Rates of rockwall slope erosion in the upper Bhagirathi catchment, Garhwal Himalaya

Rockwall slope erosion is defined for the upper Bhagirathi catchment using cosmogenic Beryllium‐10 (10Be) concentrations in sediment from medial moraines on Gangotri glacier. Beryllium‐10 concentrations range from 1.1 ± 0.2 to 2.7 ± 0.3 × 104 at/g SiO2, yielding rockwall slope erosion rates from 2.4 ± 0.4 to 6.9 ± 1.9 mm/a. Slope erosion rates are likely to have varied over space and time and responded to shifts in climate, geomorphic and/or tectonic regime throughout the late Quaternary. Geomorphic and sedimentological analyses confirm that the moraines are predominately composed of rockfall and avalanche debris mobilized from steep relief rockwall slopes via periglacial weathering processes. The glacial rockwall slope erosion affects sediment flux and storage of snow and ice at the catchment head on diurnal to millennial timescales, and more broadly influences catchment configuration and relief, glacier dynamics and microclimates. The slope erosion rates exceed the averaged catchment‐wide and exhumation rates of Bhagirathi and the Garhwal region on geomorphic timescales (103−105 years), supporting the view that erosion at the headwaters can outpace the wider catchment. The 10Be concentrations of medial moraine sediment for the upper Bhagirathi catchment and the catchments of Chhota Shigri in Lahul, northern India and Baltoro glacier in Central Karakoram, Pakistan show a tentative relationship between 10Be concentration and precipitation. As such there is more rapid glacial rockwall slope erosion in the monsoon‐influenced Lesser and Greater Himalaya compared to the semi‐arid interior of the orogen. Rockwall slope erosion in the three study areas, and more broadly across the northwest Himalaya is likely governed by individual catchment dynamics that vary across space and time. © 2019 The Authors. Earth Surface Processes and Landforms Published by John Wiley & Sons, Ltd.


Introduction
Glaciation and glacial erosion are central to the topographic evolution of high-altitude mountain belts such as the Himalayan-Tibetan orogen, by influencing rates of sedimentation and localized incision, limiting relief production and elevation, and offsetting tectonic uplift (Brozović et al., 1997;Whipple et al., 1999;Mitchell and Montgomery, 2006;Wulf et al., 2010Wulf et al., , 2011Scherler et al., 2014). The contributions of periglacial erosion at the catchment head to the denudation budgets of Himalayan glacierized catchments have been largely overlooked, with the exception of studies by Heimsath and McGlynn (2008) in the Nepal High Himalaya and Seong et al. (2009) in the Central Karakoram of Pakistan. This is surprising given that lateral slope erosion via periglacial processes are shown to exceed rates of glacial incision in other alpine settings (Brocklehurst and Whipple, 2006;Foster et al., 2008).
Periglacial weathering processes including freeze-thaw, frost cracking and ice wedging deliver large volumes of rockfall and avalanche debris to the mountain glacier sedimentary system from catchment slopes (Schroder et al., 2000;Matsuoka, 2001;Owen et al., 2003;Hales and Roering, 2005;Sanders et al., 2012;Gibson et al., 2017). The strength of coupling between rockwall slopes and glaciers affect glacier dynamics, catchment sediment flux and can dictate the relief and topographic configuration of catchment divides over time (Montgomery, 2002;Thiede et al., 2005;Moore et al., 2009). Erosion of rockwall slopes in the Himalayan-Tibetan orogen therefore has broad implications for its morphological development and the distribution of precipitation (Burbank et al., 2003;Gabet et al., 2004;Anders et al., 2006;Bookhagen and Burbank, 2006).
The distribution and rates of erosion for the Himalayan-Tibetan orogen scale with tectonics (Burbank et al., 2003;Thiede et al., 2005;Scherler et al., 2014) precipitation (Thiede et al., 2004;Grujic et al., 2006;Biswas et al., 2007;Craddock et al., 2007;Gabet et al., 2008;Wulf et al., 2010;Deeken et al., 2011;Portenga et al., 2015) and/or topography (Vance et al., 2003;Scherler et al., 2011aScherler et al., , 2014. Erosion at the catchment head is shown to outpace catchment-wide and regional landscape denudation rates, and exhibit greater or different sensitivities to local and/or regional external forcing such as shifts in climate, tectonic activity or geomorphic regime (Heimsath and McGlynn, 2008;Scherler et al., 2011a). We aim to assess the importance of periglacial processes in the Himalayan-Tibetan orogen; an essential first step is the quantification of rockwall slope erosion rates. We chose the upper Bhagirathi catchment of the Garhwal Himalaya, northern India, for this initial investigation. This region is the source area for the Ganges and it contains some of the largest glaciers in the monsoon-influenced Himalaya, including Gangotri glacier. This catchment has a well-defined glacial chronostratigraphy, comprehensive records of past and modern glacier behavior and is relatively accessible. We apply geomorphic and sedimentological methods and measure cosmogenic beryllium-10 ( 10 Be) concentrations in medial moraine sediment to calculate rockwall slope erosion rates. Moreover by comparing 10 Be concentrations along the length of the medial moraines and examining the sedimentology of the supraglacial sediment, we are able assess the feasibility of using 10 Be to determine rates of rockwall slope erosion. We compare our erosion rates to local catchment-wide erosion and exhumation records to assess the difference between rockwall slope erosion and regional landscape denudation in Garhwal. We compare slope erosion rates for upper Bhagirathi, Chhota Shigri in the Lahul Himalaya, northern India and Baltoro in the Central Karakoram of Pakistan with catchment parameters and regional climate records to help identify the factors that may be affecting slope erosion in the northwest (NW) Himalaya.

