Late Holocene Sedimentation Dynamics in the Lake Ulaan Basin, Southern Mongolia: A History of a Playa Lake

Sedimentation dynamics in the Lake Ulaan basin located in the northern margin of the Govi region, southern Mongolia show high sedimentation rates of 11.8–22.7 cm/ka in the eastern part of the basin and low rates of 3.3–5.8 cm/ka in the western part during the late Holocene. The eastern and western parts of the lake have been strongly inuenced by uvial and aeolian activities since the arid late Holocene. However, uvial sediment input was more signicantly recorded in the eastern part. Aeolian deation has been prevailing throughout the lake bank recently. Lake Ulaan reached its maximum extent before the early Holocene (Sternberg and Paillou, 2015; Holguin and Sternberg, 2016) with a water depth of ~43 m (Lehmkuhl et al., 2018a). After the early Holocene, Lake Ulaan started to decrease its area, and the drop of the lake level intensied since the middle Holocene. In the late Holocene, the лйоупы western and eastern parts were initially exposed to wind deation at 2.7–3.2 cal. ka BP and the aerial exposition continued at 0.6–1.3 cal. ka BP. In the Anthropocene, Lake Ulaan has rapidly shifted into a playa lake condition during the last ve to six decades, and it has become an open-source area of dust generation blown out by the westerly winds.


Introduction
Lake sediments can reconstruct regional environmental and climate changes during the geological period because the lake is a container, an accumulator, and a repository. Autochthonous and allochthonous lake sediments can provide records of both lake and its basin responses to climate change (Dearing 1994). The mineral component of lake sediments would re ect the intensity of erosion in the basin (Roberts 1998). Moreover, exposed paleolake deposits allow us to nd out the origin and provenance of the lake sediments (Einsele 1992).
Boundaries of natural regions in semiarid Mongolia, particularly the Govi (or Gobi) [1] region, are shifting from the south to the north by year due to an abrupt rise in average air temperature by 1.7-2.5 o C/yr (Dulamsuren 2016). The Govi region's landscape in southern Mongolia is rapidly changing from the Govi to the desertic condition. The development of the Khongor sand dune eld in southern Mongolia at 4.0 ka BP (Felauer et al. 2012) implies that the playa environment in the Govi region started to form at that time. According to Baker (2007), a playa environment is a at-bottom depression, consisting of a dry lake, in interior desert basins in arid and semiarid regions. Globally, the playa environment is periodically covered by water that slowly ltrates into the groundwater or evaporates into the atmosphere (Baker 2007; Komatsu et al. 2007).
Southern Mongolia is climatologically important because variations in the Mongolian High-Pressure System's intensity have strongly controlled Central Asian regional climate during the Quaternary Period (Owen et al. 1997). The Govi region's climate has been generally arid in the Quaternary (Berkey and Morris 1927). The average air temperature is -15°C to -25°C in January and 20-25°C in July, and the average precipitation is less than 100 mm/yr (Tserensodnom 1971). The average rainfall is 50 (Orkhonselenge et al. 2018a). Rivers, feeding lakes in the Govi region, decrease in discharges in all seasons at the present time (Davaa 2015). In the Govi region, river water temperature is expected to rise 2.5-2.7°C in 2020, 3.2-3.5°C in 2050, and 3.9-4.2°C in 2080 (Davaa 2010).
In recent years, Lake Ulaan has attracted numerous detailed studies in geomorphology, sedimentology, hydrology, and paleogeography. For instance, Lake Ulaan's water level dropped by 1-2 m since 1980 (Batnasan 1998), and the lake faced the challenge of the continuous decrease in its area due to the rising temperature over the last half-century (Orkhonselenge et al. 2018a). However, there has been little research about how sedimentation records past climates in the Lake Ulaan basin and its evolutionary history during the geological period. Up to now, little is known about the distribution and timing of the late Quaternary aeolian sediments. This study presents the late Holocene sedimentation rates in Lake Ulaan, associated uvial and aeolian processes, hydrological evolution in the past, and when the lake has shifted into the playa environment.

