John Gates¹, W. Mike Edmunds¹ and Jinzhu Ma²

1 Oxford Centre for Water Research, University of Oxford

2 Key Laboratory of Western China’s Environmental System, Lanzhou University

 

Abstract:We report here on new findings of groundwater tracer studies in the Badain Jaran Desert, Inner Mongolia, China. Shallow groundwater is ubiquitous in the desert’s low-lying areas, and supports numerousinterdune lakes despite hyper-arid conditions. With increasing pressure on the region’s water resources, a clear understanding of this hydrological system will be essential for effective water management strategies.

During field sessions in 2004 and 2005 approximately 30 groundwater samples for chemistry and stable isotope analysis were collected across the southeastern section of the Badain Jaran and areas adjacent to the main dune field to the south and southeast. In addition, one unsaturated zone profile was collected by hand auger in order to estimate direct recharge using mass balance of chloride. Profile results indicate that diffuse recharge in the Badain Jaran is low though nonzero (approximately 1 mm/yr), and somewhat variable in space and time. This result confirms that modern direct recharge is insufficient to support the desert lakes. Shallow desert groundwaters are isotopically distinct from the region’s modern rainfall as well as shallow groundwaters of locations immediately to the south of the desert. The lakes are shown to derive from shallow groundwater, related along an isotopic evaporation line with slope of 4.2.

Major ion chemistry and stable isotopes suggest a pre-modern origin for the shallow groundwaters, with recharge occurring during wetter climatic conditions during the Pleistocene or humid phases of the Holocene. Likely recharge mechanisms for these palaeo-waters include runoff from the Yabulai Mountains and direct recharge, though further research is needed to constrain these possibilities. Observed trends of decreasing TDS and less depleted isotopic compositions towards the desert interior along a piezometrically-defined flowpath from the Yabulai Mountains is consistent with this hypothesis. If this is the case, then current groundwater supply to the lakes is due to a diminishing water table which is not in equilibrium with current climatic conditions.

Keywords: groundwater recharge; unsaturated zone; Badain Jaran; chloride mass balance; stable isotopes

Introduction

The widespread water scarcity problems in Northern China have been illustrated by numerous studies over the last several years, and comprise an important focus in Chinese hydrogeological research as well as for this Congress. Depletion of water resources in the Hexi Corridor region of Inner Mongolia and Gansu Province has become evident through a range of symptoms, including diminished downstream flow in the Heihe River, falling groundwater levels in the Minqin Basin and others. In the nearby BadainJaran Desert, numerous groundwater-fed perennial lakes are found primarily in the southeastern section of the desert, and shallow groundwater can be found in interdune areas throughout. This somewhat unexpected feature, i.e. the occurrence of ‘permanent’ freshwater lakes despite arid climatic conditions, appears to be in contrast to the general trend of increasing scarcity in surrounding locations. As a result, the Badain Jaran area has attracted substantial research interest and at least one suggestion that the desert groundwaters could be developed for regional agricultural and domestic use (Chen et al, 2004).

In order to promote sustainability in water resources management and policy, any resources that are to be allocated for distribution must be well-understood in terms of quality and quantity, as well as ecosystems functions. At present, much about the sources and characteristics of Badain Jaran shallow groundwaters remain poorly characterized, and a general lack of research wells or geological detail has hindered hydrogeologicalinvestigation. Some tracer data in recent studies have been interpreted as supporting the notion that the shallow aquifer is somewhat renewable, which the freshness of the lakes also seems to imply. However, this conceptual model runs in contrast to groundwater conditions in most desert regions worldwide and must be carefully tested.

        The aim of the current study is to further refine the conceptual model of sources and rates of recharge to the shallow Quaternary aquifer of the Badain Jaran Desert using tracer-based approaches. Chemical and isotopic results from groundwater, lakes and precipitation from approximately 40 locations are discussed within the context of the physical system, previously published recharge studies, and palaeo-climate information from the region. One unsaturated zone profile is presented to illustrate rates of direct recharge. It is argued that available evidence points strongly towards a palaeo-water origin for this aquifer, although much uncertainty regarding the system still exists.

