Regional groundwater qualitative and quantitative management through hydrogeological modeling in Jangal-Geysour, Razavi Khorasan Province, Iran

Abstract

The severe discharge of aquifers is presenting many problems such as a drop in groundwater level and resulting in ground settlement and quality decline. To determine the groundwater system current conditions, the data of 15 piezometers and 11 sampling points concerning groundwater levels and quality were obtained from 1999 to 2014 in the study area of Jangal-Geysour, Razavi Khorasan Province, Iran. Concerning groundwater levels and quality data, two developed models of MODFLOW and MT3D were applied and calibrated from 1999 to 2010 and validated for the years 2010–2014. After substantial calibration and validation, four scenarios were presented. In scenario one (to balance the relative groundwater level drop condition in a five-year period) to achieve this target, the rate of withdrawals was decreased annually over 10%. To provide water demands of the public, the 10% reduction in the exploitation amount can be compensated by the groundwater sourced from the Jangal-Geysour groundwater resource field. In the second scenario, by applying the suggested addition withdrawals control method, the average moment withdrawal of 33.8 l/s was calculated for each well, and wells that exploit more than average should be controlled and the amount of overdraft subtracted from them. The third scenario studied wells used to supply the demand for agriculture, and tried to presented a suitable cropping pattern by considering allowed work hours to control and manage withdrawals. In the fourth scenario, the impact of irrigation of salty water on production decrease because of water level drop by continuing the current trend was evaluated.

Introduction

Because groundwater is the principal source of freshwater, there is the problem of overexploitation in many places that has negative results, including aquifer drawdown, poor quality, etc. (Zhou and Li 2011). Groundwater discharge is a significant menace in arid and semi-arid areas, where surface water is ordinarily temporary, and groundwater is the principal water supplier (Niazi et al. 2014). Generally, the Middle East is known for a of lack of water and rapid population growth. Therefore, water is a fundamental determinant for the future progress of this region. The obstacle presented by a lack of water for countries such as Iran, which has a semi-arid and dry climate, has long been discussed. Access to water resources is important for human consumption, agriculture, and industry. To overcome groundwater resource problems, multi-sectored oversight and administration of groundwater and surface water is necessary. Some of the effective causes of integrated technical approaches are the scientific identification of exact water supplies and the place of the plains. One of the tools used by researchers is the numerical model. The groundwater model simulates an aquifer’s natural conditions based on data that is imported to the model (Ghodrati and Sabani 2012). At present, there is insufficient knowledge of the local hydrogeological circumstances for the prosperous and sustainable utilization of the groundwater supplies. The first step is preparing a conceptual model that recognizes the behavior and natural condition of the aquifer. A conceptual model depicts the aquifer features, including hydrogeologic units, borders of system, time-varying inputs and outputs, and hydraulic transport features, including their spatial variability (Anderson and Woesner 1992; Meyer and Gee 1999). The conceptual model develops the original design of the model, and it is the fundamental theory about the system’s performance or process (Bredehoeft 2003; Sohrabi et al. 2013). Conceptual models are created according to the background or prior investigations in the area of interest, understanding of related hydrogeologic regimes, and site-specific information (Gillespie et al. 2012); therefore, they are an essential part of modeling. Furthermore, it is advised that the dilemma in groundwater modeling is extensively managed by rising ambivalence to set groundwater conceptual models (Poeter and Anderson 2005; Neuman and Wierenga 2003; Refsgaard et al. 2006; Ye et al. 2010; Højberg and Refsgaard 2005; Carrera and Neuman 1986; Rojas et al. 2008, 2010; Meyer et al. 2007). Also, neglecting the skepticism of the conceptual model results in biased forecasts and/or underestimating the ambiguity of the forecast (Ye et al. 2010). Subsequent conceptualizing of the main agents influencing the groundwater flow is fundamental to choosing the correct way to model (Bakalowicz 2006). Water resource directors proposing to simplify the securance and request usually suggest models as water administration instruments (Johnston and Smakhtin 2014). Nonetheless, complicated and nonlinear methods and data deficiency may challenge groundwater flow modeling (Rödiger et al. 2014). By applying mathematical equations that are the foundation of certain analyzing hypotheses, groundwater models represent groundwater flow, future, and transport processes. The equations are resolved by various kinds of models. Analytical models provide accurate solutions to equations that represent highly manageable flow or transport conditions, whereas numerical models supply estimate results by resolving complicated field conditions. Usually, there are not attempts to collect united data in developing countries, particularly with regard to groundwater systems. New groundwater modeling investigations, including semi-arid basins, have faced challenges and problems in evaluating recharge (Candela et al. 2013; Rödiger et al. 2014), considerations (Le Page et al. 2012), and the influence of errors on hydrogeologically intricate flow systems (Bense et al. 2013) or in describing a credible, suitable conceptual model of groundwater system (Ayenew and Tilahun 2008). Groundwater flow modeling and solute transport simulations, applying MODFLOW and MT3D, have been coupled. Numerous studies have been conducted applying MODFLOW. Jerome and Chantal (2002) used MODFLOW and MT3D to study the nitrate concentration in an aquifer. MT3D was also successfully used in an aquifer in Mashhad located in the northeast of Iran for simulation of nitrate (Siadat Moghadam 2004). Gebreyohannes et al. (2013) employed a water balance of spatially distributed model to evaluate the surface and groundwater supplies in the Geba basin. The research outcomes showed that aquifer recharge is inadequate to irrigate crops the full year, but when precipitation is deficient, in most areas, groundwater can be securely withdrawal in the summer to supply water for crop germination. Girmay et al. (2015), employing environmental isotopes and disintegrated ions, conceptualized groundwater flow in the vicinity of He et al. (2013) applied the SNESIM algorithm in connection with the MODFLOW and MODPATH codes to explore the influence of geological doubt on groundwater flow models. Further investigations were conducted by Trainor et al. 2007; Hernandez et al. 2001; Roth et al. 1998; Klise et al. 2008; Wang et al. 2016; Wen et al. 2002; Huysmans et al. 2014; P´erez-García et al. 2014. Islam et al. (2017) studied a local groundwater-flow pattern for permanent groundwater-resource administration in south Asia. The model MODFLOW-2005 was applied to model the communication between groundwater and surface water under the terms of steady-state and transient during 1981–2013 to gauge the influence of growth and climate variation on the local groundwater supplies. King et al. (2017) investigated an identification of aquifer recharge evaluation procedures in an alluvial aquifer system with an intermittent/ephemeral stream (Queensland, Australia). This research describes the significance of the conceptual hydrogeological simulation for calculating groundwater recharge in an aquifer that continues with a temporary or intermittent stream in southeast Queensland, Australia. The miss/get status of these streams is usually subordinate to transient and spatial variation, and understanding of these hydrological activities is important for the analysis of the rate of inflow to the aquifer. Niazi et al. (2017) studied the spatial pattern of groundwater recharge from stream baseflow and groundwater chloride. The full recharge across the complete watershed is computed applying the baseflow method, and next, the spatial changes of recharge are estimated utilizing groundwater chloride concentration. In this study, a regional 3D groundwater flow model was generated for evaluating the groundwater reserves in the Jangal-Geysour plain. The Jangal-Geysour plain has a cold, dry climate with long summers and short winters. The average annual precipitation of the basin is 165 mm, according to the Jangal-Jantabad meteorological station’s statistics, 60% of which occurs in winter. The requirement for water that is gradually increasing, for development of agriculture in Razavi Khorasan Province in general and Jangal-Geysour plain in particular, has caused the exploited volume of water from the aquifer (discharging) to overtake the volume of water entering the aquifer (recharging) (Velayati and Tavassloi 1991). The continuation of such a method of water exploitation has led to a continuous drop in groundwater levels and an imbalance in the input and output to the aquifer in the Jangal-Geysour plain. Its consequences include a gradual increase in groundwater salinity, and in the regions where there are two masses of saline water and freshwater (coastal and desert areas), it leads to the movement of the salty water toward the freshwater (Omole et al. 2017). Therefore, after preparation of the conceptual model and numerical model and obtaining enough information regarding aquifer behaviors of the Jangal-Geysour plain, scenarios for management and better control of the aquifer were presented.