Regional Setting
The Bhagirathi catchment is located in the Uttarkashi district of Uttarakhand, in the Garhwal Himalaya of northern India (Figure 1). Three major lithotectonic units characterize the geology of the Garhwal Himalaya: (1) Tethyan Himalaya sedimentary series; (2) the high Himalaya crystalline sequence (HHS); and (3) the lesser Himalaya sequence (LHS; Searle et al., 1997Searle et al., , 1999Vannay et al., 2004). Despite the absence of a clear shear zone, Garhwal is bounded in the north by the Tethyan Himalaya low-grade metasedimentary rocks and the south Tibetan detachment (STD) zone (Kumar et al., 2009;Srivastava, 2012). The main central thrust zone defines the southern margin of the region; the boundary between highgrade gneiss, migmatite and granite of the HHS and low-grade metasedimentary rocks of the LHS. The Jhala normal fault trends through central Garhwal and the Bhagirathi catchment, separating quartzo-feldpathic sillimanite gneiss from Harsil metasedimentary rocks (Searle et al., 1999). Maximum regional uplift rates range between 4 and 5.7 mm/a (Barnard et al., 2004a(Barnard et al., , 2004bScherler et al., 2014). Neotectonics, which include persistent microseismicity and stochastic earthquakes, greatly influence the geomorphic evolution of the region (Searle et al., 1987;Valdiya, 1991;Rajendran et al., 2000;Barnard et al., 2001;Bali et al., 2003). Detailed summaries of the geologic setting and histories of transient erosion, unroofing and exhumation for Garhwal are provided by Scaillet et al. (1995), Searle et al. (1993Searle et al. ( , 1999, Sorkhabi et al. (1996) and Scherler et al. (2014). The climate of the western Himalaya is influenced by two major climatic systems, the southwest Indian monsoon and the northern hemispheric mid-latitude westerlies (Finkel et al., 2003;Bookhagen et al., 2005;Owen, 2009). The majority of annual precipitation (1000À2500 mm/a) in Garhwal occurs between July and September; during this time the humid air masses of the Indian monsoon penetrate the high-altitude ranges of the Greater Himalaya (Burbank et al., 2003;Scherler et al., 2010;Thayyen and Gergan, 2010;Wulf et al., 2010). Rainfall magnitudes vary significantly both seasonally and across short distances (10 1 to 10 2 km) throughout the region, creating localized microclimates that are affected by the variability in terrain and geomorphic regime Barros et al., 2006;Singh et al., 2007;Srivastava, 2012).
Due to the restricted number of meteorological stations located above~5000 m above sea level (a.s.l.) in the Himalaya (Benn et al., 2012;Srivastava, 2012), climate and weather records for the upper Bhagirathi catchment are traditionally based on data from a single weather station (Mukhim, 30.6°N, 78.3°E, 1900 m a.s.l.). Mukhim station (1971À2000) records mean annual precipitation of 1648 mm and temperature of 15.5°C. An additional weather station has been established at Bhojbasa (~3780 m a.s.l., 30.9°N, 79.0°E),~4 km from the snout of Gangotri glacier. Temperatures range between -2. 3 and 11°C (2001-2009) each year. The station has documented a mean annual winter snowfall of~546 mm (Bhambri et al., 2011).
The upper Bhagirathi catchment preserves an abundance of moraines, debris flow/alluvial fans and cones, strath/fill terraces and landslides (Owen and Sharma, 1998;Burbank et al., 2003;Singh et al., 2017). Mass movements are particularly prevalent throughout the region as a consequence of glacial and fluvial erosion, heavy monsoon rains, localized storms and earthquakes, which each enhance slope instability Barnard et al., 2001Barnard et al., , 2004aBarnard et al., , 2004b.

Background
Past studies have quantified rockwall slope erosion by dating and estimating the volume of slope deposits such as talus (Andre, 1997;Curry and Morris, 2004;Hinchliffe and Ballantyne, 2009;Siewert et al., 2012), or modeling supraglacial debris flux (Heimsath and McGlynn, 2008;Gibson et al., 2017). More recently, rockwall slope erosion has been measured using cosmogenic 10 Be in medial moraine sediment (Heimsath and McGlynn, 2008;Seong et al., 2009;Ward and Anderson, 2011;Scherler and Egholm, 2017). The terrestrial cosmogenic nuclide (TCN) concentration of a substrate scales with clast size; concentrations are higher in boulders compared to coarse-fine sediment (Placzek et al., 2014). TCN-derivederosion rates from amalgamated sediment are considered to best reflect the average rate of erosion for an applicable area, and are treated as maximum estimates. Medial moraines form when rockfall and avalanche debris mobilized from the catchment slopes is exhumed to the ablation zone surface after being buried and transported englacially through the accumulation zone (Matsuoka and Sakai, 1999;Goodsell et al., 2005;MacGregor et al., 2009;Mitchell and Montgomery, 2006;Dunning et al., 2015). The 10 Be concentrations in medial moraine sediment reflect the mean surface concentrations of the source area (Ward and Anderson, 2011). For a given medial moraine sediment package, the shorter the duration of exposure of the source rockwall slopes to cosmic rays, the lower the accumulation of 10 Be, and the faster the inferred slope erosion rate. On sub-millennial timescales the 10 Be concentrations are likely to reflect slope erosion via periglacial weathering processes, whereas over geomorphic timescales (10 3 À10 5 years) the input is more likely affected by local or regional erosion rates (Gibson et al., 2017). We combine TCN methods with geomorphic and sedimentological analyses of Gangotri glacier Earth Surf. Process. Landforms, Vol. 44, 3108-3127 (2019) medial moraines to constrain rates of rockwall slope erosion for the upper Bhagirathi catchment.

Field methods
A detailed geomorphic map of the upper Bhagirathi catchment was constructed in the field and then refined using Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) global digital elevation models (GDEMs; 30-m-resolution), Landsat Enhanced Thematic Mapper Plus (ETM+; 15-mresolution) imagery, Google Earth Imagery and published topographic and geologic maps. Geomorphic and sedimentological techniques described by Benn and Owen (2002) were applied to identify and differentiate between landforms and sediment deposits.
Six major medial moraines were identified on the surface of Gangotri glacier system. Each landform originates from an area of glacier convergence and extends to a confluence with, and/or snout of Gangotri glacier ( Figure 2). The debris of each medial moraine is sourced from the rockwalls that encompass the glacier(s) above this area of convergence. Ward and Anderson (2011) argue that the trajectory of the englacial transport of rockwall debris is deeper than for debris that has been transferred onto the glacier surface at lower elevations in the catchment. The rockwall debris is entombed within the glacier ice during this transport and therefore does not mix with other sources.
The name of each moraine includes the initial term SD (for supraglacial debris moraine) and a subscript letter from A-F. The moraine sourced from the Kirti tributary catchment is referred to as SD A , for example. The SD A-C moraines are the focus of this study as they are the largest and most accessible moraines on the glacier, and extend throughout the ablation zones of Gangotri and Kirti glaciers, making them most likely to reflect rockwall slope erosion rates of the upper Bhagirathi catchment. The traditional moraine nomenclature, e.g. M 1-x , was not used to avoid confusion with Gangotri terminal and lateral moraines.
Sediment samples from each moraine were collected where possible at intervals from the snout of Gangotri to~3 km up-glacier. The moraines were inaccessible beyond this point due to major instabilities in both hillslopes and the glacier surface. The samples were taken from high relief, stable and well-defined moraine ridges, to avoid input from external sources of sediment, including lateral moraines and hillslope deposits (Supporting Information Item 1). To ensure the sampling of ≥ 10 rockwall events per sample, the sampling locations were ≥ 200 m 2 in area and sediment was collected systematically along the moraine crests every 5À10 paces. Approximately 3 kg of amalgamated sediment was collected for each sample, with a grain size range of < 3 cm (clay-coarse gravels) applying bulk sediment sampling methods of Gale and Hoare (1991). Detailed sedimentological and geomorphic descriptions of the moraine were made at each sampling location.
Two samples were collected from the SD A moraine, three from SD B and one from SD C . The samples were numbered in ascending order for each moraine, from the furthest up-glacier to the closest to the snout of Gangotri glacier. The G sup1 sample (subscript sup1), e.g., was collected at 4315 m a.s.l.,~3 km from the snout. The location of each sample was recorded using a handheld Garmin Etrex 30 global positioning system (GPS) unit and then photographed.

