A note on the lake level variations of Nam Co,south-central Tibetan Plateau from 2005 to 2019

ShiQiao Zhou

1. Key Laboratory of Tibetan Environment Changes and Land Surface Processes, Institute of Tibetan Plateau Research,Chinese Academy of Sciences,Beijing 100101,China

2.University of Chinese Academy of Sciences,Beijing 100049,China

ABSTRACT Tibetan lake levels are sensitive to global change, and their variations have a large impact on the environment, local agri‐culture and animal husbandry practices.While many remote sensing data of Tibetan lake level changes have been report‐ed,few are from in-situ measurements.This note presents the first in-situ lake level time series of the central Tibetan Pla‐teau.Since 2005,daily lake level observations have been performed at Lake Nam Co,one of the largest on the Tibetan Pla‐teau.The interannual lake level variations show an overall increasing trend from 2006 to 2014,a rapid decrease from 2014 to 2017, and a surge from 2017 to 2018.The annual average lake level of the hydrological year (May−April) rose 66 cm from 2006 to 2014, dropped 59 cm from 2014 to 2017, and increased 20 cm from 2017 to 2018, resulting in a net rise of 27 cm or an average rate of about 2 cm per year.Compared to the annual average lake level based on the calendar year,it is better to use the annual average lake level based on the hydrological year to determine the interannual lake level chang‐es.As the lake level was stable in May,it is appropriate to compare May lake levels when examining interannual lake lev‐el changes with fewer data. Overall, remote sensing results agree well with the in-situ lake level observations; however,some significant deviations exist. In the comparable 2006−2009 period, the calendar-year average lake level observed insitu rose by 10−11 cm per year,which is lower than the ICESat result of 18 cm per year.

Keywords:Lake Nam Co;lake level change;in-situ measurement;Tibetan Plateau

1 Introduction

Water-level change of the vast Tibetan lakes has become a major concern in the environmental studies of the Tibetan Plateau in recent years. One reason for concern is the sensitive nature of high-elevation lake responses to global change. The Tibetan Plateau is minimally affected by local human activities, and its lake levels are considered to be a sensitive indicator of climate change.Most Tibetan lakes have no surface outflow, with drainage basins favoring rapid runoff generation and concentration. Increases in precipita‐tion or glacier melt can therefore result in rapidly in‐creasing lake levels.Another reason for concern about lake-level changes is the direct impact on local pas‐tures. During recent decades, many lakes' levels have been rising (Luet al., 2005; Bianet al., 2006), inun‐dating large areas of high-quality grassland and caus‐ing severe damage to local agriculture and animal hus‐bandry practices (Ge and Zonggha, 2005). In addi‐tion, Tibetan lake level changes may explain the gra‐vimetry anomalies derived from GRACE (the Gravity Recovery and Climate Experiment satellite) (Jacobet al.,2012;Zhanget al.,2013).

Over the entire Tibetan Plateau, in-situ lake level observations have only been recorded at Lakes Qing‐hai (Liet al., 2007) and YamzhoYumco (Chuet al.,2012). In the high inner Tibetan Plateau, where most lakes are distributed, in-situ lake level observations are completely unavailable due to the harsh climate and environment, the difficulty in accessing the re‐gion, and the high cost of field work. With the ad‐vancement of remotely-sensed data retrieval and inter‐pretation, more and more studies have been attempt‐ing to monitor or reconstruct lake level changes over the Tibetan Plateau. Previous studies were performed using satellite altimetry data to obtain lake level changes (Zhanget al., 2011b; Phanet al., 2012; Gaoet al., 2013; Wanget al., 2013; Song, 2015a,b; Leiet al.,2017),or imagery data to acquire lake area chang‐es (Liuet al., 2009; Zhanget al., 2011a; Leiet al.,2013;Yanget al.,2017),or using both(and other data products) to further derive lake water storage changes(Crétauxet al., 2011; Kropáčeket al., 2012; Zhangetal., 2013; Wuet al., 2014; Zhanget al., 2017). These investigations contribute much to the knowledge of the global change in the context of local hydrologic variability. However, previously derived lake level changes generally lack validation with ground-based,in-situ observations.

With an area of 2, 017 km2, Lake Nam Co(90°16'E−91°03'E, 30°30'N−30°55'N; Figure 1) is one of the largest and most studied lakes on the Tibetan Plateau. In the summer of 2005, the Nam Co Station for Multisphere Observation and Research was estab‐lished at the southeast shore of the lake by the Insti‐tute of Tibetan Plateau Research, Chinese Academy of Sciences (Figure 1). In-situ lake level observa‐tions have been performed since 2005. Zhouet al.(2013) published observational data for the period 2007−2011. However, these data are discontinuous,missing lake level changes during the lake freezing periods. Moreover, lake levels are incomparable be‐tween annual observation periods. This note reports the full temporal series of lake level variations from 2005 to 2019.

