1 INTRODUCTION

The Qinghai-Tibet Plateau (QTP), known as the "Third Pole" or "Asian Water Tower", contains half of China's total area of lakes (Qiu, 2008; Yao et al., 2019; Immerzeel et al., 2020; Wang et al., 2021). More than 1 600 lakes (>1 km²) are distributed on the QTP, with a total lake area of more than 50 000 km² in 2018 (Luo et al., 2021). Amongst these lakes, more than 300 salt lakes are distributed in the Qaidam Basin (QB) and Qiangtang region on the northerncentral QTP (Wang et al., 2016). Therefore, it is also a special region of coexisting lakes and salt lakes, and dynamic changes that occur between them. Given the distinctive geographic location and climatic characteristics of the QTP, the water level, surface area and water storage of lakes across this plateau are extremely sensitive to climate change (Lei et al., 2014; Yao et al., 2018; Treichler et al., 2019; Wang et al., 2022). Rapid lake expansion in response to climate warming, increased precipitation, glacier mass loss, and cryosphere degradation (Song et al., 2014; Lei et al., 2017; Wang et al., 2019; Xu et al., 2019; Zhu et al., 2019) has become one of the most remarkable environmental changes across the QTP.

With the lake expansion on the QTP, on the one hand, several cascade lakes in Hoh Xil and typical salt lakes in the northern QB overflowed and profoundly affected plateau infrastructure, including the notable Qinghai-Tibet Railway (Wang et al., 2022). On the other hand, potassium (K) contents and total dissolved solids of salt lakes on the QTP have generally declined in the past three decades (Li et al., 2022). The QTP is a concentrated distribution area of salt lakes in China (Fig. 1), especially in the QB enriched in K, barium (B), lithium (Li), magnesium (Mg) elements in lake water chemistry (Zheng et al., 1989) has been constructed multiple industrialization resource bases of China (Zheng et al., 2014). The prevailing similar climate change and hydrological process (Chen et al., 2010; Dai et al., 2013; Xu and Yang, 2013; Lu, 2014, 2015; Du et al., 2018) have resulted in the remarkable increase in area and number of salt lakes or lakes in the QB (Fig. 1) (Zhang et al., 2010; Wu, 2014; Duan, 2018; Zhang and Liu, 2019). For example, the terminal East and West Taijinar salt lakes (Li-rich brine deposits) and Dabuxun Lake (K-rich brine deposits), replenished by the Nalenggele and Golmud Rivers from Kunlun Mountains in the southern QB, rapidly expanded (Fig. 2). A new artificial Ya Lake situated between East and West Taijinar Salt Lakes rapidly emerged over the past decade, taking on a special geomorphic landscape of coexisting lake water and Yardang. Similarly, another artificial lake in the southern Dabuxun Lake formed. These new lakes were generated due to artificial dam constructions for protecting salt lake resources from dilution and from the increasing water volume of rivers. In addition, the expansion of several lakes, such as Xiao Qaidam Lake (B-rich brine deposits) and Toson Lake supplied by Tataling and Bayin Rivers from Qilian Mountains in the northern QB, was also evident and synchronous in the past decade (Fig. 2).

As there are limited infrastructure, inconvenient transportation, and drastic climate change in the QB, it is difficult to set up monitoring equipment and obtain data, resulting in less field monitoring and research on lakes in the QB. Recent advances in lake expansion and lake storage estimation by satellite observations and measurement methods (Song et al., 2013; Yang et al., 2014, 2018; Biskop et al., 2016; Qiao and Zhu, 2019) have improved our understanding of spatiotemporal changes of lake hydrology in the QTP (Li et al., 2019; Zhang et al., 2019). However, majority of these studies have not assessed the combined effects of meteorology and hydrology at the lake-basin scale. Moreover, studies on rapid lake expansions and influencing mechanisms combined with extensive in-situ monitoring and satellite observations in the salt lake areas are scarce. Therefore, we selected lakes in the northern QB to discuss the following issues: (1) the characteristics of water level fluctuation in different lakes at multiple (day-, month-, and year-) time scales; (2) the relationship between lake fluctuation and meteorological factors (temperature, evaporation, and precipitation) based on in-situ monitoring data from two small weather stations and three water level loggers; (3) influencing mechanism of water level fluctuation of lakes at the spatial scale in this basin in combination with reported annual precipitation trend and glacier mass balance on the northern QTP (Yao et al., 2022); and (4) hydrological risk precaution of lakes in the western, central, northern-central, and eastern basins. To monitor the lake hydrological process further, three lakes in the northern QB (i.e., Sugan Lake, Xiao Qaidam Lake, and Toson Lake from west to east) were chosen on the basis of the following reasons: (1) diverse properties of these three lakes, including resource and ecological lakes; (2) relatively small lake size, increased sensitivity to climate response and ease of monitoring water level fluctuation; and (3) less human interference. This work attempted to provide an important scientific basis for understanding influencing mechanism of lake rapid expansion, hydrological risk precaution, and protection of salt lake resources in the present and the way forward.