Medial moraine sediment analysis
To elucidate the characteristics and transport histories of the medial moraine sediment and to assist with interpreting the 10 Be inventories of our samples, analyses were conducted at the University of Cincinnati (OH, USA) in the Sedimentology and X-ray Laboratories in the Department of Geology and the Advanced Material Characterization Centre (AMCC). These analyses included grain size distribution (e.g. Wentworth, 1922;Allen, 1981), shape (e.g. Hambrey and Glasser, 2003;Hambrey et al., 2008;Lukas et al., 2013), roundness (e.g. Sneed and Folk, 1958;Ballantyne and Benn, 1994), surface weathering (e.g. Sheridan and Marshall, 1987;Owen et al., 2003) and sample clay mineralogy (e.g. Chen, 1977;Moore and Reynolds, 1997). Further details of the methodologies used are provided in Supporting Information Item 2.
Beryllium-10 production rates and geochemical analysis Beryllium-10 production rates for the upper Bhagirathi catchment were calculated from an ASTER GDEM (30-m-resolution) with a revised sea-level high-latitude spallogenic production rate of 4.08 ± 0.23 Be atoms/g SiO 2 /a (Martin et al., 2017; http://calibration.ice-d.org/) and 10 Be half-life of 1.36 Ma (Nishiizumi et al., 2007), using methods of Dortch et al. (2011) in MATLAB R2017.a. The production rate for each pixel was corrected for topographic shielding and then averaged to derive the mean 10 Be production rate for the catchment. The mean production rates account for 10 Be production on the glacier surface.
Following the initial sedimentological analyses, the sediment fractions of each sample were combined, crushed, and resieved; the 250-500 μm fraction was selected for 10 Be analysis. This process minimizes the possibility of bias in the contribution of any one grain size to the geochemical analysis of the sample. Quartz isolation, dissolution, chromatography, isolation of Be and the preparation of beryllium oxide (BeO) were undertaken in the Geochronology Laboratories at the University of Cincinnati using the standards and chemical procedures described by Kohl and Nishiizumi (1992), Nishiizumi et al. (1994), and Dortch et al. (2009). The 10 Be/ 9 Be of each sample were measured using accelerator mass spectrometry (AMS) at the Purdue Rare Isotope Measurement (PRIME) Laboratory at Purdue University, West Lafayette, IN, USA (Sharma et al., 2000). Portenga et al. (2015) have demonstrated that when native 9 Be is factored into 10 Be/ 9 Be ratios, irrespective of geologic setting, the resultant 10 Be concentrations and inferred erosion rates can be significantly altered (20À400%). The native 9 Be measured in~5 g fractions of clean quartz for each sample are < 1 ppm, so no adjustment to the 10 Be/ 9 Be ratios was necessary. A procedural blank 10 Be/ 9 Be ratio correction of 3 ± 1 × 10 -15 was made for each sample. Muongenic production is negligible for the timescales of the processes characterized for this study (Brown et al., 1995;Braucher et al., 2003).
The accumulation of 10 Be between the rockwall and medial moraines (during burial, englacial transport and exhumation of sediment to the glacier surface) was calculated using the Ward and Anderson (2011) analytical model and then subtracted from the total 10 Be concentrations. Beryllium-10 accumulation during transport along the length of the medial moraine was found to be negligible. This conclusion is discussed further in the Results and Discussion sections. Rockwall slope erosion rates were calculated using the 10 Be concentrations and catchment-wide production rates by applying methods described in detail by Lal (1991), Granger et al. (1996), Balco et al. (2008) and Dortch et al. (2011).
Medial moraines in the Himalayan-Tibetan orogen including Bhagirathi are primarily composed of rockfall debris; smaller rockfall events occurring frequently across time and space, interjected by larger and less frequent events Rowan et al., 2015). Despite much of the volumetric erosion being achieved by these larger events, this bias within the sediment source in favor of small rockfall events means that the resultant erosion rates should be considered maximum estimates. Unique sediment exchange processes in the supraglacial environment (Ward and Anderson, 2011;Lukas et al., 2012Lukas et al., , 2013Lupker et al., 2012;Scherler et al., 2015), shifts in sediment source, sediment storage and remobilization, shifting geomorphic regimes and ice/snow shielding (Bierman and Steig, 1996;Scherler et al., 2014;Fame et al., 2018) are also likely to affect the 10 Be concentrations and sediment characteristics of each sample. Caution must therefore be exercised when analyzing these data.
Interpreting the 10 Be concentrations of sediment in medial moraines is made more challenging because of the variability in production rates for alpine catchments and lack of uniformity in the 10 Be inventories at the bedrock surface, which vary over temporal and spatial scales. A detailed summary of this methodology and its assumptions are provided by Ward and Anderson (2011). Topographic and geomorphic analyses ASTER DEMs and Landsat ETM+ data were used in conjunction with geographic information system (GIS) Spatial Analyst tools for additional topographic analyses including catchment and rockwall slope, 3-km-radius relief, hypsometry and aspect. These parameters are defined for the glacial-periglacial realms of the upper Bhagirathi catchment only, as this is the area that pertains to the focus of our study. Topographic analyses of Chhota Shigri catchment in Lahul, northern India and Baltoro catchment in the Central Karakoram, Pakistan was also conducted to enable comparisons between slope erosion and catchment characteristics throughout the NW Himalaya.
An adiabatic environmental lapse rate (ΔT/ΔZ) of 7°C/km, a product of both an approximate median for the orogen (Derbyshire et al., 1991;De Scally, 1997;Thayyen et al., 2005;Siddiqui and Maruthi, 2007;Bashir and Rasul, 2010;Pratap et al., 2013;Kattel et al., 2013Kattel et al., , 2015, and a lapse rate derived from the local weather stations of Mukhim (~1900 m a.s.l.) and Bhojbasa (~3780 m a.s.l.; Bhambri et al., 2011) was used to estimate the summer and winter surface temperatures throughout the upper Bhagirathi catchment. The optimum frost cracking envelope defined by Hales and Roering (2005) falls between mean annual temperatures of -8 and -3°C. Bhagirathi catchment elevations that have surface temperatures within this range were determined using the chosen lapse rate and ASTER DEM. The frost-cracking envelope and distribution of permafrost (0°C; Brown, 1970) was also calculated with respect to depth by modeling the thermal structure of the near surface of the catchment using methods outlined in detail by Anderson and Anderson (2010). A thermal diffusivity of 1.15 mm 2 /s was used as it reflects an approximate midpoint in diffusivity values for the following substrates that characterize the catchment; regolith, landforms and deposits, and bedrock. This provides an estimate for the rate of heat transfer from the surface. Despite offering a reasonable assessment of the frost cracking and permafrost distribution, these simplified methodologies involve a series of assumptions about the physics, temperature data and geologic setting . Surface temperatures of upper Bhagirathi will have varied over space and time, a condition that must be accounted for when interpreting this data. These analyses aim to determine whether slope failure in the upper Bhagirathi catchment can be influenced by surface temperature and the associated periglacial weathering processes.
ELA and snowline altitude (SA) reconstructions ELAs and ELA depressions (ΔELA) were calculated for the contemporary and past glacial stages within the upper Bhagirathi catchment using methods described by Osmaston (2005), Benn et al. (2005), Heyman (2014) and Sharma et al. (2000). To reduce the uncertainties inherent within these reconstructions, a mean ELA was calculated for each glacial stage from reconstructions derived from each of the following methods: area-altitude (AA); area accumulation ratio (AAR) with ratios of 0.4, 0.5 and 0.6; and toe-headwall accumulation ratio (THAR) with ratios of 0.4 and 0.5; Benn et al. 2005). This approach has been successfully applied in Ladakh, northern India (Dortch et al., 2010;Orr et al., 2017Orr et al., , 2018Saha et al., 2018) and the Karakoram (Seong et al., 2007), reflecting accurate estimates of the ELAs. Our study adopts methods strongly recommended by Porter (2000), whereby the mean ELA of a glacial stage also provides an estimation of the snowline altitude (SA). The aim of reconstructing ELA and SAs is to evaluate the effect of the timing and nature of glaciation on the rates of rockwall slope erosion throughout the last glacial.