Figure 1 Locations of the Nam Co station and two lake level gauges,G1 and G2

2 Methods

Two gauges, G1 and G2, were set up in the east part of the lake to measure changes in the lake level(Figure 1). Daily lake level change is manually ob‐served at Gauge G1 in the shore side water near the Nam Co station (Figure 1). Due to lake freezing in winter, annual observations at G1 start in mid-April to mid-May and proceed until early January when ex‐tensive ice coverage forms. At the beginning of the annual observations, a new lake level gauge is in‐stalled as the old one is carried away by lake ice.This annual gauge installment renders the raw re‐corded lake level values incomparable between annu‐al observation periods. Hence, lake level change val‐ues during lake freezing periods are required to cali‐brate the raw data and derive a full multi-year time series of lake level variations. To close this gap, a second gauge (G2) was established at a shore site of Zhaxi Peninsula, around 11 km from G1 (Figure 1).This gauge was linked to the bedrock features to en‐sure reliable data acquisition. Lake levels at G2 are manually recorded at the beginning and end of the annual observation period and several times within that period.Due to lake freezing,daily data is not ob‐tained at G2, and only a total lake level change is available during the lake freezing period from early January to mid-April or mid-May. The datasets from both G1 and G2 were compared for lake level chang‐es during the same period, which showed very good agreements.

3 Temporal lake level variations

Figure 2 presents the temporal lake level variations from July 2005 to May 2019 using the data from both G1 and G2.Annual lake level increase generally begins in early-June to early-July, and proceeds until late-Sep‐tember to mid-October when a maximum level is reached. This period largely corresponds to the south Asian monsoon season, which brings significant pre‐cipitation.After this time,lake level drops due to insig‐nificant winter precipitation, evaporation and subsur‐face water seepage until the lake ice melts in April(Zhouet al., 2013). As the monsoon season begins in May in most years (Baiet al., 2001), the hydrological year is determined to be from May 1 to April 30. The annual average lake levels of each hydrological year are shown in Figure 2, assuming a linear lake level de‐crease during the lake freezing period in which no dai‐ly data is available. The interannual lake level varia‐tions show an overall rising trend from 2006 to 2014,a rapid decrease from 2014 to 2017, and an increase from 2017 to 2018. The hydrological year-based annu‐al average lake level rose 66 cm from 2006 to 2014 and declined 59 cm from 2014 to 2017, and again in‐creased 20 cm from 2017 to 2018, resulting in a net rise of 27 cm or an average rate of about 2 cm per year.Among the full thirteen hydrological years from 2006 to 2018 are four separate periods with interannual lake level rise (2007 −2008, 2009 −2011, 2012 −2014 and 2017 −2018) and four other periods with interannual lake level decline (slight drop in 2006 −2007, 2008 −2009 and 2011−2012,and large fall in 2014−2017).

Figure 2 Temporal lake level variations of Nam Co by in-situ measurements from July 2005 to May 2019(black)and satellite radar altimetry from May 2005 to February 2014(blue;Crétaux et al.,2011).The red dots are the hydrological year-based annual average lake levels.The green dots are the calendar year-based annual average lake levels.The red triangles are the in-situ measured lake levels on May 1 of each year

The calendar year-based annual average lake lev‐els are also depicted in Figure 2 for comparison with those of the hydrological year-based. It is seen that while both datasets agree well with each other as a whole, significant differences exist in the years with large lake level changes, such as in 2008. The aver‐age lake level of the 2008 hydrological year is 17 cm higher than that of the 2008 calendar year.The differ‐ence between the two annual averages would result in differing interannual lake level change values. It could even cause opposite interannual lake level chang‐ing trends in the two data series, as evidenced by the data in 2008 −2010. In this period, the hydrological year-based annual average lake level decreased be‐tween 2008−2009 and increased between 2009−2010.In contrast, the calendar year-based exhibits a rise be‐tween 2008−2009 and a drop between 2009−2010.A similar situation exists in 2017−2018(Figure 2).There‐fore, the hydrological year-based annual average lake levels can better capture the multi-year lake level variations.

When the daily lake level data series and its an‐nual averages are not available, the interannual lake level changes may be obtained by comparing lake levels from the same time of different years. As the lake level variation varies from month to month(Fig‐ure 2), data from the month with the most stable lake levels or minimal variations would be desirable for the determination of the interannual lake level changes. Table 1 presents the monthly lake level variation ranges during the observation periods since 2005. It is seen that at the onset of the South Asian monsoons in May,lake levels are relatively sta‐ble compared to those in other months, with varia‐tions no more than 5 cm. Therefore, it is more appro‐priate to compare May lake levels when examining in‐terannual lake level changes with limited data.The da‐ta in Figure 2 also confirm this observation. In the fig‐ure,the lake level on May 1 of each year is highlighted to compare with the annual averages based on the hy‐drological and calendar years. It is seen that the link line of the May 1 data points, with a time lag, is roughly parallel to that of the annual averages based on the hydrological year, except for the segment be‐tween 2015−2016.This indicates that the May 1 lake level changes primarily reflected the annual average changes based on the hydrological year in most years.

