A quantitative assessment on groundwater salinization in the Tarim River lower reaches, Northwest China

JianHua Xu , WeiHong Li , YuLian Hong ,ChunMeng Wei , Jie Tang

1. The Research Center for East-West Cooperation in China, the Key Laboratory of Geographic Information Science, the Ministry of Education of PRC, East China Normal University, Shanghai 200062, China

2. State Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi, Xinjiang 830011, China

1 Introduction

Groundwater salinization, which is commonly found in coastal, arid and semi-arid areas, refers to the enrichment of a variety of dissolved substances in groundwater under the cooperative influence of human activity and natural evolution (Karroet al., 2004;Subyani, 2005). Many factors lead to continuing groundwater salinization, such as dissolution of soluble minerals, rising groundwater level, strong evaporation,scarce precipitation, large-scale exploitation of groundwater, inappropriate agricultural irrigation and excessive use of chemical fertilizers (Chourasia and Tellam, 1992; Fisher and Mulican, 1997). The study of groundwater salinization characteristics, material sources and influencing factors can provide a scientific basis for groundwater resource planning, utilization and management, and prevention of soil salinization.

In a number of chemical parameters, total dissolved solids (TDS) is an important indicator that describes groundwater salinization. It is often used to characterize the regional groundwater hydrochemical characteristics. Studies (Salamaet al., 1999; Wanget al., 2008) show that the higher the TDS’s value of shallow groundwater, the more serious soil salinization becomes when groundwater depth is lower than the critical depth of salt uprising. Otherwise the degree of soil salinization will become mild. Therefore,TDS is considered as one of the important groundwater salinization indicators.

In arid and semi-arid areas of Northwest China,groundwater salinization has become common forms of groundwater pollution. A variety of studies have defined groundwater salinization, and can be interpreted as follows: along the direction of groundwater flow, chemical composition changes from the recharge zone to discharge zone, due to an increase of TDS in the groundwater and the concentration of chemical components (Richter and Kreitler, 1993;Wenet al., 2006).

Based on monitored data from 840 samples from 2000 to 2009, this paper investigated the spatial and temporal variability of groundwater salinization in the Tarim River lower reaches by classical statistics and geostatistics.

2 Study area and data

2.1 Study area

Tarim River is located in south Xinjiang, Northwest China, with a length of 1,321 km and watershed area of 102×104km2(39°00'N–41°40'N, 74°30'E–88°30'E)(Zhanget al.,2005). It is one of the longest inland rivers around the world, and plays a key role in the development of local society and economy. The Taklamakan Desert, the second largest desert in the world, is named after the Tarim River and Kuluke Desert. It lies to the south of the river and extends across the Tarim Basin. There are Tianshan Mountains to the north of the basin and Karakorum Mountains to the south. The Tarim River lower reaches are 321 km in length from Qiala to Taitema Lake (Figure 1).

Figure 1 Location of the monitoring sites for groundwater of the Tarim River lower reaches

Downstream alluvial plain area lies in the eastern part of the Tarim Basin, and the terrain is relatively flat, with an average altitude of 825 m. Major landform types are alluvial-diluvial, alluvial, aeolian and fluvial-lacustrine sedimentation. Simultaneously, the stratum structure is simple, mainly made up of Pleistocene and Holocene quaternary strata, and the surface lithology is dominated by sand, sub-sandy soil and clay. In addition, the aquifer in this study area is quaternary, and the lithology is single with predominantly river-lake facies fine sand and silty fine sand, and eolian sand locally. Divided by water-bearing media, it belongs to a typical chinky aquifer; while divided by burial condition, it belongs to a phreatic aquifer (Chenet al., 2008; Liet al., 2010;Xuet al., 2008, 2013). Thus, the permeability and water abundance are relatively poor with the permeability coefficient varying from 1.2 to 4.8 m/d, and the single-orifice water inflow is less than 150 m3/(d·m). The study area is located in a disconnection river, from Tikanlik to Taitema Lake, which has a warm-temperate, extremely arid climate with annual solar radiation of 5,692–6,360 kJ/m2. According to observations by the Tikanlik meteorological station from 1957 to 2005, average temperature, annual precipitation and annual evaporation are 10.94 °C,34.53 mm and 2,531.45 mm, respectively. Due to years of zero flow, the groundwater table mostly dropped to 8–12 m and both the degree of salinity and mineralization are relatively high.