Medial Moraine Descriptions
Gangotri glacier system has six major medial moraines on its surface; the three investigated moraines of this study (SD A-C ) extend over 50% of the length of the glaciers' ablation zones (Figures 2 and 3; Table I). The medial moraines are composed of supraglacial diamict, and like many alpine glaciers, the debris thickness is heterogeneous over space and time (10 1 to 10 4 years), ranging from a few millimeters to several meters thick (Owen and Derbyshire, 1989;Schroder et al., 2000;Benn et al., 2012;Srivastava, 2012). The widths of the moraines range from 50 to 650 m, widening towards the snout of Gangotri glacier (Figure 3; Supporting Information Item 1). The moraine morphologies are characterized by irregular surface topographies, the result of variable diamict thicknesses, and distribution and orientation of steep relief ridges, depressions and ice cliffs. This heterogeneity contributes to variations in the surface morphology, mass balance and flow velocities of the glacier (Benn and Owen, 2002;Rowan et al., 2015;Swift et al., 2005;Haritashya et al., 2006;Hambrey et al., 2008;Gibson et al., 2017), in addition to affecting the glaciers' sensitivity to climatic and environmental change (Scherler et al., 2011a(Scherler et al., , 2011b. These supraglacial diamicts are composed of massive sandy boulder gravels with a finer sandy-silt matrix containing interstitial ice (Supporting Information Item 2). The diamicts are composed of biotite granite, tourmaline leucogranite, with some gneiss, mica schist and quartzite, reflecting the local bedrock (Searle et al., 1999;Srivastava, 2012). The subangular to very angular boulder gravels have surfaces that range from unweathered to moderately weathered. Striations or chattermarks are not present on any particle size. Large boulders (> 2À0.25 m) are located on or slightly offset from the moraine ridges with some evidence of varnish and previous toppling. The sediment samples are composed of granites and schists, with the exception of the SD A samples that include some gneiss clasts.
Finer sediment increases with proximity to the debris-ice interface, likely as a result of sorting through rainfall and meltwater flows (Hasnain and Thayyen, 1996). These fine sediments, and evidence of frost action on pebblesÀboulders, indicate active periglacial weathering processes and continued sediment production and/or clast modification by interclast attrition and abrasion across the glacier surface Benn and Evans, 2014;Benn et al., 2012). No clear englacial sediment horizons were identified in the field, despite some evidence of englacial silts and sands at the glacier surface. The exhumation of subglacial sediment to the englacial or supraglacial environments is therefore likely to be very localized . Discontinuous soil development and tundra vegetation are restricted to the stable medial moraine ridges.

SD A moraine
The SD A medial moraine extends~12 km from the Kirti tributary catchment to within~500 m of the snout of the Gangotri glacier. A medial moraine from a Kirti sub-catchment coalesces with the main tributary moraine at 4670 m a.s.l. SD A narrows in width at the confluence (~600-300 m) between Kirti and Gangotri glaciers, as a result ice deformation and shearing (Hubbard et al., 2004;Gibson et al., 2017). The SD A diamict has a unique rusty brown color, likely the result of the weathering of the local Augen gneiss or Vaikrita group gneiss bedrock slopes ( Figure 3B; Supporting Information Item 1). Sharing the same source outcrops, the medial moraine (SD E ) of Ganohim glacier also has this coloration. The G sup1 (30. 9°N , 79.09°E, 4315 m a.s.l.) was retrieved~3 km from the glacier snout. The G sup2 (30.9°N 79.08°E) was collected at 4280 m a.s. l.,~500 m downstream and northwest of G sup1 . SD B moraine SD B is the centermost, gray medial moraine of Gangotri glacier that extends~15 km from the modern ELA (4510-5160 m a.s.l.) to the glacier snout ( Figure 3C; Supporting Information Item 1). SD B is the most stable moraine of this study, with proportionally fewer ice cliffs, ice collapse features and interstitial ice within the diamict matrix. The G sup3 (30.89°N, 79.09°E) was retrieved at 4325 m a.s.l.,~300 m northeast of G sup1 ,~3 km from the glacier terminus. The G sup4 (30.90°N, 79.09°E) was collected 500 m northwest and down glacier from G sup3 , at 4285 m a.
SD C moraine SD C also extends~15 km from the modern ELA to the glacier snout. The diamict has a slightly darker gray coloration to the SD B moraine ( Figure 3D; Supporting Information Item 1). Due to access, only one sample could be retrieved from this moraine,~500 m from the glacier snout. The G sup6 (30.92°N, 79.08°E) was collected at 4130 m a.s.l., slightly upstream from the confluence between the Raktavaran tributary catchment and Gangotri trunk glacier.

Medial moraine sediment analysis
Gangotri and Kirti glacier medial moraine samples consist of medium-coarse sands and fine gravels; these fractions together account for 71 to 73% by weight of each sample (Figure 4). The variability in grain size distributions in the samples is insufficient to draw any conclusions about the transport and sorting of medial moraine sediment in upper Bhagirathi. No significant relationship is observed between grain size distribution and proximity of the sample to the glacier snout or margin (Supporting Information Item 2). Caution must be exercised however when interpreting shifts in grain size, or lack of, with distance down-glacier, as only three or fewer samples are collected from each moraine. The medial moraine samples are largely made up of angular (32À48%) and subangular (19À46%) grains, with ≤ 1% of grains considered either rounded or well rounded (Supporting Information Item 2). Grains of low sphericity constitute between 71 and 82% of each sample. The bladed (14À23%) and very bladed (30À40%) grain shape classes are the most prevalent the samples, where over 50% of grains per sample have c:a and b:a ratios < 0.3 ( Figure 5). No significant relationship can be identified between grain size and grain roundness or sphericity for any one moraine or sediment sample (Supporting Information Item 2). The covariance of clast shape and roundness indices are presented in RA-C 40 (angular, very angular) and RWR-C 40 (rounded, well rounded) plots for Gangotri glacier and previous studies in Figure 5. Distinguishing between transport pathways must be approached with care due to the pronounced overlap in facies indices, an important consideration when working in complex alpine settings (Lukas et al., 2013). The RA (85À94%) and C 40 (75À96%) indices and large proportion of bladed and extremely bladed grains suggest that the medial moraines of this study share a supraglacial transport history (Benn and Owen, 2002;Hubbard, 2004;Lukas et al., 2013). Some extraglacial and moraine control samples also record RA values greater than 80%. The RWR indices for the medial moraines range between 6 and 15%, higher than the~0% values typical for supraglacial samples. The more rounded component of the samples may reflect the input of sediment from other realms of the glacial system, clast rounding by englacial and/or supraglacial transport, or the effects of meltwater on the glacier surface.
Percentage surface weathering estimates of quartz grains range from 30 to 100%, with mean surface weathering for the samples ranging between 66 ± 22 to 78 ± 17%. A slight, yet negligible increase in surface weathering can be identified down glacier for the SD A and SD B moraines, and the glacier as a whole. Further detail of these earlier-mentioned analyses and further sediment analyses that proved less significant to the conclusions of this study are provided in Supporting Information Item 2.

