Spatiotemporal Variation of Hydrogen and Oxygen Stable Isotopes in the Yarlung Tsangpo River Basin, Southern Tibetan Plateau

Characterization of spatiotemporal variation of the stable isotopes δ18O and δD in surface water is essential to trace the water cycle, indicate moisture sources, and reconstruct paleoaltimetry. In this study, river water, rainwater, and groundwater samples were collected in the Yarlung Tsangpo River (YTR) Basin before (BM) and after the monsoon precipitation (AM) to investigate the δ18O and δD spatiotemporal variation of natural water. Most of the river waters are distributed along GMWL and the line of d-excess = 10‰, indicating that they are mainly originated from precipitation. Temporally, the δ18O and δD of river water are higher in BM series (SWL: δD = 10.26δ18O+43.01, R2 = 0.98) than AM series (SWL: δD = 9.10δ18O + 26.73, R2 = 0.82). Spatially, the isotopic compositions of tributaries increase gradually from west to east (BM: δ18O = 0.65Lon (°)-73.89, R2 = 0.79; AM: δ18O = 0.45Lon (°)-57.81, R2 = 0.70) and from high altitude to low (BM: δ18O = −0.0025Alt(m)-73.89, R2 = 0.66; AM: δ18O = −0.0018Alt(m)-10.57, R2 = 0.58), which conforms to the “continent effect” and “altitude effect” of precipitation. In the lower reaches of the mainstream, rainwater is the main source, so the variations of δ18O and δD are normally elevated with the flow direction. Anomalously, in the middle reaches, the δ18Omainstream and δDmainstream values firstly increase and then decrease. From the Saga to Lhaze section, the higher positive values of δ18Omainstream are mainly caused by groundwater afflux, which has high δ18O and low d-excess values. The δ18Omainstream decrease from the Lhaze to Qushui section is attributed to the combined action of the import of depleted 18O and D groundwater and tributaries. Therefore, because of the recharge of groundwater with markedly different δ18O and δD values, the mainstream no longer simply inherits the isotopic composition from precipitation. These results suggest that in the YTR Basin, if the δ18O value of surface water is used to trace moisture sources or reconstruct the paleoaltimetry, it is necessary to rule out the influence from groundwater.


INTRODUCTION
The stable isotopes, δ 18 O and δD, in precipitation are widely used as a fingerprint for the hydrological processes and atmospheric circulation (Fette et al., 2005;Zhu et al., 2007;Singh et al., 2013;Guo et al., 2017;Wu et al., 2019). The Tibetan Plateau (TP) is the highest and widest high-altitude region on Earth, and the uplift history and terrain atmosphere of the region are of scientific interest (Garzione et al., 2000;Rowley et al., 2001;Spicer et al., 2003;Quade et al., 2011). Because of the so-called "continent effect" and "altitude effect" of isotope composition in precipitation (i.e., the negative relationship between Δδ 18 O or ΔδD in precipitation and transport distance and elevation) producing heavier monsoonal rainfall first with transported distance, and then with orographic lifting, stable isotopes have also been used to trace moisture sources and to reconstruct paleoelevation of the TP (Garzione et al., 2000;Poage and Chamberlain, 2001;Rowley et al., 2001;Hren et al., 2009;Li and Garzione, 2017). A global network for isotopes in precipitation (GNIP) has been established by the International Atomic Energy Agency to monitor the long-term changes in δ 18 O and δD in global precipitation. However, because of the few isotope-monitoring stations in the TP, the study of precipitation isotopic variability and hydrological processes in this region is usually hampered.
Based on the consideration that stream water can provide a time-integrated record of the isotopic composition of precipitation (Gat, 1996;Kendall and Coplen, 2001), a number of studies have used the isotopes in river water as a substitute for modern precipitation to further reconstruct paleoelevation or to trace moisture sources (Garzione et al., 2000;Hren et al., 2009;Yang et al., 2012a;Bershaw et al., 2012;Kong and Pang, 2016;Li and Garzione, 2017). At present, significant negative isotope-altitude relationships have been reported: for example, the lapse rate of δ 18 O in river water is about −0.21‰ per 100 m in the northeastern TP (Bershaw et al., 2012), −0.24‰ per 100 m in the southern TP (Ding et al., 2009), −0.28‰ per 100 m in the high Himalayas of Nepal and −0.31‰ per 100 m in the Niyang River watershed (Florea et al., 2017), while it is as low as −0.36‰ per 100 m in the southern Himalaya (Wen et al., 2012) and about −0.19‰ per 100 m in the Hengduan Mountains .