2.2 Data collection and analysis

Nine groundwater monitoring sites along the Tarim River lower reaches were selected. They are Akdun (A),Yahepu (B), Yingsu (C), Abudali (D), Kardayi (E),Tugmailai (F), Alagan (G), Yiganbjima (H) and Kaogan (I). Monitoring wells, with a depth of 8–17 m,were stationed at an interval of 100 to 200 m for every monitoring site. There were 40 monitoring wells in total and they were distributed within the range of 50–1,050 m away from the watercourse. In order to analyze groundwater chemistry component change, we measured the groundwater table and collected groundwater samples regularly from each well at each site. We collected and analyzed 840 samples during the period from 2000 to 2009. The samples were stored in clean and colorless polyethylene plastic bottles, and sealed with paraffin and labeled, and placed in a shaded area until sent to the laboratory. Chemical compositions of the samples were analyzed within two days in the laboratory. The chemical analyses included total dissolved solids (TDS), pH, concentrations of CO32-,HCO3-, Cl-, SO42-, Ca2+, Mg2+, K+and Na+.

This study adopted an appropriate method to measure each parameter. The pH was measured using a pH meter (APHA, 1995). Calcium (Ca2+) and magnesium(Mg2+) were determined using the EDTA titrimetric method (APHA, 1995). Chloride (Cl-) was determined by standard AgNO3titration. Carbonate (CO32-) and bicarbonate (HCO3-) were determined by titration with HCl (APHA, 1995). Sodium (Na+) and potassium (K+)were measured by flame photometry, while sulphate(SO42-) by spectrophotometer turbidimetry (APHA,1995). Total dissolved solids (TDS) was determined by the gravimetric method (MEP PRC, 1999).

3 Methods

3.1 Classical statistics

Although groundwater salinization in the Tarim River lower reaches is affected by various factors, the correlation between groundwater salinization and parameters describing chemical components can still be established, as is commonly performed in numerous studies (Xu, 2002; Tardyet al., 2004; Giridharanet al.,2008; Chenini and Khemiri, 2009; Xuet al., 2012).For the purpose of comparison, this work also conducted regression analysis to examine the correlation among TDS and ions.

3.2 Geostatistics

Studies have shown that groundwater chemistry parameters are typical regionalized variables, which are structural as well as stochastic (Ahmadi and Sedghamiz, 2008; Wanget al., 2008; Wanget al.,2011). Thus, the spatial variability of groundwater salinization in the Tarim River lower reaches can be analyzed by the geostatistic method (Journel and Huijbregts, 1978; Marchant and Lark, 2004).

3.2.1 The semivariogram

In geostatistics (Journel and Huijbregts, 1978;Marchant and Lark, 2004), the regionalized variable,such as TDS or groundwater depth, is regarded as the value of a variable at a locationxas a realization of a random functionZ(x). This random function is assumed to be intrinsically stationary. This is a weak form of second-order stationarity and is met if two conditions hold. The first is that the expected value of the random function,E[Z(x)], is constant for allx.Secondly, the variance of the differences between the value of the variable at two different locations depends only on the lag vector separating the two locations and not on absolute locations. In general, this variance may be a function of both the direction and length of the lag vector. If the regionalized variable is isotropic, the semivariogram is purely a function of the length of the vector which we denoteh. Thus, the relationship between values from different locations is described by the semivariogram.

The semivariogram is estimated from variable values observed at sampled points,xs,s=1, . . . ,n.The method of moments estimator is the average of squared differences between observations separated by distancehas follows:

whereZ(xi) indicates the magnitude of regionalized variable, andN(h) is the total number of pairs of attributes that are separated by a distanceh.

This study used the semivariogram to reveal the spatial variability of TDS and groundwater level in each year.

3.2.2 Kriging and Cokriging methods

Based on the semivariogram, Kriging and Cokriging can be used to estimate the values of regionalized variable at unsampled locations (Goovaerts, 1997).

Ordinary Kriging can mathematically be defined as given in the following:

whereZ*Xis the estimated value, andλiis the corresponding weight of each observationZ(Xi) on the estimation. These weights are calculated to ensure that the estimator is unbiased and the estimation variance is a minimum. The nonbias condition requires that:

whereγ(Xi,Xj) is the semivariogram between sampled pointiand pointj,γ(Xi,X*) is the semivariogram between sampled point and estimated point, andμis the Lagrange multiplier of minimum condition.

