Distribution and timing of Holocene and late Pleistocene glacier
fluctuations in western Mongolia
Frank LEHMKUHL,1 Michael KLINGE,2 Henrik ROTHER,3 Daniela HÜLLE4
1Department of Geography, RWTH Aachen University, Aachen, Germany
2Department of Geography, Göttingen University, Göttingen, Germany
3Institute of Geography and Geology, University of Greifswald, Greifswald, Germany
4Institute of Geography, University of Cologne, Cologne, Germany
Correspondence: Frank Lehmkuhl <frank.lehmkuhl@geo.rwth-aachen.de>
ABSTRACT. Despite being a key location for paleoglaciological research in north-central Asia, with the
largest number of modern and Pleistocene glaciers, and in the transition zone between the humid
Russian Altai and dry Gobi Altai, little is known about the precise extent and timing of Holocene and
late Pleistocene glaciations in western Mongolia. Here we present detailed information on the
distribution of modern and late Holocene glaciers, and new results addressing the geomorphological
differentiation and numerical dating (by optically stimulated luminescence, OSL) of Pleistocene glacial
sequences in these areas. For the Mongolian Altai, geochronological results suggest large ice advances
correlative to marine isotope stages (MIS) 4 and 2. This is in contrast to results from the Khangai
mountains, central Mongolia, showing that significant ice advances additionally occurred during MIS3.
During the Pleistocene, glacial equilibrium-line altitudes (ELAs) were �500 to >1000m lower in the
more humid portion of the Russian and western Mongolian Altai, compared to 300–600m in the drier
ranges of the eastern Mongolian Altai. Pleistocene ELAs in the Khangai mountains were depressed by
700–1000m, suggesting more humid conditions at times of major glaciation than in the eastern
Mongolian Altai. This paleo-ELA pattern reveals that the precipitation gradient from the drier to the
more humid regions was more pronounced during glacial times than at present.
KEYWORDS: glacial geomorphology, glacier fluctuations, glacier mapping, mountain glaciers, paleo-
climate
1. INTRODUCTION
This paper aims to summarize the extent and timing of
Holocene and Pleistocene glaciations in western Mongolia
based on selected region in the Altai and western Khangai
mountains (Figs 1–6; see Fig. 1a for locations). Few
comprehensive studies considering Quaternary glaciation
of the entire Altai region are available (e.g. Lehmkuhl and
others, 2011; Blomdin and others, 2014).
The Altai mountains span four different countries (Russia,
Mongolia, China, Kazakhstan), with major mountain ranges
represented by the Russian Altai in the north and west, the
Mongolian Altai in the east and the Chinese Altai in the
south. Towards the southeast the Mongolian Altai extend
into the Gobi Altai. East of the Mongolian Altai, the so-
called ‘Valley of Great Lakes’ separates the Altai from the
Khangai mountains in central Mongolia, which trend
600 km from northwest to southeast (Fig. 1a).
The modern climatic conditions are of continental type
showing large temperature variations with mean winter
temperatures below –20°C and mean summer temperatures
of �20°C. The highest amount of annual precipitation
occurs in the western Altai (above 1000mm), in regions
climatically influenced by the westerlies. The eastern part of
the Altai is situated in a rain shadow and experiences more
arid conditions (200–400mm). In the basins of the Valley of
Great Lakes the annual precipitation decreases to <50mm.
Further east, in the Khangai, the annual precipitation rises
again, reaching >400mm. There is a pronounced seasonal
precipitation distribution, with 70–80% of annual precipi-
tation falling during the summer months.
The topography a nd climatic conditions of the Altai
mountains are major factors determining the characteristics
and dynamics of late Quaternary glaciations in this region.
In the western and southern Altai the alpine relief features
steep V-shaped valleys, while intramontane basins divide
the eastern Mongolian Altai into several large mountain
massifs, of which the higher parts have been intensively
shaped by glacial erosion. The summits, which lie above an
old planation surface, reach >4000ma.s.l.
In western literature, only a few results have been
published detailing historic and Holocene glacier variations
in Mongolia (e.g. Klinge, 2001; Lehmkuhl, 2012; Kamp and
others, 2013). Under modern conditions glaciation in
Mongolia is restricted to mountains above �3400ma.s.l.
At present, numerous plateau glaciers, cirque glaciers and a
few valley glaciers are found across the Mongolian Altai,
while only a single small ice field survives in the Khangai at
Otgontenger (3964ma.s.l.; Fig. 1b). In addition, Osipov and
others (2013) show a single small glacier (0.12 km2) on the
Mongolian side of Munku Sardyk (Sayan range, north of
Khövgöl lake). Krumwiede and others (2014) note that
estimates of the glacierized area of Mongolia range from
300 to 659 km2. According to Klinge (2001), the modern
glaciation across Mongolia covers an area of �850 km2.
Version 3.2 of the Randolph Glacier Inventory (Pfeffer and
others, 2014) calculated 627 km2. Currently, equilibrium-
line altitudes (ELAs) increase from 3200ma.s.l. in the
western part of the Mongolian Altai (Tavan Bogd) to
3800ma.s.l. in the southeastern Mongolian Altai and at
Otgontenger in central Mongolia.
Annals of Glaciology 57(71) 2016 doi: 10.3189/2016AoG71A030 169
© The Author(s) 2016. This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.
org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited.
http://dx.doi.org/10.3189/2016AoG71A030
Downloaded from http:/www.cambridge.org/core. SUB Gottingen, on 22 Dec 2016 at 13:44:38, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
Based on fieldwork and aerial photos the extent of the
Little Ice Age (LIA) glaciation was mapped for three regions in
western Mongolia. (1) For the Turgen-Kharkhiraa massif
(northern Mongolian Altai; Fig. 1a, No. 4) Lehmkuhl (1998,
2012) calculated ELA values for 99 glaciers and their LIA ice
extent using the toe-to-summit altitude method (TSAM; cf.