In addition to precipitation, groundwater recharge and ice/ glacier melt water entering the YTR cannot be ignored as well (Liu, 1999;Tan et al., 2021;Zhang et al., 2021). The contributions of these sources differ between different seasons and different locations: the groundwater contribution is 40% in Nugesha, 36% in Yangcun, 32% in Nuxia (Liu, 1999), and 27-40% in the middle reach (Tan et al., 2021). Furthermore, the distinctions of hydrochemical and isotopic composition between groundwater and precipitation are obvious in the YTR Basin (Tan et al., 2014;Liu et al., 2019). In the case of the YTR Basin, which is the largest river basin on the Tibetan Plateau, it is unclear as to the extent to which groundwater recharge and precipitation affect the spatiotemporal distribution of stable isotopes in the river water. Accordingly, this study documents the oxygen and hydrogen isotopic compositions in river water, rainfall, and groundwater in the YTR Basin. Spatially, major influences on the variability of the isotopic composition of river water along the channel are described and identified. Temporally, comparisons of the isotopic composition were conducted before and after monsoon precipitation. The anomalous spatial distribution of isotopic compositions of mainstream in the middle reach and its cause is revealed.

Study Area
The YTR extends from the Jima Yangzong glacier above 5,200 m elevation to the Bay of Bengal and is the highest large river in the world located south of the TP (Huang et al., 2008). The YTR Basin is a long and narrow valley covering a total area of about 240,480 km 2 in a west-east direction among the Gangdise, Nyainqentanglha, and Himalaya Mountains ( Figure 1) . The elevation of the basin gradually decreases from west to east. The upper reach is from the source to Saga, with the major tributaries Mayou Tsangpo, Chai Qu, and Jiada Tsangpo; the middle reach is from Saga to Qushui, with major tributaries being Dogxung Tsangpo and Nyang Qu; and the lower reach is below Qushui with major tributaries being Lhasa River, Niyang River, and Parlung Tsangpo. In general, mean annual precipitation decreases from about 1,000 mm in the east of the basin to about 200 mm at the headwaters (Ren et al., 2018). Some 65-80% of the annual precipitation occurs between June and September, with the exception of southeastern Tibet (Liu, 1999). The mean annual air temperature in this area is about 5.9°C with an average seasonal variation of 2.46 and 13.49°C (You et al., 2007). The mean annual evaporation in this area is 1,052 mm, which is higher in the west than that in the east (Huang et al., 2011). The mean annual discharge of the YTR Basin is ∼1,395.4 × 108 m 3 (Liu, 1999). In the upper and middle reach, groundwater is the main source; in the lower reach, the recharge pattern is a mix of rainwater and meltwater; while into the heavy rain area below the grand canyon, the river is mainly from precipitation (Liu, 1999;Yang et al., 2011).
Geologically, TP is generally regarded as a product of the Eurasian and Indian Plates collision (Gansser, 1980). The tectonic units of the TP comprise a series of east-west-trending continental blocks, including Songpan-Ganzi, Qiangtang, Lhasa, the Tethyan Himalayas, the high Himalaya, and the lesser Himalayas (Tapponnier et al., 2001;Zhang et al., 2004;Zhang et al., 2015). Due to the east-west extension of the TP, rifts trending approximately north-south cut across the YTR Basin (Armijo et al., 1986), From west to east, these rifts are Dingri-Nima (DN), Dingjie-Xietongmen-Shenzha (DXS), Yadong-Dangxiong-Gulu (YDG), and Gudui-Sangri (GS) rifts. Thus, the YTR Basin has widely distributed springs along the rifts (Wang et al., 2020;Tan et al., 2021). The terrane in the southern TP is dominated by Paleozoic-Mesozoic carbonate and clastic sedimentary rocks (Galy and France-Lanord, 1999). The bedrock in the basin mainly consists of igneous granite/granitic gneiss, or schist/other felsic volcanic and mafic volcanic rock. However, there are few if any carbonate rocks are in the upper and middle reaches of the catchments (Hren et al., 2007;Huang et al., 2011). Along the entire river course, ophiolites and ophiolitic mélanges are commonly found (GMRT Bureau of Geology and Mineral Resources of Xizang Tibet Autonomous Region, 1993).

