For effective groundwater management of a basin, it is essential that a careful water balance study be carried out. A three-dimensional transient-state finite difference groundwater flow model is used to quantify the groundwater fluxes and analyze the dynamic changes of groundwater level. After monitoring groundwater levels for 43 typical observation wells through a simulation study of the groundwater flow model with a depth of 300 m, results reveal that the study area has a lateral recharge of about 3.57 × 109 m3, which makes up 79.08% of the total recharge; total evaporation is about 1.81 × 108 m3, which makes up 3.77% of the total discharge. The balance of groundwater is negative, with a recharge and discharge difference of −2.81 × 108 m3. The correlation coefficient between the observed head and the calculated head for the simulation period is greater than 0.81, indicating the simulation results are satisfactory. The maximum groundwater drawdown is 26.59 m and the rate of the groundwater drawdown is 0.15 m/d during normal operation of the pumping well.

Introduction

Groundwater and surface water resource conversion is frequent in inland river basins in northwest China. Surface water is the main recharge source of groundwater, and the main discharge of groundwater is spring overflow and evaporation. In cases where there is less precipitation and uneven distribution through time and space, the exploitation of groundwater resources has an important ecological role (Cui et al., 2001). Groundwater research is not only related to correctly evaluating water resources and the rational arrangement of water intake engineering problems, but also to how to make full use of water resources and minimize the depletion of water resources and deterioration of water quality (Wang et al., 2010). In 1856, the French engineer Henry Darcy performed the sand tank experiment and put forward the Darcy formula, and the calculation of groundwater experiencing steady flow and unsteady flow has been studied extensively, through both physical and computer simulations. Ahmed & Umar (2009) applied Visual MODFLOW to their attempt to simulate the behavior of a flow system and evaluate a water balance, and showed that the model was the most sensitive to hydraulic conductivity and recharge parameters. Seyf-Laye et al. (2012) applied a three-dimensional groundwater flow model to evaluate groundwater potential and assess the effects of groundwater withdrawal on the regional water level and flow direction in the central Beijing area. Rao et al. (2013) applied a three-dimensional steady state finite difference groundwater flow model to quantify the groundwater fluxes and analyze the subsurface hydrodynamics in a basaltic terrain by giving particular emphasis to a well field that supplies domestic, agricultural, and industrial needs. Groundwater computer numerical simulations can be used for regional groundwater resource evaluation, forecasting, scientific management, and groundwater circulation (Shao et al., 2009). Visual MODFLOW has been widely used in scientific research, production, environmental protection, urban and rural development planning, water resource utilization, and many other industries. In the Manas River Basin, information and data are surprisingly scarce and difficult to obtain. There has been surprisingly little systematic regional research on the basin's development options and challenges using modern analytical tools that go beyond sector, country or state analysis to examine the larger strategic questions that the basin faces (Sadoff et al., 2013). The objective of this study is to establish a groundwater flow model using Visual MODFLOW Pre version 2010.1, and help quantify the groundwater fluxes and analyze the dynamic changes of the groundwater level in the Manas River Basin, China. The paper presents a comparison of the model calculated groundwater heads with field measured groundwater heads through regression plots. The results of the study can provide a valuable reference for rational utilization of watershed groundwater exploitation and water resources. We believe the basic conclusions based on the model results are robust and we have used them to develop strategic insights.

Materials and methods

Description of the study area

The Manas River Basin is located at the northern foot of the Tianshan Mountains in Xinjiang, along the southern margin of the Junggar basin, at latitudes 43 °27′–45°21′ and longitudes 85 °01′–86 °32′. The total area of the Manas River Basin is 1.98 × 104 km2, with the mountainous area measuring 0.52 × 104 km2 and the plain area 1.46 × 104 km2. The basin is far from the ocean, and the climate is dry, with a high percentage of evaporation. There are both middle temperate continental arid climate characteristics, and vertical climate characteristics, which is a typical continental climate.

