e.g., low-grade mylonite with top-to-NW sense of slip) [Sichuan Bureau of Geology and Mineral Resources (SBGMR), 1991; Xu et al., 1992; Burchfiel et al., 1995], may correspond to the normal fault predicted by models [Burchfiel et al., 1995, 2008; Meng et al., 2006; Royden et al., 2008]. Zhang  proposed a revised channel flow model, arguing that the flow of ductile lower crust drags the upper crust eastward resulting in thickening of the LMS crust through high-angle listric reverse faulting (Figure 2c).
 Low-temperature thermochronology is a potential tool for constraining the nature of the WMF and fault propagation styles across the LMS. Differential uplift along the faults may result in differential exhumation and cooling patterns, which could be readily recorded by low-temperature thermochronometers [Gleadow et al., 2002; Ehlers and Farley, 2003; Reiners and Brandon, 2006]. Most data reported by previous work in the study region have focused on the central LMS, with few on the southern part [Arne et al., 1997; Kirby et al., 2002; Godard et al., 2009; Wang et al., 2012]. In this study, new zircon (U-Th)/He (ZHe) and apatite fission track (AFT) data from both sides of the WMF in the southern LMS are reported and placed in context with previously reported data in central LMS. Further, we evaluate the effects of radiation damage on ZHe ages in the LMS, the Late Cenozoic dip-slip nature of the WMF, exhumation and active faulting patterns across the LMS, and their geodynamic implications.
2 Morphotectonic Setting and Previous Thermochronology
 The eastern margin of the TP, the southern-central LMS, forms one of the world's most remarkable continental escarpments (Figures 1b and 1c). To the east of the WMF, the target fault of this study, elevations decrease abruptly from peaks exceeding 3000 m to ~500 m in the western SB over a distance of ~50 km. Calculated topographic gradients typically exceed 5% along this mountain front and rival any plateau margins elsewhere. Also noteworthy is that elevations retain their rising trend to the west of the WMF. Furthermore, in the LMS, short wavelength (<10 km) topographic relief commonly exceeds 0.5–1 km, as suggested by frequent and high magnitude oscillations of mean elevation (Figure 1c). This kind of feature is ubiquitous in young orogens, e.g., Himalayas and Southern Alps, where valleys are deeply incised.
 The LMS is a young reactivated orogen. In Mesozoic time, it was an intracontinental fold-and-thrust belt, accommodating transpressional convergence between the Songpan-Ganze terrane, which now forms a major part of the eastern TP and the Yangtze Block (including the SB). This Mesozoic orogenic event is documented by several lines of evidence, e.g., the development of a coeval foreland basin in the SB [Liu and Dai, 1986; SBGMR, 1991; Chen et al., 1995; Li et al., 2003] and Mesozoic structures and metamorphic events along the LMS [Xu et al., 1992; Burchfiel et al., 1995; Worley and Wilson, 1996]. Cenozoic outward growth of the TP [Royden et al., 1997, 2008; Tapponnier et al., 2001; Wilson et al., 2006], following the Indian-Eurasian continental collision, reactivated structures along the LMS and formed the present thickened crust (~60 km) from a Mesozoic crust which might be thicker than average (Figures 1 and 3).
 Surface exposures along the LMS are mainly Precambrian basement (Baoxing massif and Pengguan and Xuelongbao massifs in southern and central LMS, respectively) and Paleozoic marine sediments (Figure 3). The Baoxing massif, from which most samples for this study were collected, forms a core part of the southern LMS and is composed of two Precambrian basement sheets, separated by the NW-dipping Wulong fault (Figure 3c). The eastern boundary of the Baoxing massif is the NW-dipping Xiaoguanzi fault, where deformation is characterized by top-to-SE thrusting of Precambrian and Paleozoic rocks over Mesozoic sedimentary sequences of the SB (Figure 3) [Burchfiel et al., 1995; Tao, 1999; Jin et al., 2010]. In the central LMS, similar structures were developed—the WMF and YBF mark the western and eastern boundaries of the Pengguan massif, whose composition is similar to the Baoxing massif in the southern LMS (Figure 3). Following Burchfiel et al. , the structures in southern and central LMS are linked—the Wulong fault and Xiaoguanzi fault are the southern extensions of the WMF and YBF, respectively (Figure 3c). The terms YBF and WMF have been more widely used in previous literature and will thus be used here to replace local fault nomenclature (Wulong and Xiaoguanzi faults) in the southern LMS. We note that these faults display high dip angles (50–90°) at surface [Liu and Dai, 1986; SBGMR, 1991; Burchfiel et al., 1995; Jin et al., 2010] and would best be classified as “reverse faults”. Following previous studies, the term “thrust” is used here to describe them.
 The YBF and WMF in both southern and central LMS are all long-lived fault zones since at least Mesozoic time [SBGMR, 1991; Xu et al., 1992; Burchfiel et al., 1995; Chen et al., 1995]. The focus of this study is only on the Cenozoic behavior of these fault zones, and thus, related descriptions above and below all refer to Cenozoic time, if without specific notifications.