Water Sampling and Isotopic Analysis
Two systematic sampling series were carried out in the YTR Basin (Figure 1), one in June, 2017, prior to large-scale monsoon rainfall ("BM" series) and one in September, 2017, after largescale monsoon rainfall ("AM" series). A total of 53 river water samples were collected, comprising 22 samples from the main stream and 31 samples from the large tributaries (e.g., Jiada Tsangpo, Dogxung Tsangpo, Lhasa River, Niyang River, Parlung Tsangpo, etc.). Some of the mainstream sampling sites were located at least 1 km downstream of the confluence of major tributaries to ensure that the water from all sources was fully mixed. The other mainstream samples were evenly distributed along the whole basin. The samples of the tributaries were collected before they merged with the main stream. In addition, a hot spring sample was collected from the Yangbajain (YBJ) geothermal field, central YTR Basin. The rainwater sample was from Lhasa.
River water samples were collected at a depth of 10-15 cm on the river bank. All samples were filtered through 0.45 μM cellulose membrane and then put the filtered samples into dry, clean 15 ml HDPE bottles. To prevent sample evaporation, the bottles were completely filled and immediately sealed with parafilm. Samples were frozen in a refrigerator in the laboratory and then thawed to room temperature when required for analysis. Stable δ 18 O and δD values of all samples were determined by Liquid Water Isotope Analyzer (IWA-35EP) in the State Key Laboratory of Environmental Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences. The precision of δ 18 O and δD measurements was better than ±0.1‰ and ±1.0‰, respectively. Results were reported as relative to the standard V-SMOW (Vienna Standard Mean Ocean Water).

RESULTS
The δ 18 O and δD values for river water, groundwater, and rainwater in the YTR Basin are listed in Table 1  Frontiers in Earth Science | www.frontiersin.org November 2021 | Volume 9 | Article 757094 each sampling point, the δ 18 O and δD of BM series were higher than those for the AM series, except for the Duilong Qu tributary (T8) sample, which was omitted in September. The results show that the river water is richer in heavy H and O isotopes before monsoon precipitation than after. The variation amplitude of stable isotopic composition in the upper and middle reach is less than that in lower reach. The sample with maximum amplitude variation of δ 18 O and δD values between two series is from Motuo (M29) (−4.4 and −37‰, respectively); the sample with the minimum variation is from Saga (M11) (−0.7 and −13‰, respectively).     Frontiers in Earth Science | www.frontiersin.org November 2021 | Volume 9 | Article 757094 5 the BM and AM series, δ 18 O mainstream and δD mainstream showed the same spatial variation tendency. It is notable that the trends of δ 18 O mainstream and δD mainstream were significantly different from those in the tributaries, especially in the middle reach ( Figure 2).

δD-δ 18 O Relationship
The correlation between natural water H and O isotopic compositions is usually used to identify the recharge source, circulation paths, and mixing or exchange processes (Tan et al., 2014;Ren et al., 2018). In general, δ 18 O and δD in meteoric water on a global scale have been found to fall close to the Global Meteoric Water Line (GMWL) of δD 8δ 18 O+10 (Craig, 1961). In the YTR Basin, the Local Meteoric Water Lines (LMWLs) for several sites have been reported as δD 7.9δ 18 O + 4.3 (R 2 0.98) for Bomi (Gao et al., 2011), δD 7.9δ 18 O + 6.3 for Lhasa (Tian et al., 2001b), and δD 7.2δ 18 O−15.8 for Xigaze (Ren et al., 2017a), but the exact LMWL for the entire basin is yet to be determined. The indicator, deuterium excess (d-excess δD-8δ 18 O), was defined as a measure of non-equilibrium isotopes effects (Dansgaard, 1964) to record the difference between the actual δD and the expected equilibrium values based on measured δ 18 O. The d-excess of global meteoric water is +10‰. In the low-humidity conditions, strong kinetic fractionation in evaporation causes high d-excess value (>10‰) in precipitation. Conversely, high-humidity results in a decrease in kinetic isotope fractionation, and subsequent precipitation will have a low d-excess value (<10‰) (Gat and Matsui, 1991;Gat, 1996). For the geothermal water systems, δ 18 O values of rocks and minerals are greater than those for water in general, and the O isotope exchange during the water-rock interaction increases δ 18 O in water. However, since few rock minerals contain H and the δD value is low, the isotope exchange reaction has little effect on the δD of water. As a result, the isotopic composition of the groundwater is shifted horizontally to the right on the δD vs δ 18 O diagram ( Figure 4A) when the water residence time is long enough and the water-rock interaction is significant, a phenomenon known as "δ 18 O drift" (Wang, 1991). As a result, the isotopic exchange reaction also reduces the d-excess value of geothermal water (Wang, 1991;Yin et al., 2001).
