Intense Chemical Weathering at Glacial Meltwater-Dominated Hailuogou Basin in the Southeastern Tibetan Plateau

Abstract

Climate warming has caused rapid shrinkage of glaciers in the Tibetan Plateau (TP), but the impact of glacier retreat on the chemical denudation rate remains largely unknown at the temperate glacial basins. The chemical weathering processes were examined at a temperate glacial basin (HLG) in the southeastern TP based on comprehensive data from the supraglacial meltwater, proglacial river water, precipitation and groundwater over two glacier melt seasons in 2008 and 2013. The concentrations of major ions and suspended sediments in river water exhibit a pronounced seasonality and display a close relationship with river discharge, suggesting a strong hydrological control on the chemical and physical weathering processes. Runoff chemistry is dominated by carbonate weathering and sulfide oxidation. HCO₃⁻, Ca²⁺, and/or SO₄²⁻ are the dominant ions in meltwater, river water, precipitation and groundwater. For river water, HCO₃⁻ and Ca²⁺ primarily come from calcite weathering, and SO₄²⁻ is mainly derived from pyrite oxidation. Both solute and sediment fluxes are positively related to river discharge (r = 0.69, p < 0.01 for sediments). The solute flux and yields are 18,095–19,435 t·year⁻¹ and 225–241 t·km⁻²·year⁻¹, and the sediment load and yields are 126,390 t·year⁻¹ and 1570 t·km⁻²·year⁻¹, respectively. The solute yields, cationic denudation rate (CDR; 2850–3108 Σ*meq⁺ m⁻²·year⁻¹) and chemical weathering intensity (CWI; 616–711 Σ*meq⁺ m⁻³·year⁻¹) at HLG are higher than those at most basins irrespective of the lithology, suggesting more intense weathering in the TP in comparison to other glacial basins worldwide.

1. Introduction

Climate change has resulted in rapid shrinkage of glaciers in the Tibetan Plateau (TP) [1,2,3,4,5,6,7,8,9]. Such changes in ice reserves are expected to alter the magnitude and timing of meltwater discharge and the species transport therein [10,11,12,13,14]. Meltwater runoff, affected substantially by climate warming, provides the headwaters for some large rivers (e.g., Indus, Brahmaputra, and Yangtze Rivers, which are most susceptible to the reductions of meltwater) in Asia, which potentially affect the seasonal water availability and food security of over 60 million people downstream [3]. Runoff chemistry enables us to better understand the influence of melting glaciers on the chemical denudation, biogeochemical cycling and aqueous ecosystems downstream [15,16,17]. As a result, some major and minor and trace solutes (e.g., Ca²⁺, HCO₃⁻, SO₄²⁻, Fe, Si, P, Pb, and Hg) in meltwater runoff are of particular interest due to their importance in the assessment of chemical weathering process, solute provenances, water quality and aquatic environments at glacial basins (e.g., [18,19,20,21]).

Meltwater runoff chemistry has been studied at many glacial basins in Asia. In the Muztag Ata Mountains, meltwater discharge was slightly alkaline and mainly affected by air temperature and precipitation, and the concentrations of major ions had an inverse relationship with meltwater discharge and were dominated by the water-rock interaction at Kartamak Glacier (KAG) [22]. In the Tianshan Mountains, the concentrations of major ions were influenced by many factors such as precipitation, glaciation, snow melt, frozen soil and water-rock interaction at Urumqi Glacier No.1 (UG1) [23,24]. At Koxkar Glacier (KOG), pH values in different samples followed the order: proglacial river water > supraglacial meltwater > precipitation [25]. In the Qilian Mountains, major ions, Li, Sr, and Ba, were exported predominately as the mobile monovalent or divalent ions at Qiyi Glacier (QG) [26]. In the Tanggula Mountains, the dominant ions and elements were HCO₃⁻, Ca²⁺, Fe and Sr at Dongkemadi Glacier (DG) [19]. In this basin, the electrical conductivity (EC) displayed a strong seasonality, with higher values during the early and late melt seasons in comparison to the peak flow season [27]. This can be attributed to the evolution of glacial drainage system (which is closely related to the potential for solute acquisition in glacial environments) over the glacier melt season [15,28,29]. The distribution of metal elements was featured by the mixture of the soluble metal and non-metal ligand complexes [19]. In the Himalaya Mountains, major ions were enriched after meltwater passes through glaciers at Chhota-Shigri Glacier (CSG) [30]. HCO₃⁻ and Ca²⁺ were the dominant ions, and major ion concentrations followed the orders of Ca²⁺ > Mg²⁺ > Na⁺ > K⁺ for cations and HCO₃⁻ > SO₄²⁻ > Cl⁻ > NO₃⁻ for anions, and runoff chemistry was related to the natural (mainly carbonate weathering) and anthropogenic inputs at Bilare Banga Glacier [31]. Additionally, the concentrations of major ions and some minor elements (e.g., Li, Sr, and Ba) exhibited an inverse relationship with meltwater discharge at KAG, QG and DG [19,22,26,32,33], but some elements (e.g., Fe, Al, and Cr) were not correlated with discharge at QG and DG [19,26] where major ion fluxes are positively related to discharge [32,33]. In general, meltwater runoff chemistry was dominated by carbonate weathering, sulphide oxidation, and/or sulfate and evaporates dissolution at QG, DG, Dokriani Glacier and Dudu Glacier [19,26,32,34,35], and the flushing of rainfall may have enhanced the moraine weathering at Dokriani Glacier [35]. In addition, the sediment concentration was primarily related to discharge and temperature, and the sediment yield was 411 t·km⁻²·year⁻¹ at Shaune Garang basin [36]. The Karakoram and western Himalayan glaciers produced lower sediment yields than the central Himalayan glaciers, and the hydro meteorological conditions and sediment delivery for Chorabari Glacier provided insight on the meltwater generation processes and sediment transport patterns during the Indian monsoon seasons [37]. Glacier retreat may lead to the occurrence of debris flows, which is closely related to increasing meltwater, degraded moraine, loose sediments and continuous rainfall, and should be investigated intensively in the Himalaya regions [38].