Benn and Lehmkuhl, 2000). Modern ELAs calculated by the
TSAM show consistent values as estimated from firn lines on
modern glaciers surveyed through aerial reconnaissance and
field observations. The ELA depression (�ELA) for the LIA
was calculated at �80m with respect to the present, with an
associated reduction of the glaciated area from 157.6 km2
(LIA) to 69.7 km2 (in 1991) (Lehmkuhl, 2012). (2) For the
Tsambagaravmassif (central Mongolian Altai; Fig. 1a, No. 5)
the �ELA for the LIA was calculated at only�30m, while the
total glaciated area decreased from 110.3 km2 (LIA) to
80.7 km2 (1991; Klinge, 2001). (3) The �ELA for the LIA at
the Munkh Khairkhan (southern Mongolian Altai; Fig. 1a,
No. 6) was estimated at �50m (Klinge, 2001).
During the Pleistocene, numerous large valley and cirque
glaciers existed in eastern and central Mongolia. The ice
masses formed extensive systems of interconnected valley
glaciers, constituting separate glaciation centers within the
Mongolian Altai and Khangai (Lehmkuhl and others, 2011;
Rother and others, 2014). The maximum extent of the
Pleistocene glaciations has been determined from the
distribution of glacial landforms such as large hummocky
moraines which are easily visible on aerial photographs and
satellite images (Klinge, 2001; Blomdin and others, 2014).
The spatial distribution and sedimentary composition of
moraine landforms in Mongolia vary based on relief,
geology and the tectonic situation. Using geomorphological
criteria, the Quaternary moraines can usually be differen-
tiated into three main moraine systems (Lehmkuhl and
others, 2011) comprising (1) Holocene and, in particular,
LIA moraines, characterized by a fresh morainic topography
without any sediment and soil cover; (2) the larger M1
moraines further down-valley, which can be further
differentiated into three sub-stages (M1a, 1b, 1c), of which
the oldest (M1c) shows extensive periglacial modifications
and weathered carbonate crusts; and (3) relics of the eroded
M2 moraine, which represents the oldest phase of glaciation
and is connected to relics of a high-lying terrace (T2) and
associated with dispersed erratic boulders and glacial
trimlines. The M1 moraines are usually connected to the
main terrace (T1) and/or alluvial fan system (A1) (Grunert
and others, 2000). These features typically show an aeolian
cover and kastanozem soil with a carbonate-enriched layer
that is correlated to the last glacial cycle. The M2 sequence
is tentatively correlated to the penultimate glaciation, while
the relative down-valley positions of these features indicate
an ice extent slightly larger than during the last glaciation
(Lehmkuhl and others, 2004, 2011).
Based on the altitudinal position of terminal moraines
associated with the local Last Glacial Maximum (LLGM; M1,
maximum extent of glaciers during the last glacial cycle) we
calculate that LLGM paleo-snowlines (ELAs) were at
�3000ma.s.l. in the western and central Mongolian Altai,
increasing to 3100ma.s.l. in the southern Mongolian Altai,
while data from the Khangai suggest LLGM paleo-ELAs at
3000ma.s.l. in the west and �2800ma.s.l. in the east
(Klinge, 2001; Lehmkuhl and others, 2011).
First absolute chronologies for major late Pleistocene ice
advances in Mongolia have been obtained by Klinge (2001),
Gillespie and others (2008), Lehmkuhl and others (2011) and
Rother and others (2014). However, considering the vast size
of the region there is still a lack of numerical age control for
most moraines in these areas, severely limiting the develop-
ment of a comprehensive glacial model for Mongolia.
2. LUMINESCENCE DATING METHODS
Eighteen samples from the three regions (O1–4, Z1–2, K1–3)
(Table 1) were examined by optically stimulated lumines-
cence (OSL) dating to gain information about the timing of
critical late Quaternary geomorphological and sedimento-
logical processes. Usually quartz is used in luminescence
dating studies on Holocene deposits, but, as reported by
Hülle and others (2010) and Lehmkuhl and others (2007,
2012), quartz samples from southern and central Mongolia
exhibit feldspar contaminations compromising the usability
of the luminescence signal. This problem was also observed
for the samples in this study, so the quartz fraction was
discarded and the feldspar fraction was used for dating. The
sample preparation and the applied quality criteria are
described in detail by Hülle and others (2010) and
Lehmkuhl and others (2012). For the dating procedure,
small aliquots with a grain layer of 1mm diameter of the
100–200µm grain size were used; only the three samples
from profiles K1 and K2 are polymineralic fine-grain
samples, due to their fine-grained matrix. All measurements
were carried out on automated Risø TL/OSL readers (TL-DA-
20), equipped with 90Sr/90Y �-sources for irradiation and
EMI 9235 photomultiplier tubes for luminescence detection
(Bøtter-Jensen and others, 2003). Infrared (IR) LEDs were
used for stimulation; the detection wavelength was 380–
440nm, isolated by an interference filter D410, LOT Oriel.
For 14 samples, the usual single-aliquot regenerative-dose
(SAR) protocol was used and for 10 of them the equivalent
dose was calculated with the central age model of Galbraith
and others (1999). Exceptions are made for four samples
from profiles O2 and O3, which are calculated with the
finite mixture model of Galbraith and Green (1990), due to a
high data scatter. Note that the ages are not corrected for
fading and thus have to be regarded as minimum ages.
Fading measurements from comparable samples from this
region show that a systematic age underestimation of �20%
has to be taken into account. This is not applicable for the
remaining four ‘old’ ages from profiles Z2 and K1, which
were measured with the post-IR–IR protocol using elevated
stimulation temperatures to derive a dating result that is
much less affected by fading than the conventional IR
stimulated luminescence (IRSL) signal measured at 50°C.
For this, the protocol suggested by Thiel and others (2011)
was selected, using a stimulation temperature of 290°C after
preheating at 320°C and an IR bleach at 50°C.
Dose rates are based on laboratory high-resolution
�-spectrometry results using 800 g of bulk material. The dose
rate was calculated taking variations in the water content into
account. An internal potassium content of 12.5� 0.5%
(Huntley and Baril, 1997) and an �-efficiency factor of
0.07�0.02 were assumed. The contribution of cosmic rays
to the dose rate was calculated following Prescott and Hutton
(1995). All dose rate results are given in Table 1.