In this study, most of the δ 18 O and δD values of river waters in the YTR Basin are distributed along GMWL and the line of d-excess 10‰, which indicates that river waters should be mainly originated from precipitation. The linear regression relationships between δ 18 O and δD values and surface water lines (SWLs) are δD 10.26δ 18 O + 43.01 (R 2 0.98, n 27, p < 0.01) before the monsoon precipitation and δD 9.10δ 18 O + 26.73 (R 2 0.82, n 26, p < 0.01) after the monsoon precipitation. Both the slopes and intercepts are larger than GMWL and LMWLs. The SWLs are approximate to previous studies of Hren et al. (2009) (slope≈10 and intercept≈38) and Ren et al. (2018) (slope 9.25 and intercept 24.1) in the basin. Evaporation is not the cause of steep slopes because that it would result in more isotopical enrichment and a lower δDδ 18 O slope in residual waters than GMWL (Gonfiantini 1986;Ren et al., 2018). Glacier melt is also a vital source of river water, and the points of glacier melt distributed along the GMWL inherit the isotopic composition of precipitation ( Figure 4A). Therefore, the steep slopes of SWLs cannot be attributed to the supply of glacier melt. As shown in Figure 4A, there are a few samples (mainly from the upper and middle stream) fall on the lower left-hand side of the GMWL with low d-excess ( Figure 4B), which could produce a steep δD-δ 18 O slope. Meanwhile, the samples show greater deviation more from the GMWL, indicating that they may be supplied by another source with higher δ 18 O and d-excess than precipitation.
In addition to precipitation and glacier melt water, groundwater may also be non-negligible sources of YTR water (Liu, 1999;Zhang et al., 2021). A few studies have A supply of such groundwater would lower the δ 18 O and δD values in the river water, but the slope of the SWL would not be affected; however, other groundwater samples (including YBJ) have shown a significant positive deviation of δ 18 O from GMWL and a negative deviation of d-excess from 10‰ ( Figure 4B). According to the study of Liu et al. (2019), some hot springs in the Semi and Daggyai geothermal fields have significant "δ 18 O drift" due to the mixing of magmatic fluids with higher δ 18 O values. A similar property of groundwater has been observed in other tectonic fracture zones or thermal field distribution areas (e.g., at the edge of the Guanzhong Basin, China (Ma et al., 2017), in the southeastern edge of the Eurasia (Chen et al., 2016), in Kyushu, Japan (Mizutani, 1972), and in northern Iceland (Stefánsson et al., 2019)). Consequently, from the perspective of isotopic composition characteristics, these river samples deviating from GMWL to the right and the downward trend of the d-excess 10‰ line to the downward may be due to the recharge from a particular type of groundwater with significant δ 18 O drift.

Temporal Patterns of δD and δ 18 O in the Yarlung Tsangpo River Basin
As shown in Figure 4A, the YTR waters have high δD and δ 18 O values before the monsoon precipitation and low values after the monsoon precipitation. As discussed in δD-δ 18 O Relationship, YTR waters mainly originate from precipitation and inherit its isotopic composition of precipitation. The temporal pattern of isotopic composition in the YTR Basin river waters is dominated by precipitation.
