1 Introduction

The Tibetan Plateau (TP) is the highest plateau on Earth (reaching a height of over 4,200 m and covering about 2.5×10⁶ km²), and strongly influences the climate environment of nearby regions, East Asia and the whole globe as well (Wu et al., 2015 ). TP is one of the most sensitive regions in respect to global warming, as there is an increase in warming with altitude (Holmes et al., 2009 ). Intensive land degradation, desertification and aeolian activity since the Little Ice Age (LIA, during about 1350–1850 A.D.) are widely observed over the TP (Jin et al., 2007 ; Yu et al., 2017 ). The effects of climate change, the status of land degradation, permafrost warming, glacier shrinkage, desertification, among others, on the TP has received increased attention by researchers and society (Cheng and Wu, 2007; Xue et al., 2009 ; Yao et al., 2012 ).

Some important climate reconstruction projects over the TP were carried out, in which a limited number of proxy data were used to represent this vast area (Yang et al., 2003 ). Overall warming conditions over the TP has occurred over the past few centuries as found in several lakes over the TP, and the warming rate increases with elevation, with the highest rate between 4,800 and 6,200 m (including the mountain area in this study) (Liu and Chen, 2000; Qin et al., 2009 ). However, a complete understanding of past environmental and climate conditions on the TP is still lacking. The density of data points over the TP is currently too low for identifying spatial patterns of any changes, because previous studies were performed at various locations using different climate proxies and of different temporal resolution (Bräuning, 2006; Holmes et al., 2009 ). Large deviation may arise given the strong variation of altitude and local climate over the Plateau. Data with high-confidence and high temporal interval in the central TP area is generally harder to obtain compared to other areas because most studies were concentrated on the northeastern and southern TP (Stauch, 2015). A new ice-core record was used to reconstruct the temperature variation for Geladaindong Mountain during the LIA and 20th-century (Zhang et al., 2016 ), but studies in the central TP area are still under-represented compared to other TP regions and needs more reinforcement.

Sediment deposition contains traces of past climate, environmental evolution, and responses to global change (Mügler et al., 2010 ). Sedimentary information has been used in some studies to evaluate long-period paleoclimate at different lakes including Bangong and Sumxi in the west (Gasse et al., 1991 , 1996), Nam, Siling, Cuoe and Ahung in the central area (Morinaga et al., 1993 ; Mügler et al., 2010 ; Morrill et al., 2006 ; Wu et al., 2006 ), Kuhai and Qinghai in the northeast (Shen et al., 2005 ; Mischke et al., 2010 ; An et al., 2012 ; Lu et al., 2015 ), Xinmocun in the east (Liang and Jiang, 2017), and Sugan in the Qaidam Basin of northern TP (Chen et al., 2009 ). Also, aeolian deposits have been recognized as providing information about past environments (Liu et al., 2011 , 2013, 2015; Ijmker et al., 2012 ; Yu and Lai, 2014; Qiang et al., 2016 ) (Figure 1a). However, sediment reconstruction studies are considered highly uncertain and of low resolution in chronology (Holmes et al., 2009 ).

Recently, active wind erosion and desertification to the east side of Lake Co Nag in the central TP area has received increased attention. This is due to increased aeolian activity, resulting in significant sand accumulation which could obstruct the Qinghai-Tibetan Railway (QTR) (Zhang et al., 2010 ). The sands could be mobilized under the effects of freeze-thaw and wind erosion (Xie et al., 2016 ). Recorded climate data and model projections suggest that temperature and precipitation in the area increased over the past six decades and the trend may continue in the near future (Liu et al., 2016a ). Unfortunately, research is lacking on past climate before the recording of meteorological data in the 1950s in the region of Lake Co Nag. Evaluating historic climate conditions over a longer period will help in understanding the cause and driving force of the rapid change of land surface conditions in the fragile environment on the TP.