Due to transport processes which may not be sufficient to
bleach the sample completely, the dating of moraine
sequences has proved to be problematic, as discussed in
greater detail by Fuchs and Owen (2008). To obtain robust
Lehmkuhl and others: Glacier fluctuations in western Mongolia170
http://dx.doi.org/10.3189/2016AoG71A030
Downloaded from http:/www.cambridge.org/core. SUB Gottingen, on 22 Dec 2016 at 13:44:38, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
minimum age estimates for the past glaciations, overlying
aeolian or glaciofluvial deposits of the different morphologic
sequences were sampled. In addition, four samples were
taken from fluvial sand underneath till (sections O2 and Z2)
and two from lacustrine deposits (K1) above moraines.
3. RESULTS
As described above, several Holocene and Pleistocene
stages of glaciation can be differentiated based on geo-
morphological criteria. The largest and best-preserved
moraines are associated with ice advances during the last
glacial cycle (M1). These moraine sequences are character-
ized by a fresh glacial relief featuring moraine ridges and
dead-ice depressions, while their hummocky surfaces are
littered with numerous big boulders. Stacked lateral
moraines and glacial trimlines are often found in the lower
valley portions. The main glaciofluvial terrace T1, which is
commonly covered by a layer of aeolian sediment, can be
traced to this moraine and was formed by meltwater
deposition during the M1-advance phase. In the field this
moraine system can be further divided into two sub-stages
(M1b, M1c). Further up-valley, the M1b/c moraines are often
followed by a third and much smaller moraine system (M1a),
representing terminal moraines that formed during a short
period of stagnation at the beginning of deglaciation during
the last termination.
Although there is a difference in the total glaciated areas
during the three M1 stages, no significant difference in the
elevation and distance of the associated terminal moraines
exists. This prohibits a separate ELA calculation for every M1
stage and the �ELA is delineated for the LGM as one.
3.1. Otgontenger (western Khangai)
Mapping and fieldwork concentrated on Bogdin gol valley,
west of Otgontenger peak (47°370N, 97°310 E; Figs 1b and
2). Site and sample locations for luminescence and
cosmogenic radionuclide (CRN) surface exposure dating
are shown in Figure 2. According to CRN dating efforts of
Rother and others (2014), a total of 21 samples were
collected from five moraines ranging in altitude from 2100
to 2500ma.s.l. Most samples were obtained from glacial
boulders, but in one case an ice-overridden smooth bedrock
surface was also sampled. Site context and analytical
procedures for CRN dating are described in detail in Rother
and others (2014). Table S1 (see Supplementary material
(http://igsoc.org/hyperlink/71a030_supp.pdf)) shows the
CRN results from the Otgontenger sequence using the
updated 10Be reference production rate of Heyman (2014).
This revised production rate is based on a recent compilation
Table 1. Luminescence dating results. For each sample the depth, the radionuclide contents determined by �-spectrometry, the water
content and the resulting dose rate are presented as well as the equivalent doses (determined with the central age model; only the samples
O2-1, O2-2, O3-1 and O3-4 were calculated with the finite mixture model), the number of measured aliquots (n), the relative standard
deviation of these aliquots and the resulting age estimates. Note that the ages are not corrected for signal loss due to fading and thus have to
be interpreted as minimum ages. Exceptions are the ages written in italics (Z2-1, Z2-2, K1-1 and K1-2) which were determined with the post-
IR–IR290°C protocol for which fading is relevant only to a minor degree. The U, Th and K were not measured for O1-4 and assumed from the
adjacent sample O1-3
Section Location Depth U Th K Dose rate Water
content
Equivalent dose n rsd Age
ma.s.l. cm ppm ppm % Gy ka–1 % Gy % ka
O1-1 47.6361°N, 11 1.9�0.1 8.1� 0.4 2.5� 0.1 4.1�0.1 2 11.1� 0.8 8 11 2.8� 0.3
O1-2 97.13249° E 40 2.3�0.1 10.0� 0.5 2.4� 0.1 4.2�0.2 2 14.5� 1.0 15 12 3.5� 0.3
O1-3 (1976ma.s.l.) 120 2.2�0.1 9.5� 0.5 2.5� 0.1 4.2�0.2 4 37.8� 2.41 37 44 9.1� 0.8
O1-4 155 5 110.8� 14.8 24 38 26.9� 3.8
O2-1 47.64287°N, 48 2.5�0.1 9.1� 0.5 2.3� 0.1 4.1�0.2 5 56.0� 4.3 42 51 13.5� 1.4
O2-2 97.14026° E 92 2.4�0.1 8.3� 0.4 2.3� 0.1 4.0�0.2 5 324.4� 19.0 30 29 81.8� 7.4
(1984ma.s.l.)
O3-1 47.68959°N, 35 2.1�0.1 10.5� 0.5 1.8� 0.1 3.2�0.2 6 82.2� 4.9 24 38 25.1� 2.4
O3-2 97.25225° E 50 2.2�0.1 12.5� 0.6 2.7� 0.1 4.5�0.3 1 122.2� 7.2 40 26 27.1� 2.5
(2090ma.s.l.)
O4-2 47.68843°N, 35 2.0�0.1 10.2� 0.5 2.5� 0.1 4.4�0.1 3 48.8� 10.5 11.6� 2.4
97.27974° E
(2130ma.s.l.)
Z1-1 48.5028°N, 42–48 1.5�0.1 5.7� 0.4 1.8� 0.1 3.3�0.2 1 43.1� 2.3 24 8 13.2� 1.2
Z1-2 88.9878° E 80–86 1.7�0.1 7.0� 0.4 1.8� 0.1 3.4�0.2 3 48.5� 3.4 24 24 14.2� 1.4
(2324ma.s.l.)
Z2-1 48.5327°N, 375–381 3.4�0.2 7.9� 0.5 1.9� 0.1 3.7�0.4 15 266.6� 18.1 11 15 71.2� 4.8
Z2-2 89.0181° E 405–411 2.1�0.1 7.9� 0.5 1.9� 0.1 3.8�0.3 4 366.2�113 7 10 97.6� 30.1
(2382ma.s.l.)
Z3-1 48.5396°N, 52–58 2.0�0.1 8.8� 0.5 1.9� 0.1 2.9�0.3 43 17.1� 1.0 20 10 5.8� 0.6
Z3-2 89.0307° E 70–76 3.0�0.2 11.4� 0.7 2.3� 0.1 4.1�0.4 21 42.0� 2.3 24 12 10.4� 0.6
(2455ma.s.l.)
