The Tsangpo-Brahmaputra River drains the eastern part of the Himalayan range
and flows from the Tibetan Plateau through the eastern Himalayan syntaxis
downstream to the Indo-Gangetic floodplain and the Bay of Bengal. As such, it
is a unique natural laboratory to study how denudation and sediment
production processes are transferred to river detrital signals. In this
study, we present a new
The large-scale and intense interaction between tectonics and climate make
the Himalayan range the largest sediment source to the ocean. The
Tsangpo-Brahmaputra is a major catchment draining the Himalayan range and the
Tibetan Plateau. It is an exceptional natural laboratory to better understand
how denudation processes are reflected in sedimentary load because of the
large geomorphological diversity of its catchment. A remarkable feature of
the Tsangpo-Brahmaputra catchment is the eastern Himalayan syntaxis.
Interactions between tectonic and surface processes are most likely observed
in tectonic syntaxes (Koons et al., 2013). The eastern Himalayan syntaxis has
been proposed as a typical example of how active tectonic deformation, thermal
weakening of the crust, and steep topography could be self-sustained by
intense erosion and rapid exhumation of crustal material within the framework
of the tectonic aneurysm model (Zeitler et al., 2001). Although the extent
and nature of this coupling has recently been challenged (Bendick and Ehlers,
2014; Wang et al., 2014; King et al., 2016), there is ample evidence for
superimposed rapid exhumation (Burg et al., 1998; Seward and Burg, 2008;
Zeitler et al., 2014; Bracciali et al., 2016) and active erosion (Finlayson
et al., 2002; Finnegan et al., 2008; Stewart et al., 2008; Enkelmann et al.,
2011; Larsen and Montgomery, 2012; Lang et al., 2013) in a focused area
around the Namche Barwa–Gyala Peri massif (NBGPm), where the course of the
Tsangpo-Brahmaputra is marked by a sharp bend to the south-west (Fig. 1).
Difficult access to the NBGPm means that contemporary denudation rates are
poorly constrained, but estimates range from ca. 5 to 17 mm yr
Cosmogenic nuclides have been widely used to constrain catchment-wide denudation rates (CWDRs) over a range of scales and catchment sizes (e.g., Portenga et al., 2011). The concentration of a given cosmogenic nuclide in river sediments theoretically reflects the rate at which the upstream landscape is lowering through physical erosion and chemical weathering assuming steady-state denudation (Brown et al., 1995; Bierman and Steig, 1996; Granger et al., 1996). However, it is also increasingly recognized that CWDRs are affected by a number of site-specific biases and limitations that prevent us from simply reading denudation rates from a bag of sand. Amongst others, landsliding, a dominant erosion process in actively eroding landscapes, violates the steady-state denudation assumption and may thus lead to biases in the CWDR (Niemi et al., 2005; Yanites et al., 2009). The stochastic nature of sediment supply to the channels and hydrology within the river network may also perturb the TCN signal over a range of timescales and limits our ability to use a single sediment sample as a representative and accurate representative of upstream erosion products (Kober et al., 2012; Lupker et al., 2012; West et al., 2014; Foster and Anderson, 2016). Glacial erosion is not properly accounted for, as sediments eroded under a glacier are shielded from cosmic rays and, hence, supply sediment with very low TCN concentrations independent of glacial erosion rates (Godard et al., 2012; Delunel et al., 2014). Different erosion processes also affect the final grain size of fluvially exported sediment, which may in turn lead to a grain size dependence of TCN signals that needs to be taken into account in order to derive robust CWDRs (Clapp et al., 2002; Belmont et al., 2007; Aguilar et al., 2014; Puchol et al., 2014; Lukens et al., 2016). The assumption of quartz ubiquity or content within the eroded lithology of a studied catchment is also likely to be violated in a number of cases and is also a source of biases in CWDR (Carretier et al., 2015a). Finally, large catchments, and especially floodplains, represent temporary storage compartments, in which TCN may accumulate or decay depending on storage duration and depth, which may in turn affect the TCN concentration in river sediments (Wittmann and von Blanckenburg, 2016). These limitations need to be properly evaluated in order to quantitatively relate TCN signals to surface processes and denudation rates. Since most of these limitations also apply to the Tsangpo-Brahmaputra, this study further represents a unique opportunity to evaluate and quantify the potential biases associated with the use of TCNs in large, actively eroding, and contrasted catchments.