2 STUDY AREA

The Qaidam Basin (QB) (87°48′E–99°18′E, 34°41′N–39°20′N) is located in the northern QTP and is slightly rhomboid in shape. This basin is surrounded by three large mountain belts, Kunlun Mountains in the south, Qilian Mountains in the northeast and Altyn Mountain in the northwest (Chen et al., 2010). It covers an area of 12.1×10⁴ km², with an elevation of 2 800 m above sea level and a catchment of approximately 17×10⁴ km² (Zhang, 1987). The QB is located in the junction of the midlatitude westerly belt and the East Asian monsoon system. It has a continental arid climate with rare precipitation and large evaporation. The annual precipitation in the surrounding mountainous and the center of the basin are 150–300 mm and less than 50 mm, respectively. Most moisture falls as rain during the summer months. The potential evaporation capacity inside the basin is more than 1 800 mm (Du et al., 2018). The average annual relative humidity is 30%–40%. The average annual temperature in the basin is below 5 ℃ and the daily temperature difference is often about 30 ℃ (Wei et al., 2017). The modern glacier area in the surrounding mountains is 1 693.54 km² (Zhou et al., 2021), and mainly distributed in the Qilian Mountains, the eastern part of Altyn Mountains and the northern slope of Kunlun Mountains (Yang and An, 1986; Lu, 2015).

The QB has developed multiple river systems, and annual runoff of 10 rivers is more than 1.0×108 m³ (Tan, 2014). The main rivers in the basin include Nalenggele River, Wutumeiren River, Golmud River, Tataling River, Haerteng River, Yuqia River, and Bayin River etc. (Fig. 1). The rivers flow through the surface or underground and eventually flow into terminal lakes (Du et al., 2018). Most lakes have developed salt lakes due to strong evaporation and reduced precipitation.

According to the geographical location and hydrologic systems, the terminal salt lakes are mainly distributed in the eastern-central QB, and fed by the rivers from the two major hydrologic systems of the Kunlun and Qilian Mountains (Fig. 1). The southern hydrological catchment is mainly supplied by rivers from the Kunlun Mountains. Yiliping Playa, West and East Taijinar Salt Lakes in the central QB are supplied by the Nalenggele River. These salt lakes are enriched in brine Li resource in China. Qarhan Salt Lake in the eastern QB is supplied mainly by the Golmud River and enriched in K resource in China. All the salt lakes supplied by southern hydrological catchment are resource lakes in the QB. To utilise these river water resources reasonably and protect the Li and K brine resources, some artificial dams have been built in front of the terminal salt lakes in the past 10 years (Fig. 2). The northern hydrological catchment is mainly supplied by rivers from the Qilian Mountains. Sugan Lake in the western part of the northern QB is a brackish lake supplied by the Haerteng River. Mahai Playa includes two salt lakes (Balunmahai and Dezunmahai lakes) and enriched in brine B resources. These two salt lakes are supplied by the Yuqia River. The Da Qaidam Lake is a typical B-rich salt lake in China and is supplied by thermal springs. The Xiao Qaidam Lake in the middle part of the northern QB is supplied by the Tataling River and enriched in brine B resource. Hurleg and Toson lakes in the eastern part of the northern QB are supplied by the Bayin River, these two lakes are connected to each other. Thus, Hurleg Lake is a freshwater lake, while Toson Lake is a terminal salt lake. Based on lake property in the northern QB, Mahai Playa, Da Qaidam Salt Lake, and Xiao Qaidam Lake are resource lakes, whereas Sugan, Hurleg, and Toson lakes are ecological lakes.