Geological setting
Lake Ulaan basin is underlain by reddish sandstones over rhyolite ows and andesites (Berkey and Morris 1927) and the middle to late Quaternary deposits on the Neogene molasses deposits (Academy of Sciences of Mongolia and Academy of Sciences of USSR 1990) within the Proterozoic-Carboniferous lake zone (Guy et al. 2014). Lake Ulaan is tectonostratigraphically included in the Mandal Ovoo island arc terrane consisting mainly of the middle to late Paleozoic oceanic ophiolites, tholeiitic calc-alkaline volcanic, and volcano-clastic rocks overlain by non-marine sedimentary rocks (Badarch et al. 2002). Hills to the west and south of Lake Ulaan are composed mainly of Silurian limestone and dolomite and Cretaceous basalt and clastic sedimentary rocks, while to the east and north, the bedrock is dominated by Archaean and Paleoproterozoic metamorphic complexes and Paleozoic and Mesozoic sedimentary rocks (Academy of Sciences of Mongolia and Academy of Sciences of USSR 1990).
For geological formation, Lake Ulaan covers a continental platform consisting of Cenozoic red and gray terrigenous sands, gravels, and pebbles. It borders with Cenozoic limestone and basalts and Devonian terrigenous tuff greywacke in the southeast (Academy of Sciences of Mongolia and Academy of Sciences of USSR 1990). In the lake basin, Makhbadar (2012) identi ed early Cretaceous Manlai, Khukh Shiir, Ulziit, and Kholboot Formations, consisting of light grey and grey sandstone, ooze, siltstone, argillite, conglomerate, gravels, pyroxene basalt, and andesite. Surface sediments in Lake Ulaan are derived from a ma c igneous rock in an oceanic island arc setting (Lee et al. 2013).

Geomorphological setting
Lake Ulaan is located at 1024 m a.s.l. in the eastern end of the Valley of Lakes between the Khangai and Govi Altai Mountain Ranges at the northern border of the Govi region in southern Mongolia (Fig. 1a). The modern topography of the Lake Ulaan basin formed in late Mesozoic graben-synclinal structure rift (Tserensodnom general, and it is located between relatively low hills at 1050-1110 m a.s.l. (Fig. 1b). Because this lake basin is at the lowest elevation in the Valley of Lakes (Orkhonselenge et al. 2018a), thick alluvial and lacustrine deposits were accumulated in the past (Tsegmid 1969). Lake Ulaan is hypothesized as a deposition center for the whole lake basin system in the Valley of Lakes during the Quaternary (Tserensodnom 2000). Bottom sediments of Lake Ulaan consisted of deposits in the Neogene when water level and area were almost 100 m higher and several ten times larger in size than the present (Tserensodnom, 1971). Lake beaches are at plain and swampy (Orkhonselenge et al. 2018b), and there are sand dune elds in the northwest and north (Tsegmid 1969). Dashzeveg et al. (2005) noted the Cretaceous dune eld in the Lake Ulaan basin based on cross-bedded intervals, occasionally exhibiting windripple cross lamination in the Bayan Zag and Tugrugiin Shiree stratigraphic columns.

Hydrological setting
Lake Ulaan is a terminal lake of Ongi River (Fig. 1b), draining from the Khangai Mountain Range (Fig. 1a). However, the river cannot permanently feed the lake today (Fig. 2g, h) due to the uctuating precipitation; hence the lake area is often changed. The modern Lake Ulaan is apparently divided into the eastern and western parts separated by an approximately 6 km long and 0.5-2.0 km wide arcuate N-S trending spit (Fig.   3). The spit intersects the center of Lake Ulaan, dams the eastern part, and keeps its level ~0.5 m higher than that of the western part during heavy rainfall. The water from the Ongi River enters the eastern part directly, but it feeds the western part via crossing over the spit/three small channels (or spillways) in the spit (C1 to C3 in Fig. 3) or through groundwater. Lake Ulaan holds a water resource of 0.158 km 3 (Tserensodnom 2000), but it is almost dried out during low precipitation (Fig. 2a, c). The lake has been a shallow playa lake, dry most of the year, and the lake oor was aerially exposed and vegetated (Fig. 2c, d). The lake reshrank in 1986-1989 (Batnasan 1998), 2010 (Davaa 2015), and 2015 (Fig. 2a). Lake Ulaan is generally a freshwater lake system even though it has no out ow, unlike other lakes in the Govi region (Tsegmid 1969). The lake water's freshness is related to the lake water ows through the unconsolidated soil around the lake under the highly evaporative climate condition (Tsegmid 1969). Nevertheless, according to Tsend (1965), lake water is mild, and the lake water is dominated by HCO 3 and Ca 2+ .