However, few studies have been conducted on the meltwater hydrochemistry and the sediment delivery at the temperate glacial basins in the TP, where more solutes and sediments are expected to be largely released into downstream aquatic ecosystems due to the enhanced physical and chemical weathering in the subglacial and proglacial regions as a result of increasing meltwater discharge [9,10,11,12,13,14,16,29,30,31,32,33,34]. This study aims to elucidate the hydrochemical characteristics of meltwater runoff, and then determine the seasonal controls on the runoff chemistry, solute and sediment export, and chemical weathering at the Hailuogou basin in the southeastern TP over two melt seasons. This will help to better evaluate the influence of glacial retreat on the physical and chemical denudation rates, and downstream hydrochemistry in a warming environment.

2. Materials and Methods

2.1. Study Area

The Hailuogou basin is located on the eastern slope of the Gongga Mountains in the southeastern TP (Figure 1). Hailuogou Glacier (HLG; 29°34′ N, 101°59′ E) is composed of seven glaciers, which cover 45.3% of basin area (80.5 km²) [39]. About 8.2% of the glacier area is covered by supraglacial debris [40], of which the thickness increases from several millimeters of patchy cover in the upper part of the ablation zone to over one meter at the glacier terminus [41]. The climate is dominated by the southwest monsoon in summer and the westerly jet in winter [39,42]. At the Gongga Alpine Ecosystem Observation and Research Station (GAEORS), 0.5 km east of HLG, the annual mean temperature was 4.6 °C and mean annual precipitation was 1881 mm during the period 2000 to 2008. The glacial drainage system forms by the incision of supraglacial streams, and the perennial subglacial channels are well developed [42]. As long as the conduits remain connected to the drainage network, they continue a three-dimensional meandering towards glacier bed. When the meltwater supply to an englacial conduit terminates, the ice deformation closes the conduit and any sediment deposits at the conduit floor form a sorted debris band [42]. At this basin, the bedrock is mainly composed of deep-metamorphic schist accompanied with quartzite, marble, slate, and crystalline limestone of the Permian [43]. The mineralogy of bedrock includes illite, quartz, potassium feldspar, hornblende, plagioclase, clay, pyrite, augite, calcite, kaolinite and chlorite (

Table 1. The mineral composition (%) of glacial deposits at the Hailuogou basin.

Sample	Quartz	Potash Feldspar	Plagioclase	Calcite	Pyrite	Hornblende	Augite	Clay
HLG1	27.5	19.4	24.8	N/D	2.5	16.8	3.8	5.2
HLG2	32.6	6.1	15.4	N/D	9.0	25.8	6.6	4.5
HLG3	30.4	9.2	12.2	1.8	5.0	22.9	6.7	11.8
HLG4	30.8	12.9	16.3	1.4	6.0	15.7	6.4	10.5
HLG5	31.8	8.9	14.2	1.4	8.3	16.2	6.8	12.4
Average	30.6	11.3	16.6	1.5	6.2	19.5	6.1	8.9

N/D indicates not been detected in the laboratory.

; Figure 1) [44].

2.2. Sample Collection

In 2013, the supraglacial meltwater (n = 26) was collected from 8 April to 24 November, groundwater (n = 15) was sampled from a spring lateral in the proglacial areas from 8 April to 31 July, and precipitation (n = 142) was collected at the GAEORS from 8 February to 24 November (Figure 1). The proglacial river water was sampled at a site ~1.1 km from the glacier terminus from 1 May to 30 September in 2008 (n = 144) and from 1 April to 30 November in 2013 (n = 379) (Figure 1). Field sampling was performed once daily at high flow in 2008, and twice daily at approximately minimum (09:00 h) and maximum (18:00 h) flows in 2013. River samples were immediately filtered through 0.45 μm pore size Whatman cellulose nitrate membranes using a polysulphone filter and a manual pump. After drying, the sediments on the filters in 2013 was weighed in order to calculate the suspended sediment concentrations (SSC). All samples were kept in a refrigerator, and then were transported to the State Key Laboratory of Cryospheric Sciences (SKLCS) at the Chinese Academy of Sciences (CAS) and kept in a cold room at 4 °C until laboratory analysis. The glacial deposits were collected at different sites in the proglacial area (Figure 1) and then transported to the Geochemistry Department of Test Center of Lanzhou Branch for the mineral composition analysis. Data of air temperature and precipitation were continuously observed at the GAEORS (Figure 1).

2.3. Laboratory Analysis

pH and EC/total dissolved solids (TDS) were determined by Eutech PH310 and CON410 portable meters respectively. δ¹⁸O and δ²H were measured by a LGR-DLT-100 mass spectrometer with a precision of ±0.06%. The cations (Na⁺, K⁺, Ca²⁺, and Mg²⁺) were determined by a Dionex-600 ion chromatograph using an IonPac CS-12A-HC column, 20 mM MSA eluent and CSRS-ULTRA-II suppressor, and anions (Cl⁻, NO₃⁻, and SO₄²⁻) were determined by a Dionex-2500 ion chromatograph using an IonPac AS11-HC column, 25 mM KOH eluent and ASRSULTRA-II suppressor [5,19,26]. The detection limits were ~1 μg·L⁻¹ and the precision was ±1% based on repeated measurement of a certified reference solution. HCO₃⁻ was determined by ion charge balance. The mineral composition was determined by an X-ray diffraction (XRD) in the Lanzhou Center for Oil and Gas Resources at the CAS [19,26]. The identified species and relative component contents were assessed using X’Pert High Score Plus software, and trace minerals were excluded from these analyses because semiquantitative XRD estimates involve an uncertainty of 5%. Gibbs plots and the stoichiometry and chemical reactions (e.g., [45,46,47]) were used to identify the weathering process. Principal component analysis (PCA) was performed on meltwater runoff chemistry using CANOCO 4.5 [48] to determine whether individual dissolved species were significantly controlled by meltwater discharge and to determine what species are likely derived from the same process or source.