3 MATERIAL AND METHOD 3.1 Meteorological data

The observed meteorological data of Xiao Qaidam Lake and Toson lakes (Fig. 1) were obtained from two small weather stations near the lakes, which were set up by Qinghai Institute of Salt Lakes, Chinese Academy of Sciences in June 2021, with model of Davis (VS1024AU) and an observation time resolution of 1 h. Meteorological data in this study include temperature, precipitation, and evaporation. The time sequence ranged from June 2021 to February 2022. The observation accuracy of the 0.1 ℃, 0.01 mm, and 0.01 mm, respectively.

3.2 Lake level data

The observed water level data of Sugan Lake, Xiao Qaidam Lake, and Toson Lake (Fig. 1) were measured by the water level loggers erected in the lakes, which were set up by Qinghai Institute of Salt Lakes, Chinese Academy of Sciences, in April 2021, with model of Onset (MX2001-04) and an observation time resolution of 2 h. The monitoring time ranged from April 2021 to December 2021, April 2021 to April 2022, and May 2021 to December 2021, respectively. The measured accuracy of water level data is 1×10^-16 m.

3.3 Landsat images data

A total of 23 remote sensing images of five lakes (Sugan Lake, Xiao Qaidam Lake, Toson Lake, Ya Lake, and Dabuxun Lake) in the study area were collected through Landsat 7 ETM+ and Landsat 8 OLI/TIRS. Amongst them, 15 images for Sugan, Xiao Qaidam, and Toson lakes, one each from 2017 to 2021, and 4 images for Ya Lake, including 2000, 2010, 2017, and 2021 phases, four images were from Dabuxun Lake, including 1999, 2010, 2017, and 2021 phases. All image data were from U. S. Geological Survey (https://glovis.usgs.gov/), with a resolution of 30 m. All images required that the cloud amount of cloud and thickness of snow should not be greater than 30%, and the months from September to December should be selected as much as possible to extract the image boundaries clearly and ensure temporal consistency of lake observations, which is conducive to the analysis of lake area changes.

3.4 Observational data plotting

Meteorological and water level data were directly recorded by the observation stations. These data have high precision and no abnormal value, and were directly used for lake hydrological analysis. Every serial value of temperature, precipitation, and evaporation was examined to ensure that the same set of observation at the same time interval, and redundant data were eliminated. Meanwhile, in accordance with the changing trend of the water level, temperature level, evaporation intensity, and amount of precipitation, nine days of data were selected during the observation period. Then the chronological changes of temperature, precipitation, evaporation, and water level from April 2021 to April 2022 were plotted on monthly and daily bases. The water level data from April 2021 to April 2022 were plotted to analyze the water level changes of Sugan Lake, Xiao Qaidam Lake, and Toson Lake. Meteorological data from June 2021 to February 2022 were plotted to explore the relationship between meteorological factors and water level fluctuations, and the selected daily data were plotted to conduct detailed observation of daily-scale meteorological factors and water level changes.

3.5 Landsat images preprocessing

Landsat images were processed with ArcGIS 10.8 software, and the geographic coordinate system used was GCS_Krasovsky_1940. The images were mosaicked and clipped on the basis of the range determined by the study area, and the boundaries of the lakes were vectorised. Then the area of the lake was calculated under the area calculation function of ArcGIS software. Finally, the lake boundaries of the same lake in different years were superimposed to analyze the trend of lake area changes.

4 RESULT

Sugan Lake, Xiao Qaidam Lake, and Toson Lake are distributed in Lenghu, Da Qaidam, and Delingha individual sub-basins, respectively, in the northern QB. Water level fluctuation is mainly presented in three aspects: rising time, fluctuation trend, and amplitude on the multiple (day-, month-, and year-) time scales.