The study area is located in the middle part of the basin plain, with an area of about 1.46 × 104 km2, about 0.74 of the whole area. Because of the influence of water vapor sources, topography, and latitude, the precipitation distribution is very uneven, and the vertical zoning characteristics are obvious. The annual precipitation of the middle and high mountainous areas is 400–600 mm, for the low mountainous area it is 340–420 mm, whilst for the plain area it is 110–200 mm, and it is mainly concentrated from June to August, which accounts for 68% of the annual rainfall. In the piedmont region of the alluvial fan, the groundwater is connected, interrelated, and independent, the northern edge of the fan overflow transitioning to a downstream fine soil plain area, forming a complete hydrogeological unit (Figure 1). The quaternary loose strata provide a good space for the storage and migration of groundwater. The surface water from the river is the main source of groundwater in the study area. The Manas River is the largest river within the basin, having a total length of 400 km, with an average annual runoff of 12.68 × 108 m3 (Ling et al., 2005). The river is sourced in the north Tianshan Yilianhabierga Mountain, and flows from south to north in the Junggar basin. It is a typical comb-like river system, with runoff distribution in the mountains above the mountain-pass, and loss below the mountain-pass.
Fig. 1.

Location map of the study area.

Fig. 1.

Location map of the study area.

Methodology for numerical modeling

Visual MODFLOW is a three-dimensional flow model based on the finite difference method and is capable of simulating flows in an aquifer system, as well as the interaction between the groundwater system and the surface environment. Also, it is capable of handling saturated flow with homogeneous, heterogeneous, isotropic, and anisotropic aquifer systems under steady state and transient conditions (Dawoud & Allam, 2004). Due to the subsystem aquifer soil characteristics, the water abundance and permeability show no obvious direction. Therefore, the study area was generalized to a non-homogeneous isotropic three-dimensional unsteady flow aquifer system, expressed as follows: 
formula
 
formula
 
formula
 
formula

where: D is the seepage zone; K is the aquifer hydraulic conductivity (m/d); H is the groundwater head value; W is the volumetric flux per unit value representing sources and/or sinks of water (W < 0.0 for outflow of the groundwater system, W > 0.0 for inflow), (m/d); Ss is the aquifer specific yield; H0(x, y, z) is the initial flow field head (m); t is time; n is the direction of the second boundary outward normal; H1(x, y, z, t) is the head distribution value of the first boundary (m); B1 is the first class boundary; q(x, y, z, t) is the discharge per unit width of the second boundary (m3/d); and B2 is the second boundary.

The groundwater flow model was used to compute a detailed total water balance for the study area to provide information on available water resources and demands within the modeled catchment. Actual evapotranspiration plays an essential role in the phreatic cycle, together with other processes such as precipitation, irrigation water recharge and lateral aquifer inflow/outflow. Theoretical representation of the phreatic cycle is mostly based on physical laws, particularly those of conservation of mass, Newton's laws of motion and the law of thermodynamics. According to the conditions of recharge and discharge in the study area, the water balance equation is as follows: 
formula

where: P is the precipitation infiltration recharge; QIrr. is the irrigation water recharge; QIn is the groundwater lateral aquifer inflow; QOut is the outflow of water from the northern and western boundaries; E is the groundwater evaporation for depths of less than 5 m; QWell is the artificial exploitation; and D is the amount of water storage, with units in m3.

In order to achieve effective groundwater management in the study area, a three-dimensional transient-state finite difference groundwater flow model was developed using MODFLOW-2005, which is one of the modules in Visual MODFLOW Pre version 2010.1. The simulation for transient state was solved by WHS solver. The flow chart indicating the methodology adopted is given in Figure 2.
Fig. 2.

Flow chart adopted for numerical modeling.

Fig. 2.

Flow chart adopted for numerical modeling.

Model discretization

The southern section of the study area is a phreatic saturated aquifer, with a thickness of more than 400 m which gradually thins to the north. The Wuyi highway to the north is a multilayer aquifer structure, with the upper part of the phreatic aquifer gradually thinning northward to form a perched aquifer. The lower part is a multilayer confined water-artesian aquifer. From 100 to 200 m depth, there are two to three aquifers, and within the entire 200 m, there are five aquifers (Qiao et al., 2015). Interval layers in the northern mixed structure are not complete and present an interdigitate shape, coupled with the thousands of groundwater exploitation wells throughout the aquifer. This has formed a groundwater exploitation well. Soil types of different aquifers in the study area are shown in Table 1. This paper uses a numerical simulation method to discretize the model. In the horizontal direction, the river basin is divided into 400 rows and 410 columns, a 360 × 560 m regular grid; in the vertical direction, it is divided into 10 aquifer groups (Figure 3). The benefits of such stratification are that it can give the hydrogeological parameters of different aquifers. Each layer can be divided into different parameter regions, which can be solved by different soil types in the same layer. The model uses 2011–2012 as the simulation period and has a simulation depth of 300 m. Model data inputs and sources are as follows:
  • (i) Boundary of study area: extracted from a digital elevation model (DEM) by ArcGIS10.1.