In this study, we examined an aeolian sand sequence at the shore of Lake Co Nag with optically stimulated luminescence (OSL) age dating, grain-size and geochemical analysis. Grain-size and geochemical parameters, and depositional rate of sand sediments were used to understand climate change in the central TP. Our purpose is to reconstruct the past climate and to reveal the response of land surface process (aeolian activity) to climatic change, as well as to understand the future development of desertification and aeolian sand deposition problems. We demonstrate that in this region that lacks a rich climate proxy, aeolian sediments can be used to study the climatic history and geomorphic processes of central TP.

2 Materials and methods 2.1 Study area and sampling

The study area is located near to the QTR and at the east shore of Lake Co Nag in Anduo County, Tibet (Figure 1a). Lake Co Nag is the head of the Nujiang River (also called the Salween or Thanlwin River), which runs through Southwest China to Burma and Thailand. The area is characterized by high wind and pervasive wind-blown sand activity. The average wind speed (over multiple years) is 4.1 m/s, and there are 150 days per year with high wind speeds (>17.2 m/s) (Liu et al., 2016a ). Peak wind speed is 38 m/s in this area which is significantly higher than any other regions along the QTR (Zhang et al., 2010 ).

To the east of QTR, a sandy area that occupies approximately 180 km² covers the hills that were carved by extensive seasonal gullies and streams flowing into the lake (Figure 1b). The land surface is covered by short grass in the summer, but the sandy layers are continually being stripped away by aeolian process, leaving wind erosive pits and isolated sedimentary piles that still has grass growing on top. As a result, land degradation is far worse than can be seen from satellite images. The dislocated materials can be found along gully banks and in the sand control systems at the railway sides in the downhill direction towards the lake, clearly indicating a transportation of deposition from the mountain to the lake. However, statistics (sand drift potential) on the energy of surface winds in terms of sand movement shows that the westerly wind from the lake to the mountain dominates. This means a strong interplay of sand transportation from the mountain and the lake, that is, the prevailing westerly winds moves the sands from the lakeshore to the hilly areas, while fluvial erosion and secondary-prevailing winds from the east transport sands towards the lake following the terrain downhill.

The altitude is about 4,800 m and annual mean temperature is −2.57 °C. Records from the meteorological station in Anduo County, 23 km northeast of Lake Co Nag, show that annual precipitation is 446 mm with 70% in summer, 20% in autumn, 9% in spring and only about 1% in winter. The temperature raising trend is 0.357 °C per decade from 1966–2013 and 0.491 °C per decade from 1979–2013, which is almost twice the reported global land surface air temperature increase from CRUTEM (Climatic Research Unit TEMperature) and Berkeley models (Christensen et al., 2013 ). The warming may lead to a temperature increase of top permafrost, the melting of soil ice, the drying and loosing of surface sand, and thus further surface degradation and desertification.

An isolated outcropping sandy mound approximately 4 m above the water surface was found on the eastern lakeshore sand field (91°32'22"E, 32°04'09''N) (Figure 1c). The surrounding material had been eroded to form part of the sediment along the east railway or carried down west to the low reach of the lake. In August 2014, a 2.7-m sampling profile beginning at the top of the mound was created; samples were collected using a hand shovel every 5 cm interval from the bottom of the profile. The profile was composed of purely sand sediments. In total, 57 sand samples weighing 500 g each were collected. Meanwhile, six OSL samples to determine depositional ages were collected from the profile (Figure 1c). The six samples were obtained by hammering the samplers (aluminum tubes of 30-cm long and 5 cm in diameter) into a freshly created vertical section and sealing them with black plastic tape to avoid light exposure and moisture loss.

2.2 Grain-size analysis

Grain-size and geochemical analyses are two convenient methods to study sediment features and their depositional environment (Liu et al., 2014 ; Liu et al., 2015 ). The grain-size distribution of all the samples was determined between 0.02 and 2,000 μm by a Malvern Mastersizer laser grain-size analyzer in Lanzhou University, China. The accuracy and reproducibility of the analyzer was better than 1%, respectively. Values were converted to Φ (phi) units using the equation Φ=−log₂(d), where d is the grain-size diameter in mm. Before each measurement, chemical pretreatment following the procedure in Appendix C of Konert and Vandenberghe (1997) was performed to isolate discrete particles, removal of organic matter and carbonates using H₂O₂ and HCl, respectively, and ensure an evenly dispersed suspension.