4.1 Day-scale lake level fluctuation

On the day-scale, water level fluctuation and amplitude of each lake is variable and diverse during the monitoring periods (from April to November 2021) (Fig. 3). In accordance with monitoring data of nine special days ranging eight or 12 months, water level fluctuation of lakes on the day-scale are reported in Fig. 3. On April 30, May 11, and May 23 (corresponding to spring season) (Fig. 3a), the fluctuation amplitude of Xiao Qaidam Lake is the most evident and the largest (especially from early May to late May) amongst the three lakes, and the fluctuation generally declined and reached 0.1 m. By contrast, the amplitude of fluctuation of Sugan Lake and Toson Lake takes on a relatively small range (less than 0.05 m). The fluctuation trend of Sugan Lake is gradually rising, whereas that of Toson Lake is balanced at the same interval. In addition, some distinctive fluctuation characteristics of three lakes are recorded in Fig. 3a. Based on the three-day records, the high-water level in Sugan Lake stays for 14–16 h every day, whereas that in Xiao Qaidam Lake stays for 2–4 h every day. Meanwhile, some same turning points in Xiao Qaidam Lake and Toson Lake can be observed, although the latter presents a generally small range of fluctuation on the day-scale.

On June 29, July 25, and August 17 (corresponding to summer season), the water level fluctuation of each lake on the day-scale is relatively stable. Similarly, the high-water level in Sugan Lake, Xiao Qaidam Lake, and Toson Lake stays for 14–18 h based on the three-day records. In addition, some differences in these three lakes can be observed. The amplitude of fluctuation of Xiao Qaidam Lake is the largest and reaches about 0.3 m, followed by that of the Sugan Lake, which reaches up to 0.2 m. The water level of Xiao Qaidam Lake and Toson Lake is continuously ascended, whereas that of Sugan Lake declined. The rising water level of Xiao Qaidam Lake and Toson Lake is followed by increasing precipitation of lake area.

On September 14, October 19, and November 26 (corresponding to autumn season), the water level fluctuation of each lake on the day-scale is more stable than those in the former two seasons. Some detailed records are as follows: firstly, the amplitude of fluctuation of three lakes are small (less than 0.1 m). The water level of three lakes during late November is unchangeable, due to the freezing of lakes on the QTP; secondly, the high-water level in Sugan Lake, Xiao Qaidam Lake, and Toson Lake stays for 18–20 h based on the three-day records; lastly, the water level of Xiao Qaidam Lake is generally decreased, whereas that of Sugan Lake and Toson Lake is almost paralleled and unchanged.

4.2 Month-scale lake level fluctuation

On the month-scale, hydrological process and characteristics of water level fluctuations of these three lakes are observed more clearly than those on the day-scale due to a continuous smooth change trend of the data points before and after. The water level fluctuations of three lakes during the monitoring period (Sugan Lake and Toson Lake from April to December 2021, Xiao Qaidam Lake from April 2021 to April 2022) are obviously diverse at different month intervals (Fig. 4). In general, water level fluctuation of Sugan Lake showed a stable, slow, and gradual rise-fall-rise process from April to December. Interestingly, Xiao Qaidam Lake and Toson Lake shared the similar lake fluctuations (Fig. 4) with a water level rapidly, steeply, and suddenly rising from late July to September, and showing a downwardupward-downward trend with a relatively large fluctuation amplitude.

In detail, the water level rise of Sugan Lake began from April 28 to June 28 and then decreased steadily until early October (up to the minimum point), and then the trend rose again until early December. The amplitude of fluctuation of this lake was small with no more than 0.2 m and the rising time occurred in late April during the monitoring period. By contrast, Xiao Qaidam Lake fell continuously from late April to early June, and then rose steeply from June 14 to August 22. During the rising period, water level fluctuation of this lake was drastic and rapid, and took on two steep-rapid rising and one relatively slow rising. This hydrological change of Xiao Qaidam Lake in this period kept a fresh remembrance. Afterwards, water level of Xiao Qaidam Lake dropped gradually until late March of the next year, the overall amplitude of water level fluctuation was about 0.51 m. That of the Toson Lake was relatively stable and slowly dropped from early May to midJuly, and then took a rapid and steep rise from July 22 to August 4, before declining until October 1 with overall amplitude of fluctuation of up to about 0.31 m. Afterwards, the water level of Toson Lake gradually rose until early December.

