|Mongolia belongs to one of the third regions in the world due to its pronounced continentality. Thus, its sparse water resources and fragile ecosystems make it vulnerable to climatic changes in the past and future. Nevertheless, compared with East and South Asia, the knowledge about Pleistocene, Late Glacial and Holocene climatic fluctuations in “central arid asia” (Chen et al., 2008) is still sparse and inconsistent. The aim of the present study is to deliver new data about past climate and landscape evolution in Mongolia by dating sediments with optically stimulated luminescence.
A macroscopic approach was chosen to derive a general picture of the palaeo environment from sedimentary geo-archives. Four study areas were investigated. Each of them represents an individual landscape unit with its own history, as it is an endorheic basin (Valley of the Gobi lakes), a dune field (Mongol Els, Khongoryn Els) and an area near a historical center for which a high anthropogenic impact can be assumed (Orchon Valley near Kharkhorin). Six to seven profiles were sampled in each study area to compare the data for a comprehensive interpretation.
In the first step, appropriate measurement parameters to derive reproducible ages had to be elaborated. In general, quartz is presently the mineral favored for luminescence dating, as it is the most commonly occurring mineral in sediments, very resistant to weathering and exhibits a stable signal for dating. But in this study, OSL dating of the quartz fraction is complicated due to feldspar contamination. A clear signal was observed for all samples when stimulated with IR although quartz should not be sensitive to these stimulation wavelengths. X-ray diffractometry, raster electron microscopy and tracer element spectroscopy on samples from the dune field Khongoryn Els indicate that plagioclase is adherent to the quartz grains. This is further supported by RFA analyzes by Kozlovsky et al. (2006), which conclude that peralkaline and alkaline granites are the main contributors to the sediment production in Southern Mongolia. This feldspar contamination could not be eliminated by sample preparation (additional etching, flotation etc.) or measurements techniques (“post-IR-OSL”, Banerjee et al. 2001; Pulsed OSL, Thomsen et al. 2008). Consequently, potassium-rich feldspars were used for dating in preference to quartz.
The measurement parameters for De determination were determined based on the SAR protocols for K-rich feldspar extracts described by Wallinga et al. (2000). Based on the results of preheat tests, 270 ° C for preheating of the regeneration dose as well as the test dose was chosen.
Two major difficulties had to be overcome to derive usable age results: First, at the end of the study, "optical cross-talk" was discovered to have occurred during the measurements. This term refers to the unwanted illumination of a sample during the automated measuring procedure: While measuring one aliquot in the TL / OSL Reader, the adjacent sample position is affected by the stimulation light to a certain degree. The problem was quantified by Bray et al. (2002) for quartz measurements, indicating that the effects can usually be disregarded if care is taken with the measurement sequence design. However, for our field spar measurements the sequence has to be designed in a way that the effect of “optical cross talk” for each individual stimulation accumulates for the adjacent sample. Consequently, the measured natural IRSL signal is reduced to a degree that is dependent on the used measurement equipment and the sample. Age underestimations between 0 and 50% were observed. We corrected for the effect of optical cross talk individually for every sample, based on measurements using a wider spacing of the aliquots in the reader.
Secondly, for K-feldspars, a loss in the luminescence signal called anomalous fading has been a problem in many dating studies (Aitken 1985, Balescu et al. 2001; Huntley and Lamothe 2001; Lamothe and Auclair 1999, Wintle 1973). Fading rates were determined following Auclair et al. (2003). To calculate fading corrected De values the procedure proposed by Huntley and Lamothe (2001) was used. For samples with ages> 100 ka, fading correction is complicated, as the natural signal is outside the linear part of the growth curve. Thomsen et al. (2008), Buylaert et al. (2009), Thiel et al. (2011) and Stevens et al. (in press) showed that the post-IR IRSL signal using elevated stimulation temperatures has significant potential to derive a dating result that is far more independent from fading-correction than the conventional IRSL50 ° C-signal. Therefore, the protocol suggested by Thiel et al. (2011) was selected for “old” samples, using a stimulation temperature of 290 ° C after preheating at 320 ° C and an IR bleach at 50 ° C. The general quality criteria indicate applicability of the protocol. To further evaluate the applicability of the protocol, the fading corrected results of the IRSL50 ° C-SAR and the post-IR-IR290 ° C-SAR protocols were compared for some samples in the age range 10-30 ka. In this age range, the results derived by the two methods agree within error limits.
For some samples, incomplete bleaching could also be detected. Hence, for those samples, the equivalent dose was calculated with the finite mixture model (Galbraith & Green 1990) instead of the central age model used generally for the well-bleached samples.
To calculate the exposure to radioactivity per time unit (dose rate), the concentration of uranium, thorium, and potassium was determined by laboratory high resolution gamma-spectrometry. For some of the samples, there is indication for radioactive disequilibria in the 238U-decay chain - a concluding evaluation would only be possible after determination of 238U / 234U, which was not feasible in the present study. Nevertheless, the age uncertainties caused by radioactive disequilibria appear to be negligible compared to the methodical uncertainties cited above.