Grain-size parameters including mean grain size (M_z), standard deviation or sorting (σ), skewness (SK₁), and kurtosis (K_G) were calculated based on the method of Folk and Ward (1957). Sediment depositional environment can be determined by using a statistical method based on grain-size parameters.

Linear discrimination using grain-size parameters is an effective way to distinguish aeolian sediments. The discriminant functions proposed by Sahu (1964) were applied by several authors to the sand samples taken from the Kumtagh Desert (He et al., 2009 ; Dong et al., 2011 ; Liu et al., 2014 ). The four functions used are as follows:

${Y_1} = - 3.5688{M_{\textit {z}}} + 3.7016{\sigma ^2} - 2.0766S \!\!K_1 + 3.1135{K_{{G}}}$

(1)

Y₁ is less than −2.741, 1 indicates aeolian deposition. Otherwise, proceed to Equation (2).

${Y_2}\! =\! 15.6534{M_{\textit {z}}} + 65.7091{\sigma ^2} + 18.1071S \!\!{K_1} \!+ 18.5043{K_G}$

(2)

Y₂ is less than 65.365, 0 indicates beach deposition. Otherwise, proceed to Equation (3).

${Y_3} = 0.2852{M_{\textit{z}}} - 8.7604{\sigma ^2} - 4.8932S \!\!{K_1} + 0.0482{{K}_G}$

(3)

Y₃ is greater than −7.419, 0 indicates shallow marine deposition. Otherwise, proceed to Equation (4).

${Y_4} = 0.7215M_{\textit{z}} -0.4030{\sigma ^2} + 6.7322S\!\!{K_1} + 5.2927{K_G}$

(4)

Y₄ is less than 9.843, 3 indicates turbidity current deposition; otherwise, it is fluvial deposition.

The four grain-size parameters were substituted into these functions. The results present clear differences among different sediments.

2.3 Geochemical and chronological analysis

A fully automated sequential wavelength dispersive X-ray fluorescence spectrometer (AXIOS, PANalytical B.V., the Netherlands) was used for elemental analysis at the Key Laboratory of Desert and Desertification, Chinese Academy of Sciences (KLDD-CAS). Accessories include a semi-automatic press machine, grinding mill with tungsten carbon, and closed circuit cooling unit (ZHY-401A, ZHM-1A, BLK2-8FF-R, respectively, Zhonghe Corporation, Beijing, China). The samples were crushed into a powder finer than 75 μm using a multipurpose grinder and dried in an oven at 105 °C. Next, 4 g of the dry, powdered sample was pressed into 32 mm-diameter pellets under 30 tons of pressure using the pressed powder pellet technique. The briquettes were then stored in desiccators and tested in the spectrometer. Thirty chemical elements served as tracers were analyzed including Cl, P, Ti, V, Cr, Mn, Co, Ni, Cu, Zn, Ga, As, Br, Rb, Sr, Y, Zr, Nb, Ba, La, Ce, Nd, Pb, SiO₂, Al₂O₃, Fe₂O₃, MgO, CaO, Na₂O, and K₂O. The oxide concentrations were in percentage, while other elements were in units of μg/g.

Several geochemical parameters including chemical index of alteration (CIA), SiO₂/TiO₂ and CaO/MgO were calculated based on the chemical analysis data. The CIA index is calculated as:

${\rm{CIA}} = \left[ {{\rm{A}}{{\rm{l}}_2}{{\rm{O}}_3}/\left({{\rm{A}}{{\rm{l}}_2}{{\rm{O}}_3} + {\rm{CaO}}^* + {\rm{N}}{{\rm{a}}_2}{\rm{O}} + {{\rm{K}}_2}{\rm{O}}} \right)} \right] \times 100$

(5)

where the materials are expressed by their molar fraction (m). The value of CaO* is taken from m(Na₂O) if m(CaO) > m(Na ₂O). This is done to consider CaO only from silicates while removing the content in carbonate or phosphate, as the molar fractions of CaO and Na₂O are usually equal in silicates.