4.3 Year-scale lake level fluctuation

The results of lake area change of five lakes in the northern part (Sugan Lake, Xiao Qaidam Lake, Toson Lake) and southern part (Ya Lake and Dabuxun Lake) of the QB based on annual-scale remote sensing images during the past two decades are shown in Table 1. In the past five years (2017– 2021), the area of lakes in the northern QB showed an overall expansion trend (Fig. 2e–g). Amongst them, the change rate (23.44%) and the average change rate (5.41%) of Sugan Lake in the past five years in the western part of the basin are considerably higher than those of Xiao Qaidam Lake (22.43% and 5.19%) and Toson Lake (15.97% and 3.77%) in the central and eastern part of the basin. Amongst the three lakes, Xiao Qaidam Lake expanded the most in 2018, and the lake surface expanded by 18.70% compared with the previous year. In 2019, the lake area reached the maximum value (133.35 km²), and the lake area shrank slightly since then (Fig. 2; Table 1).

In the past five years, the lake area and interannual expansion rate of Sugan Lake have gradually increased from 115.40 km² and 1.91% (2018) to 142.45 km² and 8.20% (2021), showing an increasing trend. The area of Toson Lake expanded from 159.53 km² (2017) to 185.01 km² (2021) year by year, but its annual expansion rate is gradually slowing down (Table 1). In the past 23 years (1999– 2021), the areas of Ya Lake and Dabuxun Lake in the southern QB also showed an overall expansion trend (Table 1). In the past 22 years (2000–2021), the lake area of Ya Lake has increased from 5.67 km² to 394.95 km², an expansion of nearly 70 times; in the past 10 years (2010–2021), the lake area has increased from 86.09 km² to 394.95 km², an expansion of nearly five times, the largest expansion amongst all lakes. The lake area of Dabuxun Lake expanded from 223.98 km² (1999) to 332.20 km² (2021), and the lake area showed a maximum (397.75 km²) and a minimum (213.72 km²) in 2010 and 2017, respectively. These results exhibit the characteristics of fluctuating growth.

5 DISCUSSION 5.1 Relationship between lake fluctuation and meteorological factors in the northern QB

The observation results of water level fluctuation of lakes are diverse on the day-, month-, and yearscales (Figs. 2–4). On the day-scale, no meteorological data were obtained in the spring season (Fig. 3a), water level rising of Xiao Qaidam Lake and Toson Lake is very much correlated with the beginning of precipitation from recording data in the summer and autumn seasons (Fig. 3b–c). The temperature and evaporation in the lake area should not be the main controlling factors for lake level rising. Several day records on July 25, September 14, and October 19 indicate that high temperature and strong evaporation coexisted with precipitation intervals. On the one hand, the precipitation much produces the lake level rising, although lag interval of several hours of lake rising occurred in these monitoring time. On the other hand, the water level of lakes is continuously rising along with strong evaporation and high temperature for 12–14 h every day. If no precipitation happened in the monitoring hours, strong evaporation produces a descending lake level state. These comparisons show that high precipitation intervals can induce rapid lake level rising in the subsequent hours.

On the month-scale, the difference of water level changes of Xiao Qaidam Lake and Toson Lake can be observed from August to November. This is resulting from artificial interception of water resource in Hurleg Lake because Toson Lake and Hurleg Lake are connected to each other, and Hurleg Lake is a terminal salt lake supplied by Bayin River. If artificial regulation in Hurleg Lake is excluded, the temperature and evaporation changes seem generally to be consistent with water level declining of Xiao Qaidam Lake and Toson Lake (Fig. 5). Meanwhile, the similar rising ranges of these two lakes also occurred from early June to late August (Fig. 5). The overall rising trend and fluctuating pattern of these two lakes are synchronous during this period. Although the lake level rising of Xiao Qaidam Lake and Toson Lake and temperature/evaporation of lake area occurred during the same period, the temperature or evaporation of two lake regions is relatively stable, whereas water level rising of these two lakes is rapid and drastic. In addition, the strongest evaporation corresponds to the highest water level of Xiao Qaidam Lake. These comparisons indicate that high precipitation of lake area is correlated with rapid rising of lake level from July to September (Fig. 5).