Field sampling
During the eldwork on 18 June 2018, we collected bulk samples from four sites on the exposed oor in the eastern part of Lake Ulaan (Fig. 3b). As shown in Fig. 3b, the rst two sites LU18-1 and LU18-2, are located in the lake's center, whereas the last two sites LU18-3 and LU18-4, are located toward the margins as in the case of the sites UN15-1 and UN15-2 ( Fig. 3) for the previously studied by Orkhonselenge et al. (2018b). However, at present, the sites LU18-1 to LU18-4 in the eastern part of the lake (Fig. 3b) are situated under the condition which is more permanently inundated during rainfalls than those in the lake's western part, which has almost shifted to the drier playa condition (Fig. 3). It is con rmed that our four sampling sites are included in the current lake extent but were exposed at the time of our sampling ( The sites UN15-1 and UN15-2 at the exposed oor in the western part of the lake (Fig. 3) are situated at 4-6 m lower than those sites LU18-1 to LU18-4 in the eastern part of the lake (Fig. 5a). It implies the western part consists of the unconsolidated sediments, which are easily blown by winds, i.e., the erosion rate there may have been higher than that in the eastern part of the lake since 3.2 cal. ka BP (Orkhonselenge et al. 2018b) when it faced the rst aerial exposure. Whereas, the deposition rate is higher in the eastern part of the lake ( Fig. 4b) than that in the western part of the lake (Fig. 5b, c) because the eastern part of the lake is continuously overloaded/fed by groundwater and discharge of Ongi River during the rainfall in addition to aeolian deposits today. The difference in a deposition is likely augmented by the presence of a spit that provides a physical barrier blocking the smooth transfer of runoff from the eastern part to the lake's western part (Fig. 3). The exposed oor in the western part of the lake (Fig. 2a) was covered by water in the Landsat 8 image in 2014 (Fig. 3a); however, the water coverage shifted to the east or left the western part of the lake in 2019 (Fig. 3b).

Major elements' analysis
Major elements of the lake sediments were analyzed at the Division of Radionuclide Analysis, the Central Geological Laboratory in Mongolia, using the Axios Max X-ray uorescence (XRF) spectrophotometer. The

Measurement of radiocarbon
Analysis for radiocarbon ( 14 C) dating was conducted with the Accelerated Mass Spectrophotometry (AMS) at the Institute of Accelerator Analysis Ltd. in Japan. According to the laboratory report, the graphite sample was measured against a standard of Oxalic acid (HOxII) provided by the National Institute of Standards and Technology, USA, using a 14 C-AMS system based on the tandem accelerator. The samples with the plant fragments show su cient carbon recovery values at 41.29-50.83% (Table 1).
For the calculation of 14 C age, the Libby half-life of 5568 years was used (Stuiver and Polach 1977). Percent of modern carbon (pMC) refers to a ratio of the 14 C concentration in the sample relative to 1950. Calibrated calendar age is a range of age corresponding to 14

Provenance, tectonic setting, minerals, and weathering
The discriminant functions 1 and 2 to infer provenance in the Lake Ulaan basin were plotted ( The geochemical classi cation diagram shows that the Lake Ulaan sediments are mostly composed of shale, while the sediments at the sites UN15-1 and LU18-4 in the lake's margin consist of greywacke and litharenite (Fig. 7a). The A-CN-K ternary plot shows that the Lake Ulaan sediments were derived from the granitic to granodiorite source terrain in the eastern part of the lake and from the smectite source terrain in the western part of the lake (Fig. 7b). The granodiorite source is related to the composition of volcanic clastic sedimentary rocks. The lake sediments along the side towards the plagioclase (Fig. 7b) imply the Na-rich and Ca-rich feldspars' presence.
In the Lake Ulaan sediments, the CIA range from 55.88-63.28 at the sites LU18-1 to LU18-4 in the eastern part of the lake (Fig. 7b) to 60.24-62.64 at the sites UN15-1 and UN15-2 in the western part of the lake (Orkhonselenge et al. 2018b). The CIA values in the Lake Ulaan sediments indicate the low degree of source area weathering in the lake's eastern part and the moderate to the high degree of source area weathering in the lake's western part (Fig. 7b).
The results from the source area provenance (Fig. 6a) and tectonic setting (Fig. 6b), and source rock compositions (Fig. 7a), and minerals and weathering intensity (Fig. 7b) are in agreement with the structure of the geological formation in the Lake Ulaan basin with the non-marine sedimentary rocks underlain by the oceanic ophiolites, tholeiitic to calc-alkaline volcanic and volcano-clastic rocks, bounded by limestone and basalts, and terrigenous tuff greywacke around the lake basin (see details in Section 2.2).