Materials and methods

The study area

The Geysour and Jangal plain catchment is located in northeast Iran, Razavi Khorasan Province, southwest Torbat Hydarie, and its geographical coordinates are length 58° 47′ to 59° 34′ and width 34° 33′ to 35°, and the area is approximately 4979 km² (Fig. 1). This catchment is a rectangular shape, with a length of 75 km and width of 39 km that extends from northwest to southeast. The mountainous part of the plain, with an area of 900 km², includes the low height mountains and hilly lands that have occupied part of the northern catchment. The highest altitude is in the area northwest at 1306 m above sea level, and the lowest height is at the outlet of the watershed at 837 m above sea level.

Fig. 1

Geological-hydrogeological sketch map of the Jangal-Geysour plain in southern Razavi Khorasan Province, Iran. Note that, Plms, Qcf, Qft1, and Qs,d are marl, shale, sandstone, and conglomerate; clay flat; low level piedment fan and valley terrace deposits; unconsolidated wind blown sand deposite including sand dunes, respectively. B-B′ shows the cross-section

Conceptualization

Geometry and structure

The principal geometric-structural and hydrogeological components of the Jangal-Geysour groundwater system were constructed based on geologic construction and 21 boreholes (Fig. 2). The Jangal-Geysour aquifer system, formed from fine-grained alluvium to medium and alluvium thickness, is more than 150 m. The Jangal-Geysour plain aquifer has a length of 70 km and medium width of 20 km. This aquifer is surrounded, from north to Neogene impervious, from south to Shoor River, from west to clay hills, and from the east is linked to the Geyser plain aquifer. The aquifer system is recharged by river surface flows of Kalsalar and Shastdare. Any change in entrance flood to the plain could cause many negative effects on the quality and quantity of Jangal-Geysour plain groundwater. Hard region formations in the north of the plain formed from Neogene impervious rocks. These rocks do not have the power of water absorption and do not have a role in forming aquifer and water storage, and formed part of the aquifer system. Alluvium constitutive elements are medium grain and often is composed from sand and gravel with clay (Fig. 2). The percent of clay increases toward the south and west. The northern part of the plain has a drooping state, and alluvium thickness is over 200 m. Bedrock bulges in the plain west causing a disconnection between the aquifer of the Jangal-Geysour plain and the aquifer of the adjacent plain (Mahvelat plain).

Fig. 2

Schematic cross-sections and conceptual model of the aquifer system in alluvial sediments of the modeled area and a cross section of the aquifer system in alluvial sediments

Piezometric surface and hydraulic properties

Controller observations and codification are data and information that can be used to assess the operation of the conceptual model. These data involve water tables of piezometers, hydrometer data, and exploitation wells (Izady et al. 2014). The piezometric level of 15 observation wells was measured, from July 1999 to June 2014, monthly. Heeding that, due to constant withdrawal from the numerous wells in the region, it was just desirable to consider the dynamic water level—hydraulic head data; therefore, the normal balance states are note shown, particularly in the crop season (summer), when the water demand rises due to irrigation. The spatially distributed hydrodynamic features were assessed by applying various methods (Izady et al. 2014). It involves direct and indirect methods. In the direct method, hydrodynamic properties of different points of the region were obtained using test pumping and for increasing of these points, and the good dispersion was used from an indirect method that included empirical equations, pumping tests in exploitation wells, geoelectric reports, and drilling logs. These results were used as the initial guess. Approximated hydraulic conductivity is in the range from 0.5 to 20 m/day and are greater toward the east, north, and northeast portions of the study area. Also, due to the presence of fine-grained alluvial, the hydraulic conductivity decreases near the plains outlet, and there are the same changes for specific yield as hydraulic conductivity.