The Yarlung–Tsangpo–Siang–Brahmaputra–Jamuna River system (as the trunk
river is called from upstream to downstream, from now on referred to as the
Tsangpo–Brahmaputra in this paper) is a large drainage basin of
ca. 530 000 km
In the north-eastern part of the catchment, the Tsangpo-Brahmaputra River
sharply bends to the south, leaves the Tibetan Plateau, and cuts through the
Himalayan range through deep gorges and a prominent knickpoint, losing over
2000 m of elevation in less than 100 km of channel length. The NBGPm, and
more generally the eastern Himalayan syntaxis, is marked by very steep
topography and high relief resulting from the deflection of crustal material
around the indenting Indian plate and the growth of a large antiformal
metamorphic structure that is deeply dissected by active fluvial incision
(Burg et al., 1997; Finnegan et al., 2008). This reach of the
Tsangpo-Brahmaputra is underlain by gneiss, quartzites, marbles, and generally
highly metamorphic rocks from the Higher Himalayan Crystallines (HHC) (Burg
et al., 1997). Just south of the syntaxis, the river flows over the Abor
volcanics as well as limestones and shales (Jain and Thakur, 1978; Acharyya,
2007). The large valley along the Tsangpo-Brahmaputra, south of the syntaxis,
channels moisture and precipitation northwards, which translates into
significant rainfall (> 2000 mm yr
The Tsangpo-Brahmaputra exits the Himalayan range in Pasighat, Arunachal
Pradesh, NE India, and finally discharges into the Bay of Bengal in
Bangladesh after crossing the Assam and Indo-Gangetic alluvial floodplains
for about 1000 km (Fig. 1). The main stream of the Tsangpo-Brahmaputra further
receives input from the eastern Himalayan rivers Lohit and Dibang that drain
the poorly documented Mishmi hill formations and from Himalayan rivers
draining the southern flank of the eastern Himalayan range (Subansiri,
Kameng, Manas, and Teesta rivers). The northern tributaries of the
Tsangpo-Brahmaputra drain the classical HHC and Lesser Himalayan units
composed of a variety of crystalline and metasedimentary rocks, as well as
the ca. mid-Miocene sub-Himalayan Siwalik molasses (Yin et al., 2006;
Chirouze et al., 2012). Sediment input from the southern tributaries
originating from the Shillong Plateau, as well as from the northern part of
the Indo-Burman ranges, is thought to be small (Garzanti et al., 2004). The
Indo-Gangetic floodplain and the southern flank tributaries to the
Tsangpo-Brahmaputra receive intense precipitation (up to
4000 mm yr
River sediments were sampled from the Tibetan headwaters down to the
Brahmaputra in Bangladesh, including the major tributaries. For sample
locations upstream of the Himalayan front (Pasighat, pt. 8 in Fig. 1; all points hereafter can be referred to in Fig. 1 as well),
sediments were taken from fresh sandbars. Downstream of the Himalayan front,
river sediments were dredged in the center of the channel using local boats.
A number of sampling points were revisited several years to assess the
temporal variability in the
Grain size fractions were then separated with a Frantz magnetic separator and
the nonmagnetic fraction was further treated with HCl (38 %) to remove
carbonates. This fraction subsequently underwent four to five leachings (3 to
5 days each) in H
Recent work from Portenga et al. (2015) suggests the possible occurrence of
In this study, we also included the
The procedure for the determination of basin-wide denudation rates using the
TCN data is similar to that detailed in Lupker et al. (2012) and is
summarized hereafter. All data are reported in Table S1. Basin average
denudation rates,
The basin-wide
The cosmogenic-derived sediment fluxes (
The unrecognized occurrence of
Overall, the
Ratio of
Downstream evolution of the
Catchment-wide denudation rates were calculated using Eqs. (1) and (2). The
denudation rates measured in the Tsangpo-Brahmaputra basin range from 0.03 to
> 4 mm yr
Evolution of the
The sediment fluxes that are derived from the denudation rate estimates
generally increase downstream along the main stream of the
Tsangpo-Brahmaputra and reach 1140
The main trunk river of the Tsangpo-Brahmaputra catchment drains highly
contrasted physiographic units with different denudation rates. The NBGPm in
particular shows very active erosion processes, with steep river gradients
and active landsliding that coincide with rapid exhumation (Finnegan et al.,
2008; Stewart et al., 2008; Enkelmann et al., 2011; Larsen and Montgomery,
2012; Lang et al., 2015; King et al., 2016). The NBGPm separates the slowly
eroding (< 0.3 mm yr
The Tsangpo-Brahmaputra
Downstream evolution of the total drained glaciated area
(GLIMS database; Raup et al., 2007) of the Tsangpo-Brahmaputra in blue. For
comparison, the evolution of the trunk channel elevation is plotted in grey
and the average
The presence of an intensely eroding area in the eastern Himalayan range
downstream of the NBGPm, along the lower Siang, down to the MFT, has not been
documented by previous studies. Studies suggest that most of the sediment is
produced within the NBGPm in both the long and short term.