The precipitation in the YTR Basin has a remarkable seasonal pattern. Summer precipitation contributes up to 65-80% of the annual total in this region, and it is dominated by the monsoon from the Bay of Bengal (Liu, 1999). Previous work revealed that the temporal variation of δ 18 O in precipitation (δ 18 O precipitation ) was characterized by a higher value in the dry season (October to May of the following year) and lower in the monsoon season (from mid-June to September). This is known as the "amount effect" of δ 18 O precipitation , as in the observational data from Lhasa station shown in Figure 5 (Wei and Lin., 1994;Tian et al., 1997;Tian et al., 2001a;Tian et al., 2001b;Yu et al., 2008;Yu et al., 2016;Hren et al., 2009;Gao et al., 2011;Yang et al., 2012b;Yao et al., 2013;Ren et al., 2018). As a whole, under the influence of rainfall, the δD and δ 18 O of the YTR Basin river waters are high before the monsoon precipitation and low afterward.
The δ 18 O amplitude variation of sample between two series (4.4‰, the maximum) is much small than annual variation of precipitation (∼14.5‰) in Lhasa. For one thing, the valley collects precipitation from an entire catchment over a period of time. For another, in addition to precipitation, the river may have other sources with stable isotopic composition, such as groundwater. So, the isotopic composition of river water is more stable than that of precipitation in different seasons. The variation amplitude of δ 18 O in the upper reach and middle reach is less than that in the lower reach. It could be attributed that precipitation contributes more to river in the lower reach than the upper or middle reach.

Tributaries
The δD and δ 18 O values in tributary waters (δD tributary and δ 18 O tributary ) increase gradually from west to east with high to low topographical relief for both BM and AM series (Figure 2). The spatial distribution pattern is very similar to precipitation, which indicates that the δD tributary and δ 18 O tributary values are mainly affected by precipitation.
Spatially, the δ 18 O precipitation in the YTR Basin is influenced by the "continent effect" and "altitude effect" Hren et al., 2009;Wang et al., 2000). During the summer period, moisture penetrates into the southeastern TP and is transported westward along the YTR valley, so longitude may be regarded as a convenient proxy for the transported distance of the moisture. Moreover, the terrain of the YTR valley gradually rises from east to west. As a warm and humid airmass moves westward, the moisture adiabatically cools and produces heavier monsoonal rainfall with increasing transport distance and altitude, resulting in a gradually lighter isotopic composition of precipitation Hren et al., 2009). Therefore, the δ 18 O in the precipitation decreases gradually in an upstream direction along the YTR valley. As shown in Figure 6, the annual weighted δ 18 O precipitation value is significantly correlated with longitude and altitude. The δ 18 O precipitation vertical lapse rate FIGURE 5 | Monthly precipitation amounts and weighted-average δ 18 O of precipitation at Lhasa station (data are from Gao et al., 2011).
The δD tributary and δ 18 O tributary show a similar spatial variation trend with precipitation (Figures 2, 6). It should be noted that the change rates of δ 18 O tributary value with regard to longitude and altitude (0.65‰ per longitude degree and −2.5‰/km) before the monsoon precipitation approximate more closely to annual weighted δ 18 O in precipitation (0.63‰ per longitude degree and −2.4‰ per 1 km) than after monsoon precipitation (0.44‰ per longitude degree and −1.8‰ per 1 km). The lower δ 18 O vertical lapse rate after the monsoon precipitation may be related to water sources mixed from different altitudes. During the monsoon precipitation in summer, the accumulated snows from different altitudes melt and mix to recharge the river. Although the isotopic composition of melted ice and snow inherits the precipitation influenced by the altitude effect, the mixed melted waters from different elevation change δ 18 O tributary vertical lapse rate and weaken the correlation, but do not modify the spatial patterns of δ 18 O tributary .

Main Stream
The spatial trends of mainstream δD and δ 18 O values are different from those of tributaries and precipitation. As shown in Figure 3, δ 18 O mainstream and δD mainstream increase from the upper reach to Lhaze (M15) and then decline gradually from Lhaze to Qushui (M22) in the middle reach. Below Qushui (M22), these values rise sharply along the flow direction.