The general form of Cokriging equations are:

whereuandvare the primary and covariate (secondary)variables, respectively. In the Cokriging method, theuandvare cross-correlated and the covariate contributes to the estimation of the primary variable. Generally,measuring the covariate is simpler than measuring the primary variable. For the Cokriging analysis, the cross semivariogram (or cross-semivariogram) should be determined in prior. Provided that there are points where bothuandvhave been measured, the cross-semivariogram is estimated by:

In order to reveal the spatial pattern of groundwater level and TDS, the two interpolation methods, ordinary Kriging and Cokriging were used and cross-checked for interpolation test. Our results show that the accuracy of Cokriging is higher than that of ordinary Kriging.Based on interpolation test, selecting groundwater and altitude as the first and second co-variable, we used the Cokriging method to compute the interpolation values of TDS from 2001 to 2009.

4 Results and discussion

4.1 Groundwater salinization and its related major chemical parameters

The basic statistical parameters and the t-test results for 840 samples are presented in table 1. Overall,the groundwater in the lower reaches of the Tarim River shows slight alkalinity, with a mean pH of 7.66±0.02 and a mean TDS of 5.333±0.450 g/L. As for the anions of CO32-, HCO3-, Cl-and SO42-, the contents are 0.002±0.000, 0.363±0.006, 2.047±0.233 and 1.002±0.052 g/L, respectively. Similarly, the contents for cations of Ca2+, Mg2+, Na+and K+are 0.180±0.009, 0.220±0.012, 1.299±0.140 and 0.036±0.002 g/L, respectively.

Table 1 Basic statistical analysis of groundwater salinization

Studies (Salamaet al., 1999; Wanget al., 2008)show that the higher the TDS’s value of shallow groundwater, the more serious soil salinization becomes. Otherwise the degree of soil salinization will become mild. Therefore, TDS was considered as one of the important indicators for groundwater salinization in the Tarim River lower reaches.

Stepwise regression analysis was performed by taking TDS as dependent variable and other parameters as independent variables. The regression equation is as follows:

The t-test results are presented in table 2 and indicate that the screened independent variables are Na+,Mg2+, Ca2+, Cl-and K+. The partial regression coefficient of the first four independent variables,i.e., Na+,Mg2+, Ca2+, Cl-, are all highly significant (Sig.=0.0000),as well as the independent variable,i.e., K+, shows high significance (Sig.=0.0495). This indicates that Na+,Mg2+, Ca2+, Cl-and K+are the main chemical variables to affect TDS.

The main reason why the regression of TDS exhibits high correlations with Na+, Mg2+, Ca2+, Cl-and K+is that the lower reaches of the Tarim River is a salt accumulation zone, with high salt content in the soil and groundwater (Chenet al., 2005). Moreover, due to implementation of the ecological water delivery project since 2000 (Gaoet al., 2007), the groundwater tables in the lower reaches were uplifted and the concentrations of various chemical substances were diluted. In other words, the lower the groundwater table, the higher the salinity concentration. This result is consistent with studies of Liet al. (2010) and Xuet al. (2012).

Table 2 The correlation between TDS and cations in groundwater chemical components

4.2 Spatial variability of groundwater level and TDS

Our study shows that both groundwater level and TDS have isotropic characteristics for each year from 2001–2009, and their semivariogram conform to the spherical model as follows:

whereC0is the nugget that reflects spatial heterogeneity caused by random factors,ais the maximum range of spatial autocorrelation, andC0+Cis the sill that represent total variance.

Table 3 shows the related semivariogram parameters for groundwater level and TDS. Evidently, both the nugget of groundwater level and TDS are lower,which indicates that the variable error caused by sampling and analyzing is smaller. The maximum range of the groundwater level and TDS reached 90.3 km and 81.1 km respectively, which means that ground water level and TDS presented spatial autocorrelation characteristics in a larger space.

C0/(C0+C) is the ratio of the nugget and sill. The high ratio shows that the spatial variability caused by randomness is dominant. Otherwise, the low ratio shows that the dominant spatial variability is caused by configurative factors. The regionalized variable has a strong spatial correlation whenC0/(C0+C) ≤25%. It is medium when 25%

Table 3 shows that in 2000, 2001 and 2009,C0/(C0+C) of underground water level are all greater than 25%, except for 2006. Thus, it can be seen that the underground water level shows strong spatial autocorrelation in 2002, 2003, 2004, 2005 and 2007. In the remaining years, however, underground water level shows medium spatial autocorrelation due toC0/(C0+C)greater than 25%.

