Most lakes on the Tibetan plateau are endorheic, which means that lake level changes directly reflect the variation of water balance in the catchment and the water cycling process could be recorded by the water chemistry of the lake. Obviously, researches of hydro-chemistry of the lake can be significant, as it reflects the environmental conditions of salt material accumulation or dilution in the lake and the catchment environment around the lake. Therefore, the chemical composition of the lake water provides reliable evidences for modern sedimentary processes in the lake. Many investigations have been performed to study the lake hydrochemistry in the lake e.g.  -  . Most previous work has clearly focused on the spatial distribution of the water chemistry, the geochemical whereas systematic clarification of hydro-chemical characteristics from temporal and spatial scales is reported less frequently.
As the largest inland saline lake in China, Lake Qinghai is very sensitive to climatic fluctuations and has become a hot topic of research worldwide  -  . Influenced by climate warming, the water level has fallen dramatically and reduced in area since the beginning of the twentieth century    , which has led to a series of ecological/environmental problems such as increasing salinity, deterioration of water quality, loss of grazing grassland around the lake, and desertification in the beach area (Photo 1). Therefore, research on the hydrochemistry of
Photo 1. (a) A satellite image of Lake Qinghai, which shows lake shrinkage status in 2005 and some separation of small lakes from the mother lake (northeast part); (b) Large sand dunes have been formed on the northeast bank area by northwest wind blowing sand sediments from a dried lake bed.
focusing on temporal and spatial variations of hydrochemistry, and predicts which could provide fundamental data for the study of lake sedimentary processes.
2. Study Area
Lake Qinghai is situated in the northeastern edge of the Qinghai-Tibetan Plateau, between 36˚32'N - 37˚15'N and 99˚36'E - 100˚47'E (Figure 1). Its catchment area is approximately
Climate in the catchment is cold and semi-arid, influenced by three air masses: East Asian monsoon, Indian monsoon and westerly atmospheric flow  . The average annual temperature ranges from −1.3 to 0.5˚C and the mean annual precipitation varies from 250 to
Figure 1. Map showing the location of Lake Qinghai and the distribution of sampling sites in August of 2008.
65% occurring between June and September  . The vegetation is dominated by grasses, montane shrub and alpine meadow in the catchment. There are virtually no forests in the basin, only a limited area of desert, and some irrigated farmland to the north and northeast of the lake. The pH value of the water ranges from 8.9 to 9.5, and the average salinity is approximately
3. Sampling and Analysis
Lake hydrochemistry investigation was performed in the summer of 2008. Twenty-four water samples were collected at 24 different sites that covered the whole lake area (Figure 1, sample numbered Num-01 - Num-24, respectively), including 15 water column samples (at
The concentration of bicarbonate was measured using hydrochloric acid titration in the field. All water samples were filtered through
4. Results and Discussion
4.1. Major Chemical Composition of the Lake Water
Major ions are composed of Na+, K+, Ca2+, Mg2+, , , Cl− and. There are significant variances in the proportion of major ion content at different sites based on salinity and geological background differences. As seen in Table 1 and Figure 2, is the dominant cation in the lake. The concentration ranges from 4284.3 to 4348.4 mg/L with an average of 4308.2 mg/L, which is comprises approximately 80.1% of total cations. Mg2+ concentration varies from 765.105 to 799.0 mg/L, with an average of 790.1 mg/L and comprises approximately 14.7%. The mean concentrations of K+ and Ca2+ are 271.6 and 12.0 mg/L and comprising approximately 5.1% and 0.2% of total cations, respectively. The anions are dominated by Cl−, which has a concentration varying between 6169.9 and 6282.1 mg/L (mean of 6179.3 mg/L), and comprise 60.6% of the total anions. The content of ranges from 2177.3 to 2622.8 mg/L, with mean concentration of 2419.4 mg/L and approximately 23.7% of the total anion content. The average content of the dissolved inorganic carbon (including and) is
Table 1. Major chemical compositions of the surface water in Lake Qinghai (mg/L).
Figure 2. Percentage of the major cations/total cation concentration (a) and the major anions/ total anion concentration (b) for water samples from Lake Qinghai.