Age correction
In terms of age correction, selecting the material for dating depending on shallow or deep lakes is crucial for inferring erosion, transportation, and deposition in the lake basin. In deep lakes, terrestrial organic materials are redeposited by streams (Orkhonselenge et al. 2013 In shallow Lake Ulaan, bulk sediments and plant fragments were chosen for establishing age datasets ( Table   1). The plant fragments in the sediments (LU18-2: 18-20 cm, LU18-3: 25-28 cm, and LU18-4: 34-36 cm) are dated as modern ( Table 1). The modern ages show the active aeolian and uvial erosions and rapid sedimentations at the sites LU18-3, 4 in the lake's margin than the site LU18-2 in the center of the lake (Table   2). Therefore, the same lake sediments for obtaining precise dating and reconstructing depositional environments in the past, the bulk sediments are used for age corrections as the well-preserved sediments in the shallow Lake Ulaan (Orkhonselenge et al. 2018b). As described in Section 3.3, the samples with the plant fragments from all Lake Ulaan sites show su cient carbon recovery (Table 1), i.e., the bulk sediments from Lake Ulaan can date the correct age and infer the sedimentation rates in the lake basin.
In the western part of Lake Ulaan the reservoir effect is negligibly corrected to be at 0. in the lake's marginal part ( Table 2, Fig. 8a, c). In other words, the near-surface sediments at the same 5 cm depths were deposited at 1.0-1.3 cal. ka BP in the eastern part of the lake and at 2.7-3.2 cal. ka BP at the sites UN15-1, 2 in the lake's western part (Orkhonselenge et al. 2018b; Fig. 8b, d). However, the original nearsurface sediments at 5 cm depths in the western part of the lake may have already been eroded and transported from there because the western part of the lake lies at ~4-6 m lower elevation (Fig. 5a). The  (Fig. 8).
Overall, the eastern part of Lake Ulaan shows ages 1.7-1.9 ka younger for the near-surface sediments than those in the western part, while the western part shows 2.8-3.6 ka older for the deep deposits (Fig. 8b). The 14 C ages in the eastern part of the lake indicate that the near-surface sediment at the site LU18-2 in the center is 0.3 ka younger than the sediment at the site LU18-4 in the lake's margin. In comparison, the deeper sediment at the site LU18-2 at the center is 0.8 ka older than the deeper sediment at the site LU18-3 in the margin of Lake Ulaan (Fig. 8c). It implies that the younger at the site LU18-3 and older sediments at the site LU18-4 in the margin than those at the site LU18-2 in the center of the lake show rapidly deposited pulse sediments and well-preserved old sediments.