On the year-scale, lake expansion was found in the northern and southern parts of the QB (Fig. 2). We cannot assess the relationship between meteorological factors (temperature, precipitation, and evaporation) of lake area and annual lake fluctuation in the QB because monitoring data are lacking. Recent observation data of precipitation during 1980–2018 on the QTP reports that the catchment (eastern Kunlun Mountains and Qilian Mountains) of terminal lakes or salt lakes shows a considerable increase in annual precipitation rate (Fig. 6a).

Therefore, the results and comparisons of lake fluctuation and meteorological data of day-, monthand year-scale records show that high precipitation of lake area is closely correlated with water level rising of Xiao Qaidam Lake and Toson Lake in the eastern-central part of northern QB. No meteorological data was obtained in the Sugan Lake in the northwestern part of northern QB, but water level fluctuation pattern of Sugan Lake is asynchronous with that of Xiao Qaidam Lake and Toson Lake (Fig. 4). This comparison shows that temperature or evaporation might be correlated with lake level rising of Sugan Lake.

5.2 Influence mechanisms of lake fluctuation and comparisons in the QB

Lake fluctuation characteristics of Sugan Lake, Xiao Qaidam Lake, and Toson Lake distributed in different parts (western, central, and eastern part) of the northern QB indicate that the influencing mechanisms on them are dramatically diverse. The fluctuation pattern of water level of Xiao Qaidam Lake and Toson Lake are similar, including rapidsteep rising and smooth declining pattern, high water level intervals (from July to September), and frequent fluctuation during the rising period. However, this fluctuation pattern of water level of these two lakes is obviously asynchronous and different with that of Sugan Lake, including smooth rising and declining trend, early high lake level intervals (from April to late June).

The rapid rising of the water level of Xiao Qaidam Lake and Toson Lake occurs almost at the same period with high regional precipitation in the study area. The influence of temperature on lakes is indirect, i.e., the rise in temperature reduces lake water volume by increasing lake surface evaporation and reducing inflow runoff. However, during the warming period (from June to July), the water levels of Xiao Qaidam Lake (from June 14 to August 22) and Toson Lake (from May 10 to July 22) were still rising for a long period. Therefore, temperature is not the major factor in the fluctuation variation of these two lakes. The evaporation of Xiao Qaidam Lake and Toson Lake stayed at a high level during the entire period, nevertheless, the water levels of these two lakes only dropped remarkably after midAugust. Therefore, evaporation is not the main factor for fluctuation changes of Xiao Qaidam Lake and Toson Lake. Correspondingly, precipitation can directly recharge the lake surface or indirectly recharge the lake by replenishing the mountain runoff. The monitoring data shows that precipitation in Xiao Qaidam Lake and Toson Lake was concentrated in July to August, with a total precipitation about 38.81 and 43.6 mm, respectively. After high precipitation intervals, water level of these two lakes followed and took on a rapid-steep rising by about 0.35 and 0.16 m (Fig. 4). In addition, no more glaciers from remote sensing images were reserved in the catchments of Tataling and Bayin rivers in the northern QB. Therefore, precipitation is the main influencing factor on water level rising of Xiao Qaidam Lake and Toson Lake. Interestingly, this fluctuation pattern and influencing factor (precipitation) on water level change of these two lakes are consistent with those of Qinghai Lake (Bai et al., 2019; Du et al., 2020) (Fig. 7). Paleoclimatic records of Qinghai Lake reported that warm-humid climate of this region is controlled by Asian summer monsoon (An et al., 2012). The strong Asian summer monsoon provides much precipitation during the warm period. Although the monitoring data of Xiao Qaidam Lake and Toson Lake in the central-eastern part of northern QB present a similar mirror with the observation records of Qinghai Lake on the annual scale (Fig. 7), no robust paleoclimatic records in the QB has interpreted an influence of Asian summer monsoon on the eastern QB and provided much precipitation in the study area. A study on changes of the transitional climate zone in East Asia indicated that the eastern QB and their corresponding river catchments are located in a transitional zone of arid or humid domain at different time-scales (Wang et al., 2017). This work provided an important reference that much precipitation or humid climate driven by strong Asian summer monsoon occurred in the eastern QB in the geological records. Therefore, we emphasized that Asian summer monsoon or hydrological circulation of lake-basin scale controlled water level fluctuation and fluctuation pattern of Xiao Qaidam Lake and Toson Lake in the central-eastern QB.