K1-1(fg) 48.6072°N, 30–36 2.5�0.1 9.5� 0.6 1.9� 0.1 4.0�0.6 12 231.8� 12.8 20 2 57.8� 9.1
K1-2(fg) 88.7619° E 43–49 2.7�0.1 9.6� 0.5 1.9� 0.1 4.3�0.5 9 370.4� 18.9 24 6 85.6� 10.4
(2052ma.s.l.)
K2-1(fg) 48.5981°N, 112–118 2.4�0.1 9.9� 0.6 2.0� 0.1 4.3�0.5 6 58.8� 3.0 24 3 13.6� 1.6
88.4258° E
(2091ma.s.l.)
Lehmkuhl and others: Glacier fluctuations in western Mongolia 171
http://dx.doi.org/10.3189/2016AoG71A030
Downloaded from http:/www.cambridge.org/core. SUB Gottingen, on 22 Dec 2016 at 13:44:38, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
and recalculation of the global calibration datasets, yielding
a significantly lower 10Be production rate than the
previous 10Be reference production rate of Balco and others
(2008). As a consequence, moraine exposure ages from
Otgontenger (Rother and others, 2014) increase by 9% or
between 1.3 and 5.4 ka across the period of interest
(Table S1 (http://igsoc.org/hyperlink/71a030_supp.pdf)).
CRN dating shows that moraines associated with the
maximum LLGM ice advance at Otgontenger yield mean
ages of �40 ka (C3), thus falling into marine isotope stage
(MIS) 3. Similarly, a high-lying lateral moraine at a
transfluence pass in the northwest of the Otgontenger area
(C6) also gives a late MIS3 mean site age of 41.1� 5.8 ka.
Interestingly, an ice extent very similar to the local MIS3
Fig. 2. Schematic cross sections showing the connection between ice-marginal moraines, terraces and aeolian cover beds at Otgontenger.
Fig. 1. (a) Study areas and cross section (green line). 1 = Otgontenger, 2 = Tsengel, 3 = Khurgannuur, 4 = Turgen-Kharkhiraa, 5 =
Tsambagarav, 6 = Munkh Khairkhan, 7 = Kanas Valley (Chinese Altai). (b) Geomorphic map of late Quaternary glaciation and selected
sampling sites near Otgontenger, Khangai. Background image: STRM DTM and Landsat 8 image from 22 September 2014.
Lehmkuhl and others: Glacier fluctuations in western Mongolia172
http://dx.doi.org/10.3189/2016AoG71A030
Downloaded from http:/www.cambridge.org/core. SUB Gottingen, on 22 Dec 2016 at 13:44:38, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
limits (=LLGM) was also reached during MIS2, as indicated
by a high lateral moraine near Saran nuur lake (C4), which
returned a mean site age of 25.6�1.4 ka. Following
recession from these limits, a large terminal moraine (C1,
C2), comprising multiple moraine ridges and extended
hummocky terrain, formed on the valley floor below the C4
lateral moraine. New and detailed geomorphological map-
ping of this moraine complex and its associated outwash
terraces allows the differentiation of two moraine sub-stages
which we designate as M1b1 and M1b2 (Figs 1b and 2). CRN
dating of this moraine (six boulders) returns overlapping site
ages of 16.65�1.2 ka for C1 and 18.37�1.4 ka for C2,
yielding a cumulative mean moraine age of 17.7�1.0 ka
(C1 and C2). Close to the C2 site, glaciofluvial sands,
exposed stratigraphically below the moraine, yielded con-
sistent OSL dates of >25.1�2.4 and >27.1�2.1 ka (O3),
suggesting deposition slightly before the last major ice
advance. Therefore both data sources, CRN and OSL, place
this ice advance into MIS2 between >27.1 and 17.7 ka.
Taken together, these results provide evidence suggesting
the correlation of the M1c moraines to ice advances during
MIS3 and the M1b moraines are part of the MIS2 sequence.
Three small M1a terminal moraines were identified upstream
of the main valley of Arschaani gol, which are connected to
a system of T1a terraces. Sand and silt on the T1a terrace (O4)
in front of this M1a moraine complex supplied a Late-glacial
OSL age of >11.6�2.4 ka, which should be considered a
minimum age for the M1a advance.
Further glacial sediments, probably associated with an
earlier ice advance and the oldest moraine (M2) in the
valley, are located �5 km downstream of the LLGMmoraine
(O2, Fig. 2). These deposits are covered by >1m of aeolian
sand and silt, yielding OSL ages ranging from >81.8�7.4 ka
at the base to >13.5� 1.4 ka in the upper part. These ages
are consistent with our interpretation that the M2 moraine
formed during the penultimate glacial cycle.
At Otgontenger, ELAs during the LGM were depressed to
�3000m a.s.l. The modern ELA of 3800m a.s.l. was
estimated from the small ice cap on Otgontenger and
suggests a �ELA of 800–1000m for the Khangai.
3.2. Tsengel Khairkhan
The Tsengel Khairkhan is an isolated mountain massif in the
central Mongolian Altai (48°320N, 89°010 E; Figs 1a and 2).
Our fieldwork targeted a large terminal moraine complex
located on the southern slopes of this mountain system
(Figs 3 and 4).
Significant morainic landforms and glacial deposits are
preserved in the lower portion of the trunk valley and in the
mountain foreland. These moraines show a fresh relief with
a hummocky surface and many small dead-ice depressions.
The moraine front is steep and >10m high. Here too these
accumulations can be divided into the three M1 moraine
systems, but at the base they are additionally framed by a
shallow M1–2 moraine, which can be traced to relics of the
highest glaciofluvial terrace preserved in the area. The main
LLGM moraine consists of two distinct M1c moraine ridges,
which are connected to lateral moraines found along the
valley flanks. The M1b terminal moraine is framed by the
M1c terminal moraine. This large M1bc moraine complex
encloses a small terminal moraine of the youngest M1a stage
with section Z3 situated in a dead-ice depression of the M1bc
moraine. An OSL sample from aeolian sandy silt recovered
from the base of the depression gave an age of
>10.4�0.6 ka, probably indicating aeolian deposition
during the Late-glacial warming phase and providing a
minimum age of the M1bc moraines. Older remnants of
deeply eroded M2 moraines extend >3 km downstream from
the M1 moraines, traceable by trimlines and erratic boulders
along the slopes until two terminal moraines (M2a, M2b).