Sedimentation rate
In the center of Lake Ulaan, sedimentations at 14 cm depth of the site LU18-1 occurred at 0.6 cal. ka BP, whereas the sedimentations at 7 cm depth of the site LU18-2 occurred at 1.0 cal. ka BP and 38 cm depth at 3.2 cal. ka BP (  Fig.   5b, c). In Lake Ulaan, as in other lakes in the Govi region, the sedimentation rates (Figs. 4b, 5b, c) should be a function of both sediment deposition and de ation by wind depending on the condition of water coverage and the annual precipitation input to the Ongi River (Fig. 2e, f) for its discharge and the groundwater. The center of the lake basin was more optimized for sediment preservation since it was more protected by water.
Near the center of the basin, the sedimentation has been fast during the last 600 years at the site LU18-1, where the sedimentation rate shows a massive pulse of sedimentation with an average sedimentation rate of 22.75 cm/ka (Table 3, Figs. 4b, 5b, c). In another near center of the lake, the sedimentation rate is shown at 11.85 cm/ka at the site LU18-2 (Table 3, Figs. 4b, 5b, c). In the lake's margin, the sedimentation has been fast at the rate of 19.32 cm/ka at the site LU18-3, but it has been remarkably recorded for a longer period (over 2400 years) ( Table 3, Figs. 4b, 5b, c). In the meanwhile another margin site LU18-4 of the lake, the sedimentation rate was slow (4-5 cm/yr), but for a shorter period (for 1200-1300 years), i.e., the average sedimentation rate is 4.74 cm/ka at the site LU18-4 (Table 3, Figs. 4b, 5b, c).
The average sedimentation rates in the eastern part of the lake, except for the site LU18-4 (Fig. 4b), are indicated to have been at high levels and are comparatively larger than those of 3.3-5.8 cm/ka at the sites UN15-1 and UN15-2 (Orkhonselenge et al. 2018b) in the western part of the lake (Fig. 5b, c). In both western and eastern parts of Lake Ulaan, the sedimentation rates increase from the margin toward the lake's center (e.g., from UN15-1 to UN15-2, and from LU18-4 to LU18-1) (Fig. 5b, c). The lake margins have probably been more exposed to aeolian de ations than the lake center hence recording slower sedimentations. Still, other margin sites (e.g., LU18-3) may occasionally have higher sedimentations due to massive discharge pulses during the Ongi River's ooding (Fig. 2e, f) because Lake Ulaan is a terminal lake of the Ongi River. These giant pulses of sedimentations in the eastern part of the lake imply that this part of the lake has received precipitation-derived river water in addition to the glacier and permafrost meltwaters, whereas the western part of the lake may have been fed only by glacier meltwater via groundwater (Figs. 9, 10).
The sharp division of the eastern and western parts (Fig. 3) seems to be enhanced by the spit's presence separating the two areas by obstructing the free ow of surface water from the Ongi River. As strongly

Sedimentation dynamics
The sedimentation rate in the eastern part of Lake Ulaan just near the downstream of Ongi River is remarkably higher than that of the western part (Fig. 8b) because the presently dried out Ongi River (Fig. 2g, h) may intermittently feed the lake with high meltwater discharge transporting sands (Fig. 9). The spits may block the sand transportation and water discharge, and it may slow down the uvial sedimentation rate in the western part, where the lling by water is more signi cantly delayed than the eastern part (Figs. 3, 10). The eastern part tends to be covered with the water from the Ongi River and the groundwater for more extended periods and consequently have less aeolian de ation (Figs. 9, 10). The possibility of sand transport across the ice covering the lake in the Depression of Great Lakes in western Mongolia during the long winter (Stolz et al. 2012) when most of the sand storms caused by strong west and northwest winds (Hempelmann 2010) may support the aeolian sedimentations (i.e., deposition) in the Lake Ulaan area in winter and spring.
Lakes in the Depression of Great Lakes and the Valley of Lakes nd high stands due to increased snowfall caused by a more humid climate resulting in a considerable glaciation at high elevations (Lehmkuhl and Lang 2001;Lehmkuhl et al. 2018b). This snow accumulation may contribute to the meltwater runoff in spring and early summer and the uvial sedimentations in the lake's eastern part (Fig. 9). However, the climate in the interpreted to be the lake terrain (Badarch et al. 2002). According to Lehmkuhl et al. (2018a), the silts deposited in Lake Ulaan during the humid middle Holocene may have been exposed to wind activity when the lake is dried out as a playa bed. The early to late Holocene sediments in the Lake Ulaan basin indicate the aeolian dominant sediment-transport mechanism (Lee et al. 2011). Moreover, the sediments in the Lake Ulaan basin were transported by local westerly winds blowing along the Valley of Lakes during the last 11.2 ka BP (Lee et al. 2013).
In general, the aeolian sedimentations predominated in the Lake Ulaan basin since the early Holocene. For instance, the ne aeolian sand at 4.2 m depth in a 6 m high terrace of Ongi River 55 km northeast of Lake Ulaan indicated that the deposition had already started during the late Glacial period (Lehmkuhl et al. 2018a).
However, the uvial sedimentation strengthened in the middle Holocene, and aeolian sedimentation increased during the late Holocene (Fig. 9). Some large alluvial fans or deltas may have formed around the Lake Ulaan basin in the past that are now buried by wind-derived surface sediments (Sternberg and Paillou 2015). The aeolian de ation in the Lake Ulaan basin during the Anthropocene (Figs. 9c, 10) had intensi ed since the 1950s because of the historical records of water coverage of Lake Ulaan (Table 4). Continuous de ations occurred in the Lake Ulaan basin in winter, spring, and autumn since that time, except for summer when the lake was lled due to rainfall in 1960, 1970, and 2013-2014 (Table 4, Fig. 2b, d). The lake has been an endtransfer base oor for aeolian sedimentations blown from the Depression of Great Lakes in the northwest through the Valley of Lakes, and also a continuous source for dust storms and suspension transport of silts towards the Paci c Ocean in the southeast.