Study Area

The Badain Jaran Desert lies near the centre of the Alxa plateau in western Inner Mongolia, between 39^o20’N to 41^o30’N and 100^oE to 104^oE. It approximately spans the region bounded by the Longshoumountains (maximum elevation 1963 m asl) to the south, the Yabulai mountains (with maximum elevation 1957 m asl) to the southeast, and the lowlands areas of the Gurinai grassland and the Guezi Hu wetlands (about1000 m asl) to the west and north (Hofmann, 1996). With an area of approximately 49,000 km², it is considered the second largest desert in China (Yan et al., 2001).

Most of the area of the desert is comprised of large unvegetated or sparsely vegetated dunes, strongly oriented in a SW-NE pattern. Many of the interdune areas contain groundwater-fed lakes, which vary widely in surface area (up to 1.6 km²) and salinity (measured from 1.2 to 398.2 g/L total dissolved solids (TDS)) throughout the desert (Yang and Williams, 2003). The largest of the dunes are located in the southeastern section of the desert, and are potentially the tallest in the world at over 400 m in height (Dong et al., 2004). Dune sands have a mean diameter of about 0.22 mm, and grain sizes follow a bimodal distribution characteristic of two significant source areas, suggested by Jackel (1996) to be the Heihe River alluvial fan to the west and the Qilian Shan to the southwest.

Climatically, the Badain Jaran is characterized as strongly continental (Köppen classification BWk), with mean monthly temperatures ranging from approximately -10°C in January to 25°C in July. Precipitation is influenced by the East Asian monsoon, with most rainfall occurring from July to September. Cold and dry continental air masses with temperatures below zero are dominant throughout the winter. The mean annual precipitation measured at Zhongqanzi was 89 mm from 1956-1999, and potential evaporation is approximately 2600 mm.

Little data is available regarding the hydrogeology of the region. The mainly Holocene aeolian sands that make up the desert landscape have been deposited upon older Quaternary sediments that occupy the basin depression of the Alxa platform (Ma and Edmunds, 2006). These sands also comprise the desert’s major shallow aquifer (probably unconfined conditions), which can be found at a depth of 10 m or less in many locations. Semi-confined conditions are apparent in the vicinity of some lakes where travertine islands have formed as a result of fresh groundwater being forced up under pressure. Low-porosity lacustrine sediments orinterbeded sandstone may provide the confining layers locally. Regional inputs to this aquifer and deeper groundwater systems are poorly understood.

Methods

Field data was collected in Summer 2004 and 2005 in the Badain Jaran Desert and in the nearby vicinities to the south including the Yabulai and Longshou mountain ranges. Because the study area contains only sparse settlement and therefore a limited number of wells, the sampling strategy was primarily opportunistic, aiming for as many locations as possible over a wide geographic area. Samples include shallow and deep groundwater, lakes and rainfall events. Water samples for major ion chemistry were collected in Nalgene bottles which were prepared by acid-washing (5% nitric acid) and twice rinsing with deionized/demineralizedwater. 2x30 ml samples were filtered at 0.45 um and cation samples were preserved from biological reaction by adding 1% analytical-grade HNO₃. Unfiltered 30 ml samples for stable isotope analysis were also collected at this time and stored in airtight bottles.

Major anions (Cl^-, SO₄^2-, HCO₃^-, NO₃^-) were analyzed in the Geomorphology Laboratory of the Oxford University Centre for the Environment with Dionex ion chromatography (standard error <3%). Cations were analyzed at NERC ICP-AES facility at Royal Holloway University of London with an Optima 3300RL ICP-AES. Bicarbonate alkalinity was measured by Gran titration. As an internal quality control indicator, all groundwater chemistry results were within 5% ionic charge balance. Stable isotopes (δ¹⁸O and δ²H) were analyzed at the British geological Survey Isotope Laboratory in Wallingford UK, using Isotope Ratio Mass-Spectrometry. Precision of measurement for stable isotopes was ±0.1‰ for δ¹⁸O, and ±2‰ for δ²H.

