Numerical Simulation of Groundwater Flow and ... - Springer Link

JOURNAL GEOLOGICAL SOCIETY OF INDIA Vol.83, January 2014, pp.83-92

Numerical Simulation of Groundwater Flow and Contamination Transport in Nahavand Plain Aquifer, West of Iran HOSSEIN BANEJAD1, HAMID MOHEBZADEH1, MOHAMMAD HOSSEIN GHOBADI2 and MAJID HEYDARI1 1

Department of Water Engineering, Faculty of Agriculture, Bu-Ali Sina University, Hamedan, Iran 2 Department of Geology, Faculty of Basic Science, Bu-Ali Sina University, Hamedan, Iran Email: [email protected]

Abstract: Numerical simulation of groundwater flow used for the estimation of hydraulic and hydrologic parameters which is an important tool for the management of aquifers. This study presents the results of a mathematical model developed for the simulation of groundwater flow in Nahavand plain aquifer in the southwest Hamadan province. For this purpose Groundwater Modeling Software (GMS) was used which supports the MODFLOW-2000 code. After gathering required data such as the hydrological, hydrogeological and topography maps, a 3D hydrogeological model of plain was constructed with borehole and surface elevation data. Then MODFLOW was used for simulation of flow. After initial simulation of the flow, the model was calibrated in steady state with trial-and-error and parameter estimation methods the observed head of groundwater table monitoring data of 1997. Results of calibration show that error between observed head and computed head is in allowable range. Also results of computed head with model show that groundwater flow is in the direction of the dominate slope (southeast to northwest). Finally MODPATH code which simulates advective transport of particles was used for estimation of flow path and source of contaminants. Keywords: Groundwater Flow, MODFLOW, GMS, Advective contaminant transport, MODPATH, Iran. INTRODUCTION

Numerical groundwater modeling is an important predictive tool for managing water resources of aquifers. Groundwater models can be used to test or refine different conceptual models, estimate hydraulic parameters, and most importantly for water resource management, predict how the aquifer might respond to changes in pumping and climate change (Regli et al 2003). Therefore, modeling of groundwater flow and contaminant transport has attained importance in last decades. Flow simulation models help determine the direction of groundwater flow, distribution of hydraulic heads and flow magnitudes. In essence, they help determine the most optimal locations of drill boreholes and wells and provide guidelines for maximum extraction of groundwater from wells and boreholes in a watershed (Yidana, and Ophori, 2008). Numerical groundwater model provides the accuracy in representing complex environments and can be applied to nearly all types of hydrogeological settings, and is an important tool for predicting the migration of the contamination in aquifer (Zuo et al. 2009). In this study the MODFLOW model was employed within the framework of the Groundwater Modeling System (GMS)

to simulate groundwater flow in the unconfined aquifer of Nahavand plain. The advective transport MODPATH was used to simulate pathways of particles tracking originating from sources area. The plain of Nahavand is located in the Hamedan province west of Iran. Its groundwater resources were used for agricultural and drinking purposes. By understanding the water quality of this area and contaminant transport simulation the velocity and direction of contaminants can be predicted and the dispersion of contaminants can be avoided with effective management. In addition, simulating of groundwater flow can lead us to understand about hydraulic properties such as: hydraulic conductivity, groundwater level and groundwater recharge from precipitation. STUDY AREA Location

Nahavand plain covers 285 km2. This area is located between longitudes 48°-48° 15" E and latitudes 32°-37° N. Local slope is from southeast to northwest. The highest and

0016-7622/2014-83-1-83/$ 1.00 © GEOL. SOC. INDIA

84

HOSSEIN BANEJAD AND OTHERS

Fig.1. Geographic location of the study area.

lowest locations are 1650 m and 1450 m asl (above sea level) respectively (Fig. 1). The study area is bounded by the Ardoushan, Shadmaneh, Kamarzard and Cheleh mountains on the north, and the Garin mountains on the south and west. Hydrology

consist of karstic limestone and conglomerate. The metamorphic rocks consist of crystallized limestones and shale. Generally the Northeast formations of Nahavand belong to Triassic. These formations in the north are crystallized limestones, limestone layers, metamorphic shale and Cretaceous andesite. The southwest

