In the central Himalaya, an abrupt physiographic transition at the foot of the Greater Himalaya (PT2) marks the southern edge of a zone of rapid rock uplift along a ramp in the Main Himalayan Thrust (MHT). Despite being traceable along ~1500 km of the central Himalaya, PT2 is less distinct in western Nepal, reflecting along-strike changes in MHT geometry and/or a migrating locus of midcrustal deformation, the details of which have important implications for seismic hazard in western Nepal. New mineral cooling ages (apatite and zircon U-Th/He and muscovite Ar-Ar) from a series of relief transects provide constraints on exhumation rates and histories in western Nepal. Inversion of these data using Pecube and QTQt models yields results that require rapid (~1.4–2.7 mm/yr) exhumation in the rocks near the along-strike projection of PT2 until around 9–11 Ma, followed by much slower (~0.1–0.4 mm/yr) exhumation until at least the late Pliocene. In contrast, transects from ~75 km hinterlandward are best fit by rapid exhumation rates (~1.5–2.1 mm/yr) over at least the past ~4 Myr. Midcrustal deformation in western Nepal is occurring well north of the position expected from along-strike structures in central Nepal, and a growing dataset suggests that rapid exhumation has been sustained there since the late Miocene. These new constraints on the late Cenozoic exhumation history of the western Nepal Himalaya provide key insight on the active structures behind the complex seismic hazards in the region.

The collision of the Indian and Asian plates in the central Himalaya is primarily accommodated along the Main Himalayan Thrust (MHT) [1], which is capable of generating Mw 8+ earthquakes [2]. In order to improve preparedness for future earthquakes, it is critical to develop a better understanding of the locations, geometry, and slip rates of the faults responsible for that seismic hazard.

The prevailing models for active tectonics of the central Himalaya lean heavily on the numerous studies targeting the more readily accessible regions of central Nepal. These multifaceted investigations have converged on a late Cenozoic kinematic history that includes a few distinct tectonic stages: thrusting of the highly metamorphosed Greater Himalayan Series (GHS) over the Lesser Himalayan Series (LHS) along the Main Central Thrust (MCT) and Ramgarh Thrust and/or Munsiari Thrusts from the early- to mid-Miocene, thrusting of the LHS over early foreland basin sediments in the late Miocene along the Main Boundary Thrust (MBT), and thrusting of Miocene–Quaternary foreland basin sediments over modern foreland sediments along the Main Frontal Thrust (MFT) system, which currently underlies the active fold-and-thrust belt at the orogenic front [3]. Debated aspects of this history include the longevity and total shortening along each of the above structures and the relative roles played by duplexing and out-of-sequence thrusting in the development of structures and topography (e.g., [4-9]).

In the central Himalaya, ~15–20 cm/yr (~40%) of Indo-Asian convergence has been accommodated as shortening across the Himalaya [10-14]. Most of that shortening occurs as displacement along the MHT, the basal plate boundary thrust into which earlier thrusts sole [15]. A variety of studies support a flat-ramp-flat geometry for the modern MHT, wherein the orogenic wedge deforms into a macroscale fault-bend fold above the steeper “ramp” between gentler segments up and down dip [1]. That rock uplift is responsible for sustaining the high elevations and steep slopes and rivers of the Greater Himalaya, although some studies have implicated a surface-breaking fault at the foot of the high range (e.g., [16]).

The presence of a midcrustal MHT ramp in central Nepal is supported by evidence for a strike-parallel zone of rapid late Cenozoic rock uplift generally within the Greater Himalaya and bounded on the south by a physiographic transition known as PT2 [5]. Within a ~50-km-wide belt north of PT2, hillslopes steepen to or beyond threshold angles, local relief increases, and river longitudinal profiles show a broad convexity [17-20], and millennial-scale erosion rates [11, 21, 22] and long-term exhumation rates reach maxima [7, 16, 23, 24, 24, 25].

The midcrustal ramp in the MHT is commonly assumed to be the locus of a strike-parallel belt of microseismicity [24, 26] and initiation of large earthquakes, in part because it hosts the transition zone between the deeper, creeping, and the shallower, locked segments of the fault (e.g., [1]). Geodetic observations of the earthquake cycle in central Nepal show rapid interseismic rock uplift and subsidence north and south of PT2, respectively [1, 27]. This elastic deformation is likely only partially recovered during large earthquakes and, combined with net transport over ramps on the MHT, contributes to net rock uplift of the Greater Himalaya [28]. This generalized working model was lent further support by the catastrophic 2015 Mw 7.8 Gorkha earthquake. There is broad agreement that this earthquake ruptured a ~140-km by 50-km patch of the locked portion of the fault along the MHT “flat” beneath the Lesser Himalaya of central Nepal (Figure 1; [1, 29, 30]), but there is some disagreement about whether the down-dip margin of the rupture patch is at the top of the MHT ramp or if the rupture plane extends beneath the Greater Himalaya with no ramp, but instead an out-of-sequence, surface-breaking thrust fault [8].