An unsaturated zone profile to 9.25 m was collected in the vicinity of Lake Sayin Wusu (N 39° 34’; E 102° 20’) in order to estimate direct recharge through the dune sands. The profile was obtained with an Australian-type hand auger. Samples along the profile were collected at 12.5 cm intervals and immediately sealed to prevent evaporation. The profile was cored to a depth of 9.25 m. Unsaturated zone moisture was extracted by elutriating with 30 ml of deionized/demineralized water, then centrifuging and filtering elutriates at 0.45μm. Cl^- concentrations of elutriated samples were analyzed with ion chromatography and moisture contents were determined gravimetrically by drying overnight at 110°C.

The chloride mass balance recharge estimation method makes use of the fact that Cl^- is conservative in the unsaturated zone in most locations, i.e. where no halite formations, evaporate salts etc are present. Assuming Cl^- is conservative and assuming 1-dimensional downward piston-like flow, the mass balance of Cl^- below the root zone (or zero flux plane) can be formulated as:

P*C_P = R*C_R,

where P is the precipitation rate, C_P is the Cl^- concentration in precipitation, R is the recharge rate, and C_R is the Cl^- concentration of recharging waters (or pore waters). The left-hand side of the equation represents total Cl^- inputs and the right-hand side total outputs. The left-hand side of the equation must be adjusted by adding a term for dry depositions in areas where aerosol deposition of Cl^- is an important component of the chemical balance. (In practice, owing to the difficulty in isolating the contribution of dry deposition, it is often assumed that salt contribution from dry deposition is included in samples from bulk rainfall collectors and is generally constant over time.) Based on this formulation, recharge is given by the quotient of total inputs and the measured pore water concentration, and is inversely proportional to the degree of enrichment of Cl^- due to evaporation and transpiration.

Assuming that the input parameters are known, Cl^- can also be used to estimate the residence time of water at a given depth by dividing the total mass of Cl^- in the profile above a given depth by the annual Cl^- input rate:

,

where t represents residence time, θ moisture content and z depth.. Since the calculated recharge rate is linearly related to both annual mean rainfall amount and Cl^- concentration in rainfall, accuracy in characterizing rainfall at the site is of utmost importance.

Results

i) Unsaturated zone profiles

The unsaturated zone profile is shown in Fig. 1. Cl^- concentrations measured over the profile ranged from 36 mg/L to 336 mg/L, with the maximum value within the upper 1 m and the lowest concentrations towards the middle and bottom of the profile. A shallow peak in Cl^- begins to emerge at approximately 40 cm below surface and reaches its maximum at 50 cm. This feature is common in arid unsaturated zone profiles and can be found in most profiles in the Southwestern USA (Scanlon et al, 2003) and Australia (Allison and Hughes, 1978). The peak likely represents Cl^- that has been accumulating since the last occurrence of ‘deep’ drainage, i.e. since the last time moisture has been flushed below the zone of significant moisture cycling due to evapotranspiration (zero flux plane). The peak is more than an order of magnitude lower in concentration than some documented in the USA, likely due in part to the lack of surface vegetation at this site. Very high concentration peaks in arid unsaturated zones have been attributed to zero or negative recharge throughout the Holocene, since their cumulative Cl^- amounts represent 12,000-16,000 years of accumulation (Walvoord et al, 2003).

A secondary peak of approximately 250 mg/L is shown at a depth of 2 m. This may represent a former near-surface peak which has been flushed down relatively recently. Beginning at 3 m, the Cl- concentrations begin to steady, and continue to fluctuate around a mean value of approximately 75 mg/L through approximately 7.5 m. In general, signals corresponding to individual recharge events tend to attenuate over time due to diffusion in the unsaturated zone (Cook et al, 1992), and will approach a steady mean value over time. As such, 75 mg/L is taken as the approximate mean Cl^- input signal over the period of record. Capillary fringe effects of the water table on moisture contents begin at approximately 8 m. The smaller peak near a depth of 8 m may be related to capillary effects. Alternatively, it may represent either a large Cl^- peak which has not yet attenuated to due its high concentration, or possibly a longer-term variation in the Cl^- input signal in the past due to changing climate conditions.