Thermochronological data (Finnegan et al., 2008; Stewart et al., 2008;
Bracciali et al., 2016; Salvi et al., 2017) display minimum closure ages for
various thermochronometers in a reduced area centered at the confluence
between the Po-Tsangpo and the Yarlung Tsangpo, even though this area with
high denudation rates has been suggested to extend farther southwards than
earlier recognized (Enkelmann et al., 2011). The recent erosional activity
displays a similar pattern, with a maximum landslide density in the NBGPm
around the abovementioned confluence, and which significantly decreases
further downstream along the upper Siang (Larsen and Montgomery, 2012). Our
Calculations of CWDR most commonly hypothesize that quartz is ubiquitous. However, nonuniform quartz distributions may yield significant biases in the downstream TCN-derived sediment fluxes (Carretier et al., 2015a). In our case, very low quartz concentrations in the outcropping rocks of the NBGPm might explain the apparent absence of response in the TCB-derived flux signal (Fig. 4). If the Tibetan headwaters partly drain carbonate-rich TSS (Liu and Einsele, 1994) or Linzigong volcanics, the quartz flux exported by the Tibetan Plateau would tend to be overestimated. Nevertheless, if we neglect a thin band of marbles and ophiolites along the highly deformed suture, the metamorphic or igneous, gneissic, and granulitic lithologies outcropping in the NBGPm do not present, to our knowledge, a major difference in quartz amounts compared to the Himalayan metamorphic rocks further south. In addition, such bias would not permit explanation of the very high apparent increase in sediment flux along the lower Siang (Fig. 4).
Another possible source of perturbation of TCN signals in high-altitude
environments such as the Himalayas is the input of glacial sediments.
Glacially sourced sediments have been shielded from cosmic rays due to the
thick ice cover and therefore have low
Additional processes that may affect TCN signals in steep, highly eroding
areas are stochastic events and poor sediment mixing. Landslides and other
catastrophic events may perturb the downstream TCN concentration and, hence,
bias calculated CWDRs by supplying low-TCN-concentration sediments from
previously shielded bedrock (Kober et al., 2012). Modeling studies suggest
that this effect is more pronounced for small catchments of a few tens of
square kilometers or less and the magnitude of the perturbation depends on the ratio
of landslide volume to catchment-averaged sediment flux (Niemi et al., 2005;
Yanites et al., 2009). The catchments drained by the Tsangpo-Brahmaputra
downstream of the NBGPm exceed 200 000 km
There are no clear geomorphological indications that denudation rates downstream of the NBGPm are actually higher than those in the NBGPm. Evidence for other perturbations of the TCN signal that could explain the lag in denudation rates downstream of the NBGPm such as glacial sediment input, diffuse landsliding, or recent catastrophic flood events are also missing. We therefore explore other processes in the following paragraphs.
Conceptual diagram illustrating the evolution of the sediment
Abrasion of the sediment load during fluvial transport can affect TCN
concentrations because the grain size fraction analyzed is not necessarily
representative of the entire sedimentary load (Carretier and Regard, 2011;
Lukens et al., 2016). Olen et al. (2015) suggest that abrasion of the sand
fraction during transport could affect the TCN signal in some Himalayan
rivers when sand grains produced in the headwater regions are abraded during
transport and transferred to grain sizes smaller than what is typically
analyzed. However, in the case of the Tsangpo-Brahmaputra this would imply
that sediments coming from the South Tibetan region would pass below the
analyzed grain size fraction and be overwhelmed by low-
To explore the effects of abrasion on the TCN signal in the
Tsangpo-Brahmaputra catchment, we constructed a quantitative model of fluvial
abrasion that includes the dominant landslide-derived sediment production
(Attal and Lavé, 2006). This model allows us to estimate the typical
distances over which this dilution is predicted to occur in a setting such as
the Tsangpo-Brahmaputra. The model evaluates the response of apparent TCN
denudation rates,
The volumetric flux of sediments within the TCN grain size fraction
(
The average TCN concentration of sediments along the main channel,
Figure 7 shows the apparent
These model results suggest that abrasion of landslide material is able to
induce a downstream lag in the denudation rate calculated from
Conversely, the absence of a relation between grain size and
Abrasion effects in the Tsangpo-Brahmaputra are unlikely to be restricted to
detrital TCN studies. Other detrital proxies such as thermochronological ages
of sediment grains could be affected similarly, i.e., the young cooling ages
found in the NBGPm (e.g., Zeitler et al., 2014) could be transferred to the
typically studied sand fraction of the sediment load further downstream only
after coarse landslide material is abraded. The population of zircon fission
track (FT) ages (Enkelmann et al., 2011) and
Denudation rates measured upstream of the NBGPm and on the Tibetan Plateau
are the lowest denudation rates measured in the Tsangpo-Brahmaputra catchment
(ca. 0.04 to 0.2 mm yr
Denudation rates downstream of the Tibetan Plateau, from the NBGPm to the MFT
(i.e., the Himalayan part of the Tsangpo-Brahmaputra main stream course), need
to be considered carefully because of the abrasion processes discussed above.