The east and northeast of the area include the highest specific yield values (0.2), and the south and west parts have the least values (0.03). The replenishment of the aquifer occurs through the absorption of precipitation. Also, the aquifer is recharged by infiltration toward external layers (lateral flows, in the southern and western sector), and more by the flow of water through the cavernous limestone and the contact alluvial deposits. The groundwater system is discharged by wells (artificial outflow). Water transfer in the aquifer happens due to natural circumstances (seepage, close of aquitards) and artificial reasons (multi-screened wells).

Role of surface water

Kalsalar River is the largest river of the Jangal-Geysour watershed, and it begins in northeast altitudes. Floods of this river have a significant role in aquifer recharge. The maximum moment peak discharge of river with return period 2, 50, and 100 years are 79, 362, and 422 m³/s, respectively (Velayati and Tavassloi 1991). The passage of water from the surface water is poor and negligible after the construction of a Shahid Yaghoobi dam on the Kalsalar River upstream (Roshtkhar plain). All of the agricultural water demand and drinking water is supplied by groundwater and Roshtkhar plain (Samen complex), respectively (Sabbaghi et al. 2017).

Discharge and recharge

In this area, exploitation is done only from groundwater resources by deep wells. Well density mainly is observed in the center of the plain. In the western part of the plain, the density of drilled wells is observed, which mostly are not exploitable due to high salinity. By using the groundwater balance equation, the rate of discharge, fluctuations of groundwater level, and hydrodynamic coefficient (specified yield) could calculate the amount of annual recharge. The recharge was estimated according to Eq. (1).

$$ \operatorname{Re}=\mathrm{Dis}-\mathrm{G}+\Delta \mathrm{S}+\mathrm{ET} $$

(1)

Re is the rate of recharge (m³). G is calculated by Darcy flow tools in GIS software (groundwater volume balance) (m³). ET, because of the high depth of water, is considered 0. Dis is equal to exploitation amount (wells) (m³) and

$$ \Delta \mathrm{S}=\Delta \mathrm{h}\times \mathrm{S}\mathrm{y}\times \mathrm{A} $$

(2)

A is the area of each theissen (m²) (Fig. 3). S is specified yield, and ∆h is equal to piezometer water level monthly drop (m).

Fig. 3

Theissen polygons for recharge

Determination of accuracy and adequacy of conceptual model

To determine the accuracy of a conceptual model, parameters investigation and constituent causes of the conceptual model are needed such as bedrock, water level, and quality issues such as estimation of EC, TDS, and CL. The data that we usually deal with include the types of environmental data, and therefore it is necessary to change into the continuous surface from a discrete mode that these surfaces are equal to the estimation of local changes in one point. In the result, it is required that its accuracy be evaluated. Techniques of geostatistic interpolation consider the correlation quantity of local sampling points, and it makes an estimation based on the not measured place sample situation. In the research using the geostatistics tools of GIS software, different parameters of the model were evaluated and validated. A grid with a suitable cell size to reflect the efficiency of the presented conceptual model overlays on the aquifer map. In the following, five different classes are defined as the spatial correctness for any ingredient of the conceptual model (Izady et al. 2014). The final result of every grid cell is collected as the privilege of the conceptual model elements. Finally, the precision of the conceptual model is determined due to the mediocre cells’ privileges. At this step and due to the precision level, the usefulness of modeling can be evaluated (Izady et al. 2014).

Numerical model

A numerical MODFLOW 2005 model is prepared by the conceptual model (Harbaugh et al. 2000). The model calculates the governing equations on groundwater flow in stable and unstable conditions by the numerical method of finite difference. The 3D move of groundwater in porous media demonstrated below is made using a partial differential equation.

$$ \frac{\partial }{\mathrm{\partial x}}\left({\mathrm{K}}_{\mathrm{xx}}\frac{\mathrm{\partial h}}{\mathrm{\partial x}}\right)+\frac{\partial }{\mathrm{\partial y}}\left({\mathrm{K}}_{\mathrm{yy}}\frac{\mathrm{\partial h}}{\mathrm{\partial y}}\right)+\frac{\partial }{\mathrm{\partial z}}\left({\mathrm{K}}_{\mathrm{zz}}\frac{\mathrm{\partial h}}{\mathrm{\partial z}}\right)-\mathrm{W}={\mathrm{S}}_{\mathrm{s}}\frac{\mathrm{\partial h}}{\mathrm{\partial t}} $$

(3)

Where:

Kxx, Kyy, Kzz :

hydraulic conductivity along the x, y, z axis which are assumed, (L/T)

h:

piezometric head, (L)

W:

volumetric flux per unit volume representing source / sink terms, T⁻¹

Ss:

specific storage coefficient defined as the volume of water released from storage per unit change in head per unit volume of porous material, L⁻¹, Ss = S/b, where b is aquifer thickness and unconfined aquifer S=Sy and Sy (specific yield) are the volume of water released from storage by an unconfined aquifer per unit surface area of aquifer per unit decline of the water table.

T:

Time, T

Solute transport model

To simulate advection and dispersion contaminants the MT3D numerical model was used. Groundwater flow was simulated by the groundwater flow and solute transport equations. The three-dimensional transport of dissolved solutes is represented by the partial differential equation (Zheng 1990).

$$ \frac{\partial C}{\partial t}=\left[\frac{\partial }{\partial x}\left({D}_x\frac{\partial C}{\partial x}\right)+\frac{\partial }{\partial y}\left({D}_y\frac{\partial C}{\partial y}\right)+\frac{\partial }{\partial z}\left({D}_z\frac{\partial C}{\partial z}\right)\right]-\left[\frac{\partial }{\partial x}\left({V}_xC\right)+\frac{\partial }{\partial y}\left({V}_yC\right)+\frac{\partial }{\partial z}\left({V}_zC\right)\right]\dots $$

Where:

Vx, Vy, Vz:

Seepage velocities in x, y, z directions, (LT⁻¹)

Dx, Dy, Dz:

Dispersion coefficients, (L²T⁻¹)

C :

Solute concentration, (ML⁻³)

t :

Time, T

Evaluation of aquifer quality has been done based on the result of sampling from the last period (2014) chemical analysis. In this research, Aquachem software is used to analyze the water samples. In the current research, flow simulation is done for unstable condition in the period of 15 years (1999–2014).