For the estimation of inputs for recharge rate calculations, we employ the long-term rainfall monitoring record from Zhongqanzi meteorological station (approximately 30 km distance from profile site), which has a mean of 89 mm/yr for 1957–1999. However, because of the characteristically high spatial and temporal variability in precipitation in this arid region, this mean should be treated with a degree of caution. Long-term Cl^-concentration for rainfall was estimated at 1.5 mg/L from published results on a small number of relatively intense storms by Ma and Edmunds (2006) and Hofmann (1999), as well as two storms sampled during the 2005 field investigations. Confidence in these data as representative of the true mean is boosted by the fact that 1) the variability in Cl^- concentrations across the sampled events is quite low and 2) the vast majority of rainfall received in this area results from a small number of heavy storms so that chemistry of storm events reflect well the overall yearly total. However, further monitoring of rainfall chemistry would serve to strengthen this analysis.

Using these input terms, in combination with Cl^- concentrations between the shallow peaks and the capillary fringe, a long-term average recharge rate is estimated to be 1.62 mm/yr. Applying the cumulative Cl^-equation for residence time, the surface peak represents 20 years of accumulation, and the entire profile 430 years.

ii) Groundwater chemistry and stable isotopes

Summary trilinear diagrams for major ions in groundwater are shown in Fig. 2. As reported by others (Hofmann, 1999; Yang and Williams, 2003), the region’s shallow groundwater is characterized by a high degree of variability, both with regards to TDS and percent contribution of major constituents. In general, major anion compositions of shallow groundwaters to the immediate south and southwest of the desert are dominantly Cl^- and SO₄^2- in type. This pattern transitions to no dominant type towards the desert interior, concurrent with a trend of decreasing Cl^- and TDS. Cation compositions for the two areas are similar, but with rather high overall variability. Two of the three apparent outliers represent artesian wells (<130 m depth) in the vicinity of Youqi town, and one is a shallow farm well which may be contaminated, as suggested by elevated nitrate concentrations in the sample.

Stable isotope results are summarized in a delta-plot relative to VSMOW in Fig. 3. The local meteoric water line (LMWL) is regressed on all precipitation records from Zhangye Station (r²=0.96), located approximately 250 km to the southwest of the study area (GNIP/WMO, 2004), and weighted annual means for this data set are plotted. Rainfall events sampled in 2005 fall closely along the Zhangye precipitation line, supporting the assertion that this LMWL is valid for the study area. Desert shallow groundwaters range from -5.4 to -2.0 δ¹⁸O VSMOW, and in addition to lake waters, plot along an evaporation line with slope of approximately 4.2 and LMWL intercept of approximately -12.5 δ¹⁸O (r²=0.91). This slope is fairly typical of surface water evaporation under low-moisture atmospheric conditions (Clark and Fritz, 1997). Along this lineCl^- concentrations tend to increase with increasing isotopic enrichment, a second indicator of progressive evaporation.

Contrasting with the desert groundwaters, shallow groundwaters to the south and southwest tend to group along the LMWL, primarily between -7 and -11 δ¹⁸O but as low as -12δ¹⁸O. Three exceptions to this general pattern are shown lying along the evaporation line with high Cl^- concentrations.

Discussion

The data reported here add to a growing catalogue of tracer results for the Badain Jaran Desert region (see for example Geyh et al, 1996; Hofmann, 1996; Ma et al, 2003; Yang and Williams, 2003; Chen et al, 2004; Ma and Edmunds, 2006; Yang, 2006). In general, they tend to reflect well the results of previous studies. Regarding the profile-based estimation of direct recharge, the calculated long-term average of 1.62 mm/yr is very similar to the results obtained by Ma and Edmunds (2006) of 1.33 mm/yr, 1.26 mm/yr and 0.95 mm/yr. With an annual rainfall rate of 90 mm/yr, direct recharge through the dunes is close to 1% of annual rainfall, a common relationship for arid regions. Note that these values are long-term averages, and that because of extreme rainfall variability, it is likely that positive values for direct recharge are not achieved in most years; instead occasional event-based recharge is likely the pattern. The calculated Cl^- accumulation period of 20 years since the last significant recharge event is therefore within reason.