Rockwall slope erosion rates of the upper Bhagirathi catchment
The modeled accumulation of 10 Be during the transport of sediment between the source rockwall and medial moraine ranged between 0.02 × 10 4 and 0.025 × 10 4 at/g SiO 2 (Table II; Supporting Information Item 3). The subtraction of this 10 Be from the total measured 10 Be concentrations increases the calculated erosion rates by~8.7 to 30.2% relative to uncorrected values.
The accumulation of 10 Be during transport along the moraine length is also a concern due to the high production rates (87.0 ± 11.3 to 107.3 ± 13.9 at/g/a) of upper Bhagirathi. If 10 Be were to accrue during this transport, G sup5 would theoretically record a concentration at least 0.4 × 10 4 at/g greater than G sup3 , which is located 2.5 km up-glacier. Instead a difference of~0.1 × 10 4 at/g is measured, which shows that the sampled sediment was transported down-glacier at depth in the medial moraine, permitting limited-no 10 Be accumulation. In order to test this assumption, the maximum accumulation of 10 Be along the SD AÀC moraines was estimated used methods outlined by Seong et al. (2009;Supporting Information Item 3). The estimates exceed the total sample concentrations measured, therefore indicating little 10 Be production during transport along the medial moraines. This suggests that the supraglacial sediment samples were not exposed at the surface for the entire length of each landform. The collection of multiple samples per medial moraine with a known source area is therefore recommended in order to help identify samples that do not reflect the mean surface concentrations of the catchment rockwall. To reduce the potential accumulation of 10 Be during this transport, and its affect upon the erosion rates, this approach is perhaps best applied in catchments with medial moraines < 5 km in length, particularly in areas with high annual 10 Be production.
Furthermore, G sup1 and G sup2 from the SD A moraine have 10 Be concentrations of 1.1 ± 0.2 and 1.6 ± 0.3 × 10 4 at/g SiO 2 , respectively. These concentrations infer rockwall slope erosion rates of 6.9 ± 1.9 mm/a for the G sup1 sample, and 4.3 ± 1.1 mm/a for G sup2 (Table II). The G sup3 , G sup4 , and G sup5 samples from the SD B moraine have 10 Be concentrations of 2.7 ± 0.3, 2.5 ± 0.3 and 2.6 ± 0.3 × 10 4 at/g SiO 2 respectively. An inferred erosion rate of 2.4 ± 0.4 mm/a is derived from G sup3 , 2.5 ± 0.5 mm/a from G sup4 , and 2.4 ± 0.4 mm/a from G sup5 . The SD C G sup6 sample has a 10 Be concentration of 1.5 ± 0.4 × 10 4 at/g SiO 2 and an inferred slope erosion rate of 4.3 ± 1.4 mm/a.

Topographic and geomorphic analyses
The detailed geomorphic mapping of upper Bhagirathi revealed that the identification of discrete geomorphic zones within the catchment is not possible owing to the absence of the vertical stratification of landforms (Figure 2; Supporting Information Item 4). River terraces and fans occupy elevations < 5000 m a.s.l., whereas the remaining landforms extend the full extent of the catchment.
The slopes of upper Bhagirathi catchment range from 0°to 75°, gentle (< 30°), moderate (31°À45°) and steep (> 46°) slopes occupying 38, 25 and 37% of the total catchment, respectively. The mean tributary catchment slopes of the study area range between 29.6 ± 15.7 and 40.7 ± 18.4° (Table I). Between 10 and 53% of the catchment areas are occupied by glaciers; the glacier surface slopes are included within this catchment slope analysis, which introduces a degree of uncertainty to these mean slope values. Catchment 3-km-radius relief ranges from 1.4 ± 0.4 to 2.2 ± 0.3 km.
Ideal conditions for periglacial weathering processes including frost-shattering, cryofracturing, and frost heave are present throughout the upper Bhagirathi catchment. Optimum frost cracking conditions (-3 to -8°C) within the catchment migrate from elevations of~5680À6380 m a.s.l. during the summer, tõ 3780À4480 m a.s.l. during the winter ( Figure 6A). This is consistent with the frost-cracking envelope between 4000À6000 m a.s.l. devised by Brozović et al. (1997) for the NW Himalaya. These temperatures extend to a maximum depth of~2.3 m into the near-surface, between 3780 and 4430 m a.s.l. (Figure 6B).
Peaks which exceed~6380 m a.s.l. (including Shivling, Meru and the Chaukhamba Massif) have surface temperatures < -8°C , which theoretically reduces the efficiency of periglacial weathering processes. The distribution and magnitude of these processes are affected by diurnal and seasonal cycles and climatic and microclimatic variations. Optimum frost cracking conditions over the last few glacial cycles are likely to have extended to lower and higher elevations within the catchment than the present. Transient or seasonal permafrost can occur at elevations between 3380 and 5280 m a.s.l. within upper Bhagirathi; at higher elevations the permafrost can be permanent. Transient permafrost, which exacerbates slope instabilities and mass wasting (Fischer et al., 2006), may extend from the catchment slopes to the proglacial zone of Gangotri glacier and catchment floor. Permafrost penetrates the near surface between 3080 and 3380 m a.s.l., to a maximum depth of~0.8 m ( Figure 6B).

ELA and SA reconstructions
The ELAs of contemporary glaciers range from 4880 to 5665 m a.s.l. (Table III), falling within the uncertainties of, and marginally above past estimates of 4510 to 5390 m a.s.l. (Owen and Sharma, 1998;Naithani et al., 2001;Ahmad et al., 2004, Burbank et al., 2003Srivastava, 2012;Singh et al., 2017). Gangotri glacier has retreated~30 km upstream over the past 60 ka (Owen and Sharma, 1998;Burbank et al., 2003), the ELA rising in elevation from 4095 ± 295 to 5160 ± 160 m a.s. l., providing an ΔELA of 1065 ± 295 m. Glacial studies across the Tethyan Himalaya of northern India (Dortch et al., 2011;Orr et al., 2017Orr et al., , 2018 and the Tibetan plateau (Heyman, 2014) document maximum ΔELAs between 240À290 and 280À494 m, respectively; the magnitude of Gangotri glacier retreat over this timescale stands in stark contrast to these. This rate of recession has yet to be determined in Garhwal, where local glacial stages record ΔELAs < 100 m within the past 1 ka, compared to an ΔELA of 465 ± 100 m for Gangotri glacier. A net loss in glacier volume since 1.6 ka is indicated by the heights of the ice-contact Gangotri glacial stage moraines relative to the glacier surface. Approximately 50% of the total catchment is above the modern snowline altitude (5160 ± 160 m a.s.l.).