The aforementioned model for active tectonics and topographic growth in central Nepal characterizes rather well the dominant range-front Physiography, geology, and seismicity of the central Himalaya, including a ~600-km-long segment that stretches from midwest Nepal to the Western border of Bhutan (~82.5–87° E) and a 400-km-long segment in Uttarakhand, Northern India (~77.5–81° E) (Figure 1). These long segments of along-strike similarity have led some to describe the MHT geometry as somewhat continuous across the entire central Himalaya (e.g., [26, 31]). However, there is an increasing recognition of along-strike discontinuities, as revealed by geomorphology [19, 32], bedrock geology [33], late Miocene-to-present exhumation patterns [25, 34], and geodetic and seismic observations [14, 30, 35-37].

In this contribution, we seek to provide greater clarity on the location, geometry, and history of the MHT at one of these more conspicuous discontinuities in western Nepal, a relatively poorly studied segment of the central Himalaya. Evidence from paleoseismology points to a broad “seismic gap” in Western Nepal where no large earthquake has broken the MHT to the surface at the MFT in at least 600 years [38, 39]. Meanwhile, the region is facing rapid development, including a number of large roads, bridges, and hydropower projects. Local populations are vulnerable to even smaller earthquakes, exemplified by the 2023 Mw 5.7 earthquake in Jajarkot that killed over 200 people and collapsed over 25,000 buildings (Figure 1(b)). Thus, it is essential that we develop a clearer understanding of the earthquake cycle, structural geometry, and kinematics of the MHT in Western Nepal.

1.1. Study Area

We focus on one of the more apparent along-strike discontinuities that lies in West central Nepal between ~81.5 and 82.5° E. Harvey et al. [19] described a topographic and tectonic discontinuity where, in contrast to along-strike segments, the high peaks of the Greater Himalaya retreat ~75 km toward the hinterland and where the belt of microseismicity and elevated topographic metrics that characterize PT2 and the zone of rapid rock uplift appear to bifurcate for ~100 km (Figure 1). Harvey et al. [19] identified two physiographic transitions that define the Northern and Southern branches of that bifurcation: PT2-N and PT2-S, respectively (Figures 1 and 2). In between PT2-N and PT2-S, a broad, high-elevation, low-relief landscape lies in the along-strike projection of the zone of rapid rock uplift above the MHT ramp. In Western Nepal, the physiography at PT2-N more closely resembles that of PT2 in central Nepal and Northern India, leading Harvey et al. [19] to hypothesize that the active MHT ramp and associated rapid rock uplift bend hinterlandward along PT2-N. They interpret PT2-S as the surface manifestation of a younger, midcrustal duplex driving uplift of a broad region and the preservation of a low-slope, low-relief landscape at midelevations (Figures 1(b) and 1(c) and 2). Other recent analyses have arrived at similar explanations for this along-strike discontinuity in Western Nepal (e.g., [14, 37, 40-42]). Despite that recognition, even recent seismic hazard models for Nepal generally assume the MHT ramp projects across Western Nepal with no major deviations (e.g., [43]).

As in the rest of the central Himalaya, the bedrock geology of Western Nepal is dominated by the remnants of thrust sheets that accommodated shortening since at least the Miocene and are now folded and eroded [13]. The highest part of the range (the Greater Himalayan zone, generally north of PT2-N) contains the core of the MCT sheet, which comprises highly metamorphosed gneisses, schists, and plutons of the GHS [44]. This thrust sheet extended well to the south where it overrode the metasediments of the LHS by at least 100 km along the correlative Dadeldhura Thrust (Figure 2b) [13]. The foreland margin of this thrust sheet cooled from south to north in the late Miocene [45]. The dissected remnants of this thrust sheet include, from west to east, the Dadeldhura Klippe, Karnali Klippe, and Jajarkot Klippe [45-47] (Figures 1(a) and 2).

Where exposed near PT2-N, the MCT zone dips ~20–40° to the north-northeast [44, 48]. To the south, this fault has been folded into a high-amplitude anticlinorium above a “Lesser Himalayan duplex” (LHD) active during the late Miocene and perhaps more recently (Figure 2) [46, 49] and [48, 50, 51]. Erosion of that structural culmination has exposed a window into the highly deformed LHS rocks beneath. South of this window, the GHS rocks in the Karnali Klippe have been folded into a broad synform, dipping toward the hinterland near PT2-S (Figure 2(b)).

Underlying the GHS is the LHS, exposed in the erosional window above the LHD and again south of the Karnali Klippe. As is the case in central Nepal, LHS rocks were thrust over Miocene–Pliocene foreland basin sediments (the Siwalik Formation) along the MBT, the trace of which lies ~35-km basinward of the southern margin of the Karnali Klippe [48].