4.3 The assessment of groundwater salinization

Using the Cokriging method, we selected groundwater and altitude as the first and second co-variables,TDS from 2001 to 2009 was drawn respectively. Figures 2–3 shows the spatial patterns of TDS for 2006 and 2009, respectively.

Based on the results of Cokriging interpolation for TDS, according to the classification criteria of division of groundwater salinization degree proposed by Robinoveet al. (1958), the assessment results of groundwater salinization in the Tarim River lower reaches from 2001 to 2009 are presented in table 4.TDS is basically greater than 1 g/L but less than 2 g/L in the Tarim River lower reaches, which indicates that salt stagnation pollution is more serious.The most serious salinization (3 g/L < TDS ≤35 g/L)contaminated area is mainly in the middle and lower part of the study area.

Table 3 Semivariogram and its parameters of groundwater level and TDS

Figure 3 Spatial patterns of TDS in 2009

Table 4 The area ratio of classification of groundwater salinization level

The area ratio of brackish water (1 g/L < TDS ≤3 g/L) first increased, then decreased, and at last showed a smooth trend. It increased from 71.79% to 92.31% during the period of 2001 to 2003 and then smooth in the following years. The area ratio of moderate salt water (3 g/L < TDS ≤10 g/L) presented a decreasing trend first and then basically stabilized after short fluctuations, but with a slight upward trend in 2009. In addition to more than 7% in 2004 and 2009,the area ratio of extreme salt water (10 g/L < TDS ≤35 g/L) stabilized at around 5.1% in other years. The area proportion of brine (TDS >35 g/L) was very low, and it even was 0% in 2002 and 2003.

5 Conclusions

Temporal and spatial variability of groundwater salinization in the Tarim River lower reaches were analyzed by a combined method of classical statistics,geostatistics and geographic information system. The main findings are as follows:

(1) As one of the important indicators for groundwater salinization in the Tarim River lower reaches,TDS is related with other chemical parameters, such as Ca2+, Mg2+, Na+, K+, CO32-, HCO3-, Cl-and SO42-.There is significant correlation between TDS with other related ions, such as Na+, Mg2+, Ca2+, Cl-and K+.

(2) TDS and the underground water level have characteristics of spatial autocorrelation. Although the related parameters of the semivariogram are different,both groundwater level and TDS present isotropic characteristics and conform to the spherical model for each year during 2001–2009.

(3) TDS is basically greater than 1 g/L but less than 2 g/L in the Tarim River lower reaches, which indicates that salt stagnation pollution is more serious. The most serious salinization (3 g/L < TDS ≤35 g/L) contaminated area is mainly in the middle and lower part of study area. The area proportion of brine (TDS >35 g/L)was very low.

This work was supported by the National Basic Research Program of China (973 Program; No.2010CB951003).

Ahmadi SH, Sedghamiz A, 2008. Application and evaluation of Kriging and Cokriging methods on groundwater depth mapping. Environmental Monitoring and Assessment, 138:357–368. DOI: 10.1007/s10661-007-9803-2.

APHA (American Public Health Association), 1995. Standard method for the examination of water and wastewater, 19th Ed., American Public Health Association, Washington D.C..

Chen YJ, Chen YN, Li WH,et al., 2005. The reaction of groundwater chemical characteristics to the eco-water conveyance in the lower Tarim River. Acta Geographica Sinica,60(2): 309–318.

Chen YJ, Zhou KF, Chen YN,et al., 2008. Response of groundwater chemistry to water deliveries in the lower reaches of Tarim River, Northwest China. Environmental Geology, 53:1365–1373. DOI: 10.1007/s00254-007-0746-2.

Chenini I, Khemiri S, 2009. Evaluation of ground water quality using multiple linear regression and structural equation modeling. International Journal of Environmental Science and Technology, 6(3): 509–519.

Chourasia LP, Tellam JH, 1992. Determination of the effect of surface water irrigation on the groundwater chemistry of a hard rock terrain in central India. Hydrological Sciences Journal, 37(4): 313–328.

Fisher RS, Mulican WF, 1997. Hydrochemical evolution of sodium-sulphate and sodium-chloride groundwater beneath the Northern Chihuahuan Desert, Trans-Pecos, Texas, USA. Hydrogeology Journal, 5(2): 4–16.

Gao QZ, Qu JJ, Wang R,et al., 2007. Impact of ecological water transport to green corridor on desertification reversion at lower reaches of Tarim River. Journal of Desert Research,27(1): 52–58.