Bhagirathi rockwall slope erosion
Medial moraine sediment characteristics for the lower~3 km of the ablation zone of Gangotri glacier are broadly similar, despite the discrete origins of each landform. Grain shape analysis indicates a predominantly supraglacial transport history with possible contributions from other landscape realms, i.e. moraine, extraglacial sources (hillslope deposits). Grain roundness and surface weathering is attributed to periglacial weathering processes including freeze-thaw, frost cracking and ice wedging, which dislodge angular rock fragments from the bedrock and/or regolith slopes (Benn and Lehmkuhl, 2000;Schroder et al., 2000;Benn et al., 2003;Hambrey et al., 2008;Lukas et al., 2012).
The 10 Be concentrations vary between the samples of each moraine of Gangotri glacier and between individual landforms, despite sharing similar sediment characteristics (Figure 7; Table II). No relationship is evident between 10 Be concentration and distance down-glacier or proximity to the glacier margin. The range in 10 Be concentrations of our dataset (1.1 ± 0.2 × 10 4 to 2.7 ± 0.3 × 10 4 at/g SiO 2 ) may be due to variability in the timing and magnitude of mass wasting events, the insufficient mixing of sediment, the prior or punctuated exposure to cosmic rays, and shielding by snow, ice or regolith Ward and Anderson, 2011;Heyman et al., 2011).
The SD A moraine records the lowest 10 Be concentrations (1.1 ± 0.2 × 10 4 to 1.6 ± 0.3 × 10 4 at/g SiO 2 ), corresponding to the highest inferred slope erosion rates of this study (4.3 ± 1.1 to 6.9 ± 1.9 mm/a). The close proximity of SD A to rockwall slopes and external sediment sources along the length of the landform, and the possible greater sensitivity of smaller catchments to external forcing, are two possible explanations for these lower TCN concentrations. The concentration disparity between the SD A samples may due to the sampling of isolated rockfall event(s) rather than amalgamated moraine sediment. Similarly, the sedimentology and low 10 Be concentrations of SD C G sup6 may be the result of sediment input from proximal rockwall slopes at the snout of Gangotri glacier or contributions from Raktavaran and/or Chaturangi tributary glaciers.
The SD B samples have the highest 10 Be concentrations of this study (2.5 ± 0.3 × 10 4 to 2.7 ± 0.3 × 10 4 at/g SiO 2 ), which each fall within uncertainty of each other, and record the lowest inferred slope erosion rates (2.4 ± 0.4 to 2.5 ± 0.5 mm/a). The SD B moraine extends the full length of the ablation zone of Gangotri glacier with no direct contact with the catchment slopes. Accordingly, the SD B rates are considered to be the most representative of upper Bhagirathi rockwall slope erosion, and likely captures the background erosion rates of the periglacial realms of the catchment due to mass wasting. The SD B erosion rates are therefore likely to be largely dictated by high frequency, low magnitude mass wasting events, while the SD A and SD C signals are controlled by large stochastic events. These SD B rates affirm that~2.5 m of lateral slope erosion through periglacial processes can be achieved across a single millennium in this catchment, and > 65 m when extrapolated for the whole of the Holocene. To test the robustness of our interpretations, it would be worthwhile to extend this investigation up-glacier and evaluate variations in 10 Be concentrations throughout the ablation zone, and across grain sizes.
The ELA reconstructions for the local glacial stages show that the size of the glacier and the spatial extent of the periglacial realms has decreased over the last 60 ka (Table III). This means that the slope area contributing debris directly to the accumulation zones of the glaciers have reduced, alongside the contribution of glacial erosion to the debutressing of slopes.
Studies suggest that the timing and nature of glaciation and the associated geomorphic change for the upper Bhagirathi catchment is primarily governed by climate (Barnard et al., 2004a(Barnard et al., , 2004bSrivastava, 2012;Singh et al., 2017). The magnitude and rates of rockwall slope erosion are therefore not only intrinsically linked to climate through periglacial weathering processes, but also as a result of climate-driven glaciation affecting the extent of slope-glacier coupling. Accordingly, rates of rockwall slope erosion and the contribution of the periglacial realms to the denudation budget of the catchment is likely to have fluctuated throughout the last glacial.
Garhwal landscape denudation Scherler et al. (2015) have shown that rates of fluvial incision in Garhwal during the late Pleistocene was greater than the present by a factor of~2 to 4. Whether periglacial erosion has remained constant or varied across these timescales, the influence of rockwall slope erosion on the topographic evolution of upper Bhagirathi is likely to have been maintained over time. The magnitude of this erosion is likely to affect sediment flux and the storage of snow and ice from diurnal to millennial timescales in this setting, and then more broadly influence catchment configuration and margin migration, microclimates (Bhambri et al., 2011;Srivastava, 2012), and be sufficient to limit relief and affect the architectural organization of the local fault systems (Valdiya, 1991;Sorkhabi et al., 1996;Bali et al., 2003). The frequency and magnitude of rockfall events in Bhagirathi is therefore likely to be affected, in part, by catchment-specific conditions such as geologic setting and geomorphic regime, and then external forcing such as shifts in climate or tectonism. The global intensification of late Pleistocene glaciation, for example, caused extensive mass redistribution and localized incision throughout the Himalayan-Tibetan orogen (Zeitler et al., 2001;Brozović et al., 1997;Bookhagen et al., 2005;Hewitt, 2009;Whipple, 2009). This is likely to be similar in the Bhagirathi catchment where landscape denudation can be attributed to a changing glacier mass balance over time.
Comparing lateral slope erosion with other records of landscape change is a challenge as they invariably reflect erosion or denudation through a variety of mechanisms and across different temporal and spatial scales. Moreover, TCN derived catchment-wide erosion rates reflect the net surface lowering of a catchment, which accounts for both vertical and lateral erosion. Similarly, rates of exhumation defined using low temperature thermochronology describe the ascension of rock through modeled isotherms. To better compare our dataset with these records, we calculate an approximate vertical component to our slope erosion data (Table II). Overall the rates of rockwall slope erosion largely exceed the averaged catchment-wide erosion and exhumation rates of upper Bhagirathi and the Garhwal region (Figure 7). Our erosion dataset supports the view that slope erosion of alpine headwaters can outpace the wider drainage basin (Oskin and Burbank, 2005;Naylor and Gabet, 2007), and that the distribution and magnitude of erosion can vary significantly over short distances downstream (Scherler et al., 2014). The difference in the rates of erosion and landscape denudation between these various records may be because our slope erosion dataset offers a higher resolution record of erosion (10 1 À10 4 years) than those on the catchment or mountain range scale (10 4 À10 6 years) and/or that these latter records eliminate the 'noise' in sediment flux data over time, such as single mass wasting events initiated by large and/or stochastic seismic events (Sadler and Jerolmack, 2014;Willenbring et al., 2013). The slope erosion rates of this study remain largely lower than the regional uplift rates of 4 to 5.7 mm/a (Barnard et al., 2004a(Barnard et al., , 2004bScherler et al., 2014), which may explain the preservation of high relief slopes within the study area.