2.4. Water and Solute Flux Estimates

River discharge was estimated by a glacio-hydrological model developed by Zhang and others [39]. In short, discharge from glacier area was estimated using a surface energy-/mass-balance model, which is identical to that developed by Zhang and others [40]. The model consists of two components: a surface model that computes the energy available for melting from the exchange of energy between the debris-covered/debris-free surfaces and the atmosphere, and a subsurface model that treats processes occurring in the subsurface after meltwater percolates into underlying layers [39]. The temperature and precipitation time series for each elevation band were interpolated according to the mean elevation using the temperature lapse rate and precipitation gradient [39,40]. Glaciers were separated into the debris-covered and debris-free surfaces based completely on the spatial distribution of the thermal resistance of the debris layer derived from remotely sensed data [40]. Discharge from glacier-free area was estimated by the US Soil Conservation System (SCS) curve number method [49] and the mass balance equation of snowpack [39]. The model was forced by local weather observations since 1988. The forcing data required by the model included the daily air temperature, precipitation, wind speed, relative humidity and incoming solar radiation, which is monitored by the GAEORS. Modeled discharge was calibrated and validated using observed discharge at a gauging section (Figure 1), ~1.5 km downstream to the glacier terminus.

A full description of the model components, associated optimal model parameters, and the evaluation of the model performance was provided by Zhang and others [39]. As shown in

Table 2. The model calibration and validation results for the Hailuogou basin [39].

Period	NSE	PBIAS (%)
Calibration (1994–2000)	0.86	5.0
Validation (2001–2007)	0.83	−7.0
Overall (1994–2007)	0.85	−4.5

, the results of model calibration demonstrate a close agreement between simulated and observed monthly discharge, yielding the Nash–Sutcliffe efficiency (NSE) value of 0.86 and the per cent bias (PBIAS) value of 5%. For the validation period, although the model yields a slight lower NSE value compared to the calibration period, the model generally reproduces the observed discharge well and captures the long-term trend. Overall, this model successfully replicates all key glacio-hydrological processes, with a rating of ‘very good’ on the scale of Moriasi and others [50].

Daily fluxes of solutes and sediments were calculated by daily solute/sediment concentrations multiplied by discharge. During the absent sampling days, daily solute/sediment concentrations were estimated by the best-fit regressions between the measured solute/sediment concentrations and modeled river discharge (for solutes: 0.22 < R² < 0.74, p < 0.01; for sediments: R² = 0.57, p < 0.01).

3. Results

River discharge exhibited an obvious seasonality, which showed higher variability during the high flow season (June to September) in comparison to the low flow seasons (April to May, and October to November) (Figure 2a). The temperature and precipitation displayed a similar seasonality to river discharge. Stream δ²H and δ¹⁸O values were highest during April to May, decreased to the lowest values during early August, and increased again to higher values during November (Figure 2b). pH demonstrated higher values during April to July, decreased to lower values during August or September, and increased again to higher values during November (Figure 2c). EC displayed higher values during April to May and September to November and lower values during June to August, in which its temporal variation is relatively stable (Figure 2d).

HCO₃⁻ was the dominant anion, followed by SO₄²⁻, NO₃⁻ or Cl⁻ in precipitation, groundwater and river water, and followed by Cl⁻, SO₄²⁻ and NO₃⁻ in supraglacial meltwater (

Table 3. Median (Med), maximum (Max) and minimum (Min) concentrations or values of pH, electrical conductivity (EC); μS·cm⁻¹), total dissolved solids (TDS; mg·L⁻¹), δ²H and δ¹⁸O (‰), major ions (μM), and/or suspended sediment concentrations (SSC; mg·L⁻¹) in precipitation (P), supraglacial meltwater (M), groundwater (G) and river water (R) at the Hailuogou basin (HLG) over two melt seasons in 2008 and 2013, in comparison to Urumqi Glacier No.1 (UG1) in the Tianshan Mountains, Gangotri Glacier (GG) and Chhota Shigri Glacier (CSG) in the Himalaya Mountains, and Haut Glacier d’Arolla (HGA) in Switzerland. N/A denotes data are not available.