Endorsing this model of downstream lag in the TCN response due to abrasion
effects means that the samples along the main stream of the Tsangpo-Brahmaputra
cannot be used to provide tight spatial constraints on the location and
magnitude of denudation within the NBGPm. The denudation signal is not
immediately transferred to the TCN grain size in the trunk stream and hence
TCN samples do not accurately represent the upstream denudation processes. A
first-order estimate of the denudation rates in the eastern Himalayan part of
the Tsangpo-Brahmaputra can nevertheless be made by assuming that the onset
of the steep gorges and the knickpoint marks the onset of intense landsliding
(Larsen and Montgomery, 2012) and that abrasion processes become less
significant beyond the gravel–sand transition at the Himalayan front (Dubille
and Lavé, 2015) close to the MFT. The evolution of the
Evolution of the
A very similar estimate can be made by estimating the
One important underlying assumption for these denudation rate estimates is
that the entire upper Tibetan part of the catchment is connected to the
Tsangpo-Brahmaputra trunk stream and that its entire drainage area exports
sediments at the calculated rate. A number of observations tend to suggest
that this may not be the case. The upper Tsangpo is partially dammed along
its course by several active and potentially subsiding N–S horst and graben
systems (Armijo et al., 1986), and several hundred meters of Quaternary
sediments are also stored directly upstream of the Tsangpo Gorge (Wang et
al., 2015). Internally drained areas and lakes represent additional breaks in
the sediment cascade. Furthermore, the Tibetan headwaters drain the
carbonate-rich TSS (Liu and Einsele, 1994),
which contain lower amounts of quartz than the Himalayan or NBGPm metamorphic
rocks. This would tend to overestimate the quartz flux that is exported by
the Tibetan Plateau relative to the quartz fluxes eroded downstream. In both
cases, the actual sediment flux exported by the South Tibetan part of the
catchment may be overestimated if calculated directly using
Altogether, we therefore suspect the sediment entering into the eastern
syntaxis at pt. (3) to be lower than the actual flux predicted by the TCN, but
we think that this difference remains relatively small. We conservatively
speculate here that the actual flux exported by the Tibetan Plateau part of
the catchment ranges between 50 and 90 % of the flux that would be
calculated by the TCN. The exact magnitude of this sediment flux reduction
remains to be quantified. Using a similar mixing and dilution model as
exposed above, we estimated the effect of a sediment flux exported by the
Tibetan Plateau of the catchment that is lower by 50 to 90 % compared to
the predicted flux (i.e., a sediment flux of 20 to 36 Mt yr
Downstream of the MFT, the Tsangpo-Brahmaputra enters the low-gradient
Indo-Gangetic flood plain and receives sediment from a number of significant
Himalayan tributaries with denudation rates ranging from 0.7 up to
4 mm yr
Overall, according to our TCN data, the Tsangpo-Brahmaputra receives
ca. 600 Mt yr
The
Our data also provide new constraints on the denudation across the
Tsangpo-Brahmaputra catchment. The upstream Tibetan Plateau part of the
catchment is denuding at a slow rate inferior to 0.2 mm yr
No data sets were used in this article.
The authors declare that they have no conflict of interest.
Kristina Hippe, Negar Haghipur, and Sean Gallen are thanked for the fruitful discussions and help in the lab. I. Schimmelpfennig is thanked for her help with the collection of the TSA samples. Natalie Vögeli is thanked for providing sample BRM MANAS. Réglis Braucher and the ASTER team are acknowledged for the swift measurement of TSA-16. We thank the associated editor, Simon Mudd, and editor, Joshua West, for efficient handling of the paper, as well as the two anonymous reviewers for their constructive and thoughtful comments. Maarten Lupker was supported by the ETH postdoctoral fellowship program. Jérôme Lavé and Christian France-Lanord were supported by the ANR Calimero project. Edited by: Simon Mudd Reviewed by: two anonymous referees