Rockwall slope erosion of the NW Himalaya
Comparisons between 10 Be concentrations in medial moraine sediment of the Bhagirathi glaciers with similar datasets from Chhota Shigri in Lahul (Scherler and Egholm, 2017) and Baltoro glacier in the Central Karakoram  show that the 10 Be concentrations at these other localities exceed those in our study (Figure 8). The SD B samples that are considered to Figure 5. Covariance plots of medial moraine samples from Gangotri glacier with a compilation of sediment samples from high altitude alpine catchments. Batal (Benn and Owen, 2002), Khumbu (Hambrey et al., 2008), d'Arolla (Goodsell et al., 2005) and Findelen, Pasterze, Estelette, Tasman, Vadret and Fox glaciers (Lukas et al., 2013).  Be ratios are corrected for background 10 Be detected in procedural blank (0.3 ± 0.1 × 10 -14 ). Negligible (< 1 ppm) 9 Be was detected in each sample.
c Accumulation of 10 Be during burial, englacial transport and exhumation is calculated using methods detailed in Ward and Anderson (2011;  best reflect upper Bhagirathi slope erosion measure 10 Be concentrations > 1 × 10 4 at/g SiO 2 lower than Chhota Shigri or Baltoro, and record erosion rates twice as fast. The 10 Be concentrations of the three study areas are compared with catchment parameters and regional climate records to decipher the possible drivers of rockwall slope erosion in the NW Himalaya. No significant relationship is evident between catchment area and 10 Be concentration. The catchment area does not account for the total surface area of the source rockwall slopes, a parameter that may influence these concentrations to a greater extent. Approximately 80, 50 and 53% of the total catchment area of Bhagirathi, Chhota Shigri and Baltoro respectively, are above the ELA/SA altitudes and nourish the glaciers through snow and ice avalanching. Changes to the SA over time are likely to affect the relative abundance of exposed bedrock and regolithcovered slopes and will help to moderate the slope debris flux. Similarly, there is no correlation between 10 Be and glacier area, and by association, size of medial moraine. No relationship is apparent between mean catchment or rockwall slope and 10 Be concentration, despite other studies being able to link these variables on a catchment scale and show that greater slope angles promote a larger debris flux (Finlayson et al., 2002;Burbank et al., 2003;Ouimet et al., 2009;Scherler et al., 2011bScherler et al., , 2014. Although the slopes will broadly facilitate rockfall and avalanching (Luckman, 1977;Gruber and Haerberli, 2007;Bernhardt and Schulz, 2010;Nagai et al., 2013) and the eventual evacuation of sediment from the catchment, each of the investigated catchments is also able to store extensive volumes of sediment in the form of landforms and sediment deposits (Figure 2; Seong et al., 2009). These landforms typically have gentler slopes, which transfer sediment to the glacier surface via diffusive creep processes (Carson and Kirkby, 1972). Not only will these stores of sediment have implications for the sediment flux of the catchments, but also they may influence the 10 Be concentrations measured within medial moraine sediment.  The 3-km-radius relief of the study areas exceed~1.6 km, the steeper relief catchments measuring the highest 10 Be concentrations, and therefore the lowest inferred rockwall slope erosion rates. This suggests that rates of rockwall slope erosion can in some cases be insufficient to limit catchment relief in the NW Himalaya. A more likely explanation is that the 3km-radius relief is largely dictated by the local uplift of the study areas.
A possible lithological control to slope erosion is expressed in upper Bhagirathi. The highest rates of erosion are defined by the SD A and SD C moraines, which are sourced from rockwalls composed of, in part, augen gneiss and schist respectively. Lower rates of erosion from SD B may be due to the granitic source rockwalls that have a greater rock mass strength than those of SD A and SD C , and are therefore more broadly resistant to erosion (Bhattarai and Tamrakar, 2017). However, a lithological control to erosion is less clear for the NW Himalaya, where we measure a large range of 10 Be-derived slope erosion rates within an area argued to have a relatively uniform rock mass strength (Burbank et al., 2003;Scherler et al., 2014). Investigating the jointing, structure and moisture content of the catchment walls would help to evaluate the susceptibility of the rock to failure and the ongoing damage of frost action (Hallet et al., 1991;Murton et al., 2006;Hales and Roering, 2007).
The temperature data is recovered from weather stations outlying the study areas and therefore does not accurately reflect catchment temperatures. The ranges in annual recorded temperatures prevent any correlations being made between this climatic parameter and the derived slope erosion. The high altitude setting of each study area with mean catchment elevations > 4000 m a.s.l. (Figure 8D) does however mean that the rockwall slopes of each catchment lie within the Brozović et al. (1997) 4000À6000 m a.s.l. frost cracking window for the NW Himalaya.
A tentative relationship lies between mean annual rainfall and 10 Be concentration, where higher rainfall coincides with higher rockwall slope erosion rates. This supports the extensive work on the coupling between precipitation and erosion in the Himalaya, where enhanced moisture in the monsoon-influenced Lesser and Greater Himalaya is thought to drive more rapid landscape denudation, compared to the semi-arid interior of the orogen Owen 1998, 2002;Harper and Humphrey, 2003;Bookhagen et al., 2005;Anders et al., 2006;Bookhagen and Burbank, 2006;Gabet et al., 2004;Owen, 2009). This relationship is not completely straightforward for our study areas however, as 10 Be concentrations shared by samples from Baltoro glacier (4.4 ± 0.3 × 10 4 to 11.7 ± 2.2 × 10 4 at/g SiO 2 ) and Chhota Shigri (3 to 6 × 10 4 at/g SiO 2 ) receive contrasting annual rainfall of < 500 and > 900 mm, respectively. The complex climate-topography interactions of each study area prevent a conclusive relationship between erosion and this climatic parameter from being identified. This association does suggest however that rockwall slope erosion is sensitive to precipitation and therefore the glacial-periglacial realms of Himalayan catchments are likely to respond to major climatic events and/or environmental change over time.
Studies across the Himalayan-Tibetan orogen have drawn links between erosion, climate and topography (Scherler et al., 2011a(Scherler et al., , 2011b(Scherler et al., , 2014Bookhagen et al., 2005;Bookhagen and Burbank, 2006;Gruber and Haeberli, 2007;Dortch et al., 2011). Our regional assessment of the NW Himalaya has demonstrated that no single discussed parameter provides a dominant control for rockwall slope erosion. We must therefore consider what other variables may be influencing landscape change in these high-altitude settings.
Although glacial erosion is considered secondary to periglacial erosion; glacier dynamics may affect the rates of rockwall slope erosion. Temperate glaciers, which occupy the monsoon-influenced Himalaya, erode the glacier bed through quarrying and abrasion, which exploit fractures at the base of the catchment slopes, and also generate subglacial debris which may later be incorporated into the medial moraines (Benn and Evans, 2014;Benn and Owen, 2003). Glacier retreat and changes to mass balance can therefore lead to the debuttressing of these slopes and release glacially derived sediment. These processes can affect the rockwall debris flux and either contribute to, or trigger mass wasting (Church and Slaymaker, 1989;Watanabe et al., 1998;Ballantyne, 2002aBallantyne, , 2002b. In the semi-arid Himalaya, sub-polar glaciers frozen to the bed are unlikely to further slope denudation processes and therefore would maintain low erosion rates. The velocity of Gangotri (< 5À120 m/a; Gantayat et al., 2014;Bhattacharya et al., 2016), Chhota Shigri (~20À50 m/a; Wagnon et al., 2007;Azam et al., 2012) and Baltoro (~30À160 m/a; Copland et al., 2009) may also affect the efficiency of glacial erosion and/or influence the 10 Be concentrations by largely dictating the residence time of sediment on the glacier surface. Glacial hydrology and snow blow may also affect the rates of slope erosion over time (Matsuoka andSakai, 1999, Mitchell andMacGregor et al., 2009;Scherler et al., 2011b;Barr and Spagnolo, 2015).
Studies throughout Garhwal and the NW Himalaya have underpinned a tectonic control within the distribution and magnitude of denudation, where landscape change is strongly influenced by the Indo-Asian convergence and rock uplift patterns dictated by the geometry and shortening of the Main Himalayan Thrust (Burbank et al., 2003;Thiede et al., 2005;Scherler et al., 2014). Although contributing to this work is beyond the scope of this study, persistent seismicity throughout the Punjab, Himachal Pradesh and Uttarkhand districts of northern India may introduce a neotectonic control to landscape evolution (Bali et al., 2003;Scherler et al., 2014). In Garhwal, e.g., the 1991 Uttarkashi (M 6.1; Valdiya, 1991;Owen et al., 1996;Bali et al., 2003) and 1999 Chamoli (M 6.6; Rajendran et al., 2000) earthquakes occurred during the applicable timescales of the Bhagirathi slope erosion record and may have therefore triggered mass redistribution on a sufficient scale to affect the erosion rates of our study. The mobilization and transfer of sediment from the catchment walls to the glacier surface is an example of one of the primary stages in the evacuation of sediment from a glacierized catchment. Models of sediment transfer on the catchment scale argue that a shift in sediment flux requires a set of preconditioning factors and one or more forcing factor (Ballantyne, 2002a(Ballantyne, , 2002bMcColl, 2012;Orr et al., In press). This is true of the upper Bhagirathi catchment, where the pre-existing landscape dynamics, which include catchment parameters, and transitions in climate, tectonic or geomorphic regime, are necessary to explain the nature and rates of rockwall slope erosion over time. Understanding the topographic evolution and configuration of upper Bhagirathi and glacierized catchments throughout the NW Himalaya is made particularly challenging as it involves processes that operate across a variety of temporal and spatial scales. Despite the controls of alpine headwater evolution remaining elusive, this is the first study to quantify the rates of rockwall slope erosion in Garhwal, and has helped to demonstrate the importance of rockfall processes and the lateral erosion of slopes within mountain sedimentary systems.

