Introduction

The modern glaciated area of the Tibetan Plateau covers approximately 100,000 km1,2. Glacial sediments that have been well preserved on the Tibetan Plateau were predominantly deposited after the Marine Isotope Stage (MIS) 62. The older deposits were dated back to a 10Be exposure age of 561.7 ± 54.8 thousand years ago (kyr)3 or a simulated age of 835.2 ± 241.0 kyr4, indicating that they are much younger than the eventual formation chronology (>10 million of years, Ma) of the high (>5 kilometers, km) Tibetan Plateau5,6. The old moraine sediments on the surface of the Tibetan Plateau might have been reworked or eroded, removing the evidence for the onset of glaciation7. There is a likelihood that original moraine relics are preserved in basins or gorges, but their investigation is yet to be conducted.

The Yarlung Tsangpo River (~2060 km) is the largest and longest river on the Tibetan Plateau, flowing from west to east across the plateau (Fig. 1A). The Eastern Himalayan Syntaxis, known as Namche Barwa and its highest peak towering at 7756 meters above sea level (asl), has the highest uplift rate (2–5 mm/yr) in the Himalaya8. Northward migration of the Namche Barwa has forced the Yarlung Tsangpo River to bend, ultimately leading to the creation of the Tsangpo Gorge9, one of the world’s deepest and longest gorges. Considering the enormous altitudinal difference (> 3000 m) between the Tibetan Plateau and the Indian Plain, the Eastern Himalayan Syntaxis would be expected to have undergone rapid erosion due to the headward erosion of the Yarlung Tsangpo River, causing an incision wave into the southeastern margins of the plateau. However, the preservation of thick and loose sediment (~1 km) in the Milin Valley challenges this notion, suggesting that the fluvial knickpoint did not migrate into the upstream region across Namcha Barwa after ~2.5 Ma10 (Fig. 1B). The long-term stability of the plateau’s margins poses a conundrum for scientists and has been partly attributed to moraine dams impeding river incision11. Nevertheless, this explanation is only relevant to the period after extensive glaciation on Namche Barwa. Therefore, understanding the timing of early glaciation in this area is crucial. To address this, we combine cosmogenic 26Al/10Be burial dating and facies identification of valley sediments to establish a glacial chronology.

Fig. 1: Geomorphological maps showing location of the studied area and boreholes.
figure 1

A DEM image showing the drilling locations. GP and NB represent the Gyala Peri and Namche Barwa peaks, respectively. B Chronological framework of sediments along the longitudinal profile of the Yarlung Tsangpo River.

Full size image

Geographical setting and sedimentology

The Zelongnong Glacier originates from Namche Barwa and frequently advanced and blocked the Yarlung Tsangpo River11. It experienced notable advances in 1950 and 1968, leading to temporary river blockages12. In the valley, three moraine sequences can be identified12, and remnants of the Last Glacial Maximum (LGM) moraine dams are distributed on both sides of the valley (Supplementary Fig. 1). At the Southeast Tibetan Plateau Station (29°46’N, 94°44’E, 3,326 m asl), the mean annual precipitation (MAP) between 1961 and 2008 was 835 mm, with 74.9% of the annual precipitation occurring from April to September (Supplementary Fig. 2). Moisture is derived principally from the Indian Summer Monsoon (ISM)13. Thus, the glaciers in the Eastern Himalayan Syntaxis are classified as temperate glacier, which are sensitive to temperature change12. The Yarlung Tsangpo River’s longitudinal profile features a knickpoint zone14 with main knickpoints located at Niding (P1) and Gyala (P2), developed in the sediment and bedrock, respectively (Fig. 1A, B). From the Knickpoint P1 to P2, the altitude of the riverbed drops from 2920 to 2790 m asl. Downstream of these knickpoints the river runs between the Gyala Peri peak (altitude 7294 m asl) and Namche Barwa, with an incision depth of 6 km. Relics of moraine dams are preserved between P1 and P2 (Fig. 1B and Supplementary Fig. 1). In the main valley and tributaries, such as the Nyang River, there are two wide, flat terraces (T1 and T2) composed of gravel and lake sediments (Fig. 1B). OSL and 14C dating indicate that T1 and T2 were formed by moraine dams at 9–7 and 30–20 kyr, respectively15,16,17,18.