  • (ii) Topography of the study area: DEM import model of the study area.

  • (iii) Precipitation and atmosphere temperature: from the river basin meteorological site measured data.

  • (iv) Groundwater depth data and pumping well data: provided by the Water Conservancy (representing the Department of Water Administration).

Fig. 3.

Generalized aquifers in the study area.

Fig. 3.

Generalized aquifers in the study area.

Table 1.

Soil types of different aquifers in the study area.

    Soil type
 
Layer Depth (m) II III IV VI VII 
First sandy g. m. sand m. sand m. sand m. sand clay clay 
Second 15 sandy g. m. sand m. sand m. sand m. sand m.c. sand m.c. sand 
Third 10 conglomerate sandy g. m.c. sand m.c. sand m.c. sand m.c. sand m.c. sand 
Fourth 50 conglomerate sandy g. m.c. sand m. sand clay m.c. sand m.c. sand 
Fifth 30 conglomerate sandy g. mild clay sandy gravel m.c. sand mild clay clay 
Sixth 30 conglomerate sandy g. mild clay m. sand clay mild clay medium fine sand 
Seventh 20 conglomerate sandy g. m. sand m. sand clay fine sand fine sand 
Eighth 40 conglomerate sandy g. clay fine sand fine sand fine sand fine sand 
Ninth 30 fault sandy g. clay clay clay clay clay 
Tenth 70 fault sandy g. fine silty sand m. sand m. sand m. sand m. sand 
    Soil type
 
Layer Depth (m) II III IV VI VII 
First sandy g. m. sand m. sand m. sand m. sand clay clay 
Second 15 sandy g. m. sand m. sand m. sand m. sand m.c. sand m.c. sand 
Third 10 conglomerate sandy g. m.c. sand m.c. sand m.c. sand m.c. sand m.c. sand 
Fourth 50 conglomerate sandy g. m.c. sand m. sand clay m.c. sand m.c. sand 
Fifth 30 conglomerate sandy g. mild clay sandy gravel m.c. sand mild clay clay 
Sixth 30 conglomerate sandy g. mild clay m. sand clay mild clay medium fine sand 
Seventh 20 conglomerate sandy g. m. sand m. sand clay fine sand fine sand 
Eighth 40 conglomerate sandy g. clay fine sand fine sand fine sand fine sand 
Ninth 30 fault sandy g. clay clay clay clay clay 
Tenth 70 fault sandy g. fine silty sand m. sand m. sand m. sand m. sand 

Where: sandy g. is sandy gravel, m. sand is middle sand, m.c. sand is medium coarse sand.

Grid design and boundary conditions

The study area map was digitized using the ArcGIS 10.1 software and imported into Visual MODFLOW as a shape file. We defined inactive cells to avoid unnecessary computation and reduce the model error. In addition, we input the pumping wells in the vicinity of the grid to further refine and then smooth the overall grid. In order to simulate the variation of the aquifer thickness and the actual surface topography, we imported the DEM of the study area into Visual MODFLOW, so that each layer generates a variable upper and lower elevation (Figure 4).
Fig. 4.

Three-dimensional map of the model.

Fig. 4.

Three-dimensional map of the model.

Every model requires an appropriate set of boundary conditions to represent the system's relationship with the surrounding area. General head boundaries were assigned at the eastern and southern edges of the model. A constant head boundary was simulated in the northern part to represent lateral inflow entering into the study area. A wall boundary was assigned at the western edges of the model, because there is no influence of the water table due to the flow in the study area. The evaporation boundary was set to 1,000 mm/year, based on the yearly average for the study area. Rainfall, irrigation return water and canal seepage were assigned to the recharge boundary, which is input into the model for the first layer of the aquifer. Rainfall data are from the meteorological stations on the river; the location of the meteorological stations is clearly demonstrated in Figure 1. Specific input data and sources are shown in Table 2.

Table 2.

Recharge statistics.

