Cosmogenic 10Be and equilibrium-line altitude dataset of Holocene glacier advances in the Himalayan-Tibetan orogen

A comprehensive analysis of the variable temporal and spatial responses of tropical-subtropical high-altitude glaciers to climate change is critical for successful model predictions and environmental risk assessment in the Himalayan-Tibetan orogen. High-frequency Holocene glacier chronostratigraphies are therefore reconstructed in 79 glaciated valleys across the orogen using 519 published and 16 new terrestrial cosmogenic 10Be exposure age dataset. Published 10Be ages are compiled only for moraine boulders (excluding bedrock ages). These ages are recalculated using the latest ICE-D production rate calibration database and the scaling scheme models. Outliers for the individual moraine are detected using the Chauvenet's criterion. In addition, past equilibrium-line altitudes (ELAs) are determined using the area-altitude (AA), area accumulation ratio (AAR), and toe-headwall accumulation ratio (THAR) methods for each glacier advance. The modern maximum elevations of lateral moraines (MELM) are also used to estimate modern ELAs and as an independent check on mean ELAs derived using the above three methods. These data may serve as an essential archive for future studies focusing on the cryospheric and environmental changes in the Himalayan-Tibetan orogen. A more comprehensive analysis of the published and new 10Be ages and ELA results and a list of references are presented in Saha et al. (2019, High-frequency Holocene glacier fluctuations in the Himalayan-Tibetan orogen. Quaternary Science Reviews, 220, 372–400).

modern ELAs and as an independent check on mean ELAs derived using the above three methods. These data may serve as an essential archive for future studies focusing on the cryospheric and environmental changes in the Himalayan-Tibetan orogen. A more comprehensive analysis of the published and new 10 10 Be ages from moraine boulders are used after detecting and removing outliers to maintain consistency. No bedrock sample data are used in this study. Limited statistical analyses were performed for moraines that have 2 10 Be ages before/after removing outliers. Former ELAs and change in ELAs (DELAs) are measured only for those glaciated valleys where the modern glacier-ice is present. Seventy-seven of the total 79 glaciated valleys fulfill the criteria.

Description of data collection
Approximately 500g of rock to a depth of 3.5cm from the top of each moraine boulder was collected using a hammer and chisel for 10 Be dating. Sample preparation was performed at the Quaternary Geochronology Laboratories in the University of Cincinnati. AMS measurements were performed at the Purdue Rare Isotope Measurement Laboratory at Purdue University. Raw data for ELA estimates were extracted using satellite images acquired in 26th February 2016 at https://search.earthdata.nasa.gov/search. Present and past glaciated areas were mapped (as vector layers) using Google Earth, and Landsat ETM þ images and the raw elevation data were extracted from the ASTER GDEMs (as raster layers). Data source location New 10 Be samples were collected from Sonapani glacier in the Kulti valley (32.44 N, 77.33 E) and the Parkachik valley in Nun Kun massif (34.05 N, 76.00 E). Published 10 Be ages and all ELA dataset covers a vast area including the Himalaya, Tibet, Pamir, and Tian Shan with a latitudinal range of 27.04e43.12 N and longitudinal range of 71.62e102.74 E. Note that the sample coordinates are also provided in the tables. Data accessibility Data is provided in the paper. Related research article Saha, S., Owen, LA., Orr, E.N., Caffee, M.W. (2019). High-frequency Holocene glacier fluctuations in the Himalayan-Tibetan orogen. Quaternary Science Reviews, 220, 372e400.

Value of the data
These data contain an exhaustive list of 10 Be ages and reconstructed past ELAs and DELAs of moraines across the Himalaya, Tibet, Pamir, and Tian Shan for the past 15,000 years. These data offer valuable information to anyone interested in the paleoclimatic changes in the region, especially in the past cryospheric (freshwater resources) responses to climate variability. The 10 Be data can be used/reproduced directly to recalculate exposure ages with future modifications in the dating techniques.
The ELA data are comprehensive and can be directly incorporated into numerical models that use terrestrial glaciers as a proxy for climate change. The 10 Be ages and ELAs may have the potentiality to model paleotemperatures in this high-altitude mountainous region.