Five boreholes located near the center of the Milin Valley offer records of the sedimentary sequence, extending down to the underlying bedrock (Fig. 1B). The maximum depth progressively increases downstream, ranging from 70 meters in Borehole #1 to over 500 meters in Boreholes #3 and #4. Subsequently, the depth decreases to 230 meters in Borehole #5. The bedrock depth between Boreholes #3 and #4 is expected to be greater than the depth in either Borehole #3 or #4, which aligns with the position of the confluence and is consistent with the findings obtained using an artificial neural network approach10. The prolonged stability of Knickpoint P2 indicates that the lower sections of Boreholes #3, #4 and #5 have been well preserved (Fig. 1B), which provides the opportunity to investigate the sediment accumulation process in the Milin Valley. Earlier, we reported the 26Al/10Be burial dating results of three samples collected from the bottom of Boreholes #310. Subsequent to this, in-depth investigations were conducted on the sediments retrieved from Boreholes #3, #4, and #5 in this study.

The bottom portion (283.1-566.9 m, U3-6) of Borehole #3 primarily consists of boulders, pebbly gravel and sandy gravels (Fig. 2, Supplementary Fig. 3A, Supplementary Table 1). The upper portion (0–283.1 m, U3-1 to −5) of Borehole #3 is composed of fluviolacustrine sandy gravel and fine silt (Fig. 2, Supplementary Fig. 3B), including deep lake clayey silt at depths of 72–85 m (U3-2), except for an uppermost 20 m-layer of sandy gravel. This evidence indicates a decrease in hydrological power between U3-6 to −5. The bottom portion (332.6–512.7 m, U-17) of Borehole #4 is dominated by massive, bedded large to small boulders (Supplementary Fig. 3C, Supplementary Table 1). The middle portion (220.6-332.6 m, U4-13 to −16) of Borehole #4 contains two sedimentary sequences of fluvial sandy gravel, sand, lacustrine clayey silt and massive, bedded boulders (Fig. 2). Meanwhile, the upper portion (0-220.6 m, U4-1 to −12) of Borehole #4 contains six sedimentary sequences of fluvial sandy gravel, sand and lacustrine clayey silt (Fig. 2), which are interbedded with shallow, strong dynamic lakeshore or overbank fine silts and deep-static water lake-center clayey silts (Supplementary Fig. 3D). This evidence suggests that the Yarlung Tsangpo River experienced damming at least eight times. Moving to Borehole #5, which lies between knickpoints P1 and P2, the sediment beneath the riverbed is only 230 m thick, and the overlying material has been removed by scouring. Borehole #5 is dominated by massive, bedded sands and large boulders (>1 m in diameter), intercalated with some thin lacustrine clay (U5-3) and sandy gravel (U5-5, −7, −9) beds (Supplementary Fig. 3E, F, Supplementary Table 1). A total of 91 sediment samples were collected for particle-size analysis with the aim of identifying the sedimentary facies of the boreholes. Among these, eight samples were selected for scanning electron microscope (SEM) analysis of quartz grains (Fig. 2, Supplementary Table 2). Fourteen sand samples from Boreholes #3 and #4 were collected for cosmogenic nuclide burial dating (26Al/10Be) with the aim of establishing the chronological sequences (Fig. 2 and Supplementary Table 3). A single basement bedrock sample from Borehole #5 was collected for apatite U-Th/He dating, aiming to determine the exhumation process of the Namche Barwa-Gyala Peri Massif (Fig. 2 and Supplementary Table 4).

Fig. 2: Simplified stratigraphic columns of the drill cores at sites #3 to #5, with 26Al/10Be burial ages and one bedrock (U-Th)/He age.
figure 2

Gray, yellow, and purple colors are used to depict sedimentary facies in drill core stratigraphy, classified as lacustrine, fluvial, and moraine deposits. The blue circle, red square, and orange star represent the sample locations for microstructure analysis, 26Al/10Be burial dating, and (U-Th)/He dating, respectively.