The OSL measurements were carried out using an automated RisøTL/OSL-DA-15 reader, also in the KLDD-CAS. The OSL signal from quartz was detected through one 5-mm thick Hoya U-340 filter. The IRSL signal from feldspar was detected by combining one BG-39 and one Coring-759 filters. Laboratory irradiation was carried out using ⁹⁰Sr/⁹⁰Y sources mounted within the reader with a dose rate of 0.084 Gy/s. Details on the procedures of OSL measurements can be found in Zhao et al. (2015) . OSL results are presented in Table 1.

3 Results 3.1 Grain-size and parameters

The profile samples are coarse in particle size with particles >63 μm being at least 98.3% and an average of 99.7%. No samples have particles of <16 μm. No clay and silt were found in all samples, and the soil texture of all the sediments is classified as sand. The sediments are classified into fine sand group (72.9%±4.5%), followed by very fine sand group (14.6%±4.0%) and medium sand group (12.2%±3.5%).

Deposition environment discrimination analysis was applied to all the samples. Results indicate that all samples were formed under an aeolian environment except for the third from the top sample (No. 55), which has a higher coarse particle ratio and tends to be shallow marine deposition. This outlier might be impacted by stronger lake waves in the near past compared to other layers, but overall deposition materials should still be transported by an aeolian process.

The uniform sediment environment and source are further indicated by particle parameters. The scatter plots of σ, SK₁ and K_G against M_z all show close sorting, skewness, and kurtosis of the samples except for No. 55 sample outlier (Figure 2). Mean grain sizes are also concentrated in a narrow range of about 2.5 to 2.8 Φ, corresponding to 0.17–0.14 mm. The σ value ranges from 0.50 to 0.65, representing relatively good sorting (Figure 2a). SK₁ values of close to 0 indicate symmetric distribution of the sample particles (Figure 2b). K_G values of 0.90–1.05 show that the sediments have medium ratios of middle-size particles to the tail particles (Figure 2c). The concentrated grain size, sorting, skewness, and kurtosis all indicate that the sediments were separated by wind power. As a result, the variations of aeolian material characteristics can be used to reflect past aeolian activity conditions and perhaps project future desertification trends with additional information from climate data.

3.2 Geochemical elements

For tracer elements, all the layers are rich in Co and leaching in all others (Figure 3a). The Co element concentration is six times higher than the Upper Continental Crust (UCC) (Rudnick and Gao, 2014), 10 times higher than the background concentration on the TP (Zhang et al., 2002 ), 20 times the desert sands to the north of the TP (Su and Yang, 2008), and 9 times than the average of Chinese soils (Chen et al., 1991 ). The element richness characters together with small coefficient of variation suggest that lakeshore sand deposition is derived from a single and uniform protolith, ultrabasic igneous rock (Jowitt et al., 2012 ). It can be inferred that the sediments were originally from the weathering of mountain rocks to the east of the lake, which were then transported by fluvial and wind processes toward the lake.

For major elements (expressed as oxides), the layers are significantly rich in Ca and slightly lower in Si while leaching in others (Figure 3b). The enrichment of Ca indicates a dry climatic condition with weak weathering, and Si is enriched in cold and dry conditions with less weathering and decreased leaching while its content is positively correlated with coarse particle content in sediments (Liu et al., 2013 ). However, there is no clear correlation of Si content with the coarse sediment content in the samples because the latter has uniform high values. Al is relatively stable and is separated out at the end of the chemical weathering process, and its content (Al₂O₃ ranges at 5.8%–6.5%) is clearly lower than sands in other deserts (Zimbelman and Williams, 2002; Liu et al., 2015 ).