	HLG											UG1 [51]	GG [52]	CSG [53]	HGA [54]
	P	M	G	R				R				R	R	R	R
	2013	2013	2013	Seasonal (2008)				Seasonal (2013)				Seasonal	Seasonal	Seasonal	Seasonal
	Med	Med	Med	Min	Max	Med	St dev	Min	Max	Med	St dev	Average	Average	Average	Med
N	142	26	15	144	144	144	144	379	379	379	379	217	N/A	N/A	132
SSC	N/A	N/A	N/A	N/A	N/A	N/A	N/A	36.0	3608	193	346	N/A	N/A	N/A	N/A
pH	5.90	7.72	7.89	6.77	9.26	8.47	0.68	6.81	9.26	8.15	0.31	7.61–7.74	6.64	6.6	8.1
EC	8.58	11.70	153.9	30.30	90.60	59.10	14.00	45.30	178.6	95.40	30.33	75.4–93.6	78.7	36.8	N/A
TDS	4.33	5.94	76.60	15.15	45.30	29.55	7.00	23.00	90.30	48.30	15.28	48.3–59.9	51.5	N/A	N/A
δ2H	−73.61	−119.3	−93.66	N/A	N/A	N/A	N/A	−130.8	−93.96	−112.6	8.74	N/A	N/A	N/A	N/A
δ18O	−11.42	−17.21	−13.78	−18.22	−13.57	−15.93	1.37	−18.79	−13.69	−15.91	1.13	N/A	N/A	N/A	N/A
Na+	0.582	6.109	95.96	5.643	47.91	16.05	8.304	7.174	88.39	27.40	18.18	26.96	48.43	37.91	16.09
K+	0.589	8.336	228.7	31.59	140.8	61.38	22.61	24.46	173.7	77.44	32.31	23.33	54.33	30.79	9.487
Mg2+	0.255	1.871	152.2	11.86	53.75	24.40	8.625	4.179	90.13	24.95	16.13	45.00–47.08	110.2	37.08	16.67
Ca2+	3.453	16.07	586.9	122.0	414.3	215.1	54.15	60.28	529.1	252.3	103.4	309.8–364.8	129.5	75.00	165.0
Cl−	0.451	3.253	8.714	1.113	15.77	3.422	2.177	0.580	14.70	2.766	3.008	17.78–19.17	15.33	15.89	2.444
SO42−	3.646	1.135	397.7	28.76	259.2	85.04	36.35	19.38	296.9	113.8	64.73	177.7–199.8	227.5	56.50	54.17
NO3−	2.845	0.868	9.523	0.223	9.079	2.127	1.871	0.126	4.627	1.685	1.043	15.81–18.06	5.919	3.500	6.452
HCO3−	16.03	48.21	999.7	148.9	689.6	381.7	97.46	111.8	1021	448.0	157.6	370.3–439.3	217.0	216.0	278.7

). The dominant cation was Ca²⁺, followed by K⁺, Mg²⁺ or Na⁺ in meltwater, precipitation, groundwater and river water (Table 3). The concentrations or values of EC, HCO₃⁻ and cations followed the sequence: groundwater (G) > river water (R) > meltwater (M) > precipitation (P). The sequences of pH, δ²H, Cl⁻, SO₄²⁻ and NO₃⁻ were R > G > M > P, P > G > R > M, G > M > R > P, G > R > P > M, and G > P > R > M, respectively (Table 3). The concentrations of major ions and suspended sediments (SS) in river water exhibited a pronounced seasonality (Figure 3). Major ions (except for NO₃⁻) displayed an inverse relationship with river discharge exhibiting higher concentrations during the low flow seasons (R² > 0.45, p < 0.01), and SS had a direct relationship with river discharge exhibiting the highest concentration during the high flow season (R² = 0.33, p < 0.01) in 2008 and 2013 (Figure 4).

Daily fluxes of most ions and SS also exhibited a strong seasonality, with higher values during high flow season and lower values during low flow seasons (Figure 5). Daily fluxes were positively related to river discharge (r > 0.48 and p < 0.01 for ions, and r = 0.69 and p < 0.01 for sediments), and the cumulative fluxes and discharge increased sharply during the high flow season (not shown). Total discharge was 3.52 ± 0.18 × 10⁸ m³ in 2008, which transported an estimated 19,435 ± 972 t solutes, with 50.2% from HCO₃⁻, 21.0% from SO₄²⁻, 19.1% from Ca²⁺, 8.8% from Na⁺, K⁺ and Mg²⁺, and 1.0% from Cl⁻ and NO₃⁻ (

Table 4. The flux and yields (with an estimated 5% error due to the bias between modeled and observed discharge) of major ions and suspended sediments (SS) in river water draining Hailuogou Glacier over two melt seasons in 2008 and 2013. N/A denotes data are not available.

Species	2008				2013
	Flux	Error	Yields	Error	Flux	Error	Yields	Error
	t·year−1		t·km−2·year−1		t·year−1		t·km−2·year−1
SS	N/A		N/A		126,390	6320	1570	78.5
Na+	251	12.6	3.12	0.16	225	11.3	2.8	0.14
K+	1174	58.7	14.6	0.73	1072	53.6	13.3	0.67
Mg2+	283	14.2	3.51	0.18	216	10.8	2.68	0.13
Ca2+	3712	186	46.1	2.31	3482	174	43.3	2.17
Cl−	88.1	4.41	1.09	0.05	40.2	2.01	0.5	0.03
SO42−	4073	204	50.6	2.53	3523	176	43.8	2.19
NO3−	102	5.10	1.27	0.06	20.6	1.03	0.26	0.01
HCO3−	9752	488	121	6.05	9516	476	118	5.90
Dissolved solutes	19,435	972	241	12.1	18,095	905	225	11.2

). In comparison, total discharge (3.73 ± 0.19 × 10⁸ m³) in 2013 transported 18,095 ± 905 t solutes, with 52.6% from HCO₃⁻, 19.5% from SO₄²⁻, 19.2% from Ca²⁺, 8.4% from Na⁺, K⁺ and Mg²⁺, and 0.3% from Cl⁻ and NO₃⁻ (Table 4). This equated to the melt season area weighted solute yields (the solute flux divided by the drainage contributing area) of 241 ± 12.1 t∙km⁻² in 2008 and 225 ± 11.2 t∙km⁻² in 2013. In addition, total discharge in 2013 transported 126,390 ± 6320 t suspended sediments, with area weighted sediment yields of 1570 ± 78.5 t∙km⁻² (Table 4).

4. Discussion

4.1. Controls on Runoff Chemistry

River discharge is primarily dependent on the meteorological factors, with higher discharge occurring as a result of increases in air temperature and/or precipitation (Figure 2a). The isotopic signature reflects the seasonal changes in the proportions of meltwater, precipitation and groundwater during the glacier melt season [19]. For example, the decrease of δ¹⁸O values in river water from May to August indicates the increasing proportion of meltwater and the decreasing proportion of precipitation, and vice versa for the period from August to November (Figure 6). River water is supplied by precipitation, groundwater and meltwater. This is supported by isotopic values in river water (−112.6‰ for δ²H and −15.9‰ for δ¹⁸O in 2013), which fall between meltwater (−119.3‰ and −17.2‰) and precipitation (−73.6‰ and −11.4‰)/groundwater (−93.7‰ and −13.8‰) (Table 3; Figure 6). Changes in EC highlight a temporal pattern of solute acquisition at glacial basins. The glacial drainage system evolves from a slow-inefficient distributed drainage system (delayed flow) with elevated EC during the early melt season (April to May) to a fast-efficient channelized system (fast flow) with lower EC during the peak flow season (June to September) and then to a delayed flow drainage system during the late melt season (October to November) with elevated EC again [28,29]. This limits or encourages the potential for solute acquisition in glacial environments.