 The YBF and GAF are dextral thrusts and thrusts, respectively, as introduced above. As to the WMF, while dextral movements can be demonstrated unequivocally by the present day GPS velocity field [e.g., Zhang et al., 2004a; Shen et al., 2005] and geomorphologic features [Burchfiel et al., 1995; Densmore et al., 2007], debate still continues as to whether the sense of its Late Cenozoic dip slip is top-to-SE thrusting or top-to-NW normal faulting. This uncertainty arises because within the fault zone, structures showing both thrust and normal faulting activity, e.g., low-grade mylonite with top-to-west sense of fabrics, were all developed [SBGMR, 1991; Burchfiel et al., 1995; Xu et al., 2008; Jin et al., 2010], and their crosscutting relationships have not been clearly mapped [Liu and Dai, 1986; Liu et al., 1994; Burchfiel et al., 1995; Meng et al., 2006].
 Upper crustal shortening along the LMS has been studied at both long-term and short-term timescales. Recent reanalysis of industry seismic reflection profiles and surface geology in the LMS to the SB shows a considerable amount of upper crustal shortening [Hubbard and Shaw, 2009; Hubbard et al., 2010]. Without separating the Late Cenozoic component of shortening from earlier phases however, the total shortening shows a positive relationship with topography in both the LMS and southwestern SB, implying the importance of upper crustal shortening in building up the high topography in the LMS [Hubbard and Shaw, 2009; Hubbard et al., 2010]. As quantified by Hubbard et al. , more than 45 km of shortening has occurred over a distance of ~50 km across the LMS. On a short-term scale, interseismic GPS observations indicate that upper crustal NW-SE shortening along the LMS is ~1–3 mm/yr (= km/m.y.) [Chen et al., 2000; Zhang et al., 2004a]. Field measurements [Liu-Zeng et al., 2009; Xu et al., 2009] and the GPS velocity field [Zhang et al., 2010a] indicate that co-seismic shortening and uplift during the Wenchuan Earthquake are on a scale of 100s-mm–1000s-mm.
 Previous low-temperature thermochronology studies indicate that Late Cenozoic rock uplift and tectonics in the LMS have been accompanied by a considerable amount of cooling and exhumation (Figure 4). Arne et al.  reported apatite and zircon fission track (AFT and ZFT) and muscovite 40Ar/39Ar data from the LMS, which revealed two phases of cooling in the Early Cretaceous and Miocene times respectively. Kirby et al.  applied a combination of apatite (U-Th)/He (AHe), ZHe, biotite 40Ar/39Ar, and K-feldspar 40Ar release spectra techniques to three samples from the LMS to constrain their cooling histories and suggested that the high topography in the LMS is no older than the Late Miocene or Early Pliocene. Godard et al.  reported numerous AHe and ZHe data from the central LMS, providing further confirmation for the Late Miocene exhumation, and argued for a decreased phase (~0.3 km/m.y.) since ~5 Ma. Wang et al.  carried out a systematic low-temperature thermochronology study on samples from a vertical profile located at the Pengguan massif and suggested that the LMS may have experienced two phases of rapid exhumation during ~30–25 Ma and ~15–0 Ma. Further, within the datasets of Godard et al.  and Wang et al. , there are several ZHe ages characterized by high effective uranium content, [eU] = [U] + 0.235 × [Th] [Flowers et al., 2007], and these are significantly younger than those of lower [eU] grains in the same sample and adjacent samples from similar elevations. These younger ages are probably the result of enhanced He diffusion due to high alpha-radiation damage effect [Nasdala et al., 2004; Reiners, 2005] and will be investigated further in this work. Xu and Kamp  and Wilson and Fowler  reported several AFT and ZFT data from southern LMS as well, which are broadly consistent with those reported by other studies. Studies in areas adjacent to the LMS suggested enhanced Mid-Late Miocene cooling and exhumation [Clark et al., 2005b; Enkelmann et al., 2006; Ouimet et al., 2010; Duvall et al., 2012; Tian et al., 2012a].
3 Sampling, Methods, and Results
3.1 Sampling and Experimental Strategies
 Field sampling was carried out along two sections, crosscutting the Baoxing massif in the southern LMS (Table 1 and Figure 4a). Sampled rocks were fresh Precambrian granitoids and Precambrian-Paleozoic sandstones.
Table 1. Information on Samples Reported in This Study
Samples From the WMF Hanging Wall
Samples From the WMF Footwall
 As mentioned above, previous studies suggest that the LMS has experienced several kilometers of exhumation, implying that low-temperature thermochronometers with relatively high closure temperatures, e.g., ZHe and AFT, are more useful than those with lower closure temperatures, e.g., AHe, for uncovering geological processes occurred deeper in the past. Further, the short wavelength topographic relief (>0.5–1 km) in the LMS, as indicated above, could significantly affect the closure depths of low-closure temperature thermochronometers, e.g., AHe [e.g., Braun, 2002; Reiners et al., 2003] (for details, see auxiliary material). In light of the above information, AFT and ZHe methods were applied to the samples collected. Details of experimental methods are described in previous publications [e.g., Tian et al., 2012a, 2012b] and section 1 of the auxiliary documentation.