1512 mg/L, and comprises approximately 15.7% of total anions. Characteristic coefficients of the chemical composition of the lake water are as follows: the ratio of Na+/K+ is 15.87, Mg2+/K+ is 2.91, Mg2+/Ca2+ is 65.8, Cl−/∑salinity is 0.4, /∑salinity is 0.16, coefficient of sylvite (K+/∑salinity*100) is 1.7 and the coefficient of alkalinity ((+)/∑salinity*100) is 10.3. The salinity ranges from 15292.2 to 16151.6 mg/L with an average of 15585.3 mg/L. According to O.A. Arliekin’s classification (1960)  , the hydrochemistry type of Lake Qinghai is Cl−-Na+, which is a common characteristic of salt lakes.
Concentrations of and K+ in Lake Qinghai are both quite similar to those in ocean water (Figure 3(a)), whereas other ion concentrations, particular Ca2+ are less than those of ocean water. Furthermore, there are some differences in characteristic coefficients between the lake water and ocean water (Figure 3(b)). The value of Mg2+/Ca2+ and the coefficient of alkalinity are both significantly higher in Qinghai Lake than in ocean water, probably implying that abundant calcite and aragonite deposits present in Lake Qinghai and that CaCO3 displays a relative saturation.
4.2. Mechanisms Controlling the Lake Hydrochemistry
To visualize comparison features of chemical compositions of surface water, Gibbs (1970)  proposed the boomerang envelope model to describe the dynamics of chemical compositions of surface water and classified the controlling factors to three types, namely atmospheric precipitation, rock weathering and evaporation/crystallization (Figure 4(a)). Generally, water dominated by Na+ and Cl− mainly originates from rainfall and rock weathering and has a high value of Ca2+ and. The evaporation/ crystallization dominated water is also characterized with high Na+ and Cl− concentrations. The weight ratio of Na+/(Na+ + Ca2+) or Cl−/(Cl− +) situates on the x-axis (Figure 4(b) and Figure 4(c)) and the variation of the total dissolved salts (TDS) lies on the y-axis. For increasing Na+/(Na+ + Ca2+) or Cl−/(Cl− +) with declining TDS and a descending limb, Gibbs suggested that atmospheric precipitation is the most important factor in controlling hydrochemistry. However, for increasing Na+/(Na+ + Ca2+) or Cl−/(Cl− +) with inclining TDS and an ascending limb, Gibbs thought that evaporation/crystallization is the principal control factor. For our study, data of the Lake Qinghai fall within the right-upper Gibbs boomerang envelope model with high ratios of Na+/(Na+ + Ca2+) and Cl−/(Cl− +), indicating that the chemical composition of the lake water are similar to ocean water, dominated by evaporation/crystalli- zation (Figure 4(b) and Figure 4(c)).
In Figure 5, the relationship between Na+ + K+ and Cl−, Ca2+ + Mg2+ and of the Qinghai Lake water has been shown. It is clear that the values of Na+ + K+ and Cl− in the Lake Qinghai water are close to the 1:1 equilibration line , whereas Ca2+ + Mg2+ and deviate from the 1:1 line. In general, Na+-K+ derive from evaporites and weathering products of silicates, while Ca2+-Mg2+ originate from carbonates, evaporates and silicates. is supplied by dissolution weathering of carbonate, whereas Cl− is generally derived from dissolution of evaporites   . Therefore, the two relationship
Figure 3. Comparison of chemical composition: (a) Characteristic coefficients, (b) Differences between Qinghai Lake water and ocean water. Note that data of Lake Qinghai originated from the mean value of the sampling sites in 2008 and data of ocean water refers to the research work by Sun et al. (1991)  .
lines indicate that evaporation play an essential role in controlling the hydrochemistry to a certain extent. It also implies that the human activities have a weak impact on water chemistry in
Figure 4. Plots of the major chemical compositions within the Gibbs boomerang envelope for waters in Lake Qinghai. (a), (b) and (c) show major cations and anions of the lake water in the Gibbs model. Note that TDS is calculated by the sum of sodium, potassium, calcium, magnesium, carbonate, bicarbonate, chloride and sulphate concentrations, and from which half of the bicarbonate concentration is subtracted.
Figure 5. Relationship between Na+ + K+ and Cl− (a), Ca2+ + Mg2+ and (b).