Data
Table S1 contains all the new and published 10 Be apparent moraine ages for the past 15 ka in the 79 glaciated regions of the Himalayan-Tibetan orogen (see also Supplementary item 1). We identified 128 outliers (in blue in Supplementary item 1) from a total age population of 535 in this study. Note that the ages are organized from oldest to youngest local glacial stages for each climatic zone. For an extended discussion on climatic zones, the readers are encouraged to see the article "High-frequency Holocene glacier fluctuations in the Himalayan-Tibetan orogen" in the Quaternary Science Reviews [1]. A comprehensive list of references is also provided in the article. Table 1 contains the ELAs and DELAs for the 77 glaciated regions (see also Supplementary item 2).

New 10 Be ages
We sampled multiple (2) boulders from each moraine using a chisel and hammer after carefully considering the moraine morphostratigraphy, physical setting, and surficial characteristics of moraine boulders [2]. Moraines were first mapped and grouped from oldest to youngest based on their relative position from each other (i.e., morphostratigraphy). Since Holocene moraines show similar surficial characteristics, relative dating based on the degree of weathering, vegetation cover, and soil development was not possible. We recorded the stability, degradation, and post-depositional hillslope contribution on each moraine in the field before sampling. We only sampled well-inset stable boulders with no evidence of post-depositional surface deflections, detrital cover, surface spallation, pitting, fracturing, and/or extensive weathering. Preferences were also given to boulders with well-developed lichen cover with the idea that boulders have not recently been exhumed and/or toppled allowing the steady growth of lichens. The sampled boulders have heights ranging from 0.3 to 1.3 m (Table S1). Approximately 500 g of rock was collected from the top of each boulder to a depth was 3.5 cm. Topographic shielding from the boulder surface to the horizon was measured using a compass and an inclinometer at 10 azimuth interval [2,3]. No correction for snow shielding was performed, assuming a windswept condition throughout the year [3].
Quartz extraction and 10 Be sample preparation were executed at the Quaternary Geochronology Laboratories of the University of Cincinnati [4,5]. Our sample preparation includes crushing and sieving the boulder samples to obtain a 250e500 mm particle size fraction. Subsequently, samples were leached for about 10 hours in aqua regia to remove any organics and dried for 24 hours. The dry samples were then etched in 1% HF for approximately 1 h. Since quartz is hydrophilic (sticky) in nature, the froth flotation technique was applied to remove excess muscovite and feldspar (which are hydrophobic) in the sample. Samples were then treated two to three times with 5% and 1% HF/HNO 3 . Any remaining feldspar, mica, and other heavy minerals were removed by using lithium heteropolytungstate heavy liquid separation (density 2.7 g/cm 3 ) and a Frantz magnetic separator. About 25e15 g of extracted pure quartz was dissolved in 49% concentrated HF acid after adding low background 9 Be carrier (0.495 mg/g for Kulti and 1.0459 mg/g for Parkachik) and fumed with perchloric (HClO 4 ) acid to remove fluorine atoms. In addition, to remove Fe and Ti and separate the 10 Be fraction, samples were passed through the anion and cation exchange columns using (6e1 N) HCl acid. Beryllium hydroxide (Be(OH) 2 ) gel was extracted from the 10 Be fraction by adding Ammonium hydroxide. The Be(OH) 2 was heated in an oven at 900 C for 30 minutes to form BeO, mixed with acetone, Nb powder, and then loaded into a steel target. A minimum of two blanks were prepared to assess the 9 Be carrier and laboratory background level of 10 Be for each set of samples. We measured the ratios of 10 Be/ 9 Be using the accelerator mass spectrometry (AMS) at the Purdue Rare Isotope Measurement (PRIME) Laboratory at Purdue University. 07KNSTD (standard) is used to normalize our Be isotopic. The 10 Be/ 9 Be ratios were subsequently converted into 10 Be concentrations, i.e., in atoms [2] and exposure ages (Table S1) were estimated using the available online age calculators (https://crep.otelo.univ-lorraine.fr/#/; http://hess.ess.washington.edu/; http://cronus.cosmogenicnuclides.rocks/2.0/html/al-be/).  No corrections for residual boron, radioactive decay, and muongenic production [6] were made; they are negligible for the timescale of this study. Native 9 Be in nearby (uniform) lithology is also insignificant (~0.0190 ± 0.0160 to 0.0015 ± 0.0001 ppm in Ref. [1]) to account for any adjustments in our calculated exposure ages.