Irrigation district Regiment Groundwater (104 m3Surface water (104 m3Precipitation (mm/y) Control area (m2Recharge depth (mm/y) 
Shihezi Shihezi head farm 10,230 10,569 197.9 357,017,850 905.34 
Shihezi township 293 4,165 
Xiayedi 121st 3,331.8 6,418 127.8 731,103,220 680.63 
122nd 305.94 4,993 
132nd 1,193.4 5,425 
133rd 233.46 4,396 
134th 238.68 4,219 
135th 1,847.88 3,049 
136th 2,148.12 2,618 
Mosuowan 147th 3,925 5,004 132.7 575,095,420 852.90 
148th 2,862 7,912 
149th 3,443 6,073 
150th 5,587 6,894 
Jin'an 144th 1,977 6,364 211 191,076,220 1,701.87 
143rd 5,050 15,096 
141st 624 4,283 122.8 414,087,370 634.48 
142nd 9,530 6,751 
Irrigation district Regiment Groundwater (104 m3Surface water (104 m3Precipitation (mm/y) Control area (m2Recharge depth (mm/y) 
Shihezi Shihezi head farm 10,230 10,569 197.9 357,017,850 905.34 
Shihezi township 293 4,165 
Xiayedi 121st 3,331.8 6,418 127.8 731,103,220 680.63 
122nd 305.94 4,993 
132nd 1,193.4 5,425 
133rd 233.46 4,396 
134th 238.68 4,219 
135th 1,847.88 3,049 
136th 2,148.12 2,618 
Mosuowan 147th 3,925 5,004 132.7 575,095,420 852.90 
148th 2,862 7,912 
149th 3,443 6,073 
150th 5,587 6,894 
Jin'an 144th 1,977 6,364 211 191,076,220 1,701.87 
143rd 5,050 15,096 
141st 624 4,283 122.8 414,087,370 634.48 
142nd 9,530 6,751 

In the study area irrigation districts, which are defined based on agricultural production, the annual water production is in strict accordance with the irrigation scheduling in each district's allocation. According to the irrigation schedule in different areas, the recharge amount is supplied by a different proportion of the total irrigation water. Table 2 is the statistical recharge and Figure 5 is the result of the initial head interpolation in the study area.
Fig. 5.

Initial head of the model.

Fig. 5.

Initial head of the model.

Groundwater observation and pumping wells

There are 47 groundwater level observation wells, which are automatically observed without human interference, but there are only 43 actual monitoring wells (Figure 1). These observation wells are distributed in each irrigation district, and the observation time is 08:00, which provides good data for the groundwater simulation. Pumping wells represent groundwater sinks in the model. The depth of the open bedrock zones in the wells ranges from 80 to 250 m below the ground surface. Those zones were assigned as the pumping interval in the groundwater flow model. The pumping rates of the wells were distributed over the length of the openhole interval of the pumping well, which intersects multiple model layers. As a result of the large number of pumping wells within the study area, great difficulties occurred in the model in pumping, so we conceptualized a volume of small pumping wells as a large pumping well in each regiment (Table 3). In Visual MODFLOW, the user assigns the total pumping rate to the well, and this total rate is proportioned by the model to each layer based on the length of the well screen intersecting the model layer and the horizontal hydraulic conductivity of the model layer.

Table 3.

Number of wells and pumping rate.

Irrigation district Regiment Control area (km2Actual well number Generalized number Rate (m3/d) 
Xiayedi 121st 456 241 73 5,840 
122nd 299.04 24 5,840 
132nd 463.2 69 26 5,840 
133rd 281. 53 64 5,840 
134th 222.81 32 5,840 
135th 355.73 145 40 5,840 
136th 355.73 191 47 5,840 
Jin'an 141st 207.95 95 14 5,840 
142nd 701.65 459 208 5,840 
143rd 378.72 217 110 5,840 
144th 322.26 309 43 5,840 
Mosuowan 147th 225 256 86 5,840 
148th 309 278 62 5,840 
149th 342 264 75 5,840 
150th 451 392 122 5,840 
Shihezi Shihezi city 76 167 149 5,840 
Shihezi head farm 373 433 223 5,840 
152nd 42 29 5,840 
Shihezi township 176.85 58 5,840 
Irrigation district Regiment Control area (km2Actual well number Generalized number Rate (m3/d) 
Xiayedi 121st 456 241 73 5,840 
122nd 299.04 24 5,840 
132nd 463.2 69 26 5,840 
133rd 281. 53 64 5,840 
134th 222.81 32 5,840 
135th 355.73 145 40 5,840 
136th 355.73 191 47 5,840 
Jin'an 141st 207.95 95 14 5,840 
142nd 701.65 459 208 5,840 
143rd 378.72 217 110 5,840 
144th 322.26 309 43 5,840 
Mosuowan 147th 225 256 86 5,840 
148th 309 278 62 5,840 
149th 342 264 75 5,840 
150th 451 392 122 5,840 
Shihezi Shihezi city 76 167 149 5,840 
Shihezi head farm 373 433 223 5,840 
152nd 42 29 5,840 
Shihezi township 176.85 58 5,840 