Full size image

Results

Sedimentological evidence

The massive bedded sands and variously sized boulders encountered in Boreholes #4 and #5 starkly contrast with the horizontally-bedded fluvial sands and round to sub-rounded gravels observed in Boreholes #3 and #4. The former characteristics could potentially be interpreted as indicative of either a moraine deposit or a deposit resulting from a landslide or debris flow. Landslides and debris flows are frequent in the Eastern Himalayan Syntaxis, triggered by rainfall, snow avalanches and earthquakes19. However, landslide or debris flow deposits are unstable and do not block the river for long time periods or lead to the formation of thick lacustrine deposits upstream of the landslide. For instance, the Yigong Landslide in 2000 had a volume of 1.15 × 108 m3 but only blocked the Yigong Tsangpo River for 62 days, resulting in the deposition of only a few centimeters of lacustrine sediment20. The sediments observed in Boreholes #4 and #5 exhibit similarities to the properties of LGM morainic sediments found at the foot of the Namche Barwa peak (Supplementary Fig. 1). They both contain numerous subangular and subrounded large boulders (>1 m) and clasts, which suggests subglacial erosion21,22,23. Additionally, they exhibit some poorly-formed bedding structures (Supplementary Table 1). Based on their sedimentology, these facies are likely to be interpreted as morainic sediment. In addition, the lacustrine sediments of T1 and T2 were formed by moraine dams15,16,17,18. Only prolonged and extensive glacial advances would have blocked the river for several millennia to form thick lacustrine sedimentary deposits.

Such a hypothesis can be further supported by submicroscopic photography, and particle-size and SEM data. The fine-grained fractions of the morainic sediments exhibit almost identical particle-size distributions; the fluvial and lacustrine samples are also very similar (Fig. 3A, B, Supplementary Fig. 4). The broad distributions in the moraine samples suggest that they are poorly sorted. In comparison, the fluvial samples are better sorted, showing a major peak at ~70–270 μm and a minor peak at ~10–20 μm. The lacustrine samples are the best sorted among all, with a peak at ~4–15 μm (Supplementary Fig. 4). Submicroscopic photography reveals that most of the grains in Sample 4-269, obtained from the massive bedded sands in Boreholes #4, exhibit an angular morphology (Fig. 3C). Furthermore, SEM images demonstrated that the sand grains in samples 4-269, 4-339 and 4-449 (Fig. 3E, Supplementary Figs. 5–7) exhibit a high concentration of relief features, indicating the generation of mechanical abrasions under high-energy conditions24,25,26. Straight and arcuate steps, as well as flat cleavage surfaces, are well preserved on sand grains in both samples 4-269 and 4-449. Additionally, an intact crystal habit can be observed in Sample 4-449 (Supplementary Table 2). These surface features are typical of glacial sediments26. Similarly, samples 4-339, 5-188, and 5-230 display straight steps, conchoidal fractures, flat cleavage surfaces, and other typical surface features of glacial sediments (Supplementary Table 2, Supplementary Figs. 8, 9). On the reverse sides of these samples’ particles, silica globules, solution crevasses, and crystalline overgrowths are visible, which are characteristic surface features of fluvial sediments24,25,26. Consequently, we can posit that these samples were deposited as glaciofluvial sediments. In Sample 3-280, submicroscopic photography shows that most grains are subangular to subrounded (Fig. 3D). Medium relief features can also be observed on sand grains in samples 3-213, 3-280 and 3-557 (Fig. 3F, Supplementary Figs. 10–12), suggesting deposition under strong hydrodynamic environmental conditions26. Surface features such as conchoidal fractures, crescentic percussion marks, straight/curved grooves and scratches25, V-shaped percussion cracks, solution crevasses and silica globules are well developed on these samples, which are typical of fluvial sediments24,25,26.

Fig. 3: Sedimentological analysis of typical moraine and fluvial sediments in the boreholes.
figure 3

A, B Particle-size distributions of moraine and fluvial samples; (C, E), (D, F) Sub-microscopic photographs and SEM images showing the features of coarse-grained quartz for samples 4-269 and 3-280, respectively. As-Arcuate step, Ss-Straight step, Fcs Flat cleavage surfaces, Mcf Medium conchoidal fractures, Vsc V-shaped percussion cracks, Cpm Crescentic percussion marks, Sc Solution crevasses.