The enrichment of cations between supraglacial meltwater and river water reflects the solute liberation by the subglacial bio-geochemical weathering processes as meltwater is routed through the subglacial environments, as documented at other glacial basins [26,35,55,56,57]. In addition, the concentrations of HCO₃⁻ and/or Ca²⁺, Mg²⁺, SO₄²⁻ at HLG, as well as at Gangotri Glacier (GG) and/or CSG in the Himalaya Mountains and at Haut Glacier d’Arolla (HGA) in Switzerland, were lower than those at UG1 in the Tianshan Mountains (Figure 1; Table 3). For example, the seasonal mean concentrations of Ca²⁺, Mg²⁺ and SO₄²⁻ from UG1 [51] are 1.2 to 4.9 times higher than those from HLG, CSG [53] and HGA [54]. This may be related to the intense dilution process at glacial basins which are controlled by the oceanic climate (more precipitation and higher discharge), in comparison to glacial basins which are controlled by the continental climate (less precipitation and lower discharge). The seasonal patterns of major ion concentrations at HLG (Figure 3) are similar to those at GG [52], CSG [53], HGA [54], and Longyearbreen Glacier in Svalbard [58]. However, the inter-annual variations of ionic concentrations slightly differed at HLG. For example, the discharge-weighted mean Ca²⁺ and SO₄²⁻ concentrations during 2013 were 1.1 times higher than those during 2008. This may be related to the longer duration of field sampling during the low flows (with enriched solutes) in 2013 in comparison to the short duration in 2008 (Figure 3 and Figure 4).

The relationships between solute concentrations and discharge (Figure 4) indicate a combined influence of solute acquisition and meltwater dilution. The inverse correlations suggest a strong hydrological control on meltwater runoff chemistry. The fast flow transports meltwater rapidly through a channelized drainage system in the ice-walled conduits, while the delayed flow transports meltwater slowly through a distributed drainage system in the ice-bedrock interface [15,28]. Thus, the fast flow limits, and the delayed flow encourages, the potential for solute acquisition in subglacial environments (e.g., [15,28,29]). During low flow seasons at HLG, meltwater is routed through a distributed drainage system, and the longer residence time promotes the protracted and intimate contact of meltwater with the supra-, sub- and pro-glacial sediments, debris and deposits. This leads to a higher concentration of major ions in river water. During high flow season at HLG, meltwater is routed through a channelized drainage system. When meltwater enters into the subglacial environments, the dilution of delayed flow and the duration of water-sediments/debris interaction declines. This results in lower concentrations of major ions in river water at HLG.

4.2. Chemical Weathering

The meltwater runoff chemistry is primarily related to bedrock mineralogy, and the evolution of glacial drainage system (e.g., [15,54,59]). The Gibbs plots showed that the river water and groundwater are dominated by rock weathering at HLG (Figure 7). This is consistent with previous studies for river water at UG1 [51,60] and KOG [25] in the Tianshan Mountains, at DG in the Tanggula Mountains [19], and at QG in the Qilian Mountains [26] (Figure 1). However, the clusters of precipitation and supraglacial meltwater lay between the rock weathering and precipitation end-members (Figure 7). This is reasonable because the supraglacial meltwater chemistry is dominated by the atmospheric deposition and water-debris interaction, and the precipitation chemistry may be related to surrounding crustal materials carried by wind and human activity due to busy glacier tourism at HLG [61]. This further confirmed the availability of Gibbs plots in assessing the controlling mechanisms of runoff chemistry at glacial basins. In addition, the Na-normalized ratios of Ca²⁺ and Mg²⁺ are often used to evaluate the relative importance of the carbonate, silicate and evaporate weathering (e.g., [19,62,63]). The clusters of river water and groundwater lay between carbonate and silicate end members but are close to the carbonate end member (Figure 8). This suggests that runoff chemistry may be related to a mixture of carbonate and silicate weathering products. However, the clusters of precipitation and meltwater support an influence of crustal materials, and also display an influence of precipitation and meltwater on runoff chemistry at HLG (Figure 8).

Previous studies have shown that meltwater chemical composition is typically controlled by carbonate weathering at glacial basins (e.g., [15,47,64]). Even if the bedrock is dominated by silicate minerals, trace carbonates are still preferentially weathered [47], resulting in enhanced concentration of (Ca²⁺ + Mg²⁺) versus (Na⁺ + K⁺) [65]. This is supported by high molar contribution of (Ca²⁺ + Mg²⁺) to total cations during 2008 (75.6%) and 2013 (72.6%) at HLG. The ratio of (Ca²⁺ + Mg²⁺) to (Na⁺ + K⁺) has traditionally been used to indicate the amount of silicate weathering versus carbonate weathering [65,66,67]. Elevated concentrations of (Na⁺ + K⁺) versus (Ca²⁺ + Mg²⁺) are associated with enhanced silicate weathering [68]. The molar (Ca²⁺ + Mg²⁺)/(Na⁺ + K⁺) ratio increased with the seasonal evolution of glacier drainage system during 2008 and 2013 (Figure 9a). This suggests that (Na⁺ + K⁺) concentrations decrease in relation to (Ca²⁺ + Mg²⁺) concentrations and thus silicate weathering is a negligible source at HLG. Reynolds and Johnson [69] and Brown and others [55] showed that the C-ratio [HCO₃⁻/(HCO₃⁻ + SO₄²⁻)] signifies the weathering of carbonate or sulfide or both coupled. Namely, the 1:1 line signifies the carbonation reactions and 1:2 signifies the coupled sulfide and carbonate weathering. The ratios lay between 0.5 and 1 during 2008 (ranging 0.60 to 0.91) and 2013 (0.67 to 0.93) (Figure 9b), indicating the coupled carbonate and sulfide weathering at HLG. This is supported by higher molar loading of SO₄²⁻ (10.8% during 2008 and 12.0% during 2013) only second to HCO₃⁻ (48.4% and 47.2%) and Ca²⁺ (27.3% and 26.6%). However, molar ratios of HCO₃⁻ to SO₄²⁻ are far larger than 2 (Figure 9c), suggesting carbonate rather than sulfide dominates. Thus, the carbonate dissolution and sulfide oxidation primarily affects runoff chemistry at HLG.