Model structure and calibration

Model groundwater flow was developed as a monolayer, unconfined (phereatic) and considering isotrope (Kx = Ky = Kz) and heterogenous aquifer in transient condition, for the 1999–2014 period. The steps of calibration and validation were provided from July 1999 to June 2010 and July 2010 to June 2014, respectively. There are not exact data about inflow flux from alluvial fans to the groundwater system; hence, the modeling border was formed by the Piezometric head (Fig. 1). The Piezometric data on the model boundary was utilized to assign a time-varying constant head boundary (Fig. 4). The southeast and east are an outflow path, and the southwest and west boundary are groundwater inflow routes. For borders of modeling range, a specific head-boundary was recognized (Fig. 4). The Jangal-Geysour aquifer is phereatic, and it was recognized as an only layer. The Piezometric heads on the model border were utilized to assign a time-varying constant head boundary (Fig. 4). The time parameters involving time step, time unit, and stress period were performed daily and monthly, respectively. A regular mesh and a block-centered finite difference grid were recognized with cells (250 × 250 m) that includes 89 rows and 107 columns. The model stretches from the ground surface to bedrock level and changes from 114 to 133 m. The values were extracted using a qualified digital elevation model (DEM) and bedrock maps and the inverse-distance interpolation method. (Tabios and Salas 1985). Because the extraction rates of wells in the region is high, to facilitate the model calculations, the cells were considered with dimensions smaller than expected. In the next step of modeling, the hydraulic conductivity and specific yield were computed and selected as initial values. By the observation wells and other measures of groundwater levels of withdrawal wells in July 1999, the initial head was taken. Also, using the groundwater balance equations, the recharge rates were estimated and were recognized as an input for the model.

Fig. 4

Groundwater level contour lines for the Jangal-Geysour aquifer (October 2012) along with general groundwater flow direction (m asl)

The model was calibrated in two rounds by a different collection of parameters that were directed at harmonizing simulated data with the field data; in the first step, to get the answer of the model to parameter variations (specific yield, hydraulic conductivity, and recharge rates) the model was initially calibrated by trial and error. Afterward, utilizing 15 observation wells as base points, the PEST (Parameter ESTimation) algorithm (Doherty 1998) was applied to obtain the best calibration. The accuracy of the model was investigated by various statistical parameters to determine its capability in predictions and scenarios. The parameters include root mean square error (RMSE) and normalized RMSE (NRMSE). This parameter was used because the changes of each observation wells in the groundwater level are different in the calibration and validation periods, and therefore NRMSE would be more effective, mean error (ME), and finally mean absolute error (MAE) (Table 1).

Table 1 Statistical analysis of the field and modeled data

The NRMSE for the whole plain was calculated as follows:

$$ \mathrm{Weighted}\ \mathrm{RMSE}=\frac{\sum_{i=1}^n{RMSE}_i\ast {a}_i}{A} $$

(9)

$$ \mathrm{Weighted}\ \mathrm{drawdown}=\frac{\sum_{i=1}^n\Delta {h}_i\ast {a}_i}{A} $$

(10)

$$ \mathrm{Normalized}\ \mathrm{RMSE}=\frac{Weighted\ RMSE}{Weighted\ drawdown} $$

(11)

Where A is total area, n is the number of observation wells, i points to the observation well, a is Thiessen area, Δh is the difference between maximum and minimum groundwater level swing (Izady et al. 2013).

The MT3D was calibrated by modifying the values of advection and dispersion values. Advection describes bulk transport fundamentally based on the volume flow in which the quantity is solved or the moving of solute results from groundwater flow. The groundwater flow advection will define the overall process of solute transportation. Using the numerical model of flow (MODFLOW) and the efficient porosity, the transfer of solutes based on advection was modeled, applying the field velocity for a provided time step. The value PERCEL (Courant number or number of cells in any particle) was chosen as 0.75, and MXPART was recognized as a maximum of 1,300,000 particles. It was permitted to transfer in any course inside one transportation step and frequently ranged from 0.5 to 1 (Chiang and Kinzelbach 1998). The residual parameters were used with default values for the MT3D model. To evaluate the dispersive transport of the solute, the longitudinal dispersivity (α_L) and transverse dispersivity (α_T) are needed. The distribution of the solute in a medium is called dispersivity, which is a specific characteristic of the porous medium. When the outspreading of solute is in the course of volume flow, longitudinal dispersivity is used, and when the vertical and horizontal distribution of solute is to the direction of mass flow, transverse dispersivity is used. The longitudinal dispersivity (α_L) is calculated by molecular diffusion and limited dispersion coefficient due to the heterogeneity of the medium (Domenico and Schwartz 1990; CSIRO 2003) as presented in

$$ {\alpha}_L=\left(\frac{D_d}{v}+\frac{D_L}{v}\right)+{A}_L $$

(12)

where α_L is the longitudinal dispersivity (cm), Dd is the coefficient of molecular dissemination (16 × 10^–6 cm²/s), V is the velocity of flow (6.8 × 10^–6 cm/s), DL is the dispersion coefficient (30 × 10^–6 cm²/s), and AL is asymptotic longitudinal dispersivity (500 cm, since no field data is accessible for solute transportation). Asymptotic longitudinal dispersivity is principally associated with the variety of the log-transformed conductivity and correlation length in the mean path of flow and is based on the heterogeneity of the medium (Gelhar et al. 1992). The Peclet number, P_e=∆x/α_L, was computed to determine the relevant importance of advection and dispersion, where ∆x is equal to 250 m (cell size). The prevailing factor in solute transportation between advection and dispersion is determined by the unitless Peclet number. (Huysmans and Dassargues 2005).

Future scenarios

A total of four scenarios were recognized. To anticipate the future factors of modeling, parameters must be given to the model where one of the most important of these factors is the boundary conditions that their changes are uncertain in the future. However, investigations show that inflow changes to the aquifer were almost constant. (Fig. 5). Therefore, to predict, the type of modeling boundaries changed from a specific head to a particular flow.