There are some constraints on the timing of activity of the above thrusts. The MCT was active throughout the early and middle Miocene [44] but was perhaps rejuvenated in the latest Miocene [45, 52], and the Karnali Klippe was fully emplaced by ~14 Ma, as constrained by a south-to-north cooling pattern [45] and by deposition of the Lower Siwalik foreland basin strata which derived sediment from this klippe [46]. The growth of the LHD must postdate the main phase of activity on the MCT, but it is unclear by how long. Robinson and McQuarrie [50] infer that it grew mainly during the late Miocene to early Pliocene. There is no great certainty on the direction of duplex growth, but the typical expectation is that it has been propagating to the foreland (e.g., [41, 48, 50]). The MBT appears to have initiated by ~8 Ma and has been active into the Pliocene, but this timing is only loosely constrained via signatures in the detrital record at the mouth of the Karnali River [51, 53]. Evidence for Holocene surface ruptures on the MBT suggests that it may still accommodate some of the shortening in the wedge in Western Nepal [54].

It is generally assumed that in the Quaternary, the majority of shortening has been accommodated by the MHT/MFT, but the timing of the shift of shortening from the MBT to the MFT or partitioning of strain between them remains unclear. Similarly, ongoing partitioning of shortening onto more northerly structures (out-of-sequence, surface-breaking thrusting) has been hypothesized, but not thoroughly tested [41, 52, 55]. Also of relevance to this study is the recognition by Murphy et al. [55] of an oblique dextral fault zone stepping across the range in Western Nepal (the Western Nepal fault system [WNFS]), highlighting the role of arc-parallel extension along parts of the WNFS in accommodating convergence in western Nepal since the mid-Miocene [56].

In sum, spatial relations among bedrock geology, structure, geomorphology, and seismicity of Western Nepal stand in contrast to the more laterally continuous segments along strike to the NW and SE (Figure 1) and therefore require a model for the structure of the modern MHT that is distinct from the model that seems to work for central Nepal.

Some interpretations of MHT geometry in the study area in Western Nepal include a mostly planar decollement with a single ramp positioned just south of PT2-S [13, 46], perhaps with a separate brittle–ductile transition down dip along the detachment near the LHD [35]. Robinson and McQuarrie [50] propose shifts in the position of footwall ramps and imposed uplift belts over time, but for the Pliocene-present interval, they infer a smaller MHT ramp beneath PT2-S and a long, very gently dipping segment to the north, beneath the Greater Himalaya, a model that resembles one put forth by Huyghe et al. [51] to explain detrital records at the mouth of the Karnali river. Harvey et al. [19] used an analysis of the topographic and seismic discontinuity in Western Nepal to invoke rapid uplift along a ramp along PT2-N sufficient to build a mountain front characterized by steep slopes and high topographic relief, followed by a more recent basinward stepping of that midcrustal ramp to PT2-S via footwall accretion/duplex formation. This model provides an explanation for the presence of what appears to be an uplifted, low-relief landscape and the bifurcated belt of microseismicity and elevated geomorphic metrics like slope and river steepness index (Figure 2(b)). Hubbard et al. [30] developed a model that invokes two ramps (positioned south of PT2-N and PT2-S) at different depths along the MHT in Western Nepal, rather than an active duplex (Figure 2(b)). Hoste-Colomer et al. [37] use microseismicity hypocenter data to infer active deformation at a ramp positioned near PT2-S that they liken to a ramp in the floor thrust of a duplex, as well as distributed active deformation in the main LHD zone (closer to PT2-N) (Figure 2(b)). Similarly, Olsen et al. [42] include two smaller ramps separated by flats—one near PT2-S and one near PT2-N—similar to the model from Hubbard et al. [30] (Figure 2(b)).

The aforementioned models for the MHT structure in Western Nepal remain somewhat unreconciled, but their viability can be tested using low-temperature thermochronology to constrain late Cenozoic exhumation histories across the study area. Cooling histories of rocks now at the surface are commonly used to deduce periods of exhumation driven by rock uplift, which, under the prevailing conceptual model for the central Himalaya, happens mostly above steeper ramps in the MHT and analogous thrusts [1]. Thus, in this manuscript, we present a suite of new low-temperature cooling ages aimed at constraining the spatial pattern of rock uplift and exhumation in Western Nepal, while clarifying the nature of the discontinuity in the MHT between central and Western Nepal.

2.1. Low-Temperature Thermochronology

Most existing cooling age data in the central Himalaya are apatite and zircon fission-track ages, with rate-dependent closure temperatures of ~100–140°C and 220–280°C, respectively [23, 40, 57], along with 40Ar/39Ar on muscovite with a closure temperature range of ~320–370°C [7, 58, 59] (although it is possible that the effective closer temperature is more like 450°C [60, 61]). More recent efforts have added apatite and zircon [U-Th]/He ages [7, 23], with closure temperatures of ~55–85°C and 160–200°C, respectively [62, 63].

Many previous Himalayan thermochronology studies have relied on samples collected along more accessible valley bottoms (e.g., [25, 40, 57]). Although commonly more difficult to collect, relief transects that span more than 1 km of elevation over short horizontal distances can provide more robust insights on cooling histories and changes in rates through time (e.g., [7, 64]), especially if multiple thermochronometers are used [65]. That is the approach used here.