Full size image

Based on the sedimentological evidence discussed above, we concluded that the lower (283-567 m, U3-6) and upper (0-283 m, U3-1 to -5) portions of Borehole #3 are composed of fluvial and fluviolacustrine sediments, respectively. The lower (221-513 m, U4-13 to -17) and upper (0-221 m, U4-1 to -12) sections of Borehole #4 are composed of moraine and fluviolacustrine deposits, whilst Borehole #5 is dominated by morainic deposits (Supplementary Table 1). In Boreholes #3 and #4, the sediment grain size becomes smaller upward, reflecting changes in deposition or source regions. These transitions might be related to the development of paleoglaciers in the Eastern Himalayan Syntaxis.

10Be/26Al burial ages

The ages of 2.30 ± 0.21, 1.79 ± 0.18 and 2.23 ± 0.20 Ma for samples extracted near the base of Borehole #3 (WP-1 to -3; 561-563 m) indicate that deposition in the Milin Valley had commenced by > 2.5 Ma (Figs. 1B, 2, Supplementary Table 3). Three samples collected near the transition from fluvial to fluviolacustrine sediment at 283 m yielded two similar ages of 1.14 ± 0.17 and 1.02 ± 0.19 Ma (WP-8 and −9; 275 and 270 m), and a relatively younger age of 0.61 ± 0.20 Ma (WP-7, 280 m). All three ages are roughly consistent with the ages for samples extracted from Borehole #4, within a 2σ error, further demonstrating that the deposition transition was synchronous with the glacial advance. Rapid sedimentation is suggested by samples WP-11 to -14 from Borehole #4, which have similar burial ages of ~0.75 Ma across depths of 330–450 m. Together, the 26Al/10Be burial ages of Boreholes #3 and #4 suggest that an extensive glacial advance blocking the Yarlung Tsangpo River occurred at 0.75 ± 0.19 Ma (Supplementary Fig. 13).

It is noteworthy that the paleosurface elevation of the Himalayas was over 5 km before 10 Ma6. This likely led to the development of a paleo glacier with the uplift of the Namche Barwa-Gyala Peri Massif. However, the deposition transition observed in Borehole #4 indicates that it was only around 0.75 Ma when the glacier size became sufficiently large to block the Yarlung Tsangpo River for extended periods, thereby altering the deposition environment since then (Figs. 1B, 2). The older sediments (> 0.75 Ma) underwent erosion or reworking in both Boreholes #4 and #5.

Discussion

Relationship between glacial advance and climate

The 26Al/10Be burial-dating results indicate that the earliest extensive glacial advance occurred at ~0.75 Ma (Fig. 4A), during MIS 18, which is roughly consistent with the time of maximum glaciation across the Himalayan Region. Previous study showed that the oldest moraine deposits were dated back to a 10Be exposure age of 491–423 kyr in Western Himalaya27 or a simulated age of 835.2 ± 241.0 kyr4 in the Central Himalaya, indicating that extensive glacier advance occurred since the end of the Mid-Pleistocene Climatic Transition (MPT)28. It was a time when the amplitude of ice volume increased by about 50%29,30, and the first major globally glacial event took place31. During this period, the amplitude of the Indian Summer Monsoon (ISM) weakened since around 0.92 Ma, coinciding with the expansion of Northern Hemisphere ice volumes and decreasing Antarctic temperatures32,33 (Fig. 4B–D). While there has been a reduction in effective moisture due to the weakened ISM, the glaciers in the Namche Barwa-Gyala Peri Massif still advanced in response to the temperature decrease, similar to glaciers in the monsoon-dominated Central and Eastern Himalaya34,35.

Fig. 4: Comparison of glacial chronological data in the Tsangpo Gorge with the Indian Summer Monsoon (ISM) proxy records and global temperatures.
figure 4

A 26Al/10Be burial ages from the Tsangpo Gorge. B Proxy ISM records from Heqing Basin, southwestern China31. C The LR04 benthic benthic stack27. D Change in global average surface temperatures (GASTs32). The shaded purple region represents the initiation time of glacial advance.