Lake evolution history
Like other lakes in the Govi region, Lake Ulaan experienced signi cant changes in water level and coverage area during the late Quaternary. Historical analysis of changes in lake level and size (Table 4) shows that Lake Ulaan is highly sensitive to ongoing climate changes of precipitation, air temperature, and wind velocity. Lake Ulaan has experienced dramatic uctuations of its extent over several thousand years. A giant paleolake of ~43 m depth might have existed at OSL-dated 162±13 ka, i.e., the paleolake formed after the Marine Isotope Stage (MIS) 6 (Lehmkuhl et al. 2018a; Table 4). The early time of a large lake covering an area of 19,000 km 2 being associated with wetter climates corresponded to a water level at 1285 m a.s.l. (Sternberg and Paillou 2015; Table 4). Then the lake was de ned as an intermediate-sized lake covering a surface of 6,900 km 2 at a water elevation of 1150 m a.s.l. (Sternberg and Paillou 2015; Table 4); however, the precise ages of such extended lakes are still unclear.
In the Holocene, the lake reduced to 1,700 km 2 in the area, corresponding to a water level at 1070 m a.s.l.
(Sternberg and Paillou 2015; Table 4) occurring as the stage of the lake before the present-day dry basin (Fig.  9). In the middle Holocene (Fig. 9a), Lake Ulaan occupied an area of approximately 500 km 2 in size (Lehmkuhl et al. 2018a; Table 4). In the late Holocene (Fig. 9b), Lake Ulaan may have been exposed aerially for the rst time. It coincides with the desiccated Lake Bayan Tukhum at 3.5 cal. ka BP (Felauer et al. 2012), 105 km south from Lake Ulaan. If the remobilized and redeposited sediments are taken into account, the lake has been re-exposed to wind de ations at 0.6-1.3 cal. ka BP for the sites LU18-1, 2, 4, except for deep sediments at 2.4 cal. ka BP for the site LU18-3 ( Table 2, Fig. 9c).
In the Anthropocene, Lake Ulaan may still have been a permanent lake (Table 4, Fig. 9c). The most recent high stand before the 1960s was when the lake level was maintained 2-3 m above the present dry lake bottom (Lehmkuhl et al. 2018a). The lake condition in addition to the former lakes determined by Murzaev and Bespalov before the 1950s (Table 4)  , and the former lake oor is partly covered by sedge plants (Fig. 2c). Once the largest lake in the Govi, Lake Ulaan was identi ed in 1991 but did not appear in the Landsat images since 2000 (Kang et al. 2015). According to Davaa (2015), Lake Ulaan disappeared in 2010 and recovered an area of 20.9 km 2 in 2013 ( Table 4). The frequent shrinkage of the lake since the 1950s has contributed to the aerial exposition of the lake oor to wind de ation and rapid redeposition (Figs. 9c, 10).
Although Lake Ulaan has experienced uctuations of shrinking and/or lling depending on the annual precipitation and temperature (Orkhonselenge et al. 2018a) since the 1950s, the lake has nally shifted to become a playa (Fig. 9c). This was due to the strong westerly wind effects exceeding 20-25 m/s (Amarjargal 2016) and extreme over usage of groundwater by mining operation in the basin in addition to the rapidly rising air temperature (Orkhonselenge et al. 2018a). The playa condition (Fig. 9c) is re ected in intensi ed aeolian sedimentations (Lehmkuhl et al. 2018a), chemical weathering (Fig. 7b), and the climate change from semiarid to arid conditions (Fig. 11).
Although the eastern part of the lake is occasionally lled by the pulsating sedimentations by Ongi River during the heavy rainfalls, the lake basin has been thoroughly exposed to the westerly wind de ations (Figs. 9, 10). When the eastern part of the lake lls with water by Ongi River, it feeds the western part temporarily through the only channels C2 and C3 in the spit (Figs. 3, 10). Ongi River may ll only the eastern part, but due to the spit's presence, the lake water needs to reach the spit height before the two systems start connecting to each other fully. It implies that the lling of the western part should be delayed even though the groundwater feeds it (Fig. 10). The present playa condition of Lake Ulaan (Fig. 9c) is consistent with numerous other observations showing that most lakes in the eastern Govi have been exposed aerially and dried out entirely due to the recent rapid rise in air temperature and evaporation during the last two decades (Orkhonselenge et al. 2019).