In the lower reach, the trends of δ 18 O mainstream are similar to δ 18 O precipitation and δ 18 O tributaries (Figures 2, 5). Analogous patterns of evaporation and "altitude effect" on precipitation have been observed in the main flows of other large rivers globally: For example, the Ganges River in Asia (Ramesh and Sarin, 1992), the Missouri River in the United States (Winston and Criss, 2003), the Nile River in Africa (Cockerton et al., 2013), and the Yangtze River and Yellow River in China (Li et al., 2015;Fan et al., 2017). The lower reach of the YTR valley is the major transport pathway of the monsoonal moisture with more precipitation whose δ 18 O values are relatively high, attributable to the lower altitudes and shorter transportation distance (Tributaries). Furthermore, rainwater is the main source of surface water downstream. Therefore, δ 18 O tributaries , δ 18 O precipitation , and δ 18 O mainstream have similar trends in the lower reach.
Anomalously, in the section from Saga (M11) to Qushui (M22), δ 18 O mainstream and δD mainstream firstly increase and then decrease. In particular, the decreasing tendency between Lhaze (M15) and Qushui (M22) contradicts to the expected "continent effect" and "altitude effect" on the isotopic evolution of surface water. If the local precipitation controls the isotope composition of the mainstream, the values of δ 18 O mainstream and δD mainstream would be expected to increase downstream in theory. Moreover, it is the mainstream samples from the middle reach that deviate from the GMWL line, having lower d-excess values as, shown in Figure 4, and weakening the correlation between stable isotopes composition with longitude and altitude ( Figure 3). The anomalous trend in the middle reach cannot be attributed to recharge by precipitation and melt water (δD-δ 18 O Relationship).
The phenomenon that main flows are isotopically enriched west of about 86°E longitude has been observed in other studies (Hren et al., 2009;Ren et al., 2016). They considered that the increasing influence of the westerlies would result in higher δ 18 O and d-excess of surface waters, because of the upper reach being located in the transition between monsoon-dominant area and westerlies dominant area (Hren et al., 2009;Ren et al., 2016). However, this view is not supported by our results. Firstly, the δ 18 O of the westernmost mainstream sample (M11) is lower than that of the adjacent main stream point downstream sample values (M15, M17, and M19). It is true for both two seasons, especially for BM series, where the δ 18 O of M11 is the lowest (Figure 3). Secondly, the annual weighted δ 18 O of precipitation in this section decreases from Lhasa to Dingri ( Figure 6A). Additionally, the d-excess values of mainstream waters in this  Wang et al. (2000) and Yao et al. (2013)).
Frontiers in Earth Science | www.frontiersin.org November 2021 | Volume 9 | Article 757094 8 section (∼3‰ for BM and ∼1‰ for AM) are much lower than 10‰ ( Figure 4B), which does not accord with the larger d-excess characteristics of westerly precipitation. Therefore, these results do not support that westerly precipitation is the main cause of the abnormal isotopic composition in the middle reach. The enhanced evaporation intensity from east to west will also lead to more positive δD and δ 18 O values of the upstream. However, the isotopic composition of westernmost mainstream sample (M11) is not the maximum in both two seasons. Therefore, the evaporation is not the main reason for the anomaly.
The abnormal δ 18 O mainstream trend in the middle reach is most likely related to groundwater recharge. Firstly, three roughly north-south rifts (Dingri-Nima (DN), Dingjie-Xietongmen-Shenzha (DXS), and Yadong-Dangxiong-Gulu (YDG) rifts) intersect the YTR valley in the middle reach (Wang et al., 2020). The tectonic fracture zones provide the conditions for a large amount of groundwater to recharge the mainstream (Zhang et al., 2021). Additionally, the hydraulic head of over 1,000 m could drive groundwater flow over long distances (Hoke et al., 2000;Tan et al., 2014). From the hydrochemical point of view, geothermal water has been suspected as the main source of the elevated (As) dissolved levels in the YTR (Zhang et al., 2021). Other major ion and 222 Rn data have also strongly suggested a large addition of groundwater to the YTR in the middle reach (Tan et al., 2021). In addition, the δ 18 O characteristics of groundwater show significant differences from west to east in the YTR Basin (Tan et al., 2014;Liu, 2018;Liu et al., 2019;Tan et al., 2021). The nearby groundwater in the Semi (∼86.4°E) and Daggyai (∼85.6°E) geothermal field near has obvious characteristics of 18 O drifts in the mixed recharge of meteoric waters and magmatic fluids . In Semi and Daggyai geothermal field, the δ 18 O maximum