Full size image

Influence of glacial development on the Eastern Himalayan Syntaxis

The basin-averaged erosion rate in Borehole #3 at the Milin Valley shows an increasing trend following the accumulation of fluvial sediment at ~2.5 Ma (Fig. 5). It reached its highest value of ~0.29 mm/yr at 0.61 ± 0.20 Ma, decreased to ~0.09 mm/yr at ~0.5 Ma, and finally reached its minimum value of ~0.04 mm/yr in modern times36. In Borehole #4, the basin-averaged erosion rates were ~0.10 mm/yr at ~0.75 Ma, which decreased to ~0.05 mm/yr in modern times36. This implies that the increasing regional erosion rate observed between 2.5 and 0.75 Ma can primarily be attributed to increased weathering processes, possibly caused by hanging glaciers on high peaks or frost weathering in lower mountain areas. These processes likely intensified in response to the global cooling trend from the Early Quaternary to the MPT33,37. The extensive glacier advance at ~0.75 Ma in the Namche Barwa-Gyala Peri Massif would have further enhanced the regional erosion process to its highest basin-averaged erosion rate value. However, the sedimentary changes between U3-6 to -5 of Borehole #3 indicates a decrease in hydrological power, which must be related with dramatic depositional environment change. This indicates that the subsequent accompanying moraine dam would have blocked the Yarlung Tsangpo River and altered the depositional environment (Supplementary Fig. 14). The impounding effects of the dammed lake would have subsequently reduced the basin-averaged erosion rate of the Yarlung Tsangpo River in the Milin Valley38. Consequently, the transition in sedimentary facies and the alteration in basin-averaged erosion rates indicate that the earliest extensive glacial advance in the Eastern Himalayan Syntaxis, leading to the damming of the Yarlung Tsangpo River, started at ~0.75 Ma.

Fig. 5: Comparison of 26Al/10Be burial ages and basin-wide erosion rates for Boreholes #3 and #4.
figure 5

The modern data are from ref. 35. The shaded purple region represents the initiation time of glacial advance.

Full size image

The hypothesis presented above can receive further support from low-temperature thermochronological data. These data indicate that the high exhumation center of the Eastern Himalayan Syntaxis migrated northward from 2.0 to 0.0 Ma, accompanied by a synchronous increase in exhumation rate (Supplementary Fig. 15). The U-Th/He age of 0.71 ± 0.11 Ma for the basement bedrock in Borehole #5 suggests that the Namche Barwa-Gyala Peri Massif experienced a rapid exhumation depth of ~1.3–1.6 km immediately after the extensive glacial advance at ~0.75 Ma (Fig. 2, Supplementary Table 4). This evidence suggests a relationship between glacial development and concentrated mountain exhumation. However, it is also possible that high rates of exhumation at Borehole #5 were occurring further back in time while sedimentation was occurring upstream. In this way, the zone of deposition has migrated downstream behind the zone of incision of the Tsangpo Gorge.

Factors controlling the stability of the southeastern Tibetan Plateau margins

The long and largely-continuous borehole sedimentary records strongly indicate that Knickpoint P1 has remained relatively stable since the beginning of deposition in the Tsangpo Gorge. Two potential factors that could have impeded river incision are moraine damming11 and steepening of the Tsangpo Gorge as a consequence of an increase in rock uplift rates10. Valley deposition along the upstream reaches of the Tsangpo Gorge began more than 2.5 Ma ago, while paleo-glacier became sufficiently large to block the Yarlung Tsangpo River for extended periods was recorded after ~0.75 Ma (Supplementary Fig. 15). As a result, rock uplift was the primary driving force behind the initial valley deposition, suggesting that glacial damming did not play a major role in maintaining the initial stability of the Tibetan Plateau’s southeastern margins before approximately 0.75 Ma.

Terrace T2, which formed during the last glacial period, experienced incision at ~20 kyr15,16,17,18, exposing ~250 m-thick sediments (Fig. 1B, Supplementary Fig. 14). Despite the repeated formation of glacial dams at the Terrace T1, the relatively warm climate after the last glacial epoch meant that there was a period without the presence of large, stable glacial dams that could have inhibited erosion of the river into sediments upstream of Knickpoint P2, allowing the bedrock knickpoint to advance toward the interior of the plateau. The lacustrine-alluvial cycles observed in Borehole #4 would suggest that phases of river incision have occurred at least eight times since the extensive glacial advance at ~0.75 Ma (Fig. 2). However, the U-Th/He age of ~0.71 Ma for the bedrock in Borehole #5 indicates that the bedrock between knickpoints P1 and P2 has experienced an exhumation depth of ~1.3–1.6 km (Fig. 1), which is much higher than the ~230 m thickness of moraine sediment (Fig. 2). This suggests that the erosion generated by the Namcha Barwa glaciers has been greater than any sedimentary accumulation. Therefore, we would conclude that moraine dams have had a limited effect on impeding river incision, and that it was the rock uplift of the Namche Barwa-Gyala Peri Massif that prevented knickpoint migration and maintained the stability of the southeastern margins of the Tibetan Plateau.