When evaluating the context for this result it is important to bear in mind that, since estimates are inherently point-based, extrapolation of the estimates over a large and complex landscape is problematic. However, since surficial materials are quite uniform over the desert’s extent, a high degree of recharge variability would be unexpected unless periodic runoff focuses in topographic depressions, as has been well-documented in playas in some locations (Scanlon and Goldsmith, 1997). While no evidence of runoff in the Badain Jaran is apparent, additional profiles in topographic depressions are currently being investigated as part of this project. Apart from this possibility, the profile-based estimates can be considered as upper bounds for an aerially-averaged direct recharge rate since i) these profiles were taken at non-vegetated sites and ii) evaporation from shallow water tables may occur in some interdune areas.

With potential evaporation estimates of approximately 2600 mm/yr, mass-balance considerations clearly indicate that direct recharge is not sufficient to supply the desert lakes, and that additional recharge sources would be required to do so. However, the only nearby major sources of water are down-gradient from the desert interior (i.e. Guezi Hu Wetland and the Heihe River). With exposed fissured rock surfaces, the Yabulai andLongshou mountain ranges could possibly supply mountain-front or mountain-block recharge, but under current climatic conditions this would not likely be able to sustain flow volumes implied by the lakes. Remote-source recharge from the Qilian Mountains via deep interbasin transfer has been suggested (Chen et al, 2004), but is somewhat improbable from a hydrogeological perspective and is not well supported by available data.

Stable isotope results from the desert’s shallow groundwaters show a clear distinction from groundwaters of surrounding locations to the south and southwest, and also illustrate the intimate connection with the desert lakes. The strong connection along the evaporation lines supports the hypothesis that they are supplied entirely or primarily by shallow groundwater (rather than a deeper formation), and are basically outcrops of the water table of the shallow Quaternary sand aquifer. Further, the intercept of the evaporation line with the LMWL (-12 δ¹⁸O) is significantly lighter than mean modern rainfall at approximately -7 δ¹⁸O. Ma and Edmunds (2006) argue that this depleted intercept with the LMWL is indicative of a palaeo-water origin for the groundwaters, noting that -7 δ¹⁸O matches well with expected rainfall values in this region during the late Pleistocene based upon lake sediment palaeo-environmental records. The implication is that the lake-feed groundwaters are not hydraulically related to areas to the south and are not sourced from modern rainfall in the region.

If the above interpretation is accurate, the most likely explanation for the ongoing supply of fresh groundwater is that the desert’s shallow aquifer filled during more humid climatic conditions in the past, and that flow is presently supported by some combination of i) mountain-source recharge from the Yabulai and Longshou ranges and ii) a diminishing water table which is out of equilibrium with current climate and will continue to dry. Indeed, a range of evidence including lake sediment records, palaeo-lake levels (Miscke et al, 2005) and luminescence dating of dune sands (Yang and Williams 2003; Yang, 2006) indicate that the late Pleistocene was significantly more humid in the region than at present, and that the Holocene has been marked by strong climatic variability. Radioisotope research is currently underway to test this palaeo-recharge hypothesis by establishing a chronology for the desert groundwaters.

Well-defined hydraulic head gradients from the Yabulai Mountains to the northwest (roughly mirroring surface elevation and terminating in the Gurinai) are apparent from lake and well levels, and strongly suggest an active flowpath from this mountain range feeding the desert. Geochemical results show that Cl^- concentrations and TDS generally decrease along this hypothesized flowpath, while SO₄^2- and HCO₃^- remain relatively constant. If the Yabulai range was a prominent recharge source in the past, then the observed decrease of conservative Cl^- in the downgradient direction may be consistent with the palaeo-recharge, since older waters (now down-gradient) would be expected to be less affected by evaporation during recharge than the younger, owing to the wetter climate. The Cl^- trend is mirrored by δ¹⁸O, with a strong positive correlation between the two (Fig. 4). This is similar to results from the Minqin Basin (100 km to the southeast), where the palaeowaters have been confirmed by radiocarbon analyses (Edmunds et al, in press). Similar scenarios of decreasing Cl^- concentration in the downgradient direction due to climate change have been documented elsewhere (see for example Edmunds and Smedley (2000) on the East Midlands Sandstone Aquifer, UK).

