**5. Discussion**

### *5.1. Potential Causes of Lake Expansion*

Long-term changes in the surface areas of Tuosu, Keluke, and Gahai Lakes during the study period were the result of a combination of climate change and groundwater contributions, with climate change processes mainly including increased precipitation and temperature. Expansion of Tuosu Lake was particularly obvious; hence, we used this lake as an example to analyze the reasons for lake expansion. Long-term lake evaporation was stable at 1342 mm/year, indicating that increased temperatures affect the lake surface area by accelerating glacier melting rather than promoting lake evaporation. According to long-term meteorological data, lake evaporation and runoff remained stable prior to 2003, whereas precipitation slowly increased (2.279 mm/year). However, the surface area of Tuosu Lake decreased from 159.9 km<sup>2</sup> in 1973 to 132.3 km<sup>2</sup> in 2003; thus, lake shrinkage was likely caused by the decrease in the groundwater contribution to the lake.

From 2003 to 2011, temperature and precipitation increased by 18.6% and 34.1%, respectively (Table 3), which caused an increase in runoff of 1.21 × 10<sup>8</sup> m3/year and 0.15 × 10<sup>8</sup> m3/year at Delingha station, respectively. Temperature increases during this period mainly promoted lake expansion by accelerating glacial melting, which resulted

in more surface runoff into the lake. This reflects the contribution of glacial meltwater to lake expansion. The rates of temperature, precipitation, and runoff increase were similar between the two periods of 2003–2011 and 2011–present; however, the expansion rate of Tuosu Lake increased rapidly between these periods, from 1.22 km2/year to 3.38 km2/year. Thus, the increase in runoff caused by accelerated glacial melting cannot fully explain the observed lake expansion, which implies that groundwater was an important reason for rapid lake expansion.


**Table 3.** Variation of factors affecting the Tuosu Lake surface area before 2003 and after 2003.

Note: units of runoff, precipitation, evaporation, and lake surface area are 10<sup>8</sup> m3/year, mm/year, mm/year, and km2, respectively.

When glacial meltwater flows through the piedmont plain, some infiltrates the ground as subsurface runoff, which is an important contributor of water to the lake. However, not all of the groundwater contribution to the lakes is derived from glacial meltwater, as groundwater collected near the lakes showed greater isotopic depletion than glacial meltwater in the basin. This indicates the existence of other water sources with more depleted isotopes, which may be related to the mechanism of rapid lake expansion. Several earlier studies have revealed the important role of groundwater in lake expansion and shrinkage [8–11]. For example, the groundwater contribution flux estimated by the radon isotope (222Rn) is 0.55 − 2.49 × 10−<sup>4</sup> m3/(s × m); however, this flux was only measured at a certain time [11]. In future studies, long-term observations of groundwater contributions are required to determine and predict the effects of groundwater on lakes.

### *5.2. Sources of Groundwater Contribution to Lakes*

Deuterium and oxygen-18 isotopes revealed that confined groundwater is characterized by significant isotopic depletion. The origin of depleted confined groundwater in alpine arid basins is controversial. According to 14C-dating of groundwater, it is generally believed that glacial meltwater generated a large amount of recharge after the last glacial period [46,47], or some studies have suggested that meteoric precipitation during glacial and interglacial periods recharged the confined groundwater [48–50]. 14C-dating of groundwater age strictly requires that endogenous CO2 from no other sources is dissolved in the groundwater system, which can dilute the 14C concentration in the groundwater, resulting in overestimation of the significant groundwater age. However, Qaidam Basin is an active geological environment containing multiple crisscrossing fractures. Thus, mantle-derived endogenous CO2 with low 14C activity can migrate upward through active structures such as fault zones and dissolve into the groundwater, leading to significant overestimation of groundwater age [51,52] by tens of thousands of years. As such, tritium was used to identify the groundwater renewal cycle in this study. The half-life of tritium is only 12.3 a, and the background value of tritium in natural groundwater systems is generally less than 1 TU. After 1952, global nuclear explosion tests caused a peak in the atmospheric tritium concentration. Therefore, groundwater systems with tritium values >5 TU are considered to have a groundwater renewal cycle of several decades. The confined groundwater samples in this study exhibited high tritium concentrations of between 8.5 TU and 15.6 TU, which demonstrates that confined groundwater is rapidly circulated and recharged by modern water since the global nuclear explosion tests. Therefore, confined groundwater in the

study area may be recharged by water sources with more depleted isotopic signatures from other areas.