Fig. 5

The changes of groundwater inflow and outflow

In the first scenario, the target has zero balance and prevention of drop in groundwater level as for current recharge and the decrease of withdrawal in a 5-year period.

In the second scenario, the method proposes to control overdraft. If withdrawal volume of wells is divided by their number, and this volume is divided by defined allowed work hours of the pump by Razavi Khorasan Province regional water for this studied area, then we will have an average moment withdrawal for every well. Therefore, the withdrawal of every well in the area is specified. Agriculture wells that are not actually able to be exploited for more than that, they do not need to control, and wells that can have more exploitation must be controlled.

In the third scenario, because the studied area wells are used to supply the demand for agriculture, a suitable cropping pattern has been presented by considering allowed work hours to control and management of withdrawals.

In the fourth scenario, because an enormous decrease of water quality is evident in the studied area, and since, with the decrease of water level, water quality reduces too, the impact of irrigation of salty water on production decrease as for water level drop by continuing the current trend was evaluated. Finally, production decrease percent was estimated at 2021. To evaluate the impact of the rate of irrigation of salty water on the decrease of production, the first step was to determine the electrical conductivity of irrigation water (ECiw) and plants that are cultivated in the area, and by ECiw, soil saturated electrical conductivity (ECe) was estimated.

$$ {\mathrm{EC}}_{\mathrm{e}}=1.5\ {\mathrm{EC}}_{\mathrm{iw}} $$

(13)

Then, as for Table 9 two parameters, a and b were calculated for every plant. a is salinity tolerance threshold (mmhos/cm) and b is production decrease percent, per increase each unit salinity (%)(mmhos/cm). Finally, the percentage decrease of products than production potential in the condition of non-saline is obtained from Eq. (12) (Alizadeh 2009).

$$ \mathrm{Y}=100-\mathrm{b}\ \left({\mathrm{EC}}_{\mathrm{e}}\hbox{--} \mathrm{a}\right) $$

(14)

Results and discussion

Figure 6 presents the grade of components of the conceptual model. It shows the overall precision level of every grid cell due to the recommended method in Table 1 (Izady et al. 2014). In regard to Fig. 6, the overall precision of the conceptual model for the Jangal-Geysour plain has an “excellent” to “fair” status. The white cells (class A) are well distributed along the aquifer. Most cells have class B conditions where it can be inferred that the extended conceptual model is ready for numerical simulation.

Fig. 6

The spatial accuracy level of each grid (1 × 1 km) for the recognized conceptual model

Map of groundwater level and flow direction

As for the map of groundwater level in June 2014 (Fig. 7), three different regions could be recognized due to potential curves, including the west region (region of kalsalar river path), the central region, and the east region. In the west region, the groundwater level starts at 845 m and gradually decreased in the flow direction and the outlet area is at 835 m. The hydraulic slope for inflow and outflow mass is 0.6 × 10^–3 in this region. In the center region, withdrawals of more than enough create closed curves in the center of the studied area. Groundwater flow is from north and south toward the center while it is gradually driven to east side. The highest groundwater level in the north and south region is 837 m and the lowest level is 814 m, where hydraulic slope for inflow and outflow mass is equal to 2.35 × 10^–3. In the east region, the groundwater level begins at 835 m and gradually decreases toward the southeast, and northeast the groundwater level is at 808 m in the output region. The hydraulic slope is 2.7 × 10^–3 for inflow and outflow mass in this region.

Fig. 7

Groundwater level flow direction in February 2014

The areas of Jangal and Geysour have 100 deep wells with annual exploitation of approximately 85 MCM (million cubic meters), the maximum withdrawal 85 l/s, and the least 4 l/s (Fig. 8b). Also, the rate of work hours of the pump of wells is approximately 6000 h annually (Fig. 8a). Figure 9 shows the initial guess of recharge.

Fig. 8

Well work hours work (a) and annual average withdrawal (b)

Fig. 9

Recharge initial guess (conceptual model result)

Physico-chemical concentrations

Sources of the groundwater mineralization

The result of the analysis of the delegate water samples over the region showed that pH varied from 7.5 to 8.8 and was in the allowable pH range from 6.5 to 8.5 of WHO (2011) for drinking water. Hence, it would be safe to recommend that the great movement of ions in the groundwater, as implied by the average content of whole dissolved solids (TDS) of 4961 mg/l (Fig. 10a), is almost affected by the water confined through deposition of the Kalsalar River sediments. Rooki et al. (2017) qualitatively studied the groundwater of Gonabad plain and stated that pH varied from 8.18 to 8.2.

Fig. 10

The map of electrical conductivity (a) and dry residue TDS (b) for the Jangal-Geysour aquifer in February 2014

Electrical conductivity (EC) shows the rate of dispersion of dissolved solutes, which ranged from 3200 μS/cm in the area northeast to 15,000 μS/cm in the area northwest; see the EC distribution in Fig. 10b. The dispersion pattern implies northward and eastward, putting the solutes in the western and south-southwest parts of the area, where there is the highest concentration. Rooki et al. (2017) qualitatively studied the groundwater of the Gonabad plain and stated that EC varied from 3594.9 to 3653.2 μS/cm. Joodavi and Zare (2009) qualitatively studied the groundwater of the Feyz-Abad plain and reported EC changes between 1320 to 15,120 μS/cm. The Piper diagram (the result of sampling in February 2014) shows (Fig. 11b) that in the groundwater system, CL⁻ > SO₄²⁻ > HCO₃⁻ are often prevailing at various sites: The recent pattern shows that chloride-rich connate water or saline source is restricted in some places. The main cations are described principally by Na⁺ > Mg²⁺ > Ca²⁺ trend. According to the figure, maximum K⁺ is between 0.7 to 13.5 meq/l, Na⁺ change between 36 to 95.8 meq/l, the maximum Ca²⁺ and Mg²⁺ are 24.5 and 41.3 meq/l. The minimum Ca²⁺ and Mg²⁺ are 1.95 and 6.4 meq/l, respectively. Anions for different places, including CL⁻, HCO₃⁻, and SO₄²⁻ change between 27.14 to 117.2, 1.8 to 6.34, and 17.2 to 42.3, respectively (see Fig. 11a).