Table 1 Monthly lake level variation ranges(cm)during the observation periods.Note that only partial data is available inJuly 2005 and Decembers 2009−2010,2012−2013 and 2015−2017

4 Comparison with remote sensing results

The average lake level rise rate of Nam Co based on remote sensing data was reported to be 25 cm per year (2003 −2009; Zhanget al., 2011b), 31 cm per year (2000−2009; Kropáčeket al., 2012), 23 cm per year (2003−2009; Phanet al., 2012), 18 cm per year(2003 −2009; Wanget al., 2013), 25.8 cm per year(2003−2009, Songet al., 2014), and 33 cm per year(2003 −2009; Wuet al., 2014). Differences are seen between these results, although all of them are based on the same highly accurate ICESat data acquired by laser altimetry and linear regression, except for the one by Kropáčeket al. (2012), which used multi-al‐timetry data. Comparisons can be made between the remote sensing results and gauge observations for 2006−2009.The net lake level rise by gauge observa‐tion in 2006−2009 was 33 cm for the annual averages based on the calendar year (32 cm for the hydrologi‐cal year)and 31 cm for May 1 lake levels,correspond‐ing to respective average rates of 11 cm and 10 cm per year.These rates are much lower than the above remote sensing results of 2003−2009.Nevertheless,when look‐ing at the ICESat data of Nam Co (Kropáčeket al.,2012),the net lake level rise in 2006−2009 is shown to be much lower than in 2003−2006,with the former be‐ing 54 cm and the later 123 cm. This net increase of 54 cm in 2006−2009 corresponds to an average rate of 18 cm per year,which is still higher than the gauge-ob‐served average rate of 11 cm or 10 cm per year.

Apart from the above, a time series of lake level variations of Nam Co has also been retrieved and has been updated from satellite radar altimetry (Crétauxet al., 2011,http://www.legos.obs-mip.fr/soa/hydrologie/hydroweb/). Some of those results (2005 −2014) are shown in Figure 2 for comparison with the gauge mea‐surements. It is seen from Figure 2 that both time se‐ries overall agree well with each other, indicating good retrieval of the lake level fluctuations by radar altime‐try. Large deviations, however, are also apparent at oc‐casional time points or during occasional periods.

It is noted that most of the gauge-observed lake level data from 2005−2009 have been used to validate the study results of remote sensing by Zhanget al.(2011b; 2019), Kropáčeket al.(2012) and Wanget al.(2013). More recently, the gauge data from 2005 −2012 was also used to evaluate the accuracy of the sat‐ellite altimetry results (Songet al., 2015a,b). These six papers, however, drew only on the data from G1(Figure 1). The data from G2 was not used for un‐known reasons.As the data from G1 is only relatively accurate for the annual observation period of April or May to early January, the reported interannual values of lake level change are incorrect. This data disconti‐nuity was clearly pointed out only in the work by Leiet al.(2017). Therefore, the published multi-year lake level time series shows just the raw data of one obser‐vation site,which does not represent the real temporal lake level variations.

5 Conclusions

This work provides the first long-term detailed insitu lake level observations from the inner Tibetan Pla‐teau,which are valuable for validation of remote sens‐ing and hydrological modeling results. The interannu‐al lake level variations clearly show an overall rising trend from 2006 to 2014, a rapid decrease from 2014 to 2017, and a large increase from 2017 to 2018. The annual average lake level based on the hydrological year (May−April) rose 66 cm from 2006 to 2014, de‐creased 59 cm from 2014 to 2017, and increased 20 cm from 2017 to 2018, resulting in a net rise of 27 cm or an average rate of approximate 2 cm per year.Compared to the annual average lake level based on the calendar year,it is better to use the annual aver‐age lake level based on the hydrological year to deter‐mine the interannual lake level changes.When the an‐nual averages are not available, the time when the lake level is annually stable should be selected for comparison.This level-stable time occurs during May for Lake Nam Co. Remote sensing results agree well overall with the in-situ lake level observations; howev‐er, some significant deviations exist. In the comparable period 2006−2009, the average lake level rise rate by in-situ observation was 10−11 cm per year, which is lower than the ICESat result of 18 cm per year.

Acknowledgments:

This work was supported by the National Natural Sci‐ence Foundation of China (Grant No. 41771092), the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant No.XDA20020102), the National Key Research and Development Program of China (2017YFA0603101), and the Second Tibetan Plateau Scientific Expedition and Research Program(Grant No. 2019QZKK0202). The author also wishes to thank the Nam Co station staff for their extensive time commitment and dedication to collect the lake level data.