Model parameter and calibration

The various aquifer parameters, such as hydraulic conductivity, specific yield and specific storage, were estimated and assigned to different layers, using data derived from previous studies (Yang et al., 2011; Yu-Jiao et al., 2012). Figure 6 is the recharge boundary of the model. In order to facilitate the input of the observation well data, the initial head interpolation is applied to the elevation water level interpolation. The actual method has the following steps: first, the groundwater depth is interpolated; then the interpolation data are derived; finally, we subtract the interpolation results from the elevation of the corresponding points, giving us the groundwater level.
Fig. 6.

Boundary conditions of the model.

Fig. 6.

Boundary conditions of the model.

The groundwater flow model calibration procedure involves adjusting model parameters so that the simulated results provide an acceptable match to the observed conditions while reasonable parameter values are maintained. Through a combination of automatic model parameter identifications and manual parameter tuning methods, we analyze the fitting effect of the actual groundwater flow field and simulation flow field, taking the groundwater level of the 43 long-term observation wells as the basis for parameter calibration. The calculated results for the water balance are used as the standard for model parameter verification. Table 4 shows the hydrological parameters of each type of soil after calibration.

Table 4.

Hydrogeological parameters for different soil types.

Soil types Conductivity (m/s) Specific yield Specific storage (1/m) 
Sandy gravel 8.68 × 10−4 0.1 1.00 × 10−5 
Middle sand 4.05 × 10−4 0.06 1.00 × 10−5 
Clay 5.79 × 10−7 0.3 1.00 × 10−5 
Medium coarse sand 5.21 × 10−4 0.15 1.00 × 10−5 
Mild clay 2.31 × 10−6 0.3 1.00 × 10−5 
Medium fine sand 3.47 × 10−4 0.12 1.00 × 10−5 
Fine sand 2.31 × 10−4 0.11 1.00 × 10−5 
Conglomerate 8.68 × 10−4 0.1 1.00 × 10−5 
Fine silty sand 5.79 × 10−4 0.07 1.00 × 10−5 
Soil types Conductivity (m/s) Specific yield Specific storage (1/m) 
Sandy gravel 8.68 × 10−4 0.1 1.00 × 10−5 
Middle sand 4.05 × 10−4 0.06 1.00 × 10−5 
Clay 5.79 × 10−7 0.3 1.00 × 10−5 
Medium coarse sand 5.21 × 10−4 0.15 1.00 × 10−5 
Mild clay 2.31 × 10−6 0.3 1.00 × 10−5 
Medium fine sand 3.47 × 10−4 0.12 1.00 × 10−5 
Fine sand 2.31 × 10−4 0.11 1.00 × 10−5 
Conglomerate 8.68 × 10−4 0.1 1.00 × 10−5 
Fine silty sand 5.79 × 10−4 0.07 1.00 × 10−5 

Results and discussion

Figure 7 gives the simulation results of the model in a series of 24 days, 120 days, 240 days, and 300 days of groundwater flow field. From the simulation results in Figure 7: from January to April, the agricultural wells stop pumping, so groundwater flow is stable and less affected by human disturbance. In the winter, air temperatures are generally below 0 °C and evaporation intensity weakens, and the groundwater is supplied by lateral and vertical snowmelt recharge, so the groundwater level begins to rise slowly. In April, snow melts completely, and agricultural activities begin, with thousands of agricultural motor pumped wells actively pumping; accordingly, groundwater level decreases rapidly until groundwater depth reaches the maximum value in August. As pumping capacity decreases, the groundwater level begins to rise slowly, but there is a ‘salt washing moisture conservation’ irrigation concentration in October, with a large quantity of pumping and a small decrease in groundwater level. After that the groundwater level rises slowly until the initial water level is recovered. The groundwater levels in the Mosuowan, Anjihai, Xiayedi, and Shihezi irrigation districts are almost the same at the end and the beginning of the year, indicating that in the study area, throughout the year, groundwater level undergoes a cyclical process, and the simulation is reasonable for the year 2011.
Fig. 7.

24, 120, 240, and 300 days of groundwater simulation flow field for the study area.