Mediterranean climate is the main source of precipitation in the area (Hamedan Regional Water Organization). In this study the average monthly precipitation data from 19972008 collected at Vasj weather station was used for groundwater recharge estimation. Figure 2 presents rainfall data collected at Vasj weather station located in the study area. Fig.2 (a) and (b) show the monthly precipitation in 1997 and the average monthly precipitation from 1997 to 2008, respectively. Total precipitation in 1997 was 407.7 mm/year. The average monthly precipitation from 1997 to 2008 was 412.1 mm/year. The yearly temperature is +13.2 °C with moderately warm summers in August with an average temperature of 23.67 °C and moderately cold winters, coldest in February with an average temperature of -2.1 °C. Gamasiab river, the largest river in Hamedan, flows in this area in a SE-NW direction. Geology

The mountains in the area have different types of sedimentary and metamorphic rocks. The sedimentary rocks

Fig.2. Precipiation data in Vasj weather station, (a) monthly precipitation for 1997, (b) average monthly precipitation from 1997 to 2008 JOUR.GEOL.SOC.INDIA, VOL.83, JAN. 2014

SIMULATION OF GROUNDWATER FLOW AND CONTAMINATION TRANSPORT IN NAHAVAND AQUIFER, IRAN

85

Fig.3. The geological cross-section of the study area.

area consists of light gray to dark grey limestones and crystallized limestones. Tectonism that affected these formations have lead to the development of joints and fissures in all direction which are filled with clay and calcite. The geological cross-section of the study area shows that plain’s bedrock consists of limestone, dolomite limestone and pyroclastic rocks with recrystallized limestone (Fig. 3). This cross-section and the location of formations beneath the plain, shows that the plain’s thickness is about100m and it is supposed that this thickness is uniform throughout the plain. Hydrogeology

In this study the data collected from 19 logs of drilled water wells in the region was used. Unfortunately the groundwater monitoring system of the study area was poor and did not contain any data about pumping rates and hydraulic conductivity. The depth of wells is varying between 15 m to 70 m and data collected from logs of wells show that Nahavand plain consists of clay and silt, sand and gravel, coarse sand, and lime and schist. Clay, silt and sand, and gravel are predominant. METHODOLOGY

In this study, the MODFLOW model was employed to study the groundwater process of the hydrogeological system of Nahavand plain unconfined aquifer. MODFLOW is a finite difference groundwater flow model that simulates three-dimensional steady and transient state flows in heterogeneous layered aquifer systems, and predicts flow paths using particle tracking post-processing program (MODPATH, Pollock, 1994). Data used in the model include hydrological data such as: precipitation and hydrogeological properties including geological formations, a topographic map, location of well logs, land use map and soil hydrological group map. Then, the study area model domain was identified. The conceptual model included location of Gamasib river, recharge from precipitation and the measured water level at a number of JOUR.GEOL.SOC.INDIA, VOL.83, JAN. 2014

monitoring wells. Distribution of hydraulic properties such as: hydraulic conductivity was entered using 3D hydrogeological model. Hydraulic properties and infiltration recharge, assigned to model, were adjusted to calibrate the model. The model was calibrated for steady state by trialand-error and PEST (Parameter Estimation) methods. Forward and reverse particle-tracking analysis by the particle-tracking package MODPATH was conducted to investigate the potential contamination source and their down-gradient effects. The path line flow was simulated by tracking a particle of water through the model domain according to the velocity field computed from cell-by-cell flow term produced by MODFLOW. Estimation of Recharge

Groundwater recharge estimation is a subject of active research. It has received considerable attention from hydrogeologists both because of its importance for longterm sustenance of groundwater systems and also because of the challenges associated with its estimation (Manfreda et al. 2005). Recharge estimation in arid and semi-arid environments is especially important because groundwater sources are usually the most reliable water sources available for human use (Yaouti et al. 2008). Based on results obtained from observation wells, the aquifer of the plain is an unconfined aquifer, and is therefore capable of receiving direct recharge flux from precipitation. In this study the water balance technique, proposed by Yaouti et al. (2008), was used to estimate the recharge: R = P–Q–RET–ÄW

(1)

Where R is recharge (mm/year), P is precipitation (mm/ year), Q is net runoff (mm/year), RET is real evapotranspiration (mm/year), and ÄW is change in soil moisture storage that is considered negligible in semi-arid regions. In this study the evapotranspiration was considered negligible, because there is no information available about evapotranspiration. Runoff was estimated based on the CN I method, proposed by Patil et al. (2008), which is

86


modifications of the original NRCS-CN (natural resources conservation services curve number). NRCS-CN method is one of the most widely used methods for quick and accurate estimation of surface runoff from ungauged watersheds. The NRCS-CN method is represented as:

Development of the Numerical Model

The equation that describes the three dimensional movement of groundwater of constant density through a porous earth material under equilibrium conditions is the partial differential equation (Don et al. 2006).