         In addition to increased baseflow derived from mountain sources, it may be that direct recharge within the desert has also played a more prominent role during wetter and/or cooler climatic periods, since a higher ratio of precipitation to evaporation would be experienced. The net effect would be that wetting fronts would be able to pass below the zero flux plane more frequently, though denser vegetation in these periods may have a moderating influence. A higher direct recharge rate would also help to explain the high chemical variability apparent in the shallow groundwaters.

It is recognized that the scenario described above is not consistent with the conclusions of some previous studies. Based on Mg²⁺ concentration, Hofmann (1999) argues for two separate aquifers for the northern and southern areas of the eastern lakes region, one supported by direct recharge and one by regional flow from the Yabulai Mountains. He estimates that approximately 30 mm/yr direct recharge would be necessary to support the northern lakes. In view of the global literature on arid zone recharge as well as the profile results presented here, this value is not likely to be achieved under current climate conditions. Also in contrast are the tritium results of Yang and Williams (2003), which indicate a mean residence time of less than 100 years for the shallow groundwaters. One possibility is that some of the waters that were sampled had interacted with modern atmospheric tritium subsequent to recharge. Many of the sampling locations were springs and shallow wells close to the lakes, which may be subject to atmospheric interaction owing to the shallowness of water tables in the interdune areas and nearby lakes. Finally, while the proposition of Chen et al (2004) that snowmelt feeds the region through deep faults needs to be tested further, the notion that groundwater rises through hundreds of meters of low moisture-content sands to supply the near-surface pore water runs counter to a great deal of physical vadose zone research to date.

Conclusion

In this paper new tracer data was presented to help establish a sound conceptual model for groundwater recharge to the Badain Jaran shallow sand aquifer. The results build on previously published data for the region, and were generally in good agreement with the earlier data. It was shown that direct, diffuse recharge through the Quaternary sands is low though nonzero, on the order of 1 mm/yr. As approximately 1% of local rainfall this is consistent with arid recharge studies elsewhere. Stable isotope ratios indicate that the lakes are fed by shallow groundwater, which is not related to local precipitation and most likely is palaeowater in origin, and therefore non-renewable on human timescales. Major ion chemistry of the groundwaters is consistent with this hypothesis, though high spatial variability complicates interpretation.

The geographical source(s) of the palaeowater cannot be identified at this time, but possibilities include i) direct recharge under more humid conditions ii) runoff from the Yabulai Mountains and iii) waters from theQilian Mountains/Tibetan Plateau fed by interbasin flow through fault systems. Research activities currently in progress including radioisotopes and groundwater modeling will help to refine this conceptual model. A clear understanding of the region’s hydrological systems will be valuable information for water resources planning and environmental protection goals.

Acknowledgements

We are grateful to Dr. George Darling (for stable isotopes) and Mr. Adam Young (ICP-AES under NERC Grant OSS/301/0905) for analytical support for this project. Research funding provided by Environmental Change Institute of Oxford University. Additional thanks to Huang Tianming, Li Xianghu and Ding Zhenyu for field and laboratory support.

 

References

[1]      Allison, G.B. and Hughes, M.W., 1978. The use of environmental tritium and chloride to estimate total recharge to an unconfined aquifer. Australian Journal of Soil Sciences, 16: 181-195.

[2]      Chen, J.S., Li, L., Wang, J.Y., Barry, D.A., Sheng, X. F., Gu, W.Z., Zhao, X.. and Chen, L., 2004. Groundwater maintains dune landscape. Nature, 432: 459-460.

[3]      Clark, I. and Fritz, P., 1997. Environmental isotopes in hydrogeology. CRC Press/Lewis Publishers, Boca Raton, 328 pp.

[4]      Cook, P.G., Edmunds, W.M. and Gaye, C.B., 1992. Estimating paleorecharge and paleoclimate from unsaturated zone profiles. Water Resources Research, 28: 2721-2731.

[5]      Dong, G., Gao, Q. & Zou, X., 1995. The climatic change during the last Pleistocene in the Badain Jaran Desert. Chinese Science Bulletin, 40: 1214-1218.