Fig. 7.

24, 120, 240, and 300 days of groundwater simulation flow field for the study area.

In the study area, as shown in Figure 7, the overall groundwater depth is not the same in different regions. The groundwater depth in the south of the study area is higher than that in the north, and that in the east is higher than that in the west. This phenomenon has a great relationship with the topography of the study area, which tilts from the southeast to the northwest. We find that the groundwater level contours in the study area are relatively dense and in the irrigation district, represented by a point in the outer radius. There is a great relationship between the agricultural activities in the irrigation district. In addition, we have a volume of small pumping wells conceptualized as a large pumping well in each interior regiment, which is also a major contributor to this relationship.

From Figure 8 the correlation coefficients between the calculated and observed heads in the 1st, 4th, 8th, and 10th month were above 0.81 (above 0.96 in the first four months without pumping wells). The hydrogeological parameters of the model can be used as the theoretical parameters for the basin in the future, suggesting the simulation results are satisfactory.
Fig. 8.

Calculated head and observed head correlation coefficients.

Fig. 8.

Calculated head and observed head correlation coefficients.

Drawdown calculation

Groundwater depth is very important to the ecological stability in arid and semi-arid regions (Currell et al., 2012). The calculation time of groundwater drawdown is the normal operation of the pumping well. The calculation of groundwater drawdown rate is the groundwater depth at the end of the normal operation of the pumping well minus the initial groundwater depth. From Table 5, we see that Shihezi and Jin'an irrigation districts' groundwater depths are larger than those of the other irrigation districts. During the normal operation of the pumping well, from the results of the study, we see that the maximum depth of groundwater is greater than 18.65 m, the maximum groundwater drawdown is greater than 14.19 m, and the maximum groundwater drawdown rate is 0.07 m/day. The groundwater flow field distribution in the four irrigation districts differs, as it generally decreases from southeast to northwest. The Jin'an and Shihezi groundwater depths have a maximum depth of greater than 40 m, the Mosuowan groundwater depth is greater than 26 m, and the Xiayedi has a maximum depth of more than 18 m. This variation is greatly related to geography, groundwater recharge, and transport under gravity flow from high to low groundwater levels, such that elevation largely determines the groundwater depth distribution. However, the largest groundwater drawdown in numerical terms is found in the Xiayedi irrigation district, which has a significantly greater drawdown rate than the other three irrigation districts. The drawdown rate correlates well to the Xiayedi hydrogeological conditions and geographical location. The Xiayedi irrigation district is the largest of the four, the Gurbantunggut boundary is quite extensive, and the groundwater discharges into the desert freely. Furthermore, groundwater is not recharged timely and effectively. According to the hydrogeological section, the Xiayedi irrigation district groundwater aquifers have clay layers at different depths, blocking groundwater vertical recharge and leading to increasing groundwater depth over time, as the drawdown rate increases.

Table 5.

Groundwater drawdown.

Irrigation district Groundwater depth (m) Maximum depth reduction (m) Reduction rate (m/d) Time scale (days) Pumping quantity (104 m3
Mosuowan 3.34 ∼ 26.55 16.25 0.07 232 15,817 
Xiayedi 3.02 ∼ 18.65 26.59 0.15 177 9,299.28 
Jin'an 3.17 ∼ 44.1 18.23 0.08 228 17,181 
Shihezi 2.31 ∼ 65.55 14.19 0.08 177 17,478 
Irrigation district Groundwater depth (m) Maximum depth reduction (m) Reduction rate (m/d) Time scale (days) Pumping quantity (104 m3
Mosuowan 3.34 ∼ 26.55 16.25 0.07 232 15,817 
Xiayedi 3.02 ∼ 18.65 26.59 0.15 177 9,299.28 
Jin'an 3.17 ∼ 44.1 18.23 0.08 228 17,181 
Shihezi 2.31 ∼ 65.55 14.19 0.08 177 17,478 