4.3. Spatial Distribution of Chemical Composition of the Lake
4.3.1. Horizontal Distribution of Chemical Composition of Surface Water
Many factors such as lake basin morphology, hydrodynamic characteristics of the lake, catchment conditions around the lake, chemical properties of inflows could probably impact the horizontal distribution of the lake chemical composition. Changes in surface water chemical composition at 24 sampling sites in Lake Qinghai are listed in Table 2. Major ion concentrations and salinity have no obvious spatial variation in the surface
Table 2. Changes in chemical composition of the surface water in Lake Qinghai (mg/L).
water, with a standard deviation (STDEV) of 0.43 - 351.7 mg/L, which may be related to the supply of water systems in the vicinity of Lake Qinghai throughout the year and lake currents generated from frequent wind actions.
4.3.2. Vertical Variation in Chemical Composition of the Lake
The variation of chemical composition in vertical water profiles is likely influenced by the lake basin morphology, water depth and hydrodynamic characteristics. Table 3 lists the vertical variation of major ions contents and salinities of 15 water profiles collected in August, 2008. In general, the concentrations of carbonate and bicarbonate anions have more obvious changes in the vertical water profiles than other main ions, with standard deviation (STDEV) of 23.2 - 93.6 and 54.1 - 312.9 mg/L, respectively. Changes of Na+, K+, Cl− and are small and almost equal with depth in every profile. The salinity fluctuations with water depth at different sites of the lake have a STDEV range from 19.8 to 303.8 mg/L in 15 water profiles and can easily be divided into three types of change trends. (1) The salinity of surface layer is relatively high in the water profiles of Num-02, Num-10, Num-14 and Num-21 (Figure 6(a)), which is probably because these profiles are nearby the lakeshore and bay, resulting in strong evaporation and relative difficulty of transporting surface soluble substances to the bottom. Additionally, there is no recharge of freshwater directly into the above sampling sites. (2) The salinity of bottom water is comparatively high in the water profiles of Num-01, Num-03, Num-05, Num-22 and Num-23 (Figure 6(b)). These sites are usually located in the centre parts of the lake with excessive accumulation of soluble salts in the bottom water. Moreover, large river inflow (e.g., Buha R., Haergai R. and Heima R.) would dilute chemical substances of the surface water layer to some extent. (3) The salinity has no obvious trend in vertical direction at other collected sites (Figure 6(c)).
4.4. Temporal Variation of Hydrochemistry
4.4.1. Variation and Impact Factors of Hydrochemistry from 1960 to 2008
The increased salinity was accompanied by an overall downward trend in water level of the Qinghai Lake during the stage from the 1960s to the 2000s (Figure 7(a), Figure 7(b)). The salinity was
Table 3. Vertical variation of chemical composition in Lake Qinghai (mg/L).
Figure 6. Fluctuation trends of salinity with water depth of 15 profiles in Lake Qinghai: (a) the salinity of the surface water layer is relatively high, (b) the salinity of bottom water layer is comparatively high, and (c) the salinity has no obvious trend in vertical direction.
Figure 7. Water level fluctuations of Qinghai Lake from 1960 to 2008 (a); Salinity variations of the Qinghai Lake from 1960s to 2000s (b); Changes in cation (c); Anion concentrations of the Qinghai Lake in 1961-1962, 1985 and 2008 (d). The data sources: Wang and Dou (1998)  ; Lanzhou Branch, Chinese Academy of Science (1994)  ; Yi et al. (2010)  .
However, the lake has increased in the most recent eight years, and has continuously risen nearly
Figure 7(c) and Figure 7(d) show that almost all of the ion concentrations changed as a result of the water level decline of the past 40 years. concentration experienced a large change from 525 mg/L to 1201 mg/L, suggesting that the alkalinity of the lake gradually increased. The amount of K+ and gently increased, whereas Ca2+ and Mg2+ remained unchanged. In addition, the content of Na+ and Cl− had a certain degree of increase (1050 and 904, respectively) from 1961 to 2008. Both Na+ and Cl− had been the dominant cation and anion in the Qinghai Lake during the past 40 years, indicating that the water chemistry had been kept in Cl−-Na+ type, dominated by evaporation/crystallization through the entire time period.