Behind the M2b terminal moraine, >1m of aeolian sand is
accumulated (Z3). Dating samples recovered from this
deposit at 40 and 70 cm depth returned OSL ages of
13.2�1.1 ka and 14.2�1.4 ka, suggesting an onset of dust
trapping by developing vegetation during the Late-glacial
phase. Approximately 100m outward from the western
margins of the M1 terminal moraines, two M2 lateral
moraine ridges are located. Between the moraines a river
channel cuts >10m into the M2 moraine (Z2). Two samples
were taken for OSL (post-IR) dating, providing ages of
>97.6�30.1 ka from a layer of interbedded fluvial sand/silt
3.4m below the surface, and an age of >71.2�4.8 ka from
the till below that layer, indicating that the M2 moraines
date to the penultimate glaciation, although residual signals
due to incomplete bleaching in these depositional environ-
ments need to be considered.
The modern ELA in this area is at �3420ma.s.l., while
the reconstructed ELA during the LLGM was 3120ma.s.l.,
resulting in a �ELA of 300m.
3.3. Khurgan nuur
The Khurgan and Khoton nuur are glacial lakes located in
the western Mongolian Altai (48°400N, 88°480 E; Fig. 1a,
No. 3) formed in a glacial trough associated with a very
large Pleistocene valley glacier extending from the Tavan
Bogd (the highest peak of Mongolia; 4374ma.s.l.) in a
southeasterly direction. In the downstream portion of the
trough, voluminous morainic sediments filled up an
intramontane tectonic basin from which the Khovd river
cuts a steep and narrow gorge across the outer mountain
range (Fig. 5). Figure 6 shows an overview of the geo-
morphological setting and the locations of the examined
sediment sections in this study area.
Within the hummocky terrain of the very extensive ice-
marginal zone, several systems of terminal moraines can be
identified. An important feature is the incorporation of large
amounts of glaciolacustrine clay and silt into the moraine,
evidencing that a large proglacial lake filled the glacial
trough before and during the late Pleistocene ice advances
and associated retreat stages. During glacial advances these
sediments were pushed together, forming glaciotectonic
moraine complexes. Profile K1 is situated in laminated
lacustrine sediments which are exposed within the M1c
moraine. These lake sediments show micro-fault structures
and are likely to have been transported en bloc, possibly as
a frozen slab. OSL (post-IR) dating of these sediments
yielded ages of 57.8�9.1 ka (silt) and 85.5�10.4 ka (clay),
consistent with our morphostratigraphic findings which
indicate that formation of the M1c moraine at Khurgan nuur
occurred during the early part of the last glaciation (MIS4).
Khoton nuur lies a few kilometres upstream of Khurgan
nuur and is dammed by a cross-valley terminal moraine
which itself is connected to a glaciofluvial outwash fan that
extends into the glaciolacustrine sediments of Khurgan nuur.
At the northern side of Khoton nuur, very extensive sets of
stacked lateral moraines and kame terraces are found (see
Fig. 5). Section K2 is situated down-valley of Khoton nuur,
where the river cuts a 13m deep channel across the M1a
Lehmkuhl and others: Glacier fluctuations in western Mongolia 173
http://dx.doi.org/10.3189/2016AoG71A030
Downloaded from http:/www.cambridge.org/core. SUB Gottingen, on 22 Dec 2016 at 13:44:38, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
moraine. Located 1.2m below the surface, a layer of silt and
sand was dated (OSL) to >13.6� 1.6 ka, supporting our
interpretation that the M1a moraine formed during the Late-
glacial phase.
The calculation of the LGM �ELA for this area is
complicated, because large valley and plateau glaciers,
originating from different glaciation centers, coalesced
within the Khoton–Khurgan trough. At present, ELAs de-
crease eastward from about 3400 to 3220ma.s.l. (at Khovd
gol, 3790ma.s.l.), while the elevation of the intermontane
basin itself limits the lowest position of the maximal terminal
moraine to 2050ma.s.l. This would indicate a general late
Pleistocene (LLGM) ELA at �2900ma.s.l., resulting in a
�ELA of 400m for Khurgannuur glacier.
3.4. Results from other study sites in the Mongolian
Altai
Mapping by Lehmkuhl (1998) in the Turgen-Kharkhiraa
mountains (Fig. 1a, No. 4) found almost no moraines
preserved between the above-described LIA limits and the
late Pleistocene LLGM ice margin. Based on luminescence
dating of aeolian cover beds overlying the local moraines,
alluvial fans and lake terraces, Grunert and others (2000)
suggested two major ice advances in the area and correlated
them to expanded glaciation during MIS2 and 4. Pötsch and
others (2015) presented new CRN dating results from the
Kharkhiraa gol valley supporting this view. Their results
indicate major ice advances during early MIS4 (74–71 ka)
and MIS2 (25–20 and 18–17 ka). During the late Pleistocene,
Fig. 4. Schematic cross sections showing the connection between ice-marginal moraines, terraces and aeolian cover beds at Tsengel
Khairkhan.
Fig. 3. Geomorphic map of late Quaternary glaciation and sampling of the Tsengel Khairkhan. Background image: SRTM DTM and
Landsat 8 image from 9 June 2013.
Lehmkuhl and others: Glacier fluctuations in western Mongolia174
http://dx.doi.org/10.3189/2016AoG71A030
Downloaded from http:/www.cambridge.org/core. SUB Gottingen, on 22 Dec 2016 at 13:44:38, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
local glaciers extended down to 1950ma.s.l. in the north
and to 2250ma.s.l. in the south. The modern ELA in the
Turgen-Kharkhiraa is at �3550m and we calculated a
spatially averaged LGM �ELA of �520m (430–600m).