At HLG, HCO₃⁻ and Ca²⁺ are primarily derived from calcite weathering (Table 1). This is confirmed by the correlations between Ca²⁺ and HCO₃⁻ (R² > 0.80, p < 0.01) and the ratios of Ca²⁺ to HCO₃⁻ (0.41 to 0.88) that are distributed around 1:2 (which denotes calcite dissolution [52,53]) during 2008 and 2013 (Figure 9d). The Ca-Mg carbonate also contributes HCO₃⁻ and Ca²⁺, which is supported by relatively excess of Ca²⁺ (the ratios above the 1:2 line) during calcite weathering (Figure 9d) and the ratios of (Ca²⁺ + Mg²⁺) to HCO₃⁻ that are close to 1:2 (which denotes Ca-Mg carbonate dissolution [53]) during part of the periods (Figure 9e). Another source of HCO₃⁻ is oxidation of organic carbon (OC; i.e., OC + O₂ + H₂O → HCO₃⁻), which characterizes the solutes derived from the decomposed biota and microbial activity [70,71]. Previous studies have shown that the glacially-derived OC primarily comes from in situ primary production in the supraglacial, subglacial and proglacial environments and from the external terrestrial (e.g., soils and plant) and anthropogenic sources (e.g., [13,14,72]). However, attempting to quantify the contribution of OC to HCO₃⁻ is hampered by other sources, such as the carbonate dissolution or atmospheric sequestration of CO₂ (e.g., CaCO₃ + CO₂ + H₂O → Ca²⁺ + HCO₃⁻ + OH⁻) [70]. Plagioclase, hornblende, augite and clay (e.g., montmorillonite, vermiculite) may also contribute Ca²⁺ to runoff chemistry (Table 1). Thus, C1 in Figure 10 reflects the geochemical reactivity of ubiquitous minerals at HLG.

SO₄²⁻ mainly results from pyrite (Table 1) because pyrite is the most common sulfide mineral at glacial basins (e.g., [52,58,73]). The sulphate minerals may also contribute SO₄²⁻, which is confirmed by the close relationships between (Ca²⁺ + Mg²⁺) and SO₄²⁻ during 2008 and 2013 (R² > 0.66, p < 0.01) (Figure 9f). The molar ratios of cations to Cl⁻ is usually applied to distinguish solutes from atmospheric source after Sharp and others [59]. Compared to the molar ratios in sea water (Na/Cl = 1, K/Cl = 0.2), higher molar ratios of Na/Cl and K/Cl from HLG during 2008 (4.8 for Na/Cl and 17.8 for K/Cl) and 2013 (9.2 and 27.4) indicate less contribution of atmospheric source. In addition, the clusters of Cl⁻, NO₃⁻ and SO₄²⁻ imply an influence of human activity (C2; Figure 10). Recent studies showed that the atmospheric environment in the TP is closely related to the combustion of fossil fuel and biomass from the Indo-Gangetic Plain [74]. Additionally, at Baishui Glacier No.1 basin in the southeastern TP, NO₃⁻ and SO₄²⁻ in snowpack are mainly derived from the wet deposition and anthropogenic input [61]. This source is supported by PCA (Figure 10), where significant correlations exist between Cl⁻, SO₄²⁻ and NO₃⁻ during 2008 (0.37 < R² < 0.39, p < 0.01) and 2013 (0.19 < R² < 0.62, p < 0.01), because NO₃⁻ has been suggested mainly from the input of acid aerosols (e.g., [58,71,75]). Thus, C2 in Figure 10 represents the contributions of the sulfide oxidation and atmosphere and/or human activity.

4.3. Solute/Sediment Export

During the high flow season, although solute concentrations are lower, solute load is higher at HLG. The temporal pattern of solute exports can largely be attributed to meltwater flux. Namely, the changes in the daily and cumulative solute fluxes are consistent with the variations in the daily and cumulative discharge respectively (not shown). This suggests that solute export is dominated by meltwater runoff at HLG. The solute yields at HLG ranges from 225 to 241 t·km⁻²·year⁻¹ during 2008 and 2013 (Table 4), which is far larger than those from most glacial basins worldwide (

Table 5. The solute yields (kg·km⁻²·year⁻¹), cationic denudation rate (CDR; Σ*meq⁺·m⁻²·year⁻¹) and chemical weathering intensity (CWI; Σ*meq⁺·m⁻³·year⁻¹) at the Hailuogou basin, in comparison to other glacial basins worldwide. N/A denotes data are not available.