values of groundwater are −11.8 and −15.3‰, respectively . Their locations exactly coincide with the area where δ 18 O mainstream values are abnormally high, from Saga to Lhaze (Figure 7). However, the groundwater sampled in the range of approximately 88-92°E has been observed to be depleted in D and 18 O and may have originated from paleo-precipitation during a cooler time and contributes 27-40% of the river flow (Tan et al., 2021). Near the DXS and YDG rifts, the δ 18 O minimum values of groundwater are −21.5‰ and −21.4‰, respectively (Tan et al., 2021). Just in this section (from Lhaze to Qushui), the δ 18 O value of mainstream is decreasing. In addition, for the tributaries, the δ 18 O values of Jiada Tsangbo (T10), Dogxung Tsangbo (T16), and Xiang Qu (T18) before the confluence are lower than that of the main stream in the middle reach (Figure 7), so the inflow of the tributaries will lead to the isotopic composition declining as well. Therefore, the following conclusions can be drawn: from the Saga (M11) to Lhaze (M15) section, the recharge of D and 18 O enriched groundwater will result in the increase of the isotopic composition of the mainstream water; from Lhaze (M15) to Qushui (M22), the isotopic composition of the mainstream decreases due to the combined action of depleted D and 18 O groundwater recharge and tributaries import.

Implications
The stable isotopic characteristics of modern precipitation provide vital information to indicate moisture sources and to reconstruct paleoaltimetry. In some regions where GNIP observations are sparse, δ 18 O values of river water have to be used to alleviate this problem (Rowley, 2007;Timsic and Patterson, 2014;Xu et al., 2014;Li and Garzione, 2017). However, the water in a large river is generally collected from two main sources: 1) recent precipitation through surface runoff or channel precipitation or by rapid flow through the shallow subsurface and 2) groundwater recharge. The relative contribution of these sources differs in each watershed (Ogrinc et al., 2008). Therefore, the stable isotopic characteristics of river Frontiers in Earth Science | www.frontiersin.org November 2021 | Volume 9 | Article 757094 9 water are affected by meteorological and hydrological factors. In this study, the spatial distribution of δ 18 O mainstream in the YTR middle reach is significantly different from that of tributaries and local precipitation. Due to the groundwater recharge, δ 18 O mainstream in the section from Lhaze to Qushui is characterized by inverse isotope-elevation and isotope-moisture transport distance relationships that could produce significant misestimates of paleoaltimetry and moisture sources. Thus, when reconstructing the paleoaltimetry and tracing moisture sources using river water as a substitute for modern precipitation, it is necessary to consider that groundwater may also influence the distribution of stable isotopes in river water, especially near tectonic fracture zones where geothermal water may be characterized by remarkable high or low δ 18 O values.

CONCLUSION
In the YTR Basin, most of the river waters originate in precipitation and inherit the δD and δ 18 O characteristics of precipitation. Temporally, under the predominance of precipitation, the isotopic composition of river water is high before the monsoon precipitation (mid-June) and low after the monsoon precipitation (mid-September). Spatially, the δD and δ 18 O values in tributary water increase gradually from west to east and conform to the "continent effect" and "altitude effect" of precipitation. For the mainstream, rainwater is the prime source of surface water in the lower reach, with the result that the δD and δ 18 O variations are normally elevated. Anomalously, in the middle reach of the mainstream, the δ 18 O and δD firstly increase and then decrease. From Saga to Lhaze, the groundwater is characterized by high δ 18 O and low d-excess afflux causes the δ 18 O mainstream to be more positive. Then, from Lhaze to Qushui, the decrease in isotopic compositions of the mainstream is attributed to the combined action of the D and 18 O depleted groundwater and tributaries import. As a result, due to the recharge of groundwater with remarkable differences in isotopic composition, the mainstream no longer simply inherits the characteristics of tributaries or precipitation: therefore, when reconstructing paleoaltimetry and tracing moisture sources by using δ 18 O of river water as a substitute for modern precipitation, it is necessary to rule out the influence of groundwater, especially near the tectonic fracture zones whose condition is conducive to groundwater drainage, and the δ 18 O of the groundwater varies significantly.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material; further inquiries can be directed to the corresponding authors.