Fig. 11

Statistical parameters graph (a) of chemical components and piper graph for sampling points (b)

The study area is divided into 11 Theissen based on sampling points (Fig. 12). In the groundwater samples, sodium and magnesium are the dominant cations, and chloride and sulfate are the dominant anions; and most groundwater is characterized as Na–Mg–Cl– and followed by Na–Mg–Cl–SO₄ water types caused by possible plagioclase weathering in the basin. The weathering rate of plagioclase was evaluated in the groundwater system of a sandy, silicate aquifer created subsequent to the Wisconsin Glacial Stage (Kim 2002; Amiri et al. 2020).

Fig. 12

Map of spatial trends in the physical or chemical characteristics of each sample by plotting pie

Correlation analysis (Amiri & Berndtsson 2020) (Fig. 13) revealed that the TDS is principally provided by NaCl separating and moving from the deep salt bed, as well as HCO₃ from infiltrating precipitation water that identifies the shallow groundwater system. Using the software Sigmaplot, the significance in the correlative investigation (Sohrabi et al. 2017) was automatically calculated; in Fig. 13, an asterisk (*) corresponds to significance at 0.05, or 5%, for all coefficients at 0.79 ≥ r ≤ 1.0. Thus, for P′ ≤ 0.05.

Fig. 13

Correlative analysis of the analyzed geochemical data

Results showed a strong correlation at r = 0.97 (i.e., 97%) and r ≥ 0.83 or 83% between CL⁻ and TDS, and also the sodium ion (HCO₃ and K), respectively (see Fig. 13). Therefore, Cl and Na ions make up the main ingredients of the TDS. The typical concentration of calcium in groundwater changes from 10 to 100 mg/l (Nag 2009). Carbonate rocks, i.e., limestones and dolomites (dissolved by carbonic acid) and also a chemical breakdown of calcic-plagioclase feldspars, and pyroxenes are the major causes of calcium in groundwater (Ganyaglo et al. 2010). The TDS, is at variance with calcium (Ca) due to a weak correlation at r = 0.31; therefore, showing various sources from which the K₂SO₄⁻ rich water and KHCO₃ rich water transfer, from deep and near-surface zones, respectively. There is a strong correlation with each other at 0.9 (90%) between K and HCO₃ that indicates the effect of chemical dissolution is applied by precipitation water on the calcareous shale.

Correlations of Ca and HCO₃ with the Mg ion are weak at −0.07 ≤ R ≤ 0.5 (Fig. 13), indicating a lack of dolomite (CaMgCO₃), and demonstrating the high molar ratio of Ca²⁺/mg²⁺, which is usually higher than 1 for every representative sample. Mayo and Loucks (1995) expressed the lack of dolomite can be considered if the molar ratio of calcium to magnesium ions is higher than 1. The scarcity of gypsum (CaSO₄) is also assumed due to irrelevant correlations of Ca²⁺ and SO₄²⁻ at r = −0.14 (Fig. 13). Hence, the dissolution of the calcareous sediments through the act of infiltrating water is presumed to have had a significant impact on the chemistry of shallow groundwater.

Calibration and validation of models

Parameter identification

To set the hydrogeological parameters kh and Sy for a suitable match (between simulated data and field data), a method of trial-and-error was applied. Table 2 shows the rate of fit between simulated data and field data by statistical parameters and the result is in an adequate range. The minus of ME (mean error) described that the simulated head was larger than the measuring head. The root mean square error (RMSE) represents the standard rate of calibration for the hydraulic head, which is fewer than 10% of the simulated head in the groundwater system (Tsou et al. 2006), and the corresponding standard is reflected in TDS. The head’s changes in the aquifer are approximately 799 to 855 m, where the result due to the RMSE parameter is in an agreeable range of 5.6 m or less. Furthermore, the TDS at chosen sites are changed from 2661 to 8865 mg/l, where the result is in an adequate RMSE of 620 mg/l or less. Anderson and Woesner (1992) and Moriasi et al. (2007) stated that the RMSE is frequently analyzed as the most suitable calibration symbol. Asghar et al. (2002) stated that the minus mark of modelefficiency (MEF) shows the remarkable changes between the field data and modeled values. If the value of MEF = 0, the simulation indicates a weak performance. Also if MEF = 1, the simulation values and the field value is matched accurately. As a result, the values presented a great fit between field data and simulated values. The mediocre coefficient of determination (R²) for the piezometric head was estimated as 0.99 and 0.95. This rate was earned in the condition of calibration and validation, respectively (Figs. 14a and b), for MODFLOW and MT3D models, respectively, which showed a good performance between the simulated and field data. Hagos (2010) reported the result of calibration by the PMWIN model for Raya Valley, Ethiopia. The rates of ME, MAE, and RMSE were 1.4 m, 7.8 m, and 10.7 m, respectively, with the R² of 0.97 and also, in similar study regions, Izady et al. (2007) stated that the RMSE and R² for the Neyshabour plain is 0.378 and 0.937, respectively. Mokhtari et al. (2012) stated that the RMSE and R² for the Shistar plain are 1.43 and 0.996, respectively.