We analyzed a new suite of apatite and zircon [U-Th]/He and muscovite 40Ar/39Ar cooling ages for Western Nepal. Samples were collected along seven relief transects: four rising above the Humla Karnali and Chuwa Khola rivers in the Greater Himalaya of far northwest Nepal, well north of PT2-N, one from just north of PT2-N along the Mugu Karnali river, and two from the more distal part of the Karnali Klippe to the south, closer to PT2-S (Figure 3(f)). Transects typically rise ~1.5–2.5 km above main stem rivers to adjacent ridges over as short a lateral distance as was feasible. Care was taken to only sample bedrock that was in its original position and was relatively unaltered by weathering or obvious hydrothermal flow. Samples were crushed and processed to isolate fractions containing apatite, zircon, and muscovite following standard mineral separation techniques. Geochemical analysis and resulting ages were determined using standard procedures (analytical details can be found in Supporting Information Text S1). For each apatite and zircon sample, multiple aliquots (grains) were analyzed. Ages are reported as weighted means of accepted aliquot ages calculated in IsoplotR [66], with standard errors calculated while allowing for “random effects” from nonanalytical sources of error.

2.2. Thermal History Modeling with QTQt

For the few samples that yielded ages with all three dating techniques, we used QTQt (67; to explore the likely range of T-t histories that can explain the observed cooling ages. We refer the reader to Gallagher [67] and Abbey et al. [68] for explanation of the QTQt methodology. In summary, it can be run in the forward mode to predict cooling ages given a thermal history or in the inversion mode to constrain thermal histories based on known ages.

We used QTQt in the inversion mode to determine the likely range of thermal histories that could reproduce the cooling ages from those samples. Single-grain ages, age uncertainties, and grain dimensions were supplied for both apatite and zircon U-Th/He. A T-t constraint box from the Ar/Ar cooling age was established to force the T-t paths through that constraint (T = 350–450°C, t = measured age ± analytical error). Other parameters including which radiation damage models were used and inversion scheme can be found in the online supplementary information.

2.3. Estimating Exhumation Rates from Cooling Ages

Cooling ages collected along relief transects in rapidly uplifting mountain belts are commonly interpreted in age-elevation space because in a simple system, where rocks are advected vertically through horizontal isotherms, the slope of a line regressed through age-elevation data can be interpreted as a time-averaged, local exhumation rate. However, this approach is complicated by the relief-induced perturbation of isothermal surfaces for lower temperature thermochronometers like apatite [U-Th]/He and fission track (e.g., [62, 65]). To effectively account for this effect while interpreting our cooling age dataset, we use the Pecube (version 3) modeling package, which allows the user to define a topographic surface into which the modeled particles are advected [69, 70].

The Pecube forward model first uses imposed kinematics to back calculate where rocks at the surface today would have been at the start of the model run, then it solves the heat-transfer equation in three dimensions as it advects those particles back to the surface. In the absence of any specific constraints to the contrary, we maintain a steady topography throughout the model duration with the acknowledgment that changing relief and drainage reorganization could have a nontrivial impact on predicted ages and on age-elevation trends where exhumation has been slow [65, 71]. Values assumed for other boundary conditions are derived from the literature and listed in the supplementary materials.

To home in on realistic exhumation rate scenarios, we use an inversion scheme that uses the neighborhood algorithm (Sambridge, 1999) to iteratively sample a parameter space for a suite of forward-model runs that attempt to minimize the misfit between modeled and measured ages [69]. The total misfit (μ) is calculated as:

μ= i=1n(mioiσi)

where n is the number of data points, and for each data point i, oi is the observed age, mi is the modeled age, and σi is the 1-σ error on the measured age. During a first iteration, random values of the parameter(s) to be optimized are generated, run as forward models, and compared to observed ages. Subsequent iterations sample an increasingly restricted parameter space defined by the best-fitting parameters in the previous iterations. After a sufficient number of iterations, the model converges on a best-fit scenario and parameters for the input cooling ages.

Although Pecube can be used to constrain the geometry, slip rates, and timing of specific faults within a model domain, deriving meaningful results from such an effort requires a good spread of cooling age constraints across the model domain, sufficient constraints on thermal parameters of the involved crust, and rather simplified kinematic scenarios. Because of the limited spread of cooling age constraints and likelihood of a rather complex kinematic history on important faults in the study domain, we chose to take a more conservative approach that focuses on constraining exhumation histories for several distinct blocks across the study area.

The study area was divided into six model “blocks,” sized to fully encompass 1–2 relief transects and surrounding topographic features that could influence isotherm depth: four blocks north of PT2-N containing one transect each (SRT1, SRT3, SRT4, and SRT5), an eastern block encompassing the transect closest to PT2-N (JRT-1), and a southern block encompassing the two transects closest to PT2-S (JRT3 and JRT4) (Figure 3(f)). For each block, we impose vertical advection of the model block through the modern topography. We use hole-filled digital elevation models from the Shuttle Radar Topography Mission (90 m nominal resolution [72]) downsampled to a 450-m grid during the model runs. Transects with young (<~4 Ma) cooling ages were inverted with a single free parameter, an uplift rate (i.e., exhumation rate, ER1), characterizing the entire model run. Transects with older ages were inverted with three free parameters: ER1, a transition time (t_change) at which a second uplift rate (ER2) replaces the original and continues through the present to allow for changing exhumation rates through time. Not all reported ages were included in the inversion; where apatite ages were older than zircon ages, they were not included due to the disproportionate impact of those less certain age constraints on the inversion results.