The influence of atmospheric temperature on groundwater depth

From Figure 9, the comparison between the observed head of different irrigation district observation wells and calculated head from the model indicate that the annual variation of groundwater depth in some areas is basically consistent with the atmospheric temperature variation. Figure 9(a), 9(b), 9(c) and 9(d), respectively, represent the Mosuowan, Xiayedi, Jin'an and Shihezi irrigation districts. In general, when the atmospheric temperature increases, evaporation increases, and the shallow groundwater begins to evaporate, resulting in the decline of groundwater level; when the atmospheric temperature decreases, evaporation becomes weak, the evaporation of shallow groundwater is reduced, and the groundwater level is increased. The effect of evaporation on the groundwater should have a ‘limit depth’. Related studies believe that the Manas River Basin evaporation ‘limit depth’ is 6 m (Huang et al., 2007). Evaporation for groundwater depths greater than 6 m is very small, but in this case, the evaporation in the groundwater cycle acts as a driving force, promoting the exchange of aquifer water and indirectly having an impact on deep groundwater depth. In order to explore which aquifer the evaporation affects, SPSS18.0 software is used to analyze the correlation between the atmospheric temperature data and the groundwater depth data of the first and second aquifers, using a 2-tailed test.
Fig. 9.

Comparison of observed head and calculated head.

Fig. 9.

Comparison of observed head and calculated head.

From Table 6, a significant correlation between the simulated atmospheric temperature and the first layer aquifer groundwater depth is found, with a sig. <0.01, but atmospheric temperature and the second layer aquifer groundwater depth has a value of sig. >0.01, indicating they are not significantly correlated. These results show the evaporation of groundwater directly influences the first layer of the aquifer.

Table 6.

Variable correlation analysis.

Layer Variable Correlations Temperature Head 
First Temperature Pearson correlation 0.328** 
Sig. (2-tailed)  0.000 
365 365 
Head Pearson correlation 0.328** 
Sig. (2-tailed) 0.000  
365 365 
Second Temperature Pearson correlation 0.096 
Sig. (2-tailed)  0.068 
365 365 
Head Pearson correlation 0.096 
Sig. (2-tailed) 0.068  
365 365 
Layer Variable Correlations Temperature Head 
First Temperature Pearson correlation 0.328** 
Sig. (2-tailed)  0.000 
365 365 
Head Pearson correlation 0.328** 
Sig. (2-tailed) 0.000  
365 365 
Second Temperature Pearson correlation 0.096 
Sig. (2-tailed)  0.068 
365 365 
Head Pearson correlation 0.096 
Sig. (2-tailed) 0.068  
365 365 

**Correlation is significant at the 0.01 level (2-tailed).

Water balance analysis

The groundwater balance of the study area is controlled by climate variables, agricultural production, and the characteristics of the watershed. Under natural conditions, in a groundwater aquifer or groundwater aquifer system which has a relatively stable environment, the interaction between atmospheric water, surface water, soil water and groundwater is basically stable. In different regions, groundwater balance factors are different. A large number of irrigated areas use both surface water and groundwater, with irrigation return recharge and artificial exploitation of the main revenue and expenditure items. The characteristics of the groundwater balance in different periods are also different, for example the change of the average annual groundwater balance is equal to zero. By using the equilibrium relationship of groundwater, the groundwater resources can be estimated and the conversion relationship between various water resources studied. In the groundwater flow model, we set the study area to a water balance calculation area. Exported model generated calculation data are shown in Table 7. Water balance calculation results show that surface water recharge is about 9.45 × 108 m3, accounting for 20.92% of the total water recharge. Evaporation is about 1.81 × 108 m3, accounting for 3.77% of the total discharge. Taking into account the water-saving irrigation measures and plant transpiration, we anticipate this figure to be much larger than the calculated value, but in terms of evaporation, this is large. The southern piedmont alluvial fan and the eastern boundary have a lateral recharge, which is about 3.57 × 109 m3, accounting for 79.08% of the total recharge. The groundwater sources in the study area are primarily sourced by upstream glacier melt water in the form of lateral recharge. On the whole, groundwater balance is negative, with a recharge and discharge difference of −2.81 × 108 m3.

Table 7.

Contributions to water balance in the study area.

Water balance Contribution (m3Total (m3Percentage of total (%) 
Recharge Surface inflow 9.45 × 108 4.52 × 109 20.92% 
Lateral inflow 3.57 × 109 79.08% 
Discharge Lateral outflow 4.02 × 109 4.80 × 109 83.78% 
Pumping quantity 5.98 × 108 12.46% 
Evaporation capacity 1.81 × 108 3.77% 
Water balance Contribution (m3Total (m3Percentage of total (%) 
Recharge Surface inflow 9.45 × 108 4.52 × 109 20.92% 
Lateral inflow 3.57 × 109 79.08% 
Discharge Lateral outflow 4.02 × 109 4.80 × 109 83.78% 
Pumping quantity 5.98 × 108 12.46% 
Evaporation capacity 1.81 × 108 3.77% 