Conclusion
Rockwall slope erosion has been defined for the upper Bhagirathi catchment by measuring 10 Be concentrations in sediment samples from three medial moraines of Gangotri glacier system. The concentrations are corrected for accumulation of 10 Be between the source rockwall and the medial moraine. Accumulation along the length of the medial moraines is found to be negligible. The 10 Be sample concentrations (1.1 ± 0.2 to 2.7 ± 0.3 × 10 4 at/g SiO 2 ) therefore reflect rates of slope erosion only. The slope erosion of the upper Bhagirathi catchment is best reflected by the SD B moraine rates, which range from 2.4 ± 0.4 to 2.5 ± 0.5 mm/a. These rates affirm that~2.5 m of lateral slope erosion through periglacial processes can be achieved across a single millennium in this catchment, and > 65 m when extrapolated for the whole of the Holocene. Slope erosion is therefore sufficient to affect sediment flux and glacier dynamics in upper Bhagirathi, in addition to helping set the pace of topographic change at the catchment head.
The rockwall slope erosion rates (2.4±0.4-6.9±1.9 mm/a) exceed the averaged catchment-wide (0.1 ± 0.001 to 5.4 ± 0.5 mm/a; Figure 7) and erosional exhumation (1.5 ± 0.5 mm/a; Figure 7) rates of Bhagirathi and the Garhwal region, indicating that erosion at the headwaters can outpace downstream reaches and the wider catchment. A possible explanation is that the high-altitude periglacial settings of upper Bhagirathi have a greater sensitivity to external forcing such a shift in climatic conditions, than the wider catchment or mountain range. The variance found between the rates of landscape denudation may also be due to the difference in the nature and resolution of the erosion records.
Rockwall slope erosion rates are higher in upper Bhagirathi compared to the catchments of Chhota Shigri in the Lahul Himalaya and Baltoro glacier in the Central Karakoram. Comparisons were made between the erosion datasets of these three Shigri (Wagnon et al., 2007) and Baltoro (Mihalcea et al., 2006(Mihalcea et al., , 2008. (H) Annual rainfall for Bhagirathi (Bhambri et al., 2011, Srivastava, 2012, Chhota Shigri and Baltoro (TRMM rainfall record [1998À2005]; Bookhagen and Burbank, 2006 study areas with catchment parameters and regional climate records, including catchment and glacier area, mean elevation and slope, 3-km-radius relief and annual temperature and rainfall. A tentative relationship is evident between erosion and precipitation, where more rapid slope erosion was recorded in the monsoon-influenced Lesser and Greater Himalaya, compared with the semi-arid interior of the orogen. No other individual catchment attribute was found to offer a dominant control on the rates of slope erosion in the NW Himalaya. We were unable to confidently link rockwall slope erosion with climate-topography. We conclude that rockwall slope erosion in the three study areas and then more broadly across the NW Himalaya is likely governed by individual catchment dynamics, which vary across space and time. The frequency and magnitude of rockfall and avalanche events is therefore determined by a set of preconditioning factors unique to each catchment, and one or more local and/or regional forcing factor. By continuing to decipher the rates and controls of rockwall slope erosion, we will improve our understanding of the role and importance of periglacial processes in the morphological development of mountain ranges and contribute to future studies of sediment flux and wider landscape change across the orogen.

Supporting Information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Figure S1.1. Images of the sampling locations for the investigated medial moraines of the Gangotri glacier system. White line denotes the moraine ridge. Figure S1.2. Cross-sections of the sampling locations for the investigated medial moraines of the Gangotri glacier system. Figure S2.1. Photomicrographs of the G sup4 sample from the SD B medial moraine. A) Image of moraine surface at the G sup4 sampling site. B) Image of fine-coarse sand sediment fractions taken using digital microscope. C) SEM image (200x mag) of silt-clay fraction. Figure S2.2. SEM images of Gangotri medial moraine samples. G sup1 i) 350x mag ii) 500x mag iii) 1000x mag. G sup2 i) 80x mag, ii) 350x mag, iii) 500x mag. G sup3 i) 120x mag, ii) 200x mag, iii) 650x mag. G sup4 i) 150x mag, ii) 120x mag, iii) 250x mag. G sup5 i) 120x mag, ii) 350x mag, iii) 500x mag. G sup6 i) 100x mag, ii) 200x mag, iii) 1000x mag. Figure S2.3. Particle size distribution of Gangotri medial moraine samples (see Figure 4). Figure S2.4. Mean weight percentages per ½ Φ interval of medial moraine samples from Gangotri glacier and comparisons derived from Owen et al. (2003). These comparisons include supraglacial debris from Rakhiot, Chungphur and Glacier de Cheilon  in addition to Glacier de Tsidijore Nourve (Small 1983), Breidamerkurjokkull, Sore Buchananisen, and the Glacier d'Argentiere (Boulton 1978).