 Five AFT ages, from the WMF hanging wall, range between 5.3 ± 1.0 and 8.1 ± 2.2 Ma, whereas two samples (BX24 and BX30) from the WMF footwall yield apparently older AFT ages of 11.7 ± 1.6 and 10.6 ± 1.0 Ma (Table 2). Combining AFT data reported by Arne et al. , Xu and Kamp , and Wilson and Fowler  in the southern LMS and eastern TP, a clear age-elevation relationship is shown (Figure 5). The AFT age-elevation relationship for the data from the eastern TP indicates a low exhumation rate of ~0.03 km/m.y. Track length distributions for samples from the southern LMS and eastern TP differ markedly. Length distributions of samples from the southern LMS yielding young Late Cenozoic ages are characterized by an “undisturbed basement” type distribution [Gleadow et al., 1986] with long mean length (13.93 ± 0.14 µm), whereas those from the eastern TP yielding Mid Cenozoic ages are characterized by skewed distributions [Gleadow et al., 1986] with relatively shorter mean lengths (12.30 ± 0.24 µm).
Table 2. New Apatite Fission Track Results From the Southern LMSa
Brackets show number of tracks counted or confined lengths measured.
Standard and induced track densities measured on external mica detectors (g = 0.5) and fossil track densities on polished internal surfaces.
Age data determined by YT using zeta = 389 ± 5 for dosimeter glass Corning-5.
Lengths measured following 252Cf irradiation.
Samples From the WMF Hanging Wall
7.6 ± 1.3
5.3 ± 1.0
8.1 ± 2.2
7.5 ± 1.1
7.4 ± 0.8
14.03 ± 0.33 (14)
Samples From the WMF Footwall
11.7 ± 1.6
13.93 ± 0.14 (100)
10.6 ± 1.0
 In the WMF hanging wall, single-grain ZHe ages from eight samples range between ~7 and 13 Ma (Table 3). However, in the footwall, ZHe ages from five samples are older and range between 13 and 33 Ma, except for two anomalously younger ages of 11.4 ± 0.7 and 8.3 ± 0.5 Ma in samples BX30 and BX58, respectively. These younger ages are characterized by high [eU] (>1000 ppm) and high alpha dosage (>3 × 1018 α/g) (Table 3 and Figures 6a and 6b). The ZHe ages-elevation plot for samples from both sides of the WMF shows no or only a weak correlation (Figure 6b).
Table 3. New Single-Grain Zircon (U-Th)/He Results From the Southern LMS
FT is the α-ejection correction after Farley et al. .
Weighted means at 95% confidence level calculated using Isoplot V3.25 [Ludwig, 1991].
Effective Uranium content, [eU] = U + 0.235 × Th ppm [Flowers et al., 2007].
Alpha dosage, which is used to quantify radiation damage in nature grains [e.g., Nasdala et al., 2004], is calculated using equation 2 of Murakami et al.  and a zircon formation age of 844 ± 6 Ma [Yan et al., 2008] from central LMS.
High radiation damage grains with [eU] >1000 ppm and alpha dosage >3.0 × 1018 α/g.
 ZHe datasets for the LMS, reported by this work, Godard et al. , and Wang et al. , include several zircon grains with [eU] >1000 ppm and Neoproterozoic formation ages [e.g., SBGMR, 1991; Fu et al., 2013; Yan et al., 2008]. This [eU] value corresponds to an alpha dosage (as a measure of alpha-radiation damage), of >3 × 1018 α/g [e.g., Nasdala et al., 2004], as calculated using equation (2) of Murakami et al.  and a zircon formation age of 844 ± 6 Ma [Yan et al., 2008]. Such grains are probably not suitable to be used directly to interpret cooling histories of their host samples using the present He diffusion parameters derived from Fish Canyon Tuff zircon [Reiners et al., 2004]. Below we qualitatively analyze the effects of alpha-radiation damage on ZHe ages from the LMS and then interpret the geological significance of the data sets.
4.1 Influence of Radiation Damage on ZHe Ages
 Two grains, from the WMF footwall in the southern LMS, yield evidently younger ZHe ages (11.4 ± 0.7 and 8.3 ± 0.5 Ma of samples BX30 and BX58, respectively) than other grains from the same samples with lower [eU] and alpha-radiation damage (22–27 Ma and 13–22 Ma of BX30 and BX58, respectively) (Table S3 and Figures 6a and 6b). Compared to other samples from the same fault-wall, these two ages are also significantly younger by >10 Ma. These two grains are characterized by high [eU] (>1000 ppm) and high alpha dosage (>3 × 1018 α/g). Previous studies suggest that high radiation damage in zircons, as measured by alpha dosage higher than 2–4 × 1018 α/g, would increase the He diffusion rate and lead to anomalously younger ZHe ages [Nasdala et al., 2004; Reiners, 2005; Guenthner et al., 2013]. Therefore, these two relatively younger ZHe ages reflect a clear influence of high radiation damage on He diffusion and are not amenable to be used directly to interpret the thermal histories of their host samples.