In spite of the described methodical difficulties, with an average relative error of 12%, the resulting ages are comparatively precise. Additionally, a certain confirmation for the reliability of the ages is provided by the following facts / arguments: a) The only 14C-age which is suitable as age control fits well to the IRSL-ages of the profile “ME-S”. b) The ages of almost all profiles show a reasonable chronostratigraphy without inversions. c) Results of K-Feldspar samples that were measured with different methods (SAR-IRSL50 ° C, Post-IR-IRSL290 ° C, Post-IR-YOSL260 ° C in Freiberg) agree within error limits.
The deduction of geomorphological processes from the derived IRSL-ages is thus feasible, keeping in mind that within the identified phases of activity the accuracy of ages is unassured.
Some of the aspects of landscape reconstruction can be summarized as follows: In the endorheic basin of the lake Bayan Tochom (study area “Khongoryn Els”) an evolution of the surface of the alluvial fans reaching into the basin during LGM and later could be determined . The aeolian sediments within the basin are younger than ~ 3 ka. For this time period, increasing aridity is corroborated in the other study areas, too. Within the recently active dune fields, silty sediments could be identified that are remnants of interdune-ponds - contrary to the interpretation in the field, IRSL dating could prove that they do not represent relicts of high lake levels of the Bayan Tochom.
In the northeastern part of the study area “Khongoryn Els”, a 23 m high section of alternating dune sand and silty water-lain sediments was investigated (profile US). The data imply that the basal aeolian sediments were deposited 27 ka ago. Samples between 20 and 15 m depth could not be taken, but it is assumed that this part of the section represents an extremely arid time period with an intensive remobilization of sand around LGM. The major part of the profile was accumulated in a rather short period around ~ 15 ka. The temporal differentiation implies that the stratification of aeolian and fluvial sediments is not caused by long term climatic variations. It rather represents arid conditions with episodical fluvial activity.
At the transition from Pleistocene to Holocene, the dune field was probably intersected by the river Khongoryn Gol. At that time, sedimentation at the profile US ended, but fluvial and flood sediments were deposited at the northern side of the dune field (Profile KG). The ages of lacustrine sediments in the terminal playa of the river imply that around 8 to 7 ka, the basin was filled with water. A dune overlying the section US was deposited during the Late Holocene (<1.5 ka) and reflects ongoing aridity in this region.
In the lake basin of the Oroog Nuur all sediment profiles are of Holocene age. During the middle Holocene, around 7-6 ka, the lake level was 23 m above the present lake level, documented by IRSL-ages from sediments within and above a beach wall. At ~ 4.6 ka the water level must have been at least 13 m lower, as aeolian sediments were deposited at the 10 m level at that time. Since 2 ka, aeolian sedimentation due to increased aridity was prevalent. Similar to the results in the Bayan Tochom basin, silty sediments of modern age, left by interdune-ponds, could be identified.
Interpretation of the age results of a beach wall in the Adgin Tsagaan Nuur lake basin is not straightforwad: Coarse sediments at the base of the profile which are probably of fluvial origin were deposited 33 ka ago, samples on top of the profile implicate that the formation of the beach wall was more or less concluded cicra 12 ka.
The investigations undertaken in the third study area at the eastern rim of the dune field Mongol Els have revealed a complex interaction of aeolian and fluvial / lacustrine morphodynamics during Late Pleistocene and Holocene. Lake formation was identified for early MIS 5 and for the middle Holocene. During the youngest Holocene humidity decreased. No more permanent ponding effect of the dunes was detectable. Only flooding by irregular high discharge of the rivers can be proved in the lowest part of the floodplain in the dune foreland.
At two profiles that were exposed by the river Savhan Gol, post-IR-IR290 ° C-ages imply a high geomorphodynamic activity during the transitions from MIS 7 to MIS 6 and from MIS 6 to MIS 5.
In the fourth study area, terrestrial records from a mountainous steppe region in the upper Orkhon Valley close to Kharkhorin and the ancient Mongolian capital Karakorum were investigated. The sedimentation of sandy sediments and gravel in this area started in the Late Glacial. During the Early Holocene, a phase of pedogenesis was identified, interrupted by the covering with sand. A further phase of soil development probably occurred during early to middle Holocene. Geomorphological activity in the Orchon Valley is enhanced in the late Holocene. Increased sedimentation on terraces as well as on the slopes is probably of aeolian as well as colluvial-alluvial origin. Remobilization of sands is further accelerated due to human impact, especially during the reign of Genghis Khan (~ 0.8 ka).
A significant shortcoming of the results presented is their point distribution. This only partly enables an overall landscape reconstruction in the study areas. Achieved results would have to be verified systematically in further research projects. Furthermore, a wider variety of methods such as micromorphology, grain size analysis or scanning electron microscopy would ideally be applied to characterize the sediments.
Taking these constraining conditions into account, the process described dynamic is linked to a tentative area-wide palaeoclimate reconstruction. The climatic phases which can be relatively reproducibly determined include:
a) A comparably dry late Glacial (18 ka until circa 11/10 ka) with episodic precipitation
b) A climatic optimum during the middle Holocene (8 ka up to circa 5/4 ka) with high lake levels, water-filled depressions as well as soil formation in the Orchon Valley.
c) A change to three conditions in the time span between circa 5 ka and 3 ka with increased colluvial activity.
d) Increased aridity since circa 3 ka reflected by dune activity at several study sites.
e) Amplified anthropogenic influence in the working area of the Orchon valley around about 0.8 ka.||English|