Holocene climate changes
In Mongolia, there is a common trend with the warm and humid early Holocene, the humid early to middle Holocene, the arid middle Holocene, and the humid late Holocene reconstructed paleoclimate records  During the early Holocene, the paleoclimate in the Govi region is described relatively well ( Table 5 The late Holocene climate in the Govi region is recorded from lakes as arid ( shrinkage of the shallow lake (Fig. 9b, c), a phenomenon seems to be continuing up to the modern time. The dry climate in the Lake Ulaan basin (Figs. 9b, c, 11 Fig. 12) Lake Ulaan may have been exposed to wind erosion since 3.2-2.7 cal. ka BP (Fig. 9b), and aeolian de ation has strengthened particularly since 0.6-1.3 cal. ka BP (Fig. 9c). The phenomena in the Lake Ulaan basin is in agreement with the desiccation of Lake Bayan Tukhum at 3. In the Anthropocene, the Govi region's climate is indicated as a dry (Table 5 Fig. 11) has caused the playa lake condition observed today (Fig. 9c). The trend is con rmed by a distinct tendency towards drier conditions since the 1980s reconstructed for the eastern central High Asia By the end of this century, potential evaporation will rise by 200-300 mm/yr, with an increase in annual average surface temperature by 5-6 o C, and the Govi region will extend 600 km toward the north (Dulamsuren 2016). This means that the present arid Govi region will shift into the dry desert, and the semiarid steppe regions will become arid Govi regions. This trend has been previously noted by Davi et al. (2015), showing that the current drought conditions in Mongolia associated with annual average temperature increase bỹ 2°C over the last 60 years have resulted in the expansion of desert areas from the warm and arid southern Mongolia towards central and northern regions of the country. The dry climate in the Govi region may rapidly contribute to the playa environments covering the semiarid steppe in a few years and result in a shortage of surface water resources, especially lakes and rivers. This trend has been con rmed by Lake Ulaan sediments showing the climate around the lake basin has been signi cantly shifted from the semiarid into the arid (Fig.   11).

Conclusions
Late Holocene sedimentation records in Lake Ulaan at the eastern end of the Valley of Lakes at the northern border of the Govi region, southern Mongolia, show a higher sedimentation rate of 11.8-22.7 cm/ka in the eastern part of the lake and a lower sedimentation rate of 3.3-5.8 cm/ka in the western part of the lake. The lake has historically experienced a long-term evolution of a large lake since the initial formation after MIS 6 (Lehmkuhl et al. 2018a), and around that time, the lake reached the maximum size. In the context of lake retreats, the lake started to drop in the area before the early Holocene, strengthened the decreasing trend since the middle Holocene, and faced exposure to wind de ation in the late Holocene. The rst aerial exposure occurred at 2.7-3.2 cal. ka BP and frequently exposed at 0.6-1.3 cal. ka BP. At the present time, the substantial westerly wind-induced aeolian de ation occurs throughout the lake bank even though the eastern part of the lake receives the pulsating uvial sedimentations during heavy rainfall, and both parts receive aeolian air fall sedimentations. Today, Lake Ulaan has been remarkably reduced in extension and has rapidly shifted into a playa lake in the last ve to six decades, and it has become an open-source area of dust generation blown out by the westerly winds.

Declarations
The authors declare that they have no relevant nancial or non-nancial interests to disclose. The authors have no con icts of interest to declare that are relevant to the content of this article.