Changes in major ion concentrations, salinity and water level observed are closely related to climate factors including temperature, precipitation, evaporation and wind speed in the Qinghai Lake catchment recorded from 1960 to 2008. Correlations can be observed between the annually lake level change and the mean annual precipitation (Figure 8(a)), the annual mean evaporation (Figure 8(b)), the annual mean temperature (Figure 8(c)) and the annual mean wind speed (Figure 8(d)). Results show that the variation of lake level was highly positively correlated to precipitation and negatively to evaporation, with correlation coefficients of 0.81 and −0.72, respectively. How- ever, correlation coefficients between lake level and temperature, as well as between
Figure 8. Relationship between annual changes of the lake level (ΔH) and the mean annual precipitation (P), mean annual mean evaporation (E), mean annual temperature (T), and mean annual wind speed in the catchment of Lake Qinghai from 1960 to 2008. Note that: ΔH data refers to work by Yi et al. (2010)  ; Data of climate proxies are calculated by meteorological observation data at Gangcha station.
lake level and wind speed are 0.15 and −0.57, respectively. This would imply that the water level is mainly influenced by precipitation and evaporation (namely dry or wet climate). Additionally, wind speed could play an important role in evaporation process. However, the variation of lake level had little association with fluctuation of temperature, which would also exert some influence on the evaporation.
As seen in Table 4, temperature had a remarkable warming trend in the 1990s. It was 0.3˚C - 0.4˚C higher in the 1970s than the 1960s and nearly 1.0˚C higher in the 1990s than the 1960s. In the case of precipitation, obvious fluctuations occurred, with an increasing trend in the 1960s and 1980s and a decreasing trend in the 1970s and 1990s. Precipitation increased gradually in the most recent 10 years at a rate of
4.4.2. Change Trends in Hydrochemistry of Lake Qinghai in the Future
To understand lake hydrological processes in the future under climate warming conditions, a combined model was created to simulate the lake level changes in the future  . The model is composed of the lake level change model (LLCM) and a statistical downscaling model (SDM). The former is a combination of the SWAT Model  , which is designed to simulate hydrological processes in the catchment and the water balance model (WBM). The latter is derived from the output of SRES A2 scheme from the MPI-OM1 GCM model  and in a Grid of ECHAM 5.0  near the Lake Qinghai catchment as the large scale climate information. Climate scenarios for the future 50
Table 4. Inter-annually variations of water level, salinity of the Qinghai Lake and climatic proxies in the catchment.
Data sources: Yi et al. (2010)  ; Meteorological Observation Station at Gangcha; and Lanzhou Branch Chinese Academy of Science (1994)  .
years (2000-2030) in the catchment of Lake Qinghai are obtained using the Statistical Downscaling Model    . Based on established databases from selected data of the meteorology, hydrology, EDM, vegetation, soil features and land use, social characteristics and water demands and so on, the water balance and water level changes of Lake Qinghai in the future were simulated. To facilitate the calculation, the catchment is divided into 62 sub-basin and 155 hydrological response unions (Figure 9). The SWAT model for modeling the hydrological processes in the catchment had been checked by the runoff observation data from 1990 to 2000 with a relation coefficient of 0.95. The lake water balance model had been checked by observational data of lake level changes from 1981 to 2000 with a relation coefficient of 0.92. The final simulation result shows that the lake level will display an upward trend in coming years and will reach approximately
We investigated the hydrochemistry of Lake Qinghai including chemistry and spatial distribution characteristics as well as its evolution. Na+ and Cl− were the dominant cations and anions in the lake water, and hydrochemistry type was Cl−-Na+ with an obvious characteristic of a saline lake. For this study, the Gibbs plot illuminated that the dominant mechanisms responsible for controlling chemical composition of the lake
Figure 9. The catchment of Lake Qinghai (a) and the dividing scheme for sub-hydrological units of Lake Qinghai basin (b). (Liu et al., 2007)  .
Figure 10. Trends of simulated water level changes of Lake Qinghai in the future (Liu et al., 2007)  .
water were evaporation/crystallization. The increase of salinity was accompanied by an overall downward trend in water level of Lake Qinghai from the 1960s to 2000s. This might be attributed to the balance of precipitation and evaporation in the catchment. Based on the simulated model, the lake level will rise gradually in the coming decades with precipitation increasing under climate warming scenarios (Liu et al., 2007). This could be benefit to the ecological environment and biodiversity in the catchment.
Many thanks are given to Qinghai Lake National Nature Reserve and Mr. He Yubang and Zhang Hu for kind help in field investigations. The authors are also grateful to the Ministry of Science and Technology for the program financial support (No. 2006FY 110600).