It is interesting to note that two distinct synchronous changes are found from the profiles of many elements, a stronger one at about 90 years before present (B.P.) and a weaker one at about 380 years B.P. (Figure 4). Si content at the two periods are found to be significantly lower than in adjacent samples, while other linked elements such as Mn, Co, Ca and Fe are clearly higher. This feature is not found in other tested elements. The corresponding changes of other oxides including MgO, CaO, and Fe₂O₃ can be responses to SiO₂. Because Si existed as oxides, leaching was easier in warm or humid conditions due to enhanced weathering. These changes may indicate significant warm or wet periods for the corresponding sedimentary layers. The temperature condition was generally known as colder, especially during the LIA, than present; thus, we can assume that the two strong synchronous profile changes were attributed to strong wet conditions, which may have occurred before the dated time because of lagged responses of geochemical elements.

3.3 Geochemical parameters

CIA reflects the chemical weathering of silicate minerals. A higher CIA mirrors a humid climate and stronger geochemical weathering with more leaching of elements such as Na, K, and Ca. Results show that CIA values are low for all the samples and vary modestly from 49.8 to 53.5, indicating low level chemical weathering through the whole profile (Figure 5). However, they are still higher than the weathering intensity of sands collected from deserts (Dong et al., 2011 ), which can be explained by the relatively high sediment water content due to their closeness to the lake. A decreasing tendency was found from the bottom to the top of the profile, indicating recent weather condition is drier or colder than the past. This slightly decreasing trend before about 150–160 years B.P. was caused by a cold and dry environment during the late LIA, while the more clearly decreasing after the end of LIA till recent time was due to a significant drying caused by warming climate (Yang et al., 2002 ; Jin et al., 2007 ; Ijmker et al., 2012 ; Gou et al., 2015 ).

Ti content is usually stable because it mainly exists in stable minerals. Si accumulates in cold and dry environments, as mentioned above. Therefore, an increased SiO₂/TiO₂ ratio indicates a colder or drier climate, while a decreased ratio reflects enhanced weathering. The SiO₂/TiO₂ values of the profile generally increases with fluctuation from the bottom of the profile, and reflects a drier recent climate compared to about 400 years B.P. (Figure 5). However the ratio was slightly reduced before 150–160 years B.P. and quickly increased afterwards, corresponding to a relative stable or lower dryness condition during the late LIA with a cold and rapidly drying trend after LIA with warming temperatures since about 150–160 years B.P. (Zhang et al., 2016 ).

The Ca²⁺/Mg²⁺ ratio is also a good indicator of an arid climate, because these two elements are typical in arid and semi-arid environments, though Ca²⁺ migrates easier than Mg²⁺. Therefore, an increasing CaO/MgO ratio represents a drier environmental changing trend. Variation of the CaO/MgO profile is synergetic to and more significant compared to that of the SiO₂/TiO₂ profile, indicating the consistency of these geochemical parameters (Figure 5).

3.4 Change of deposition rate

The deposition rate of sediments can also be analyzed using the OSL dating and core depth information to trace past aeolian environments due to its sensitivity to alternations of aridity and wetness, and to related atmospheric circulations (Lu et al., 2005 ; Long et al., 2016 ). The transportation distance was short in the study area so that the blown sand can reach the sedimentary site in one wind event or a short period. Thus, the deposition rate can imply the richness of loose surface material and general climate conditions at the time of sand accumulation. Using the six OSL dating values and sample depths, we obtained the time span and depth change between two adjacent samples; thereafter, the rate of sand deposition change (cm/a) at five different depositional periods can be estimated.

Results show that the deposition rate of sand sediments increased quickly in the past four centuries. The rate was 0.29 cm/a in the period of 380–240 years B.P., but with a nearly 6 times increase to 1.63 cm/a in the most recent period of 60–20 years B.P.. This significant change can be expressed by a power function with high confidence level (R²=0.90), in which x is time (years B.P.) and y is the deposition rate (Figure 6). The change of rate indicates that aeolian activity became much more prevalent during the tested period with significantly increased sands available at the land surface. However, there was a stagnation of deposition rate change between 140 and 90 years B.P., which means the erosibility of surface materials was low although it was under a warming background. This can be explained by the wet period before 90 years B.P. as indicated in section 3.2, which may increase surface moisture and plant growth, while reducing aeolian activity over numerous years in mostly winter and spring dry seasons.