The Tsambagarav is an isolated mountain massif (central
Mongolian Altai; Fig. 1a, No. 5) bordered by active faults. In
the northwest, river incision at the mountain front provides a
10m section exposing a sequence of fanglomerates. At the
base a �1m layer of fluvial sand separates the fanglome-
rates from underlying till. This layer was IRSL-dated to
57.0�17.0 ka (Klinge, 2001) and places the till into the
beginning of the last glacial cycle (M1c; MIS4) or the pen-
ultimate glaciation (M2). The Pleistocene �ELA was 575m.
At Munkh Khairkhan (southern Mongolian Altai; Fig. 1a,
No. 6) the glacial morphology resembles the morphostrati-
graphic patterns described from previous sites (Klinge,
2001). At the northern slope an intermontane basin without
outflow was blocked by M1 and M2 lateral moraines. The
recent lake has cut a section of 4.4m into fine clastic lake
deposits alternating with peat layers. The radiocarbon
analysis of organic material yielded calibrated 14C ages
between 28.9 and 42.2 ka BP (Klinge, 2001), in which the
higher lake levels indicate moist, warm conditions during
MIS3. During the LLGM the �ELA was �500m.
In summary these findings contribute more OSL dates to
support the large glacier extension during MIS3 at Otgon-
tenger, and previously undocumented LLGM during MIS2
and 4 in western Mongolia.
4. DISCUSSION
4.1. Extent of modern, Holocene and late Quaternary
glaciations
For the wider study area and under present conditions,
glaciation is more extensive in the Russian Altai, with a
Fig. 6. Schematic cross sections showing the connection between ice-marginal moraines, terraces and aeolian cover beds in the area
surrounding the Khurgan nuur.
Fig. 5. Geomorphic map of late Quaternary glacial landforms and sampling sites in the surrounding of the Khurgannuur area. Background
image: SRTM DTM and Landsat 8 image from 31 August 2014.
Lehmkuhl and others: Glacier fluctuations in western Mongolia 175
http://dx.doi.org/10.3189/2016AoG71A030
Downloaded from http:/www.cambridge.org/core. SUB Gottingen, on 22 Dec 2016 at 13:44:38, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
modern ELA between 2660 and 3000m. In the central Altai,
only two larger valley glaciers are present on the eastern
side of Tavan Bogd, with the glaciated area becoming
limited eastwards towards the Mongolian Altai due to
reduced precipitation in the lee of the Russian Altai. In the
Mongolian Altai separated and smaller glaciation centers,
each comprising plateau, cirque and smaller valley glaciers,
indicate that the modern ELA rises eastwards from 3300m to
>3800m.
Extensive Holocene moraines, including the morpho-
logically prominent LIA ice margin, are common features
throughout the Altai region. For the study areas discussed in
this paper, the �ELA during the LIA was 80–30m. However,
the total glaciated area was nearly twice as large as current
values. There is also evidence for the trend of smaller �ELAs
and ice extension during the LIA in the more arid regions of
the Altai.
In many parts of the formerly glaciated areas of the
Mongolian Altai the Late-glacial M1a terminal moraines are
rare and often missing, while there are several valleys
preserving extensive retreat stages in the Khangai moun-
tains, most prominently in valleys on the eastern slopes of
Otgontenger. In between these two regions the maximum
ice advance reached terminal ice positions within the limits
of the mountain massifs, but in some cases glaciers reached
the foreland areas (e.g. Tsengel Khairkhan, Ikh Turgen).
Pronounced valley glaciers (>50 km) occurred only in the
Tavan Bogd area (Khurgan nuur, Tsagaan gol). The max-
imum ELA lowering during the LGM is recorded for the
western Altai and Khangai regions (up to 1000m), while the
�ELA in the Mongolian Altai was 300–600m (Fig. 7).
4.2. Timing of late Quaternary glaciations and
paleoclimatic implications
Widespread remnants of the M2 moraines indicate a general
trend of slightly greater ice extent during the penultimate
glaciation (compared to the last glacial cycle). However,
more detailed mapping and sampling is still required to
accurately delineate the extent of ice and numerical
chronology for this period. In contrast, ice extent and
morphological characteristics of glacial features generated
during the last glacial cycle are well documented. Recent
advances in the numerical dating of glacial moraines in
Mongolia demonstrate that the timing of formation of the
M1c moraine, which commonly represents the LLGM, falls
into MIS4 for western Mongolia and into MIS3 for central
Mongolia. The underlying pattern shows that in the
Mongolian Altai large ice advances occurred during MIS4
(74–71 ka) and during MIS2 (25–20 and 18–17 ka), while
there appears to have been no significant ice expansion
during MIS3.
For the Russian Altai, Lehmkuhl and others (2007)
published four luminescence age determinations from sand
and silt layers between till units ranging from 28 to 19 ka
suggesting deposition during MIS2. In addition, Reuther and
others (2006) showed that outburst floods from formerly
glacier-dammed lakes occurred down the Ob river which
date to 15.8� 1.8 ka (10Be), suggesting that MIS2 glaciers
had retreated from their maximum extents at this time.
In the Kanas river valley (Chinese Altai; Fig. 1a, No. 7)
late Quaternary glacier fluctuations generated four different
moraine systems and associated glaciofluvial deposits. The
largest glacial advances occurred during MIS4 (�73 ka),
mid-MIS3 (52–38 ka) and MIS2 (28–16 ka) (Zhao and others,
2013). According to their results, the modern ELA lies
between 3020 and 3360ma.s.l. and the LIA moraines are
located �200m below the modern glacier margins,
suggesting a �ELA of �100m. During the LLGM (MIS4)
the �ELA was about 600–700m.
In northern Mongolia, in the Darhad basin, the major
advances of the LGM occurred during MIS2 (17–19 ka) and
MIS3 (35–53 ka) (Gillespie and others, 2008). For the
Otgontenger area in the western Khangai, CRN data
presented by Rother and others (2014) supported by new
OSL data (this study) suggest major ice expansion during
MIS3 (40–35 ka) and during MIS2 (at �23 and 17–16 ka).