Materials and methods

Particle-size and micromorphological analysis

Sample pre-treatment for particle-size analysis followed the procedure developed by ref. 39. The sediment was first treated with dry-sieving to remove coarse fragments (>2 mm). The remaining sedimentary material of the samples then underwent treatment with 30% H2O2 to eliminate any organic material, and 10% HCl to remove carbonates. Subsequently, the samples were deflocculated using a 0.05 M solution of (NaPO3)6 before being measured using a Malvern Mastersizer 3000 laser grain size analyzer.

For micromorphological analysis, sample pre-treatment was performed at the Institute of Geology, China Earthquake Administration. In the sandy samples, coarse-grained fractions (>63 μm) were first dry-sieved. Magnetic minerals were then separated using electromagnetic separation, and heavy minerals were removed by heavy liquid separation. The quartz crystals without inclusions were subsequently separated through handpicking.

Cosmogenic nuclide dating

The samples for 26Al/10Be burial dating collected from borehole cores weighed 2 ~ 3 kg. The pretreatment procedures for 26Al/10Be burial dating were conducted following the method of40. To isolate pure quartz grains, coarse-grained fractions (250–500 μm) were first separated from bulk samples using mechanical dry sieving. The quartz-rich extracts were treated with 10% hydrochloric acid and 30% hydrogen peroxide to remove carbonate and organic matter, respectively. Then the samples were etched with a HF/HNO3 solution at a temperature of 80 °C in an ultrasonic bath for ~9 h to remove meteoric 10Be and other matter. The dried samples were etched again using the same process, and this etching-drying process was repeated six times. It should be mentioned that the residual impurity was removed using the heavy liquid method and hand separation under the microscope, prior to the final etching procedure. Aliquots of ~1 g for each sample were dissolved in HF/HNO3 before being measured and analyzed by an Inductively Coupled Plasma Optical Emission Spectrometer (ICP-OES) at the Institute of Geology and Geophysics, Chinese Academy of Sciences for 27Al and other elements, to assure purity (Supplementary Table 5). The remaining pure quartz was spiked with 0.25 mg of 9Be in a weak HNO3 carrier solution, then heated, and subsequently dissolved. Fluorides were removed by fuming with HNO3/HClO4. The resulting samples were added to pure water and HClO4, heated and dried two times, then rinsed out with HCl. Afterwards, Al3+ and Be2+ were isolated from the samples using an ion exchange resin. Having been neutralized by ammonia water, Al(OH)3 and Be(OH)2 were obtained. Their solution was separated gravitationally, rinsed, transferred to a quartz crucible and dried. The samples were then baked for 5 min at 750 °C, yielding Al oxide and Be oxide powders. These final samples were sent to the Cosmogenic Nuclide Lab of the European Center for Earth Sciences, Environment Research and Education (CEREGE) in France for target making and the measurement of 10Be and 26Al contents. The ratios were measured relative to the standard NIST4325 with a nominal 10Be/9Be ratio of 2.79 × 10−11, and to SM-Al-11 with a nominal 26Al/27Al ratio of 7.401 × 10−12. Our process blank 10Be/9Be and 26Al/27Al ratios were 6.3 × 10−15 and 7.2 × 10−16, respectively. The corresponding 10Be atoms were subtracted from the reported sample results. Analytical uncertainties as well as uncertainties in the carrier and blanks were propagated using standard error-propagation methods.