Giridharan L, Venugopal T, Jayaprakash M, 2008. Evaluation of the seasonal variation on the geochemical parameters and quality assessment of the groundwater in the proximity of River Cooum, Chennai, India. Environmental Monitoring and Assessment, 143: 161–178. DOI:10.1007/s10661-007-9965-y.

Goovaerts P, 1997. Geostatistics for Natural Resources Evaluation. Oxford University Press, New York.

Journel AG, Huijbregts CJ, 1978. Mining Geostatistics. Academic Press, New York.

Karro E, Marandi A, Vaikmäe R, 2004. The original of increase salinity in the Cambrian-Vendian aquifer system on the Kopli Peninsula, Northern Estonia. Hydrogeology Journal, 12(4):424–435.

Li WH, Hao XM, Chen YJ,et al., 2010. Response of groundwater chemical characteristics to ecological water conveyance in the lower reaches of the Tarim River, Xinjiang, China.Hydrological Processes, 24(2): 187–195. DOI:10.1002/hyp.7430.

Marchant BP, Lark RM, 2004. Estimating semivariogram uncertainty. Mathematical Geology, 36(8): 867–898.

MEP PRC (Ministry of Environmental Protection of the People’s Republic of China), 1999. Water Quality—Determination of Total Salt—Gravimetric Method. China Environmental Science Press, Beijing, China.

Richter BC, Kreitler WC, 1993. Geochemical Techniques for Identifying Sources of Groundwater Salinization. CRC Press,New York, pp. 14–26.

Robinove CC, Langford RH, Brookhart W, 1958. Saline water resources of North Dakota. In: Water Supply. US Geological Survey, pp. 1428.

Salama RB, Otto CJ, Fitzpatrick RW, 1999. Contributions of groundwater conditions to soil and water salinization. Hydrogeology Journal, 7: 46–64.

Subyani AM, 2005. Hydrochemical identification and salinity problem of ground-water in Wadi Yalamlam basin,Western Saudi Arabia. Journal of Arid Environment,60(1): 53–66.

Tardy Y, Bustillo V, Boeglin JL, 2004. Geochemistry applied to the watershed survey: hydrograph separation, erosion and soil dynamics. A case study: the basin of the Niger River,Africa. Applied Geochemistry, 19(4): 469–518. DOI:10.1016/j.apgeochem.2003.07.003.

Wang Y, Ma T, Luo Z, 2011. Geostatistical and geochemical analysis of surface water leakage into groundwater on a regional scale: a case study in the Liulin karst system, northwestern China. Journal of Hydrology, 246: 223–234. DOI:10.1016/S0022-1694(01)00376-6.

Wang YG, Xiao DN, Li Y,et al., 2008. Soil salinity evolution and its relationship with dynamics of groundwater in the oasis of inland river basins: case study from the Fubei region of Xinjiang Province, China. Environmental Monitoring and Assessment, 140: 291–302. DOI: 10.1007/s10661-007-9867-z.

Wen XH, Wu YQ, Su JP,et al., 2006. Groundwater salinization characteristics and mechanism in Ejin Basin. Journal of Desert Research, 26(5): 836–841.

Xu JH, 2002. Mathematical Methods in Contemporary Geography. Higher Education Press, Beijing, China, pp. 37–105.

Xu JH, Chen YN, Li WH, 2008. Grey modelling the groundwater level dynamic in the lower reaches of Tarim River affected by water delivery from upper reaches: A demonstration from Yingsu section. 2008 IEEE International Conference on Fuzzy Systems, 2008, vol. 1–5, pp. 43–48.

Xu JH, Chen YN, Li WH,et al., 2012. Statistical analysis of groundwater chemistry of the Tarim River lower reaches,Northwest China. Environmental Earth Sciences, 65(6):1807–1820. DOI: 10.1007/s12665-011-1161-2.

Xu JH, Chen YN, Li WH,et al., 2013. The dynamic of groundwater level in the lower reaches of Tarim River affected by transported water from upper reaches. International Journal of Water, 7(1–2): 66–79. DOI: 10.1504/IJW.2013.051979.

Zhang YM, Chen YN, Pan BR, 2005. Distribution and floristics of desert plant communities in the lower reaches of Tarim River, southern Xinjiang, People’s Republic of China. Journal of Arid Environments, 63(4): 772–784. DOI:10.1016/j.jaridenv.2005.03.023.