 The Godard et al.  ZHe data set, derived from four vertical profiles and several additional isolated localities from the central LMS (Pengguan and Xuelongbao massifs) (Figures 1, 3, and 4), includes eight analyses with high [eU] (>1000 ppm) and high alpha dosage (> 3.0 × 1018 α/g) (Figures 6c and 6d and Table S1). Seven of the eight analyses are contained in two vertical profiles (East Pengguan and Wenchuan profiles) in the WMF footwall and yield ages of 8–15 Ma, which are significantly younger than other grains with lower [eU] content from the same sample or nearby samples, which range from ~15 to 30 Ma. A further high [eU] grain from the Minjing profile is also located in the WMF footwall. This grain shows an extremely high [eU] (43,214 ppm) and alpha dosage (129 × 1018 α/g) and yields an age of 7.5 ± 0.6 Ma, while other samples at similar or lower elevations yield older ages (9–20 Ma, except for one grain of 3.2 ± 0.3 Ma).
 The Wang et al.  ZHe data set, derived from a vertical profile located at the WMF footwall, includes as many as 33 analyses from six samples with high [eU] (>1000 ppm) and high alpha dosage (> 3.0 × 1018 α/g) (Figures 6e and 6f and Table S1). For most cases, the high [eU] ZHe ages are clearly younger than lower-[eU] ages for the same sample or adjacent samples at similar elevations. For example, in sample LME-15, three high [eU] grains yield ZHe ages between ~28 and 35 Ma, whereas two lower [eU] grains produce ages between ~39 and 48 Ma. In sample LME-17, five high [eU] grains produce ZHe ages between ~21 and 43 Ma (of which four out of the five are between ~21 and 35 Ma), whereas one lower [eU] grain gives an age of 49.6 ± 1.7 Ma. In sample LME-18, 13 high [eU] grains produce ZHe ages between ~20 and 53 Ma (of which eleven out of the 13 are between ~20 and 28 Ma), whereas three lower [eU] grains yield ages between ~35 and 80 Ma.
 What is noteworthy is that the abnormally young ages are all from the WMF footwall in both southern and central LMS. This probably indicates that those grains have similar He diffusion properties but differ significantly from those of Fish Canyon Tuff zircon standard [Reiners, 2005; Guenthner et al., 2013]. Therefore, the younger ZHe ages should not be used together with other lower [eU] grains to construct age-elevation relationships for constraining exhumation and thermal histories of the LMS.
 Even after excluding the young age anomalies from ZHe datasets of the WMF footwall, considerable intrasample age dispersion is still apparent (Figure 6). This dispersion could result from several factors, e.g., grain size variation, U and Th zoning, and micro-inclusions [e.g., Reiners, 2005]. However, only in a situation of slow cooling would the influences of such factors be significant [e.g., Flowers et al., 2009; Guenthner et al., 2013]. Therefore, the presence of strong radiation damage effects and the high ZHe age dispersion observed indirectly suggest that prior to Late Miocene rapid cooling, the WMF footwall may have cooled slowly through the ZHe partial retention zone, approximately 130–200 °C [e.g., Reiners et al., 2004].
4.2 Onset of Rapid Cooling/Exhumation
 Consistent with previous studies [Arne et al., 1997; Kirby et al., 2002; Godard et al., 2009], data presented here from the southern LMS suggest a Late Miocene phase of rapid cooling/exhumation with no indication for an earlier phase, as proposed by Wang et al.  based on data from the Pengguan massif in the central LMS. First, the relatively long and “undisturbed” AFT length distributions in southern LMS samples, yielding Late Miocene AFT ages (~5–11 Ma), suggest rapid exhumation through the AFT partial annealing zone during this time [Gleadow et al., 1986]. Further, in the WMF hanging wall, in both southern and central LMS, ZHe ages range between ~6 and 13 Ma with most lying between ~10 and 12 Ma, suggesting that enhanced cooling commenced at or before Late Miocene time (Figure 6). Last, both the presence of evident radiation damage effects and the high dispersion of ZHe ages in the WMF footwall indirectly suggest that prior to Late Miocene rapid cooling, the WMF footwall may have cooled slowly through the ZHe partial retention zone (as indicated previously).
4.3 Spatial Age Distribution and Exhumation Pattern
4.3.1 Spatial Age Distribution
 To facilitate comparison of spatial cooling/exhumation variations across the LMS, low-temperature thermochronology data compiled in Figure 4 were projected onto two NW-SE swaths across the southern and central LMS, respectively. Along the southern LMS swath (Figure 7a), AFT ages decrease westward from ~100 Ma in western SB to ~11–12 Ma in the WMF footwall, ~4–11 Ma in the WMF hanging wall, and then increase to ~10–60 Ma in areas further to the west. ZHe ages decrease from ~13–25 Ma (excluding younger age anomalies due to effects of high radiation damage) to ~6–13 Ma from the WMF footwall to the hanging wall. A few previously reported ZFT ages also follow this trend: a ZFT age from the WMF hanging wall gives an age of ~12 Ma, whereas two ZFT determinations further west are of Early Mesozoic age [Xu and Kamp, 2000].