Table 2 The results of statistical analysis of the field and modeled data

Fig. 14

Scattergram of measured versus modeled heads in MODLOW (a) and modeled TDS in MT3D (b)

Transport of salts

The longitudinal dispersivity was estimated in the range 1–10 (Table 3). Ahmad (1974) estimated the rate of longitudinal dispersivity between 1.89 and 5 m for the groundwater system of the Indus Basin of Pakistan. Gelhar et al. (1992) analyzed several types of research and announced that longitudinal dispersivity was between 3 and 15.24 m, and also, due to soil type and hydraulic conductivity, the rate of horizontal and vertical transverse dispersivity was determined to be 0.45 and 0.015 m, respectively. Engesgaard et al. (1996) stated the changes of the values of longitudinal dispersivity from 1 to 10 m. The transverse (horizontal and vertical) dispersivity in the research of Shieh et al. (2010) was calculated between 0.01 and 10 m. Narayan et al. (2003) calculated two parameters of longitudinal dispersivity and transverse dispersivity at 2.5 and 0.5 m, respectively. The dispersivity handled in the MT3D model is presented in Table 4. The value of TRPT (ratio of the horizontal transverse dispersivity to the longitudinal dispersivity) and TRPV (ratio of the vertical transverse dispersivity to the longitudinal dispersivity) obtained were 0.11 and 0.1, respectively (Sun 1994). The molecular diffusion coefficient (DMCOEF) is usually very poor and negligible (explains the diffusive mass of a solute in the water of higher concentration toward less concentration) compared to the mechanical dispersion, and it was 0.1 (Kori et al. 2013; Chiang and Kinzelbach 1998). The value of the Peclet number was determined to be 655. For such a high value of Peclet number, solute transportation is overcome by the advection, that is, the conveyor of solute based on the mass flow of water (Huysmans and Dassargues 2005; Fiori 1996). Kardan and Banihabib (2016) calculated the parameters of ratio of horizontal to vertical distribution, vertical diffusion length, effective molecular diffusion coefficient, and longitudinal diffusion to be 0.5, 0.2, 1 m, and 20, respectively.

Table 3 Dispersivity values used in MT3D

Table 4 Value of parameters for dispersive transport

Time series plots of the simulated heads and observed heads in Wells a1, a6, a7, a15, a11, and a5 are shown in Fig. 15 where, the field data (head) show higher variability than the simulated heads. The six observation wells are placed in various plain areas and show the variations in the aquifer.

Fig. 15

Comparison between computed (blue line) and measured (red line) groundwater level in different observation wells. The starting data of the simulation is 7/1/1999–6/1/2014. Comparison is made with monthly value

The change of hydraulic conductivity (K) (Fig. 16a) is between 0.5 to 20 m/day with an average of 6 m/day, and the variation range of specific yield (Sy) is between 0.01 to 0.2 with an average of 0.1 (Fig. 16b). The model considers nine zones for hydraulic conductivity that differ in terms of hydraulic properties. The rates of hydraulic conductivity are specified by applying the zonation capacity of the Layer-Property Flow (LPF) package of MODFLOW. The 1, 2, 4, 7, and 9 zones of the model consist of clay, low-level piedment fan, and vally terrace deposits, which are mainly of local importance (Fig. 16a), and hydraulic conductivity was estimated in the range of 0.5–3 m/d. The 3, 5, and 6 zones are low-level piedment fan and vally terrace deposits.

Fig. 16

(a) and (b) show the changes of hydraulic conductivity (K) and specific yield (S), respectively

The modeling area is divided into east and west parts. The western part includes the geology unit Qcf (clay flat). The rate of recharge (Fig. 17) as for features of the geology formation is not suitable. The eastern part includes geology unit Qft2 (rich groundwater aquifers), and this part also includes the density of exploitation wells and agriculture grounds. The rate of recharge (Fig. 17) is appropriate; according to the result of the model, the most and the least recharge rate is related to Hozekaram (a10) and Bandozbak (a2) Theissen, respectively.

Fig. 17

The rate of annual recharge average (MCM) (1999–2014)

Changes in simulated head and TDS in the period of 1999–2014 are presented in Fig. 18. The outcomes showed that the groundwater quality of the north of the plain would not be changed by a drop of groundwater level, due to the constant concentration of TDS passing with time (see Fig. 18, the changes in groundwater level (a, b, c) and TDS (d, e, f) for September 1999, February 2006, and September 2014). Groundwater quality rate in the center, lower part, and southern will be seriously influenced by salinity because of the imbalance between inflow and outflow of the groundwater system and geology structure. This overflowing aquifer withdrawal is causing a decline in water quality due to both horizontal and vertical saline interference (Asghar 2014, Mekonnen et al. 2016). Prinos (2016) reported the various pathways for saltwater interference into the freshwater region in Florida near the good field. Khan et al. (2008) reported a decline of aquifers due to over-extraction of groundwater influenced the mass saltwater interference into the freshwater zone. The farmers in the area extracted water from the depths with no quality control. The constant withdrawal of such inadequate quality groundwater has two consequences: in the first step, secondary salinization that ultimately reduces crop productivity (Shakoor et al. 2015; Basharat 2012). The diffusion of the TDS, principally from the south and west toward the center of the study area, agrees with the groundwater flow route (compare Figs. 7 and 18), which implies a high concentration gradient. Therefore, due to the distribution of TDS concentration, it is possible to state the constant charging of salinity from across the study area.

Fig. 18

The changes of groundwater level (a, b, c) and TDS (d, e, f) for September 1999, February 2006, and September 2014

Water budget

By drawing the Theissen map and water table fluctuations of 15 observed wells, the unit hydrograph of the plain shows a water table continuous drop in the plain aquifer (Fig. 19b). The linear graph is descending, and points of maximum are not specified. Change of curve slope shows the time of withdrawals decrease in cool and wet periods. Also, this curve shows the rate of groundwater level annual average drop, 0.88 m, in the period of 14 years. That explains the aquifer negative balance. In the region of the case study, major rainfall happens from October to April. Therefore, this period could be considered as a wet period and May–September as a dry period. The reaction of the aquifer in the wet period is water table decline where all of them show water table fluctuation in during the year.