For the one-parameter model runs, the misfit function takes the form of a concave-up parabola, with the best-fit model(s) at its minimum. For model runs with three parameters, the misfit function is defined by a surface for any two parameters, rather than a simple parabola. Uncertainties in model inversion results are difficult to adequately quantify given the possible ranges of all model parameters, including those that are not included as free parameters in the inversion. Because of the futility of accurately representing them, we do not attempt to report uncertainty limits on the inversion results. That said, each inversion does yield a “region” of low misfit, which we report. We focus our discussion on the spatiotemporal variability of those low-misfit exhumation rates, rather than on the exact magnitude of said rates.

3.1. Apatite and Zircon U-Th/He Cooling Ages

Of 57 bedrock samples that were processed for analysis, 39 apatite and 47 zircon U-Th/He ages were produced (Figure 3; Table 1; reduced and raw results in online supplementary Tables S2–S4 ). Although many samples yielded large, pristine, equant grains, some contained only smaller, pitted grains or grains with numerous inclusions. Some samples simply did not yield any dateable apatites and/or zircons despite repeated rounds of mineral separation. We note that rocks of the LHS were least likely to yield datable apatites or zircons. We observe that those apatites with the best optical characteristics tend to produce tighter and more geologically reasonable age clusters, lending greater confidence in those samples. Samples relying on fragmentary, inclusion-rich apatites can yield inaccurate ages due to a failure to capture the total parent isotope concentrations and the challenge of making alpha-ejection corrections [73]. Those samples tend to have greater dispersion among aliquots and are commonly inconsistent with adjacent ages in a given transect or ages from different mineral phases in the same sample. Nonetheless, in the inversion runs, we opted to include all but the most obvious outliers that would have exerted an undue “pull” on the inversion. In general, we find that the zircons produce much lower age dispersions than do the apatites, apparently due both to greater crystal integrity and the fact that any inclusions in a measured zircon dissolve with the grain in hydrofluoric acid, whereas inclusions in apatites will dissolve only if that phase is soluble in nitric acid.

Our U-Th/He cooling ages show striking spatial variation across the study area (Figure 3). The four transects in the Greater Himalaya north of PT2-N (SIM11-RT1, SIM11-RT2, SIM11-RT3, and SIM11-RT5) exhibit relatively young apatite (~1–5 Ma) and zircon (~2–4 Ma) cooling ages throughout (Figures 3(a)–3(c)). The transect collected from the hanging wall just above PT2-N (JUM13-RT1: Figure 3(d)) also exhibits young cooling ages (~2–4 Ma for apatite; ~2–5 Ma for zircon), although these ages are slightly older than those to the northwest. Ages increase substantially for the two transects on the Karnali Klippe between PT2-N and PT2-S (JUM13-RT3 and JUM13-RT4: Figure 3(e)) to ~10–12 Ma for zircon and ~6–11 Ma for apatite. Despite significant lateral differences in the cooling age, age-elevation trends are very steep in all transects, implying rather rapid erosion and cooling of a 1.5- to 2.5-km-thick column of rock at the time when the cooling ages were set.

3.2. Muscovite Ar–Ar Cooling Ages

Cooling ages for Ar–Ar in muscovite showed a similar pattern across the study area (Figures 3(a), 3(c), 3(d), and 3(e)). Eight samples were analyzed, generally from the top and bottom of each transect. Samples along and north of PT2-N yielded ages of ~7–10 Ma. Ages from within the Karnali Klippe were closer to 15–18 Ma. Some samples, especially those collected near and north of the MCT, showed signs of excess argon which requires that they be interpreted as maximum ages. Detailed results can be found in the supplementary information.

3.3. Interpreting Exhumation Rates

Qualitative interpretation of these distributions of cooling ages suggests that, after the late Miocene (~8 Ma), rapid exhumation has been mostly restricted to north of PT2-N. No obvious “kinks” in age-elevation relationships are present (Figures 3(a)–3(e)), suggesting that abrupt changes in exhumation rate were not directly recorded during passage of the sampled rock columns through closure isotherms. Inversions of the observed cooling ages for each transect permit quantitative estimation of exhumation rates in these locations. Although inversions were performed in both Pecube and QTQt, results were similar and are summarized below.

Relief transects from north of PT2-N in the Greater Himalaya (SIM11-RT1, SIM11-RT2, SIM11-RT3, and SIM11-RT5) have been cooling at ~50°C/Myr and, based on our modeling, require rapid exhumation rates of ~1.5–2.1 mm/yr (Figures 3(a)–3(c) and 4; Table 2). Data from near PT2-N ~40 km to the southeast (transect JUM13-RT1) are not as well constrained but are also consistent with ~1.6 mm/yr exhumation rates (Figure 3(d)). Both transects from the Karnali Klippe (JUM13-RT3 and JUM13-RT4) require very rapid exhumation (~1.4–2.7 mm/yr) before ~9–11 Ma followed by an almost order-of-magnitude decrease to ~0.1–0.4 mm/yr ever since (Figures 3(e) and 4).