4 Discussion 4.1 Source of lakeshore depositions

Gain-size analysis with parameters such as mean grain size, sorting, skewness and kurtosis, together with special geochemical characters such as Co richness and absence of Zn, indicates that lakeshore deposition was transported by the aeolian process and their same origin was the weathering of local mountain rocks east of the lake. This is consistent with other studies that show local and small-scale source of the TP sands (Ijmker et al., 2012 ). These results also agree with a previous study on the sources of lakeshore sands based on composite element fingerprinting analysis (Liu et al., 2016b ).

Alpine meadows at different degraded levels can be seen throughout the mountains, and aeolian material is plentiful although the natural weathering process is slow in this cold area. The mountain surface was devoid of developed soil during the field survey, and the top several meters of sands may have been totally removed by wind (Liu et al., 2016a ). The dominant wind direction was westerly; however, the downhill transportation of sands from the eastern mountain area should have a significant lower threshold, which means the movement of sands toward the lake is much easier with secondary prevailing easterly wind. The accumulated aeolian sands rest at or beside the gullies can also be carried by floods during the rainy season and deposited along the lakeshore.

A similar sand material transportation and the interplay of fluvial/aeolian processes may be common to the many other lakes over the TP, as a large part of this high plateau is influenced by the westerlies.

4.2 Recent climate change

Several geochemical parameters including CIA and ratios of SiO₂/TiO₂ and CaO/MgO indicate that climate at Lake Co Nag during the past four centuries was impacted by LIA and significant drying after the end of LIA in the 20th century. LIA cold temperature resulted in low weathering which is reflected by a slightly reduced CIA. The frozen surface, low evaporation, and high soil moisture in the sand surface that resulted in a slight reduction in dryness is reflected by ratios of SiO₂/TiO₂ and CaO/MgO. After LIA, temperatures started to increase over the TP with a significant warming in the 20th century (Kang et al., 2015 ; Zhang et al., 2016 ). Warming in the dry winter and spring seasons can increase evaporation and reduce surface soil moisture, which leads to higher erodibility of surface sands, while the change of precipitation in the wet season has a limited effect. This is reflected by geochemical elements with lower CIA, higher SiO₂/TiO₂ and CaO/MgO values.

Recorded meteorological data for the past half century show that precipitation and temperature increased at the same time; however, the increased rainfall occurred mainly in the summer rainy season, while in the dry winter and spring with strong wind activity, dry and bare surfaces received little additional water (Liu et al., 2016a ; Zhang et al., 2016 ). Another hypothesis is that the added plants that responded to the increase of summer rainfall may intercept more shifting sands and lead to an increase of sand depositional rate. However, aeolian activity was mainly dominant in winter and spring, while the ephemeral short grass in summer has little affect as no biological residual remained in the studied sand profile. Even increasing precipitation in summer might not be consistent in other areas, and it remains difficult to evaluate the precipitation change trend in the face of limited observational data in the TP (Conroy et al., 2017 ). This is because the summer was dominated by the Indian monsoon and the winter and spring were controlled by the westerlies, and the two climate systems have different patterns.

A similar conclusion in the southern TP was found where temperature instead of precipitation is the primary factor on local evapotranspiration and effective moisture (Klinge and Lehmkuhl, 2015). As a result, the dry season became warmer and drier, resulting in more available sands and stronger aeolian activity over the eastern mountain area and more sands transported west to the lake. This increasing richness of surface sands caused by the warming in the dry seasons was verified by the change of deposition rate of sands at the lakeshore of Lake Co Nag, which increased nearly six times at about 60–20 years ago compared to 380–240 years ago. This trend might have ceased in some wet periods with higher surface moisture conditions in the dry seasons.

The increase in the deposition rate of sands with the warming trend in this area suggests that the strengthening of surface sand erodibility is sensitive to climate change. Significant surface aeolian activity and desertification may be triggered by climatic forces in the central TP, resulting in problems including strong sand depositional at the two sides of the QTR. This condition may be worse in the future as there is high confidence that temperatures are projected to increase (Christensen et al., 2013 ). Thus, more attention and action is needed to address the challenging future of increased sand and dust.