Fig. 19

(a) shows the calculation grid for the 3D numerical model and water balance of the Jangal- Geysour groundwater system plain in the period of July 2011 - June 2012 gained for unsteady conditions: inflows (positive value) and outflows (negative value) in the modeling range represented in MCM, and (b) is showing aquifer hydrograph of modeled area

By representing the groundwater system water balance, it is safe to estimate the water budget rate (Fig. 19a and Table 5). From July 2011 through June 2012, the groundwater system shows a negative equilibrium of over 45.6 × 10^–6 m³. The water budget simulated by the model for July inflow factors involves groundwater flow mass and surface recharge (infiltration of rainfall, infiltration of drinking, industry, and irrigation), and outflow factors include evaporation from the aquifer, groundwater outflow mass, and withdrawals). The rate of recharge and characteristics of discharge are estimated at over 43.6 and 89 MCM, respectively. Figure 19a shows the rate of water budget components. Groundwater inflow mass from boundaries obtained by the result of the model and the rest of infiltration volume belong to infiltration of rainfall, irrigation, drinking, and industry.

Table 5 The factors of modeled area inflow and outflow in the period of July 2011–June 2012

The unit hydrograph of Bandozbak (Theissen a2, see Fig. 20a) observed well shows the least water table drop over 5.1 m and the unit hydrograph of Hozekaram (Theissen a10, see Fig. 20b) observed well shows the greatest water table drop over 29 m in a 14 year period. Despite that, the rate of recharge than discharge is more in the Bandozbak area but the decrease of water quality is significant in this area because in addition to aquifer formation caused by the booster of water quality decrease, by drops of the water table, there is a change of hydraulic slope and inflow mass with high salts from the west and southwest. In the Hozekaram area (Fig. 20b), the rate of discharge than recharge is higher, but a decrease of quality in this area is very low because there is a low change of water quality in depth as for the type of area formation.

Fig. 20

Recharge, discharge grap, and observation well hydrograph of Hozekaram (b), Theissen a10 and Bandozbak (a), Theissen a2 with the most and the least groundwater level drop, respectively

Management scenarios

Four management scenarios for the prepared model were presented. The target of scenario A was balancing the relative to the aquifer drop condition in a 5-year period, and the rate of withdrawals was decreased annually over 10% and the model was run afresh using a new withdrawal rate (Fig. 21 and Table 6). Based on the scenario applied, the level of groundwater is almost fixed, and the water budget will be in relative balance. In scenario B, by using the suggested addition, withdrawals control method, the average moment withdrawal calculated and subtracted 33.8 l/s for each well, and wells to exploit more than average should be controlled and the amount of overdraft subtracted from them. Allowing working hours (3977.389 h) and suggested withdrawal save the rate of 27.54 MCM yearly and add to aquifer resources. With the persuasion of farmers, benefits to use of the pressurized irrigation method and increase of irrigation efficiency could compensate for decreased withdrawal (Fig. 21 and Table 6). In scenario C, according to Table 7, the most requirements of water occur during 21 June–21 July and 22 July–21 August months that respectively is 17.09, 17.47, and for four months 21 November–20 March water requirement is zero, due to rainfall and the yearly wet season (according to the water national document). The gross irrigation requirement calculated by dividing irrigation requirement on irrigation efficiency (46.7% water national document) in column 5 of Table 6 applies a 20% deficit irrigation in maximum requirement months. Hydromodul of irrigation by assuming full hour working on wells (744 h) calculated 1.37 l/s/ha on 20 April–21 August. Finally, well hour working in the other wells was computed by dividing gross irrigation requirement (80%) on hydromodul of four months by maximum working hour. If the withdrawal is according to presented working hours monthly and the model is run again with the new condition of exploitation, the the result of groundwater level hydrograph will increase (Fig. 21 and Table 6). Scenario D (Table 8) investigates groundwater level changes and quality parameters with the continuation of the current conditions process, in the year 2021. The result shows the descending changes in groundwater level and water quality decrease. The groundwater level on 21 September is equal to 817.24 (Fig. 21 and Table 6). The tolerance threshold of kitchen garden productions and water salinity will have the largest decrease percent.

Fig. 21

The changes of groundwater level for scenario A, B, C, and D

Table 6 Water budget of scenario A (September 2019–February 2020), B and C (September 2022–February 2023)

Table 7 Water requirement of cropping pattern and working hours of wells for scenario C

Table 8 The result of salty water influence quantitative evaluation on product decrease (scenario D)

Conclusions

In the research, a regional groundwater qualitative and quantitative management through hydrogeological modeling in Jangal-Geysour, Razavi Khorasan Province, Iran, was developed. This research presents a considerable contribution to the recognition of the groundwater resources interest of the area, especially when there is no earlier qualitative and quantitative analysis. The model combines modern science and hydrogeological data for the area. It considers all relevant contributions to groundwater flow, including lateral inflow and outflow, as well as recharge, discharge mechanisms, and qualitative changes of groundwater during the modeling period. Also, the model was generated to discover the most effective ingredients of the water budget and to recognize surface-water/groundwater system relations.

The investigations show that hard formation of the area except Cretaceous carbonate rocks is inadequate in terms of groundwater resources and does not have any role in saving and supplying area water and formations of clay and Neogene marl have a direct effect in area lands water quality degradation. The area alluvial aquifer has significant resources and annually is exploited by wells over 65 MCM. That is more than the annual recharge. Because further exploitation of the aquifer is expected, the area is faced with the continuous drop in groundwater level and decrease of groundwater resources. The groundwater resources in an area can present severe restrictions in agriculture and have caused the destruction of agricultural soils. The water type in this area is Na–Mg–Cl. To save an aquifer, a quantitative and qualitative method of exploitation management should replace the current exploitation policy. The presented model depicts the foundation for future simulations. This model remains uncertain due to the incomplete information of the system to completely control results; however, it can now give valuable data on the overall development of the system under various stress conditions.

The forthcoming investigation will enhance the geological and hydrogeological rehabilitation of the groundwater system by implementing novel lithostratigraphic data from wells and boreholes, and more piezometric, pumping and tracer experiments will develop an awareness of the hydraulic and dispersive characteristics of the aquifer. Adjustments of exploitation rate in groundwater of area for groundwater level drop decrease or its consolidation with recommendations of suitable methods for the irrigation efficiency increase, change of cropping pattern, and correction of withdrawals by considering aquifer annual recharge should be made.