Basin	Type	Solute Yields	CDR	CWI	Reference
Tuva Glacier	Antarctic	N/A	163	308	[76]
Austre Broggerbreen	Arctic	23,000–28,000	240–270	208–286	[56]
Erdmannbreen	Arctic	16,000	190	235	[56]
Erikbreen	Arctic	31,000	320	627	[56]
Hannabreen	Arctic	30,000	320	400	[56]
Finsterwalderbreen	Arctic	N/A	210–440	250–1257	[73]
Scott Turnerbreen	Arctic	16,000	160	308	[75]
Longyearbreen	Arctic	23,535	322	940	[58]
Midre Lovénbreen	Arctic	41,000–47,000	450–560	300–431	[76]
Lewis River	Arctic	N/A	94	132	[77]
Kuannersuit	Greenland	15,900	680–850	272–340	[70,71]
Mittivakkat	Greenland	N/A	270	83	[76]
Watson River	Greenland	N/A	38–56	100–138	[63]
Tungufljót	Icelandic	98,000	720	343	[78]
Hvítá-S/W	Icelandic	80,000–123,000	650–1100	361–524	[78]
Worthington Glacier	N American	15,000	1600	208	[79]
Berendon Glacier	N American	N/A	947	256	[80]
S Cascade Glacier	N American	14,100	676–930	173–282	[69,79,81]
Batura Glacier	Himalayan	N/A	1460	730	[66,82]
Chhota Shigri Glacier	Himalayan	17,400	750–935	214	[30,53,66]
Dokriani	Himalayan	9700	462–4200	385–646	[62,83]
Gornergletscher Glacier	Alpine	N/A	450	321	[84]
Haut Glacier d’Arolla	Alpine	50,000–61,000	64–685	298–376	[59]
Dongkemadi Glacier	Alpine	14,911	185	189	[19,85]
Glacier No.1	Alpine	43,240–52,329	577–703	N/A	[51]
Hailuogou Glacier	Alpine	231,700–241,428	2850–3108	616–711	This study

; [76,77,78,79,80,81,82,83,84,85]). For example, the yields from HLG are 2.3 to 16.0 times higher than Midre Lovénbreen (41,000–47,000 kg·km⁻²·year⁻¹) in Arctic [76], Kuannersuit (15,900 kg·km⁻²·year⁻¹) in Greenland [70,71], Tungufljót (98,000 kg·km⁻²·year⁻¹) in Icelandic [78], Worthington Glacier (15,000 kg·km⁻²·year⁻¹) in North American [79], Chhota Shigri Glacier (17,400 kg·km⁻²·year⁻¹) in Himalayan [30,53,66] and Haut Glacier d’Arolla (50,000–61,000 kg·km⁻²·year⁻¹) in Alpine [59]. Although the cation denudation rate (CDR) at HLG (H-CDR; 2850 to 3108 Σ*meq⁺·m⁻²·year⁻¹) is within the global range of currently published CDR (94–4200 Σ*meq⁺·m⁻²·year⁻¹; Table 5), H-CDR is larger than those from almost all glacial basins except for Dokriani Glacier (462–4200 Σ*meq⁺·m⁻²·year⁻¹) in the Himalaya Mountains (Table 5). If we normalize H-CDR by annual specific discharge (4.37 m in 2008, and 4.63 m in 2013), the chemical weathering intensity (CWI) at HLG (H-CWI) ranges from 616 to 711 Σ*meq⁺·m⁻³·year⁻¹, which is also larger than those at most glacial basins worldwide irrespective of the lithology (Table 5). The higher yields, CDR and CWI at HLG may be caused by greater abundance of carbonate (e.g., calcite with the proportion of 1.5% in glacial deposits; Table 1), in comparison to other basins such as Haut Glacier d’Arolla (Switzerland) with 0.6% of calcite [59], suggesting that chemical weathering at glacial basins in the TP is far higher than that at other glacial basins worldwide. Given limited field data, further estimates of solute export from HLG would be helpful to better quantify the variability in the catchment-scale hydrodynamics and solute acquisition processes. Noticeably, the CDR at alpine basins (185–4200 Σ*meq⁺·m⁻²·year⁻¹) is larger than those at polar basins (except for Icelandic; 38–850 Σ*meq⁺·m⁻²·year⁻¹) (Table 5). This should be related to higher melt rates and more precipitation as well as the polythermal- or warm-based drainage system for alpine glaciers [51,82,83,84,85], in comparison to lower rates and less precipitation as well as cold-based drainage systems for studied polar glaciers [70,71,76]. Both daily mean suspended sediment load and yields from HLG are higher than from Chhota Shigri Glacier (135 t·day⁻¹ for SSL and 3.0 t·km⁻²·day⁻¹ for SSY; [86]), Dunagiri Glacier (47 t·day⁻¹ and 2.6 t·km⁻²·day⁻¹; [87]), Changme Khangpu Glacier (18 t·day⁻¹ and 4.0 t·km⁻²·day⁻¹; [88]) and Shaune Garang Glacier (43 t·day⁻¹ and 1.1 t·km⁻²·day⁻¹; [37]) and lower than from Dokriani Glacier (447 t·day⁻¹ and 28 t·km⁻²·day⁻¹; [89]) and Gangotri Glacier (11,673 t·day⁻¹ and 21 t·km⁻²·day⁻¹ [52]) in the Himalaya Mountains (

Table 6. Daily mean suspended sediment load (SSL) and yields (SSY) from Hailuogou Glacier in comparison to nearby Himalayan glaciers. N/A denotes data are not available.

Glacier	Basin Area (km2)	Glacierized Area (%)	SSL (t∙day−1)	SSY (t·km−2·day−1)	Reference
Shaune Garang	38.1	N/A	43	1.1	[37]
Gangotri	556	51.4	11,673	21.0	[52]
Chhota Shigri	45	25	135	3.0	[86]
Dunagiri	17.9	N/A	47	2.6	[87]
Changme Khangpu	4.5	N/A	18	4.0	[88]
Dokriani	16.1	45	447	27.8	[89]
Hailuogou	80.5	45.3	346	4.3	This study

). The difference between basins is related to many factors such as river discharge, glacierized area, rock debris, basin geology, weather conditions and sampling methods. For example, the suspended sediment concentrations and fluxes are generally controlled by river discharge at many glacial basins [30,36,38,52], and field sampling only during the high flow season will lead to the high sediment fluxes and yields at a glacial basin (Figure 4).