Although numerous thermochronologic studies have probed the cooling histories of Himalayan rocks during the past 40 years, spatially dense arrays of vertically extensive relief transects like those presented here are rare. The seven, relatively closely spaced relief transects encompassed by this study in Western Nepal provide both striking contrasts among sections and reassuring redundancy of results between spatially related sections. As a consequence, robust comparisons can be made between the exhumation histories within different physiographic and tectonic domains of Western Nepal.

4.1. New Constraints from This Study

Our derivation of best-fit exhumation rates from measured cooling ages provides constraints against which we can consider of the viability of various tectonic models for Western Nepal (Figure 2). The first-order constraints are that:

  1. Exhumation has been rapid and rather steady in the Greater Himalaya along and north of PT2-N since ~8 Ma (Figure 4). Thrusting along the MCT at PT2-N is generally thought to have ceased in the middle Miocene [49], yet the rapid exhumation rates north of PT2-N require more recent or ongoing deformation near the MCT zone to drive rapid rock uplift into the Plio-Pleistocene. Other recent results suggest that MCT in the hinterland was ductily deforming as late as ~6 Ma [74] and that ductile, midcrustal deformation could have recently built the structural relief on the Dolpo anticline. Hence, it is possible that deformation and rapid exhumation have been occurring in that tectonostratigraphic position since the late Miocene (Figure 4).

  2. Exhumation was rapid above the rocks of the Karnali Klippe (PT2-S) around ~16–10 Ma and has been considerably slower since then. Although the area in question sits in the along-strike projection of the inferred MHT ramp from central Nepal and Northern India, these older cooling ages (from transects JUM13-RT3 and JUM13-RT4; Figure 3(e)) suggest that relatively little uplift or exhumation has occurred here since the late Miocene. Several workers have invoked an MHT ramp at depth near the position of PT2-S (e.g., [19, 35, 37, 42, 46, 49, 75]). However, to the degree that the exhumation rates presented here track rock-uplift rates, they lend no support for the presence of a long-lived MHT ramp near the modern PT2-S in Western Nepal with shortening and uplift rates akin to those inferred above the ramp in central Nepal. These data could have important implications for seismic hazard models for Nepal that currently put the main MHT ramp in that location.

  3. Available thermochronological data neither require nor preclude recent (i.e. <3 Ma) rejuvenation of tectonism/exhumation along PT2-S. It remains possible that the uplift inferred by Harvey et al. [19] has indeed recently elevated what was once a low-relief, low-elevation landscape to mid-elevations, introducing a transient erosional state in the region. However, if that is the case, ensuing exhumation has apparently been insufficient to bring apatites within a developing partial retention zone to the surface, meaning that post-8 Ma exhumation has likely not exceeded ~1.3 km. Even at the higher end of the derived exhumation rates (~300 m/Myr) for southern sample transects, it would take ~5 Myr to erode away ~1.3 km of rock (a rough estimate of depth to the base of the apatite [U-Th]/He partial retention zone). Modern erosion rates are likely faster where incision along the tributaries of the Karnali and Tila rivers has eaten into the low-relief landscape, but even at rates as high as 0.5 mm/yr, it would still require ~2.6 Myr of exhumation to unroof apatites from the top of a partial retention zone. As a thought experiment, if we assume a ramp dip of 10° with 3 km/Myr of slip along it (half the overthrusting rate [58], a ramp positioned near our southern transects could have formed no more than 3 Myr ago; otherwise, it would have produced 1.5 km of uplift across those 3 Myr, which, if incision along the Karnali river kept pace with uplift, would have brought rocks with younger cooling ages to the surface (a gentler dip and/or slower slip rate would make this threshold proportionally older) (Figure 5).

4.2. Reconciliation with Foreland Basin Records

The detrital record from the Mio-Pliocene Siwalik sediments near the mouth of the Karnali and Narayani rivers provides insight on an integrated denudation and catchment history of Western Nepal and central Nepal, respectively [40, 53, 61]. Through analysis of Siwalik strata in Western Nepal, van der Beek et al. [76] derived a coarse exhumation history from detrital apatite fission-track ages. Lag times between cooling ages and depositional ages reveal a time-varying exhumation rate between 1.0 and 1.5 mm/yr from ~7 Ma to the present, with an acceleration around 6.5 Ma. These rates are broadly consistent with those from the five relief transects in our study that lie north of PT2-N and suggest that the fast uplift rates we see there today may stretch back into at least the latest Miocene. The apatite fission-track record from modern Karnali sediments yields a central age of 8.0 ± 1.4 Ma, whereas the same analysis in the Narayani yields a central age of 1.8 ± 0.4 Ma [76] (Figure 1(c) for location of these rivers).