 Along the central LMS swath (Figure 7b), a similar pattern is apparent. AFT ages decrease from ~40–168 Ma in the western SB to ~3–57 Ma (mostly between ~3 and 27 Ma) in the central LMS and eastern TP. ZHe ages decrease from ~194–446 Ma in the western SB to ~3–138 Ma (mostly between ~8 and 50 Ma) in the WMF footwall (excluding younger age anomalies due to effects of high radiation damage), and to ~5–13 Ma in the WMF hanging wall. From the eastern to western side of the Pengguan massif, ZHe ages decrease from ~60–138 Ma to ~3–20 Ma and then increase to ~20–80 Ma, suggesting differential cooling within the massif. AHe ages increase from ~3–5 Ma on the southeastern side of the Pengguan massif to ~8–22 Ma at the northwestern side of the Pengguan massif, before decreasing to ~2–3 Ma at the WMF hanging wall. ZFT ages from the central LMS are Late Mesozoic, whereas those from the eastern TP range between latest Cretaceous and Eocene. These systematic lateral variations in thermochronological ages in both southern and central LMS indicate differential cooling/exhumation between the western SB, the WMF footwall (i.e., YBF hanging wall), the WMF hanging wall, and the eastern TP.
4.3.2 Calculating Exhumation Rates
 The relationship between closure depth and age, to first order, yields an estimate of time-averaged exhumation rate, which can be calculated by relating each age to a cooling-rate-dependent closure depth through a series of equations with assumed parameters. Calculations of closure depths of different thermochronometers were carried out using the one-dimensional method of Brandon et al.  and Reiners and Brandon . This method assumes a steady-state one-dimensional upper crustal section with thermal diffusivity Κ (here assumed to be 20 km2/myr), thickness L (assumed to be 20 km, the thickness of the possible detached upper crust), and heat production HT (assumed to be 5 °C/m.y.). The upper and lower boundaries of this crustal section are held at surface temperature TS (assumed to be 10 °C) and TS + g0L, respectively, where g0 (assumed to be 20 °C/km) is the geothermal gradient prior to the onset of exhumation. Exhumation occurs at a constant rate ė, and removal of material from the top is assumed to be balanced by accretion below (e.g., crustal thickening), resulting in a constant advection rate toward the surface. For details, see section 2 of the supplementary information or Brandon et al.  or Reiners and Brandon . Further, assuming a syn-exhumation steady-state topography, closure depth for each sample was adjusted to account for local topographic effects by adding the difference between sample and mean local elevation (for a 10 km averaging window) to each closure depth determined by the approach cited. Calculated exhumation rates from the data compiled in Figure 7 were plotted onto the southern and central LMS swathes, respectively (Figure 8).
4.3.3 Spatial Exhumation Pattern
 Exhumation across both the southern and central LMS is evidently controlled by the structures (Figure 8). Along the southern swath (Figure 8a), calculated exhumation rate, in the western SB, to the east of the YBF, is as low as <0.1 km/m.y. Rates increase from 0.3–0.5 km/m.y. to 0.5–1.0 km/m.y. from the WMF footwall to the hanging wall. Further west, exhumation shows a westward decrease to 0.1–0.4 km/m.y. within a distance of ~40 km. To first order, the central LMS swath shows a similar pattern with minor differences (Figure 8b). Along this swath, the exhumation rates increase from <0.1 km/m.y. to 0.1– > 1.0 km/m.y. from the western SB to areas between the YBF and WMF. To the west of the WMF, exhumation shows a possible westward decrease from 0.6–1.0 km/m.y. to 0.4–0.6 km/m.y. (with few exceptions) within a distance of ~30–40 km. Within the Pengguan massif, a ~30 km wide section bounded by the YBF and WMF shows exhumation rates exhibiting a domed pattern—rates increase from 0.1–0.5 km/m.y. on the eastern side to 0.4–> 1.0 km/m.y. before decreasing back to 0.1–0.4 km/m.y. on the western side.
 The multithermochronology data approach in this study and the previously reported crustal heat flow studies provide an opportunity to monitor the influence of potential shortcomings with the one-dimensional method used above. The method does not account for any lateral heat advection by short wavelength topographic relief, which may effectively affect the closure depths of thermochronometers of relatively lower closure temperatures, e.g., AHe (see auxiliary material) [Braun, 2002; Ehlers and Farley, 2003; Reiners et al., 2003]. Further, the time-averaged rates calculated by this method may underestimate the post-Late Miocene exhumation of samples with the relatively older ZFT and ZHe ages in the WMF footwall. These older ages possibly and even likely record some pre-Late Miocene relatively slow exhumation histories, as explained in sections 4.1 and 4.2. This is clearly shown by the systematically lower ZFT exhumation rates (Figure 8b). However, exhumation rates calculated from AHe, AFT, and ZHe ages of each structural unit are consistent (Figure 8).
 Assuming a pre-exhumation geothermal gradient of 20 °C/km, calculations suggest that exhumation at a rate of ~0.3–1.0 km/m.y. (most of the final calculated exhumation rate in the LMS) would elevate the syn-exhumation geothermal gradient to ~23–30 °C/km (Figure S1). The syn-exhumation gradient is consistent with the present geothermal gradients in the LMS and eastern TP, which range between 22 and 32 °C/km, as constrained from thermal logging results of two deep (>4.5 km) boreholes (Figure 1) and two-dimensional thermal regime modeling along a crustal section from the western SB to the eastern TP [Xu et al., 2011a, 2011b]. The compatibility between modeled syn-exhumation and present geothermal gradients supports the thermal assumptions (see auxiliary materials for details). We acknowledge that changes in the assumed values of this model would lead to changes in the calculated exhumation rates. However, the change would be systematic for all data points, and thus, the spatial exhumation pattern would not change.