Previous research [12] has revealed an enormous amount of missing water in Qiangtang Basin, with a leakage water volume of up to 540 × 10<sup>8</sup> m3/year, which is related to tectonic activity such as earthquakes. Leakage occurs in six major rift valleys in the southern part of Qiangtang Basin and is transported to other basins by underground runoff. Although the drainage area was not identified in previous literature, this leakage was likely discharged to surrounding areas at lower elevations, causing groundwater levels to rise and lakes to expand. Delingha, Qinghai Lake, Taklamakan Desert, and Hexi Corridor, located on the northern edge of the Tibetan Plateau, have a relatively low altitude and an active geological environment with frequent earthquakes, providing suitable conditions for the remote discharge of groundwater. Moreover, earthquakes are known to increase groundwater discharge [53–55]. In 2003, groundwater release induced by earthquakes was observed in Qinghai Lake (https://hydroweb.theia-land.fr/hydroweb, accessed on 27 November 2021), Taklimakan Desert [43], and the Hexi corridor [44], resulting in the emergence of new lakes, the expansion of existing lakes, and an increase in groundwater levels. An earthquake with a magnitude of 6.1 can affect areas as far as 80 km from the earthquake source [56]. The 2003 Ms 6.4 earthquake that occurred in the northwestern part of the study area (Figure 1) was only 57 km away from Tuosu Lake; thus, it very likely led to an increase in groundwater contribution to the lake. The endorheic lakes in the study area represent places of convergence for surface water and groundwater, and stable isotope analysis showed that groundwater is an important contributor to the lakes. Outflow from the lake to groundwater is limited, which in turn supports that lake evaporation is the main output of lake water balance. The distinctly depleted H and O isotopes in the confined groundwater indicate remote discharge from a high-altitude water source with a more depleted isotopic signature. Therefore, it is speculated that either the 2003 Ms 6.4 earthquake in the northwest of Delingha or the 2001 Ms 8.1 earthquake in the Kunlun Mountains were possible mechanisms for expansion of the lakes in the study area. Earthquakes enhance crustal permeability and keep fractures open [43,44], which promotes the groundwater contribution to lakes and in turn causes rapid lake expansion.

### *5.3. Uncertainty in Lake Evaporation Calculations*

Uncertainty in the calculated evaporation values is derived from the lack of solar radiation monitoring data at the Delingha station, which was replaced by data from the nearby Golmud station in this study. Because the intensity of solar radiation is mainly related to latitude, and the latitude difference between Delingha and Golmud stations is only approximately 1◦, the uncertainty caused by solar radiation data can be ignored. Uncertainty also originates from the lake surface temperature simulated by the improved lake water temperature model (air2water), with a deviation of ±0.55 ◦C, which also has a limited impact on the calculation results. Combined with remote sensing and meteorological data, evaporation from Tuosu Lake is 1333 mm/year [57], which only differs by 0.7% from the value calculated in this study. Earlier studies have reported substantial variation in evaporation between lakes, even if the lake surface areas are similar. For example, Laguo and Yang Lakes have only a 1.3% difference in surface area but a 24.4% difference in lake evaporation [57]. The improved Penman–Monteith model considers the effect of lake surface area and depth on evaporation, leading to more accurate calculation results.

### *5.4. Mitigating the Environmental Effects of Lake Expansion*

Lake expansion likely affects the groundwater runoff process, leading to geological and environmental problems. For example, Gahai Lake is located at the southeastern edge of the alluvial fan and has no inflow from surface runoff. Water from the Bayin River seeps into groundwater in the middle reaches, with weak groundwater runoff in the southeast direction being one of the sources of Gahai Lake. The Gahai irrigation area (D) is located in this groundwater flow path (Figure 8). The geological formation in front of the alluvial

fan is mainly composed of fine-grained sediment, which has a strong water-blocking effect. The water level of Gahai Lake has been rising continuously since 2000, which slows down the groundwater discharge rate and enhances groundwater level rises in the surrounding area. After 2006, the groundwater level increased at a rate of 0.5 m/year. In 2012, the groundwater level came close to the surface and overflowed, rising as high as 11 m in some areas [58]. This directly led to problems such as foundation collapses, and soil salinization occurred in the vicinity of Gahai Lake, which threatened the lives and livelihoods of nearby residents. Therefore, measures should be taken to mitigate the effects of lake expansion in the study area.

**Figure 8.** The distribution of irrigation area in Delingha. A, B, C, D, F represents Huaitoutala, Gebi, Delingha, Gahai, and Zelinggou irrigation area, respectively. The blue arrow represents the direction of weak groundwater runoff after the Bayan River leakage.

Furthermore, when the water level of Keluke Lake rises, the water can be discharged to Tuosu Lake through the Lianshui River. The Huaitoutala irrigation area (A) and nearby villages are located northwest of Keluke Lake (Figure 8), at higher elevation than both Keluke Lake (by 36 m) and Tuosu Lake (by 48 m). Therefore, the future expansion of Tuosu Lake will have little impact on the nearby irrigation areas and villages but is likely to promote expansion of the marsh in the northeast of Tuosu Lake. Considering the long-term trend of climate warming, a rise in groundwater levels in the vicinity of Gahai Lake is inevitable. Therefore, to ensure the safety of residents and continued operation of the irrigation area, drainage channels can be excavated at the end of the alluvial fan to divert groundwater to the Bayin River. Despite the potential for further increases in the water level of Tuosu Lake after channel excavation, the lack of villages and farmland around the lake makes this an appropriate managemen<sup>t</sup> solution.