Our bedrock cooling ages provide valuable context for these catchment-averaged results in terms of their likely provenance. The cooling age distributions from modern Karnali sediments were originally interpreted as being dominated by apatites and zircons from the Karnali/Dadeldhura Klippe, although our cooling age dataset argues for faster Plio-Pleistocene exhumation north of PT2-N. In central Nepal, in contrast, the Narayani sediments show a strong Greater Himalayan affinity. The temporal trend in the Karnali data suggests that the Mio-Pliocene Siwalik detrital record in Western Nepal is consistent with derivation of sediment from the GHS north of PT2-N, whereas the modern sediments show a stronger component from the older Karnali Klippe and Lesser Himalayan zone. This apparent shift in sediment source areas may be related to the progression of incision into the high-elevation, low-relief landscape that lies between PT2-S and PT2-N (Figure 1). Overall, the combination of detrital and in situ cooling ages is consistent with sustained, rapid exhumation north of PT2 in central Nepal and rapid exhumation shifting from north to south in Western Nepal.

Copeland et al. [61] present an interpretation of cooling age data (muscovite ArAr and ZFT) from Karnali sediments that simplify Western Nepal’s kinematic history into an one-dimensional model. That model includes rapid exhumation and cooling from ~400 to ~200°C around 16–13 Ma, followed by relative quiescence ever since. This interpretation is consistent with what we see in the area of the Karnali Klippe but does not capture the ongoing deformation and rapid exhumation occurring in the hinterland. Today’s rapidly eroding areas along and north of PT2-N will appear much younger (yielding ~2–3 Ma Zr U-Th/He and ~7–9 Ma muscovite ArAr cooling ages) compared with southern areas yielding cooling ages that are 2–4 times older. Such spatial complexity is difficult to discern when diverse source areas are amalgamated by detrital samples delivered by large rivers with spatially variable modern erosion rates.

4.3. Comparison with Central Nepal

The cooling ages and interpreted exhumation histories presented here help to fill a broad gap of low-temperature cooling ages in western Nepal. It is instructive to compare this dataset to comparable data from ~400 km to the east in central Nepal near Kathmandu, an area that has seen a greater number of studies. In central Nepal, abundant muscovite Ar–Ar cooling ages show a northward-decreasing age trend, with the youngest ages found in the high-relief zone just north of PT2 (Figure 6(a)). In Western Nepal, a similar northward-younging trend is visible, with the cooling age minimum found in the relatively high-relief area north of PT2-N (Figure 6(b)). This spatial pattern provides more support for the interpretation that PT2-N is the geomorphic expression of subsurface deformation of a scale most closely relatable to that occurring above a ramp near central Nepal’s PT2 [19, 30].

4.4. A Revised Model for Active Tectonics in Western Nepal

The young cooling ages and geomorphic indices north of PT2-N indicate that rapid uplift and erosion have continued in the hinterland until at least 2 Ma and likely continues into the present [19, 77]. This belt of rapid uplift is consistent with of advection over a midcrustal ramp that is positioned well north of the trend of the main MHT ramp in central Nepal and may have been a locus of rapid uplift since the latest Miocene (e.g., [52]). In a study focused on the Dolpo anticline even farther into the hinterland, Cannon and Murphy [20] argue that duplexing within the ductile portion of the MHT zone is responsible for rapid rock uplift and the buildup of structural relief along the Dolpo anticline, which could explain the rapid rates seen in transects SIM11-RT1 through SIM11-RT5. However, transect JUM13-RT1 lies well outside of the Dolpo anticline and still shows rapid Plio-Pleistocene rock uplift and exhumation. Thus, it appears that the belt of uplift imaged by our cooling age study is distinct from one related specifically to the Dolpo anticline.

One way to reconcile our new observations is to propose that the rapid rock uplift along PT2-N could be related to the oblique-slip WNFS that runs roughly parallel to PT2-N in NW Nepal [36, 55]. Along with the Gurla–Mandhata–Humla fault to the northwest [78], the WNFS may represent the southeastern most extension of the dextral Karakoram fault. In this context, it is possible that the rapid uplift at PT2-N in Western Nepal is a result of oblique thrusting over a related cross-range structure in the midcrust (Figure 7). Such a geometry could lend further credence to the proposal that the WNFS is the surface expression of a major discontinuity separating the Western and Eastern Himalaya [41, 55]. One problem with this model is that it is not particularly spatially consistent with the northern belt of microseismicity that should theoretically be recording stress accumulation at a creeping-locked transition (e.g., [24]; Figure 1(d)). Our tentative interpretation is that the northern belt of seismicity reflects ongoing deformation within the LHD.

We can explain the apparent rapid exhumation at 16–9 Ma detected in those rocks now sitting near PT2-S as the result of rapid cooling as those rocks passed over a more northerly ramp in the active decollement at the time before being translated laterally southward to their present position. In this context, the WNFS and PT2-N appear to trace an occasionally surface-breaking [56], oblique thrust that initiated by the early Pliocene, disrupting a cooling age pattern that otherwise was set by late Miocene emplacement and cooling of the MCT and Dadeldhura/Karnali Thrust sheets (e.g., [45]). Such out-of-sequence thrusting along PT2-N/WNFS would be consistent with interpretations of contemporaneous thrusting along or near the MCT just north of PT2 in central Nepal [5, 8, 16, 23]. The high-elevation, low-relief landscape along PT2-S, then, could be the result of a more recent deformational episode related to a smaller, intrawedge Plio-Pleistocene duplex ramp initiating near PT2-S as proposed by Harvey et al. [19] and supported by more recent articles [14, 37].