 The exhumation pattern across the LMS shows a clear correlation with structures (Figure 8). This can be conservatively summarized by the following points: (1) Abrupt increase in exhumation rate across the YBF. (2) Exhumation rates in the WMF hanging wall were higher than those in the footwall. What is worth noting here is that the ZHe, AFT, and AHe ages constraining the higher exhumation rate in the WMF hanging wall are all less than ~10–12 Ma in both southern and central LMS, suggesting that differential exhumation occurred across the fault in Late Cenozoic time. (3) To the west of the WMF, exhumation rates decrease rapidly within a distance of ~30–40 km. (4) In the WMF footwall in central LMS, exhumation shows a domal pattern. Below, we focus on the first three observations and discuss their implications for structural and tectonic models for the LMS.
5.1 Spatial Exhumation Constrained by Different Methods
 Different patterns of erosion across the LMS have been reported by previous studies [e.g., Godard et al., 2010; Kirby and Ouimet, 2011; Liu-Zeng et al., 2011]. Godard et al.  estimated the erosion rate in the eastern TP using an empirical lithology-dependent relationship (higher erodibility for sediment and lower for crystalline basement) between fluvial shear stress and erosion rates, and their results suggest a relatively slow erosion rate (<0.5 km/m.y.) in basement rocks (Pengguan and Xuelongbao massif) of the LMS front and relatively higher rates (0.5–1.0 km/m.y.) in the broader LMS hinterland of Paleo-Mesozoic sediments (>100 km width west of the WMF). Using an empirical relationship between stream-steepness and erosion rate, Kirby and Ouimet  calculated the erosion rate in eastern TP from stream-steepness. The results show that enhanced exhumation at a rate of 0.2–0.33 km/m.y. occurred in the Longmen Shan and areas ~60 km west of WMF. What should be noted is that the erosion rates used by Godard et al.  and Kirby and Ouimet  for calibrating the empirical models are a mixture of long-term and short-term rates, which are constrained by low-temperature thermochronology and cosmogenic isotope data, respectively. Liu-Zeng et al.  reported river sediment load data and delineated a decadal erosion pattern of the eastern TP, and the results show that erosion rate decreased abruptly from 0.4–0.7 mm/yr in drainage basins on crystalline basement and Paleozoic sediment in the LMS to 0.1–0.23 mm/y in basins on Mesozoic sediments and plutons to the west of the LMS.
 The spatial exhumation pattern constrained by individual low-temperature thermochronology data in this study differs from that reported by previous studies as follows. (1) To the west of the WMF, exhumation rates decrease rapidly from 0.5–1.0 km/m.y. to 0.1–0.6 km/m.y. over a relatively short distance of ~30–40 km (Figure 8). This finding is similar to the distance reported by Liu-Zeng et al.  but markedly shorter than that reported by Godard et al.  and Kirby and Ouimet . (2) Our thermochronology ages and exhumation rates derived from basement rocks and sediments belonging to the same tectonic unit are similar. For example, in the WMF hanging wall, basement rocks and Paleozoic sediments yield comparable thermochronological ages and million-year exhumation rates (Tables 2 and 3; Figures 4 and 8). This observation differs from that suggested by Godard et al.  who reported a dependence of millennial erosion rate on lithological variations.
5.2 Out-of-Sequence Thrusting in the LMS
 The first two observations in the spatial exhumation pattern, summarized at the end of section 4.3, are consistent with a LMS structure, which is characterized by out-of-sequence thrusting (Figures 2a and 2c). (1) Abrupt increase in exhumation across the YBF suggests that this fault forms the main thrust boundary between the SB and LMS, and this is supported by structures and topography across this fault zone (Figure 3). (2) The Late Cenozoic increase in exhumation from the WMF footwall to the hanging wall requires the WMF to be a Late Cenozoic active thrust, so as to transport mass upward to maintain the higher exhumation rate and topography in on the hanging wall side. Considering the Late Cenozoic thrusting along the GAF, it is apparent that the structural pattern of the LMS is characterized by three parallel thrusts with apparent dextral slip. Therefore, the structures of the LMS are out-of-sequence thrusts that developed or remained active in the hinterland of a fold-and-thrust belt [Morley, 1988].
 Short-term earthquake observations suggest that more frequent and larger earthquakes occur in the LMS hinterland than in its corresponding foreland (western SB), and this also supports an out-of-sequence thrusting structure [Liu-Zeng et al., 2009]. The out-of-sequence thrust faulting of the LMS provides a preliminary structural test for models explaining the Late Cenozoic crustal thickening and uplift in the LMS and eastern TP (Figure 2).