Conclusions

There are daunting challenges in the Manas River Basin but there are also many opportunities for cooperation and productive management of rivers and water resources. From the simulation results, we suggest that the generalized aquifer system accuracy and modeling process provides a true and reliable case. Visual MODFLOW can be used to study the groundwater cycle in the river basin. It reveals the inner mechanism of the water resource evolution in the river basin, and puts forward a scheme of water resource management in the irrigation districts. Because the water also transports mass, groundwater numerical simulation can not only solve the flow field within the river basin water, but also study watershed soil erosion and sediment transport, as well as the migration of pollutants, for which this study lays the foundation. We make three suggestions for the basin based on the results of the study, as follows. (1) Increasing supply of surface water and reducing demand of groundwater can reduce groundwater depletion to increase sustainability. Increasing water storage can help resolve the temporal disconnections between supply and demand. (2) There is a limit to the efficiency of irrigation systems because of the potential for soil salinization. Perhaps, irrigated crops could be converted to rainfed crops to reduce groundwater depletion. (3) Temporal disconnects between water supply and demand can be managed through conjunctive use of surface water and groundwater and increased water storage.

Acknowledgements

We acknowledge the support from the National Natural Science Fund Project (U1203282) (51269026) (41361096) (41201113), Innovation Team of Bingtuan (2014CC001), Science and Technology Support Program (2013BAC10B01) (2014BAC14B01), Shihezi University Outstanding Young Project (2012ZRKXJQ08), and the Scientific Research and Innovation projects of graduate students in Xin Jiang (XJGRI2014051) (XJGRI2014056) (XJGRI2015043).

References

References
Cui
Y.-L.
Shao
J.-L.
Han
S.-P.
(
2001
).
Ecological environment adjustment by groundwater in Northwest China
.
Earth Science Frontiers
8
,
191
196
.
Huang
M.
Cai
H.
Huang
Z.
(
2007
).
Research on the evaporation of underground water in different burying depth conditions on the Loess area
.
Journal of Northwest A & F University (Natural Science Edition)
35
,
233
237
.
Ling
G. P.
Wu
L. T.
Feng
Z. S.
Qian
L.
(
2005
).
Dynamic simulation to ground water resource in Manas plain
.
Journal of Hydrodynamics
20
(
5
),
648
653
.
Qiao
L. I.
Zhou
J. L.
Gao
Y. X.
Ming-Liang
D. U.
Cheng
F.
Feng-Xiu
L. I.
Meng
Q.
(
2015
).
Groundwater hydro-geochemistry in Plain of Manasi River Basin, Xinjiang
.
Geoscience
29
(
2
),
239
244
.
Sadoff
C.
Harshadeep
N. R.
Blackmore
D.
Wu
X.
O'Donnell
A.
Jeuland
M.
Whittington
D.
(
2013
).
Ten fundamental questions for water resources development in the Ganges: myths and realities
.
Water Policy
15
,
147
164
.
Seyf-Laye
A-S. M.
Mingzhu
L.
Djanéyé-Bouindjou
G.
Fei
L.
Lyutsiya
K.
Moctar
B. L.
Honghan
C.
(
2012
).
Groundwater flow and contaminant transport modeling applications in urban area: scopes and limitations
.
Environmental Science and Pollution Research
19
,
1981
1993
.
Shao
J.
Zhao
Z.
Cui
Y.
Wang
R.
Li
C.
Yang
Q.
(
2009
).
Application of groundwater modeling system to the evaluation of groundwater resources in North China plain
.
Resources Science
31
,
361
367
.
Wang
H.
Lu
C.
Qin
D.
Sang
X.
Li
Y.
Xiao
W.
(
2010
).
Advances in method and application of groundwater numerical simulation
.
Earth Science Frontiers
17
,
1
12
.
Yang
G.
Chen
F.
He
X.
(
2011
).
The groundwater evolution trend and driving forces analysis of the plain areas in the Manas River Basin
.
Journal of Shihezi University (Natural Science)
2
,
023
.
Yu-Jiao
D. U.
Xin-Lin
H. E.
Yang
G.
Wang
Z. W.
Huang
D.
Wang
W.
(
2012
).
An evaluation of groundwater resources of Mosuowan irrigation district of the Manas River by using the water balance
.
China Rural Water & Hydropower
9
,
63
65
.