By contrast, Sugan Lake is mainly supplied by the perennial Haerteng River. The upper reach of this river has 8.94 km³ of glacier reserves (Han, 2019). From late April, the glacier-snow in the catchment of Haerteng River has been melting slowly with the increasing temperature, which results into increasing river runoff and stably rising water level of Sugan Lake. This interpretation has been demonstrated from news reports in 2021 that water level of Sugan Lake rose in late April, even overflowed on the northwest bank and submerged the equipment of enterprises. According to synchronously rising time and increasing temperature in late April, gradual rising trend (Fig. 4) and snow/glacier developments in the upper reach of Haerteng River (Fig. 6a), we conclude that glacier and snow melting associated with increasing temperature is responsible for water level rising of Sugan Lake.

The influencing mechanisms of lake fluctuation in the northern QB have been demonstrated by a linear trend of increasing precipitation rate and mass balance of glaciers (Yuzhufeng) during 1980–2018 on the northern QTP (including the QB) (Fig. 6a–b) (Yao et al., 2022). Increased precipitation rate and reduced glacier development were observed in the eastern part of the northern QB; by contrast, reduced precipitation rate and further glacier development in the river catchments in the western part of the northern QB were recorded. Based on the above discussion, the fluctuation characteristics and pattern of lake level in the northern QB are divided into eastern and western parts. This conclusion is essential for the management of water resources and protection of salt lake resources in the arid area.

5.3 Hydrological risk precaution and utilization of water resource in the QB

On the basis of the increasing precipitation rate and glacier mass balance on the QTP, the precipitation in the eastern part is higher than that in the western part of the QB (Yao et al., 2022) (Fig. 6a). Therefore, theoretically, the expansion amplitude/rate of water surface of lakes in the eastern part should be greater than those of the western part of the basin. This inference is inconsistent with the actual observation that the expansion amplitude and rate of Toson Lake in the eastern part are smaller than those of Sugan Lake in the west part (Fig. 8; Table 1). This result is mainly due to their differences in the topography of lakes basin and influencing factors of water level fluctuation amongst the three lakes in the northern QB. The lakes selected in this study are all terminal lakes that developed at the end of each river system, the lake basins are mostly broad "shoal-shaped", and the shapes of the lake basins are basically similar. However, the connection between Toson Lake and its upstream Hurleg Lake is equivalent to doubling the water storage area of the lake, resulting in a reduction in the amplitude and rate of its lake surface expansion. In addition, the lake area of Sugan Lake in the western part of the QB is mainly controlled by the recharge of glacial meltwater under the influence of increasing temperature, whereas the lake areas of Xiao Qaidam Lake and Toson Lake in the eastern part of the basin are affected by the regional precipitation recharge. Therefore, the time and amplitude of lake water level fluctuations in the eastern and western parts of the basin are asynchronous and out of phase (Figs. 4–5).

In addition to the asynchronous characteristics in the east-west direction, the lake area change in the QB also has remarkable differences in the northsouth direction (Fig. 6a; Table 1). The amplitude and rate of lake surface changes in Ya Lake and Dabuxun Lake recharged by the Kunlun Mountains water system in the southern QB were much higher than those in the northern QB recharged by the Qilian Mountains water system (Fig. 8). This result is mainly due to the establishment of artificial dams by salt lakes companies to intervene in the distribution of water resources, which has largely accelerated the water surface expansion of Ya Lake and Dabuxun Lake (Mao et al., 2018; Tian et al., 2018). In addition, according to the existing observational data, the Kunlun Mountains in the southern part of the QB have more precipitation than the Qilian Mountains in the north (Yao et al., 2022). Given that the average elevation of the Kunlun Mountains is considerably higher than that of the Qilian Mountains (Fig. 1), atmospheric precipitation is mainly distributed in high-altitude mountainous areas. Part of the atmospheric precipitation in this area is accumulated and stored in the form of ice and snow, and the other part is directly supplied to the downstream lakes by runoff. During periods of high temperature, both atmospheric precipitation and glacial meltwater in the upper high-altitude mountains superimposed to replenish the terminal lakes in the southern QB. Therefore, the total material input rate of the southern lakes was higher than that of the northern lakes, resulting in higher expansion amplitudes or rates of Ya Lake and Dabuxun Lake compared with that of the lakes in the northern QB.