 Several studies have proposed that ductile normal sense of slip along the WMF may be associated with lower crustal “channel flow” in Late Cenozoic time [e.g., Burchfiel et al., 1995, 2008]. Evidence for ductile normal faulting along the WMF in both the southern and central LMS is preserved by a low-grade mylonite belt, showing top-to-NW sense of fabrics [SBGMR, 1991; Burchfiel et al., 1995; Xu et al., 2008]. However, its timing is still equivocal. Several informally reported Mid Cenozoic 40Ar/39Ar ages were considered as chronological constraints to the mylonization [Hames and Burchfiel, 1993; Burchfiel et al., 1995]. However, previously reported mica 40Ar/39Ar ages from the mylonite belt fall between Late Jurassic-Cretaceous time [Arne et al., 1997; Xu et al., 2008; Yan et al., 2011], suggesting the top-to-NW mylonization, is more likely a Late Mesozoic event.
 Based on the finding of an enhanced cooling phase during ~20–30 Ma at the WMF footwall in central LMS, recent studies [Wang et al., 2012; Oskin, 2012] speculated that thrusting along the WMF and faults further west was active in Oligocene time. As introduced and discussed above, data from both the WMF hanging wall and footwall in both the southern and central LMS indicates that in Late Cenozoic time, thrusting along the WMF has also been active; and this kind of activity may have persisted to the present day, as suggested by the relocation of historical earthquakes, with plenty possibly related to this fault [e.g., Huang et al., 2008; Wang et al., 2011].
5.3 Tectonic Models for Forming the LMS
 Based on the relationship between tectonic uplift, surface uplift and exhumation, the tectonic rock uplift rates are equal to exhumation rates, if the topography is in a steady state [England and Molnar, 1990; Willett and Brandon, 2002]. In the case, where the topography is transient, rock uplift and exhumation rates are unrelated.
 However, if the eastern TP is in, or even close to, a topographic steady state, the third observation in spatial exhumation pattern (westward decrease in exhumation rate), summarized at the end of the section 4.3, indicates that the rate of tectonic uplift decreases westward from the WMF to the interior of the eastern TP. Such a pattern is consistent with a crustal configuration where the steeply dipping thrusts in the LMS merge gradually into a gentle detachment seated at a depth of ~20–30 km (Figures 2a and 2c). This crustal architecture has been proposed by several studies [Lin, 1994; Liu et al., 1994; Burchfiel et al., 1995; Wang, 1996] and is supported by a variety of recent geophysical observations [e.g., Tang et al., 2008; An et al., 2009; Xu et al., 2009; Hubbard et al., 2010; Robert et al., 2010b; Zhang et al., 2010a]. For example, recent studies suggest that most of the earthquake slip, associated with the Wenchuan Earthquake, occurred in the shallow crust of the LMS along steeply dipping fault planes, which merge into a subhorizontal detachment fault extending for tens of kilometers beneath the LMS and eastern TP [Xu et al., 2009; Zhang et al., 2010a; Wang et al., 2011].
 The crustal architecture supported by this study cannot discriminate between the upper crustal shortening model (Figure 2a) and modified lower crustal flow model (Figure 2c) but it is not consistent with the lower crustal flow and extrusion model (Figure 2b). Considering the large amount of Cenozoic upper crust shortening proposed [e.g., Hubbard et al., 2010], we argue that it is not necessary to involve lower crust flow processes to explain the formation of the LMS. Similar conclusions have also been proposed by recent studies from the northeastern TP [e.g., Lease et al., 2012].
 Below, we synthesize available structural kinematic data of major strike-slip faults (Ganzi-Xianshuihe and Kunlun faults), bounding the Songpan-Ganze terrane (the major part of the eastern TP) (Figure 3a), to update the upper crustal shortening model for explaining the Late Cenozoic buildup of the LMS and eastern TP. In the updated model (Figure 9), thrusting along the LMS was triggered by the eastward extrusion of the upper crust of the Songpan-Ganze terrane along giant bounding strike-slip fault systems (the sinistral Ganze-Xianshuihe fault to the south and the sinistral Kunlun fault to the north). Such an extrusion occurred at a depth of ~20–30 km along a gently NW-dipping detachment, where deformation is probably ductile due to its relatively lower strength [Molnar, 1988]. The extruded rocks were backstopped, piled up, and thrust upward by either a NW-dipping crustal ramp [Hubbard et al., 2010; Liu-Zeng et al., 2011] or a series of listric-reverse faults [e.g., Liu et al., 1994; Burchfiel et al., 1995] beneath the LMS, so as to produce thickened crust and a large amount of uplift and exhumation therein.
 Regionally, the kinematics of large strike-slip faults bounding the Songpan-Ganze terrane support the updated model. A minimum time (~13 Ma) for initiation of ~60–80 km of sinistral transport [Wang et al., 1998, 2008] on the Xianshuihe-Ganze fault zone is well defined by the emplacement of the Konga Shan granitic intrusions at 12.8 ± 1.4 Ma (U/Pb zircon) [Roger et al., 1995], by coeval deformation dated by Rb/Sr (10–12 Ma) and 40Ar/39Ar ages (3–12 Ma) of syn-kinematic deformed rocks within the shear zone [Roger et al., 1995; Zhang et al., 2004b]. Similarly, a minimum time of initiation (~15 Ma) for the sinistral Kunlun fault zone, with slip offset of ~60–100 km [Yin et al., 1999; Fu and Awata, 2007], is constrained by Miocene and Pleistocene shoshonitic volcanism within pull-apart basins along the fault [Yang et al., 2002; Jolivet et al., 2003].