[6]      Dong, Z., Wang, T. and Wang, X., 2004. Geomorphology of the megadunes in the Badain Jaran Desert. Geomorphology, 60(1-2): 203.

[7]     Edmunds, W.M., Ma, J., Aeschbach-Hertig, W. & Darbyshire, F. (in press) Groundwater recharge history and geochemical evolution in the Minqin Basin. Applied Geochemistry.

[8]      Edmunds, W.M. and Smedley, P.L., 2000. Residence time indicators in groundwaters: The East Midlands Triassic Sandstone aquifer. Applied Geochemistry, 15: 737-752.

[9]      Geyh, M.A., Gu, W. and Jäkel, D., 1996. Groundwater recharge study in the Gobi Desert, China. Geowissenschaften, 7–8: 279–280.

[10]  IAEA/WMO, 2004. Global Network of Isotopes in Precipitation. The GNIP Database. Accessible at: http://isohis.iaea.org.

[11]  Hofmann, J., 1996. Lakes in the SE part of Badain Jaran Shamo, their limnology and geochemistry. Geowissenschaften, 14(7-8): 275-278.

[12]  Hofmann J., 1999. Geoökologische Untersuchungen der Gewässer im Südosten der Badain Jaran Wüste (Geo-ecological investigations of the waters in the south-eastern Badain Jaran Desert).

[13]  Berliner Geographische Abhandlungen, 64.

[14]  Jackel, D., 1996. The Badain Jaran Desert: its origin and development. Geowissenschaften, 7/8: 272-274.

[15] Ma, J., Ding, L., Jiawu, Z., Edmunds, W.M. and Prudhomme, C., 2003. Groundwater recharge and climatic change during the last 1000 years from unsaturated zone of SE Badain Jaran Desert. Chinese Science Bulletin, 48(14): 1469-1474.

[16] Ma, J. and Edmunds, W.M., 2006. Groundwater and lake evolution in the Badain Jaran desert ecosystem, Inner Mongolia. Hydrogeology Journal, in press.

[17]  Mischke, S., Demske, D., Wunnemann, B. and Schudack, M., 2005. Groundwater discharge to a Gobi desert lake during Mid and Late Holcene dry periods. Palaeogeography, Palaeoclimatology, Palaeoecology,225: 157-172.

[18]  Scanlon, B.R., Keese, K., Reedy, R.C., Simunek, J., and Andraski, B.J., 2003. Variations in flow and transport in thick desert vadose zones in response to paleoclimatic forcing (0–90 kyr): field measurements, modeling, and uncertainties. Water Resources Research, 39(7): 1179.

[19]  Walvoord, M.A., Phillips, F.M., Tyler, S.W. and Hartsough, P.C., 2003. Deep arid system hydrodynamics 2. Application to paleohydrologic reconstruction using vadose zone profiles from the northern Mojave Desert. Water Resources Research, 38(12): 1291.

[20]  Yan, M., Wang, S., Li, B. and Dong, G., 2001. Formation and growth of high megadunes in Badain Jaran Desert. Acta Geographica Sinica, 56: 83-91.

[21]  Yang, X., 2006. Chemistry and late Quaternary evolution of ground and surface waters in the area of Yabulai Mountains, western Inner Mongolia, China. Catena, 66(1-2): 135-144.

[22]  Yang, X. and Williams, M.A.J., 2003. The ion chemistry of lakes and late Holocene desiccation in the Badain Jaran Desert, Inner Mongolia, China. Catena, 51(1): 45-60.

 

Figures

		Figure 2b.   Trilinear diagram for major cations in the study area.

 

 

 

	Figure 3.   δ-δ plot of stable isotope results. Sizes of data markers are scaled to the sample’s corresponding Cl- concentration. Lakes data appear courtesy Mr. Adam Young (unpublished data).

 

 

 

	Figure 4.   Relationship between Cl- and δ18O for Badain Jaran Desert shallow groundwaters (circles), shallow groundwaters in the vicinity of Yabulai and Youqi to the south and southwest of the desert (squares), and rainfall (triangles). Linear regression line shown for desert groundwaters (r2=0.6).