Published 10 Be ages
For consistency, we followed a strict procedure while compiling the published 10 Be ages. This includes only using moraine boulder ages, excluding any bedrock ages from the analysis (Table S1; Supplementary item 1). 10 Be ages that do not follow the moraine morphostratigraphic order as outlined in the original literature were excluded. Slip rate studies on moraines that only dated pebbles/ cobbles were also not used in this compilation. Only studies that used the standard [4,5] 10 Be extraction procedure are targeted. Since published studies used different standards (e.g., LLNL3000, S555, NIST_Certified, NIST_27900, KNSTD, 07KNSTD) to normalize their Be isotopic measurements (Table S1), a correction factor is used whenever required while recalculating the ages [7]. We used 5 cm as the maximum depth of sample collection and zero erosion rates for studies that did not report any such information. Using the raw data provided in the original literature, we therefore recalculated all the published 10 Be ages following the same parameters (Table S1; Supplementary item 1).
We performed several statistical treatments if > 2 concordant boulder ages are available for a moraine. We applied reduced chi-squared (c 2 ) statistics to assess the distribution of ages. Any age population with c 2 > 1 likely had outliers, and further statistical treatment was performed. Chauvenet's criterion [14] was used to detect outliers and highlighted in blue in Table S1. Outliers for new 10 Be ages were only removed if convincing field evidence supported our statistical results (e.g., possible recent hillslope deposits, shallow burial, and/or toppling). For published studies, we relied on statistical treatment and the recommendations in the original studies to detect and remove outliers. Mean moraine ages (local glacial stages) are reported using arithmetic means ± 1s, weighted mean ± 1s, and peaks in the probability distribution (Table S1).

Equilibrium-line altitudes (ELAs)
Present and past ELAs were determined using area-altitude (AA), area accumulation ratio (AAR), and toe-headwall accumulation ratio (THAR) methods for each glacier advance in 77 glaciated valleys (Table 1; Supplementary item 2) [15]. Additionally, the modern maximum elevations of lateral moraines (MELM) were used to evaluate the modern estimated ELAs derived using the above three methods (Table 1). ELAs and DELAs are only measured for those glaciated valleys where the modern glacier-ice is present.
Raw data for ELA estimates were extracted using satellite images acquired in 26th February 2016 at https://search.earthdata.nasa.gov/search. Present and past glaciated areas were mapped as vector layers using Google Earth and Landsat ETMþ images in ArcGIS 10.5 (Supplementary item 2). In addition, we used ASTER GDEMs to prepare Hillshade and Slope maps (Spatial Analyst Tools in ArcGIS) to further aid in outlining modern glaciated areas and paleo-ice extents. Paleo-ice extents were defined on the satellite images using moraine positions in the individual valley (Supplementary item 2). The vector layers/maps of the modern and the past glacier extents are then used to extract the DEM values and converted into ASCII flies. The ASCII files were inserted into the Read ArcGrid program developed by Professor David Nash of the University of Cincinnati to generate the glacier's hypsometry. The Read ArcGrid program calculates basic statistics, including Elevation Relief Ratio (hypsometric integral), for a matrix of elevations. Using the steps outlined in Ref. [15] and a combination of AA, AAR, and THAR ratios, we finally measured the ELAs. Different combinations of AARs (e.g., ranging from 0.45 to 0.80) and THARs (e.g., varies from 0.3 to 0.6) were used depending on the glacier setting, physical characteristics, and climate (Table 1). We obtained these ratios from the published literature for each distinct climatic zone (see Ref. [1] for details on climatic zones and references therein). In Table 1, we report the (arithmetic) mean ELA and DELA with ± 1s uncertainty.