The burial and exhumation history prior to the final deposition of the sediment was imported to allow 26Al/10Be burial age calculation. In a previous study, we tested the difference between different burial conditions and produced similar dating results10. An instantaneous burial to depths where no more production of cosmogenic nuclides could occur after deposition was assumed for our age calculations. Assuming steady erosion rates at the sediments’ source(s), the measured 26Al and 10Be concentrations in samples can be given by:

$${N}_{10,m}=\frac{{P}_{10}(0)}{{{{{{{\rm{\lambda }}}}}}}_{10}+{{{{{{\rm{\rho }}}}}}}_{s}E/{\Lambda }_{n}}\, \cdot \, {e}^{-{t}_{b}{{{{{{\rm{\lambda }}}}}}}_{10}}+{N}_{10,{pb}}$$
(1)
$${N}_{26,m}=\frac{{P}_{26}\left(0\right)}{{{{{{{\rm{\lambda }}}}}}}_{26}+{{{{{{\rm{\rho }}}}}}}_{s}E/{\Lambda }_{n}}\, \cdot \, {e}^{-{t}_{b}{{{{{{\rm{\lambda }}}}}}}_{26}}+{N}_{26,{pb}}$$
(2)

where tb is the burial time from when the sediments were deposited; E is the surface erosion rate where the sediment originated, which is referred to henceforth as the paleo-erosion rate; Pi is the mean basin-wide nuclide production rate at source; λi is the radioactive decay constant; ρs is the density of the source rock; Λn is the attenuation length for neutron flux; and Ni,pb is the is the nuclide production after burial, which is assumed to be zero.

Assuming rapid source erosion, and without taking the post-burial nuclide production into account, the combination of Eqs. (1) and (2) can yield the 26Al/10Be burial age and paleo-erosion rate. The 26Al/10Be burial age and paleo-erosion rate of each sample was calculated using the ‘TCNCal’ program developed by ref. 41, and the results are listed in Supplementary Table 3.

U-Th/He dating of apatite

Detrital minerals of the bedrock Sample YJ-1 from Borehole #5 were processed at the Berkeley Geochronology Center, using standard density and magnetic susceptibility techniques. The euhedral apatite crystals (≥ 80 μm) without inclusions were separated by handpicking, and three grains (YJ1-x, -y, -z) were selected for measurement (Supplementary Table 4). The laboratory techniques employed to measure apatite isotopes were conducted following the methods described in refs. 42,43. The individual crystals were first wrapped in platinum tubes and heated in a vacuum at 1050 ± 50 °C. He isotopic analysis was determined using a quadrupole mass spectrometer. The crystals were then dissolved in nitric acid, and the U, Th, and Sm isotopes were analyzed using multicollector inductively coupled plasma mass spectrometry. The exhumation depth was calculated using an apatite closure temperature of ~65–80 °C with a cooling rate of 10–100 °C/Ma44, and a mean thermal gradient of 50 °C/km in the Eastern Himalayan Syntaxis45. The calculation of exhumation depth can be given by:

$${{{{{\rm{Exhumation}}}}}}\,{{{{{\rm{depth}}}}}}={{{{{\rm{Closure}}}}}}\,{{{{{\rm{temperature}}}}}}/{{{{{\rm{Mean}}}}}}\,{{{{{\rm{thermal}}}}}}\,{{{{{\rm{gradient}}}}}}$$

Thermal modeling

To invert exhumation rates from thermochronometric ages, we used the inversion approach of46. We inverted existing apatite fission-track, zircon fission-track, zircon-helium and biotite 40Ar/39Ar thermochronometric data (original data summarized in Supplementary Table 6) using the method described in ref. 46. This approach converts the thermochronometric age to an exhumation rate based on the depth to the closure isotherm, this being the integral of erosion rates from the cooling age to the present. However, we modified the code to incorporate the age from Borehole #5, which is telling us about exhumation from depth to the surface over a time interval in the past. To do this, we recognised that the distance to the closure isotherm is equal to the integral of the erosion rates from the cooling age to the age of the overlying sediments. The kinetic parameters for helium diffusion in apatite and zircon were taken from refs. 47,48, for fission track annealing in apatite and zircon from refs. 49,50, and for argon diffusion in biotite from refs. 51. An a priori erosion rate (2 km/Ma) and its variance (2 km/Ma) were used to initialize the inversion. The a priori was iteratively updated to a posterior erosion rate that maximizes the fit to the data using the non-linear least squares method. Using this approach, thermochronometric data are spatially correlated by imposing a spatial correlation length (10 km), ensuring that erosion rates vary smoothly in space. Earlier, the exhumation history over the past 6 Ma in 2 Ma intervals in the in the eastern Himalaya Syntaxis has been reported52. The exhumation history over the past 2 Ma in 1 Ma intervals is further investigated with more detailed data in this study.