To investigate the temporal behavior of the delivery of solutes and sediments, the dates corresponding to 10%, 50% and 90% delivery of major ions and sediments were calculated (

Table 7. Comparison of the dates of cumulative percentage distribution of major ion and suspended sediment (SS) fluxes with discharge (DIS) at the Hailuogou basin over two melt seasons in 2008 (before the slash) and 2013 (after the slash). Negative and positive values denote the days for solutes and SS later and earlier than discharge delivery, respectively. N/A denotes data are not available.

	Dates for Cumulative Percentage of Species Flux and Discharge			Lead Days of Species Flux Relative to Discharge
	10%	50%	90%	10%	50%	90%
DIS	5 May/7 May	28 Jul/26 Jul	6 Oct/2 Oct	NA/NA	NA/NA	NA/NA
SS	NA/3 Jun	NA/14 Jul	NA/10 Sep	NA/+27	NA/−12	NA/−22
Na+	5 Feb/12 Mar	29 Jun/23 Jul	26 Nov/26 Oct	−90/−56	−29/−3	+51/+24
K+	3 Mar/4 Apr	18 Jul/23 Jul	6 Nov/13 Oct	−63/−33	−10/−3	+31/+11
Mg2+	29 Feb/20 Mar	19 Jul/29 Jul	8 Nov/21 Oct	−66/−48	−9/+3	+33/+19
Ca2+	16 Mar/1 Apr	23 Jul/2 Aug	30 Oct/19 Oct	−50/−36	−5/+7	+24/+17
Cl−	4 Feb/18 Feb	27 Jun/7 Jul	27 Nov/10 Nov	−91/−78	−31/+19	+52/+39
SO42−	24 Feb/5 Mar	19 Jul/7 Aug	10 Nov/3 Nov	−71/−63	−9/+15	+35/+32
NO3−	28 Jan/19 Feb	15 Jun/30 May	3 Dec/13 Nov	−98/−77	−43/−57	+58/+42
HCO3−	21 Mar/12 Apr	23 Jul/29 Jul	28 Oct/14 Oct	−45/−25	−5/+3	+22/+12

). In general, the 10% delivery of all ions was 25 to 98 days later than that of discharge in both 2008 and 2013. However, the 50% delivery for all ions in 2008 and for Na⁺, K⁺ and NO₃⁻ in 2013 were 3 to 57 days later, and for HCO₃⁻, Ca²⁺, Mg²⁺, SO₄²⁻ and Cl⁻ in 2013 were 3 to 19 days earlier than that of discharge. Additionally, the 90% delivery of all ions were 11 to 58 days earlier than discharge in both 2008 and 2013 (Table 7). In addition, the 10% delivery of sediments was 27 days earlier than that of discharge, but the 50% and 90% deliveries were 12 and 22 days later than discharge, respectively. This is consistent with the delivery patterns of sediments from Gornergletscher Glacier in Switzerland [90] and from Gangotri Glacier in the Himalaya [91]. The later delivery suggests that the solute and sediment exports are dominated by meltwater, and the early delivery indicates that their exports are controlled not only by meltwater but also by others factors (e.g., precipitation). After glacier meltwater enters downstream, it is often used for the irrigation and drinking in the arid and/or semiarid regions (e.g., [91,92,93]). Thus, the delivery pattern of some solutes (e.g., N, P, Fe, As, Cd, Pb, and Th) adsorbed on the sediment surface [20] has important implications for the planning and management of water resources and water quality in human settlements downstream [18,19,20,21].

5. Conclusions

The chemical weathering processes and downstream solute and sediment delivery were examined at the Hailuogou Glacier basin (HLG) in the southeast Tibetan Plateau (TP) over two glacier melt seasons in 2008 and 2013. The dominant ions were HCO₃⁻, Ca²⁺ and SO₄²⁻ in supraglacial meltwater (M), proglacial river water (R) and groundwater (G), but in precipitation (P) the dominant ions were HCO₃⁻ and SO₄²⁻. Concentrations of HCO₃⁻ and cations followed the sequence: G > R > M > P, while the sequence of SO₄²⁻ was G > R > P > M. Major ions and suspended sediment concentrations in river water exhibited a pronounced seasonality, with higher values during low flow for major ions and during high flow for suspended sediments. Concentrations of major ions and suspended sediments exhibited negative and positive relationships with river discharge respectively, suggesting a strong hydrological control on meltwater runoff chemistry and suspended sediment transport. Runoff chemistry was dominated by carbonate weathering and sulfide oxidation, with less influence of oxidation of organic carbon and anthropogenic input. HCO₃⁻ and Ca²⁺ in river water were mainly derived from calcite weathering, and SO₄²⁻ primarily came from pyrite oxidation. The solute and sediment fluxes also exhibited a strong seasonality and were positively related to river discharge. Solute flux ranged from 18,095 to 19,435 t·year⁻¹, with 50–53% from HCO₃⁻, 20–21% from SO₄²⁻ and 19% from Ca²⁺, and solute yields were 225–241 t·km⁻²·year⁻¹. The sediment load and yields were 126,390 t·year⁻¹ and 1570 t·km⁻²·year⁻¹, respectively. The cationic denudation rate (CDR) ranged from 2850 to 3108 Σ*meq⁺·m⁻²·year⁻¹, and chemical weathering intensity (CWI) was 616–711 Σ*meq⁺·m⁻³·year⁻¹. The solute yields, CDR and CWI at HLG were higher than those at most glacial basins worldwide, suggesting an intense chemical weathering at monsoon-dominated glacial basins in the TP.