4.3 Evidence of significant wet periods

The wet/dry episodes over the TP are explained by the weakening of monsoons from East Asia since 4.2 cal ka B.P. (Mügler et al., 2010 ; Chen et al., 2016 ). This weakening of Asian summer monsoon can lead to strengthening of aeolian activity on the TP and vice versa (Stauch et al., 2016 , 2017). The coordinated changes of deposition rate and geochemical profile indicate a significant wet period between 90 and 140 years B.P., which led to a cessation of deposition rate increase and abrupt changes of many elemental contents. There was another older wet period at about 380 years B.P. that can be inferred from a similar geochemical profile change, although we have no deposition rate data to verify this because no older age dating values were available. Thus, mutual verification of physical and chemical information indicates that geochemical data, together with supporting data from grain size and age dating data, can be used to detect ancient climate events.

The recent wet period between 90 and 140 years B.P. was confirmed by a wet phase during 1883–1906 according to historical archives (Lin and Wu, 1986). Some studies revealed a similar pronounced wet condition, for example during 1890–1900 in the southeast (Li et al., 2017 ), 1891–1913 in south-central (He et al., 2013 ), and 1886–1913 in north (Wang et al., 2013 ) of TP. In addition, sometimes a relative dry period was also shown at various locations over TP (Chen et al., 2009 ; Griessinger et al., 2011 ). This assumption can hardly be verified by other high-precision precipitation reconstruction studies, because proxy data with high temporal resolution and reliability like those from tree rings are not available in central TP close to our study area. However, results in this study still raise the possibility of understanding the historical climate and environmental evolution in the remote plateau area. Future evidence may still come from soil or lake sediments, as they can provide accurate long term records of past climate (Holmes et al., 2009 ).

5 Conclusions

The TP is sensitive to climate change and is therefore a good indicator of climate change, but available paleoclimate information with high temporal precision is rare in the remote and harsh central TP. In this study, grain-size analysis with size parameters and deposition environment discriminant, geochemical element analysis with deposition profile change and chemical parameters, and sediment age dating with deposition rate estimation of a sand profile were combined to examine the environment and climate change with seasonal differences at the shore of Lake Co Nag in central TP during the past four centuries.

The physical and chemical factors support each other in showing changes of aeolian sediments and climatic forcing. The results suggest an agreement with the broader picture of LIA and 20th century environmental evolution. This study adds data to the overall picture of past climate and geomorphic history of the TP in showing that when other proxies are not available, aeolian sediments can reflect climatic variation and surface process with multi information including physical, chemical, and dating data.

Some facts in this study can be drawn as follows:

(1) The depositional rate of aeolian sands increased by six times during the past four centuries resulting from increasingly active surface sands, and problems related to aeolian activity such as surface erosion, blown-sand deposition and desertification. This type of activity may worsen in the future under a projected warming climate which may occur in other cold and high altitude areas. The source of the east lakeshore sediments of Co Nag was the mountain area to the east from ultrabasic igneous rock.

(2) Collaborative analysis using chemical, depositional and meteorological data indicates different seasonal effects of climate variation on geomorphic processes. Aeolian activity is very sensitive to warming in dry seasons in the study area. The dry seasons of winter and spring are warmer and drier compared to 400 years ago although the summer season might have received more precipitation. Climatic conditions before and after the LIA were clearly different as indicated by geochemical parameters including CIA and ratios of SiO₂/TiO₂ and CaO/MgO.

(3) Two significant wet periods might have happened between 90–140 years B.P. and about 380 years B.P. in the central TP, which led to abrupt changes of the geochemical profile and pauses of deposition rate increase.

Acknowledgments:

We thank Pro. Hui Zhao and Dr. JiangLin Wang at the Northwest Institute of Eco-Environment and Engineering, Chinese Academy of Sciences for discussions on past climatic events in the Tibetan Plateau. This study was supported by the National Science Fund of China (41501008), the China Postdoctoral Science Foundation (2014M550518), the Youth Innovation Promotion Association (2016373), and the "Light of West China" Program of the Chinese Academy of Sciences.