 To explain the thickened crust beneath the LMS and eastern TP, the updated model presented here requires large amounts of upper crustal shortening along the plateau margin. Several previous studies, based on interseismic GPS data [Chen et al., 2000; Zhang et al., 2004a] and the observation of lacking of a frontal Cenozoic foredeep [e.g., Burchfiel et al., 1995], have suggested that upper crustal shortening along the eastern TP margin was very limited. However, the recent study of Hubbard et al.  using seismic reflection profiles and surface geology suggested that more than 95% of shortening across the LMS (i.e., >45 km of shortening over a distance of <50 km in central LMS) has occurred and is more than sufficient to thicken the crust from an average of ~30 km to its present value of ~50–60 km [Hubbard et al., 2010]. Worthy of note here is that the shortening calculated by Hubbard et al.  should be considered as an integration of that occurred in both Mesozoic and Cenozoic time.
 The updated model presented here explains various geological and geophysical observations in and around the LMS. First, the proposed underlying gentle crustal detachment explains the westward decreasing exhumation pattern to the west of the WMF. Second, recent seismologic studies suggest lower P-wave velocity along the trajectories of the three fault zones in the LMS, and that these low velocity belts merge together at a depth of ~20 km beneath the eastern TP [An et al., 2009; Robert et al., 2010a, 2010b]. These seismic signals are consistent with the structural model proposed here (Figures 2a and 9). Third, the steeply NW-dipping thrusts beneath the LMS account for the slow horizontal shortening (1–3 mm/yr) [Chen et al., 2000; Zhang et al., 2004a] and the steep surface foliations [Shen et al., 2009]. Lastly, >95% of focal depths of historical earthquakes are shallower than ~20 km, suggesting stress and strain accumulation and relaxation in response to upper crustal shortening therein [Huang et al., 2008; Zhang et al., 2010a; Wang et al., 2011], as indicated by focal mechanisms of historical seismic events, e.g., the M 6.1 earthquake on 22 Sep, 1989 [Dziewonski et al., 1990] and M 7.9 Wenchuan Earthquake, 12 May, 2008 (Figures 1b and 9).
 Results presented in this work lead to the following conclusions.
 Enhanced cooling/exhumation in the southern LMS commenced in Late Miocene time.
 The dextral YBF thrust, across which exhumation increased abruptly from <0.15 km/m.y. in the SB to 0.2–1.0 km/m.y., acts as the main boundary structure between the SB and LMS. In Late Cenozoic time, the WMF, whose hanging wall experienced a greater magnitude of exhumation than the footwall, has behaved as an active dextral thrust. To the west of the WMF, exhumation decreased from 0.5–1.0 km/m.y. to 0.1–0.6 km/m.y. within a distance of ~30–40 km. These 106 year scale observations suggest that active thrusting along the LMS is of an out-of-sequence style.
 The exhumation pattern across the LMS favors a crustal configuration where the steeply NW-dipping out-of-sequence listric-reverse faults in the LMS merge gradually into a gentle detachment at a depth of ~20–30 km beneath the eastern TP (Songpan-Ganze terrane).
 The crustal geometry and kinematics, indicated by this study, cannot discriminate between an upper crustal shortening model and a modified lower crustal flow model but are not consistent with a lower crustal flow and extrusion model. However, taking into account the considerable amount of Cenozoic upper crust shortening, we tentatively argue that it is not necessary to involve lower crust flow processes to explain the formation of the LMS.
 An updated upper crustal shortening model for explaining the Late Cenozoic crustal thickening and uplift in the LMS is presented. It is argued that shortening in the LMS was triggered by eastward extrusion of the upper crust of the Songpan-Ganze terrane along a detachment fault at a depth of ~20–30 km. The extruded rocks were backstopped, piled up, and thrust upward by steeply dipping, out-of-sequence thrusts beneath the LMS, so as to build up the thickened crust and high elevations in the LMS and eastern TP. The model is consistent with a number of geological and geophysical observations.
 The effect of alpha-radiation damage on ZHe ages is recognized in Precambrian zircons in the LMS. The effect leads to relatively young ages for grains with high [eU] (>1000 ppm) compared to those with lower [eU] content. The current understanding of zircon He diffusion systematics does not allow these ages to be interpreted robustly within the context of regional cooling histories.
 Funding for this research was provided by China National Oil-Gas Major Special Projects (2011ZX05008-002). The University of Melbourne thermochronology laboratory receives infrastructure support under the AuScope Program of NCRIS. YT received support from IPRS and MIRS scholarships at the University of Melbourne. YT is grateful to Abaz Alimanovic for assistance with (U-Th)/He dating; to Sergei Medvedev and Chris Beaumont for discussions on channel flow models; and to Chuangqing Zhu, Ming Xu, Zhaokun Yan, Zhiwu Li, and Zhonghua Tian for their assistance during seasons of fieldwork. Constructive reviews by Richard Lease and Eva Enkelmann and editorial work by Todd Ehlers and Andrew Carter are gratefully acknowledged.