In the larger context of understanding the signature of late
Quaternary glaciations across central Asia, an increasing
number of numerical dates from recent studies reveals the
irregular pattern of the extent and timing of the LLGM
(Rother and others, 2014). In high- to mid-latitude areas,
maximum late Pleistocene glacier advances are commonly
recorded during the global temperature minima associated
with MIS2 and MIS4, while at lower latitudes, particularly
where glaciation was limited by sparse precipitation, LLGM
limits were often reached during the warmer but wetter
conditions of the MIS3 interstadial. The hydrological limi-
tation in these semi-arid to arid regions of Mongolia can also
be observed in the development of permafrost features.
Although mountainous areas in western Mongolia generally
Fig. 7.West–east cross section of the Altai showing modern and Pleistocene glaciations and ELAs. The profile is along the transect shown in
Figure 1a (green line). Key sites shown in Figure 1.
Lehmkuhl and others: Glacier fluctuations in western Mongolia176
http://dx.doi.org/10.3189/2016AoG71A030
Downloaded from http:/www.cambridge.org/core. SUB Gottingen, on 22 Dec 2016 at 13:44:38, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
feature a mean annual air temperature below 0°C, which
supports periglacial forms and processes, they occur
exclusively at sites where additional groundwater is avail-
able under present conditions.
Local climatic conditions and their past variability have
exerted a major influence on the dynamics of glaciation in
the Altai region, particularly where the moisture supply is
the critical limiting parameter. To fully understand the
spatial and chronological pattern of glaciation in Mongolia
we consider it necessary to include past lake level
fluctuations, which can serve as a proxy for reconstructing
humidity variations and provide, at times of high lake
stands, an important source of additional humidity. For
example, elevated lake levels have been documented for the
Late-glacial phase and MIS3 periods across a wider region of
central Asia and particularly in Mongolia (Grunert and
others, 2000; Komatsu and others, 2001). Similarly,
evidence for high lake levels during these phases was found
at Tsetseg nuur (46°360N, 93°110 E), in the eastern Mon-
golian Altai, based on IRSL-dating of sandy beach walls
giving an age of 31.1� 3.8 ka, and a radiocarbon age on a
mollusk yielding an age of 13.2�0.3 cal. ka BP (Klinge and
Lehmkuhl, 2013). At Khar Us nuur (47°510N, 92°010 E), the
authors found aeolian sediments beneath a beach wall. The
beach wall represents a paleo-lake level �20m higher than
at present. An IRSL-dating of the aeolian sediment shows
17.1�3.6 ka and arranges this higher lake level into the
Late-glacial phase.
The rain-shadow effect in the Mongolian Altai explains
the increased aridity of this region and the limited extent of
glaciation, while the Russian and Chinese Altai receive
more precipitation, which is mainly delivered by the zonal
westerlies. Further to the east, in the Khangai, the air masses
may have absorbed additional humidity supplied from large
paleo-lakes. This combination may serve as an explanation
for the occurrence of Late-glacial phase and MIS3 stages in
central Mongolia.
5. CONCLUSION
Modern and LIA glaciers are restricted to mountains
reaching above 3400m. Coincident with the rising aridity
the ELA increases by 600m from the western part of the
Mongolian Altai to the east. Compared to its modern extent,
glaciation during the LIA was twice the size, with a �ELA of
30–80m.
Clear morphostratigraphy, geomorphology, and develop-
ment of distinct aeolian cover sheets provide evidence for
different Pleistocene glacier stages.
ELA reconstructions provide evidence for different paleo-
environmental conditions. There are regional differences
due to paleoclimatic gradients showing more aridity
(indicated by low �ELAs), particularly in the lee of the
(western) Altai. In addition, no recessional moraines of Late-
glacial phases are preserved, which indicates a lack of
humidity during the rapid warming after the LGM.
New dating results (OSL) provide an improved framework
for late Pleistocene glaciations in the Mongolian Altai and
Khangai. The timing of maximum glacial advances in the
Russian Altai, the Khangai and the Darhad basin appears to
be approximately synchronous, but differs from the pattern
observed in the Mongolian Altai and the Chinese Altai. This
supports the view that paleoclimatic conditions in central
Asia varied substantially.
It appears that glacial advances were common in
Mongolia during MIS4, MIS3 and MIS2. More numerical
results are required to understand glacial development in
Mongolia during the last glacial cycle.
ACKNOWLEDGEMENTS
We thank the Deutsche Forschungsgemeinschaft (HI 1398/
3-1 and LE 730/31-1) for financial support. Fieldwork was
supported by the Institute of Geography of the Mongolian
Academy of Sciences (D. Dorjgotov, A. Chimegsaikhan).
A. Ehrig and D. Falk helped to produce the maps. We thank
D. Falk, A. Hilgers, S. Pötsch, G. Stauch, M. Walther and
A. Zander for support during the fieldwork and scientific
discussions. In addition, we thank Graham Cogley and two
anonymous reviewers who helped greatly to improve the
paper.
REFERENCES
Balco G, Stone J, Lifton N and Dunai T (2008) A simple, internally
consistent, and easily accessible means of calculating surface
exposure ages and erosion rates from Be-10 and Al-26
measurements. Quat. Res., 69, 242–249
Benn DI and Lehmkuhl F (2000) Mass balance and equilibrium-line
altitudes of glaciers in high mountain environments. Quat. Int.,
65/66, 15–29
Blomdin R and 11 others (2014) Glacial geomorphology of the
Altai and Western Sayan Mountains, Central Asia. J. Maps, 10
(doi: 10.1080/17445647.2014.992177)
Bøtter-Jensen L, McKeever SWS and Wintle AG (2003) Optically
stimulated luminescence dosimetry. Elsevier, Amsterdam
Dunai T (2001) Influence of secular variation of the geomagnetic
field on production rates of in situ produced cosmogenic
nuclides. Earth Planet. Sci. Lett., 193, 197–212
Fuchs M and Owen LA (2008) Luminescence dating of glacial and
associated sediments: review, recommendations and future
directions. Boreas 37, 636–659
Galbraith RF and Green PF (1990) Estimating the component ages
in a finite mixture. Nucl. Tracks Radiat. Meas., 17, 196–206
Galbraith RF, Roberts RG, Laslett GM, Yoshida and Olley JM (1999)
Optical dating of single and multiple grains of quartz from
Jinmium Rock Shelter, Northern Australia: Part I, Experimental
design and statistical models. Archaeometry, 41, 339–364
Gillespie AR, Burke RM, Komatsu G and Bayasgalan A (2008) Late
Pleistocene glaciers in Darhad Basin, northern Mongolia. Quat.