The fluctuation amplitudes, timing, high/low lake level, expansion rate, and influencing factors of lakes surface of different lakes in the QB vary. Therefore, special water resource management and warning programs of hydrogeological disaster must be formulated for the catchments of different lakes. The Qaidam Basin was divided into four regions in this study (Ⅰ, Ⅱ, Ⅲ, Ⅵ, separated by green dotted lines). In the western part of the QB, the salt lakes are rich in mineral resources and the vegetation coverage is extremely low. Therefore, it is mainly based on salt lakes industrial activities, supplemented by a small amount of animal husbandry, and almost no agricultural activities. The results show that the lake surface expansion in the western region is mainly controlled by the recharge of glacial meltwater under the influence of increasing temperature. Therefore, the reinforcement and maintenance of hydrogeological disaster prevention facilities should be conducted before the temperature rises (that is, in April) in this area (Fig. 9), to reduce the damage to industry and animal husbandry caused by sudden changes in hydrological conditions. In addition to numerous salt-lake industries in the central area of the basin, agricultural and animal husbandry activities occur. Observations of lake water surface fluctuations in the southern-central parts of the basin show that the hydrological changes in this region are affected by atmospheric precipitation and glacial meltwater. Therefore, the prevention and management of hydrogeological disasters must comprehensively consider the potential disaster risks caused by atmospheric precipitation and glacial meltwater under the influence of warming, and the prevention of hydrogeological disasters should be strengthened from June to September every year. The water surface expansion of lakes in the northern-central part of the basin is mainly affected by atmospheric precipitation. Therefore, the comprehensive management of water resources and the prevention of hydrogeological disaster risks should be strengthened during the period from July to September. The eastern part of the QB is mainly dominated by agricultural and animal husbandry activities, and its hydrological changes are mainly affected by atmospheric precipitation. Therefore, the prevention of hydrogeological disasters and the comprehensive management and utilisation of water resources should be strengthened during the period of concentrated atmospheric precipitation (that is, from June to September).

6 CONCLUSION

On the basis of field observation data and remote sensing images, this study comprehensively analyzes the variation characteristics of lake water level and area in different spatial and temporal resolutions of lakes in the Qaidam Basin, discusses the influencing mechanism of lake hydrological changes, and proposes effective suggestions on the prevention of hydrogeological hazards and efficient utilization of water resources in lakes in arid areas. The main conclusions are drawn as follows:

1) Regardless of the east-west direction or in the north-south direction, the water level fluctuations of the lakes in the Qaidam Basin show asynchronous characteristics in terms of timing, amplitude, and rate.

2) The fluctuation of lake water level in the Qaidam Basin is mainly controlled by the recharge of glacial meltwater under the influence of temperature in the west, and of atmospheric precipitation in the northern-central and eastern parts. By contrast, the south-central regions are simultaneously affected by the meltwater of glaciers in the high-altitude mountainous areas and atmospheric precipitation.

3) The formulation of the hydrogeological disaster prevention strategy and the high-efficiency utilization plan of water resources in the Qaidam Basin should be based on the factors affecting the water level fluctuation of lakes in different regions according to local conditions.

7 DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.

8 ACKNOWLEDGMENT

We would like to thank Dr. Haicheng WEI, Yongsheng DU, and Zhanjie QIN from the Qinghai Institute of Salt Lakes, Chinese Academy of Scienses, for their helpful suggestions and field contributions about arrangement of monitoring requirements and scientific project. Similarly, we also thank the reviewers and editor-in-chief for their constructive comments to improve the paper.