An alternative explanation for the rapid exhumation rates observed north of PT2-N is that they are related to ongoing growth of the LHD [13]. However, that deformational mechanism was previously interpreted to be mostly dormant and its floor thrust ramp positioned well to the foreland relative to our hinterland transects (Figure 3) that show rapid cooling rates into the Pleistocene. Hence, in order to have set the younger cooling ages presented here, the LHD would have to be propagating toward the hinterland much farther and more recently than previously thought, instead of mostly propagating toward the foreland during the late Miocene and Pliocene [50].

Although ambiguity remains regarding the origin of the low-relief landscape and belt of seismicity along PT2-S, rapid rock uplift in Western Nepal along a northward bend (PT2-N) that diverges from the trend of PT2 across central Nepal and Northern India has important implications for the local MHT geometry. The destructive 2015 Mw 7.8 Gorkha earthquake ruptured near the top of the MHT ramp and propagated over a broad locked zone within the flat, but it failed to break the surface. Similarly, the 2023 Mw 5.7 Jajarkot earthquake that occurred during review of this article appears to have initiated in a similar position with respect to the main MHT ramp (Figure 1(d)). Because the latter event occurred near the study area, the analysis of this event and its aftershocks may yield further insights into the structural transition from the broader zone of midcrustal deformation in Western Nepal to the more distinct MHT ramp in central Nepal.

In any case, if the MHT ramp is deflected northward in Western Nepal as our cooling age data seem to insist, then MHT ruptures may be able to initiate much farther into the hinterland in Western Nepal than even recent seismic hazard models suggest (e.g., [43]). If there is a second ramp at PT2-S, it could serve as an important barrier to westward rupture propagation [30, 76, 79]. Even in the case that there is no ramp at PT2-S, the even greater width of the shallow “locked” zone in Western Nepal would increase the likelihood that ruptures fail to make it to the surface at the MFT, biasing the paleoseismic record away from recording large earthquakes at the range front and instead resulting in more internal deformation of the range and shifting zones of stress accumulation from the base of the locked zone toward the surface [80].

Using the densest array of [U-Th]/He relief transects yet reported in the central Himalaya, we show that the topographic and seismic discontinuity in Western Nepal indeed reflects a major transition in the modern geometry of the MHT. Specifically, the midcrustal ramp in the MHT reported along nearly 1000 km of the central Himalaya retreats as much as 75 km toward the hinterland in Western Nepal between ~81.5 and 82.5° E, a nuance that is largely unrepresented in existing seismic hazard models for Nepal. Exhumation rate histories derived from inversion of our cooling age data help refute some prior hypotheses for the geometry of the MHT in the study area, including the inference of the absence of a ramp or the presence of a long-lived ramp positioned at the southern “physiographic transition,” PT2-S. Rather, our results suggest that oblique slip across the more northerly MHT ramp (near PT2-N) could be related to the recently identified cross-structure (the WNFS), driving localized rapid (~1–2 mm/yr) uplift along its hanging wall for at least the past ~4 Ma. In contrast, 40–60 km closer to the foreland in the along-strike projection of the MHT ramp from central Nepal, rocks just north of PT2-S record a period of rapid exhumation around 9–11 Ma, apparently followed by relative quiescence ever since. We interpret this cooling history as recording passage over a more northerly midcrustal ramp around 9–11 Ma followed by slow erosion during mostly lateral transport along the upper MHT flat into the Pliocene. Although geomorphologic data suggest a period of renewed uplift on the scale of 1–1.5 km in this southerly zone near PT2-S, it has not been enough to exhume an obvious apatite or zircon helium partial retention zone into the range of a modern thermochronology relief transect. Together, these new spatiotemporal constraints on active midcrustal deformation at the apex of the range have important implications for our understanding of the structure of the MHT and possible seismogenic zones in the Western Nepal seismic gap.

Supporting cooling age analysis methods and detailed results can be found in the supporting information accompanying this article.

The author(s) declare(s) that there is no conflict of interest regarding the publication of this paper.

NSF EAR [0819874].

The authors acknowledge the field support of Earth’s Paradise Treks and Expeditions for assisting with field logistics as well as analytical support provided by P. Gans, D. Stockli, M. Grove, and J. Hourigan which was crucial for providing the cooling ages presented in this study. The authors also thank the reviewers of this article (Tianyi Shen and Zhiyuan He) and Xavier Robert and Renaud Soucy la Roche for constructive comments on an earlier version of this article.

Exclusive Licensee GeoScienceWorld. Distributed under a Creative Commons Attribution License (CC BY 4.0).

Supplementary data