Res., 69, 169–187
Grunert J, Lehmkuhl F and Walther M (2000) Palaeoclimatic
evolution of the Uvs Nuur Basin and adjacent areas (Western
Mongolia). Quat. Int., 65/66, 171–192
Heyman J (2014) Paleoglaciation of the Tibetan Plateau and
surrounding mountains based on exposure ages and ELA
depression estimates. Quat. Sci. Rev., 91, 30–41
Hülle D and 7 others (2010) OSL dating of sediments from the Gobi
Desert, Southern Mongolia. Quat. Geochronol., 5, 107–113
Huntley DJ and Baril MR (1997) The K content of the K-feldspars
being measured in optical dating or in thermoluminescence
dating. Ancient TL 15, 11–13
Kamp U, Krumwiede BS, Mc Manigal KG, Pan CG, Walther M and
Dashtseren A (2013) The glaciers of Mongolia. INSTAAR Occas.
Pap. 61
Klinge M (2001) Glazialgeomorphologische Untersuchungen im
Mongolischen Altai als Beitrag zur jungquartären Landschafts-
und Klimageschichte der Westmongolei. Aach. Geogr. Arb., 35
Klinge M and Lehmkuhl F (2013) Geomorphologic map of Tsetseg
Nuur Basin, Mongolian Altai – lake development, fluvial
sedimentation and aeolian transport in a semi-arid environment.
J. Maps, 9, 361–366
Lehmkuhl and others: Glacier fluctuations in western Mongolia 177
http://dx.doi.org/10.3189/2016AoG71A030
Downloaded from http:/www.cambridge.org/core. SUB Gottingen, on 22 Dec 2016 at 13:44:38, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.
Komatsu G, Brantingham PJ, Olsen J and Baker V (2001)
Paleoshoreline geomorphology of Böön Tsagaan Nuur, Tsagaan
Nuur and Orog Nuur: the Valley of Lakes, Mongolia. Geo-
morphology, 39, 83–98
Krumwiede BS, Kamp U, Leonard GJ, Kargel JS, Dashtseren A and
Walther M (2014) Recent glacier changes in the Mongolian Altai
Mountains: case studies from Munkh Khairkhan and Tavan
Bogd. In Kargel JS, Leonard GJ, Bishop MP, Kääb A and Raup B
eds Global Land Ice Measurements from Space. Praxis-Springer,
Berlin, 229–239
Lehmkuhl F (1998) Quaternary glaciations in central and western
Mongolia. Quat. Proc., 6, 153–167
Lehmkuhl F (2012) Holocene glaciers in the Mongolian Altai: an
example from the Turgen-Kharkhiraa-Mountains. J. Asian Earth
Sci., 52, 12–20
Lehmkuhl F, Klinge M and Stauch G (2004) The extent of late
Pleistocene glaciations in the Altai and Khangai Mountains. In
Ehlers J and Gibbard PL eds Quaternary glaciations – extent and
chronology, Part III: South America, Asia, Africa, Australia,
Antarctica. Elsevier, Amsterdam, 243–254
Lehmkuhl F, Frechen M and Zander A (2007) Luminescence
chronology of fluvial and aeolian deposits in the Russian Altai
(Southern Siberia). Quat. Geochronol., 2, 195–201
Lehmkuhl F, Klinge M and Stauch G (2011) The extent and timing
of Late Pleistocene Glaciations in the Altai and neighbouring
mountain systems. In Ehlers J, Gibbard PL, Hughes PL and
Hughes PD eds. Developments in Quaternary Science: extent
and chronology – a closer look, 15, 967–979
Lehmkuhl F, Hülle D and Knippertz M (2012) Holocene
geomorphic processes and landscape evolution in the lower
reaches of the Orkhon Valley (northern Mongolia). Catena 98,
17–28
Osipov EU, Ashmetev AP, Osipova ON and Klevtsov EB (2013)
Novaya inventarizatsiya lednikov v yugo-vostochnoj chasti
Vostochnogo Sayana. Led i Sneg, 123, 45–54 [in Russian]
Pfeffer WT and 75 others (2014) The Randolph Glacier Inventory: a
globally complete inventory of glaciers. J. Glaciol., 60(221),
537–551 (doi: 10.3189/2014JoG13J176)
Pötsch S, Rother H, Lorenz S, Walther M and Lehmkuhl F (2015)
Timing of late Pleistocene glaciation in Mongolia: surface
exposure dating reveals a differentiated pattern of ice advances.
Geophys. Res. Abstr., 17
Prescott JR and Hutton JT (1995) Environmental dose rates and
radioactive disequilibrium from some Australian luminescence
dating sites. Quat. Sci. Rev. 14, 439–448
Reuther AU, Herget J, Ivy-Ochs S, Borodavko P, Kubik PW and
Heine K (2006) Constraining the timing of the most recent
cataclysmic flood event from ice-dammed lakes in the Russian
Altai Mountains, Siberia, using cosmogenic in situ 10Be.
Geology 34(11), 913–916
Rother H, Lehmkuhl F, Fink D and Nottebaum V (2014) Timing of
late Pleistocene glaciations in the Khangai Mountains of
Mongolia: evidence for a MIS 3 glacial maximum. Quat. Res.,
82, 297–308
Thiel C and 6 others (2011) Luminescence dating of the Stratzing
loess profile (Austria) – testing the potential of an elevated
temperature post-IR IRSL protocol. Quat. Int. 234, 23–31
Zhao J, Yin X, Harbor JM, Lai Z, Liu S and Li Z (2013) Quaternary
glacial chronology of the Kanas River valley, Altai Mountains,
China. Quat. Int. 311, 44–53
Lehmkuhl and others: Glacier fluctuations in western Mongolia178
http://dx.doi.org/10.3189/2016AoG71A030
Downloaded from http:/www.cambridge.org/core. SUB Gottingen, on 22 Dec 2016 at 13:44:38, subject to the Cambridge Core terms of use, available at http:/www.cambridge.org/core/terms.