(2) Where P is the total rainfall, Ia the initial abstraction, Q the direct runoff, and S the potential maximum retention or infiltration. The two extremely dry and wet scenarios, which may produce runoff, were not considered in the original NRCS-CN method since its concept is that runoff occurs only after fulfilling the initial abstraction Ia requirements. But the modified CNI method is based on the concept of zero initial abstraction (Ia = 0). The CNI method is represented as: (3) The potential maximum retention storage S of watershed is related to a CN, which is a function of land use, land treatments, soil type and antecedent moisture condition of watershed. The CN is dimensionless and its value varies from 0 to 100. The S-value in mm can be obtained from CN by using the relationship: (4) In this study, feature classes of land use and hydrologic soil group were used as the input data to estimate CN values. These feature classes were integrated with Hec-GeoHMS, GIS extension, and CN values were calculated. After calculation of CN values, The S-value was obtained from equation 5. Recharge (or effective precipitation) is the amount of water that percolates downward in the unsaturated zone towards the water table after subtracting the portions of water that are subjected to runoff, evapotranspiration and soil storage. For estimating the average recharge from 1997 to 2008, first, the cumulative annual rainfall for this period was calculated (412.1mm/year). Then the amount of runoff was calculated with equation 4, and the runoff values lay between 226 to 377 mm/year. Finally, the average recharge was calculated by subtracting the portion of water that is subjected to runoff from cumulative annual rainfall, and the recharge values lay between 34 to 185 mm/year or 0.0000952 to 0.0005068 m/day.

(5) Where K xx , K yy , and K zz are the hydraulic conductivities in the x, y and z directions respectively, which are assumed to be parallel to the axes of hydraulic conductivity (LT-1), h is the potentiometric head (L). The finite difference code, MODFLOW (McDonald and Harbaugh, 1988) was chosen to solve Eq. (1) for hydraulic heads in the area. The MODFLOW model, which has been improved and verified, has been used for 20 years. The model is extremely accurate and its suitability has been verified (Laronne Ben-Itzhak and Gvirtzman, 2005). In MODFLOW model, layers can be simulated as confined or unconfined. Flow associated with external stresses, such as wells, areal recharge, evapotranspiration, drains, and rivers, can also be simulated (Harbaugh et al. 2000). Some assumptions of the model are stated as: The density of the fluid is constant; water movement can be in three (orthogonal) directions (x, y, z) and properties within a cell are assumed to be homogeneous (El-Bihery, 2009). In this work the computer software, GMS 7.1, which supports the groundwater numerical code MODFLOW was utilized. GMS is a graphical interface software and is developed by the Environmental Modeling Research Laboratory of Brigham Young University (Environmental Modeling Research 2004). It includes a set of pre/post processing tools for helping users in characterization of the model domain, model conceptualization, mesh and grid generation, geostatistics and post-processing of output (Froukh, 2002). 3D Hydrogeological Conceptual Model

3D Hydrogeological conceptual model was developed by simplifying 19 logs of drilled water wells in the region. According to the data collected from logs clay, silt and sand, gravel are predominant layers of the plain. Logs of wells did not specify location of bedrock. For specifying the location of bedrock, the geological cross-section of the region was used. According to this cross-section, lower layers of the plain consist of recrystallized limestone which JOUR.GEOL.SOC.INDIA, VOL.83, JAN. 2014


87

is placed at a depth of approximately 100 m, and has low permeability. Based on this assumption, recrystallized limestone layers were selected as bedrock and with hydrogeological conceptual model 100 m thickness was estimated for the bedrock. The procedure described is as follows. At first, data collected from logs of wells were entered into the GMS and after the definition of different layers; a 3D geological model of plain was developed. Then the initial 3D numerical analysis network was considered based on data availability and hydrogeological conditions. For model discretization initial network was constructed with 150×100 cells in direction x, y, respectively. The model was divided vertically into 5 layers to obtain an improved resolution and for the benefit of better graphical illustration of results. This network has 91506 nodes and 75000 cells. Figures 4(a), (b) and (c), shows a 3D geological model, a 3D numerical network analysis and the range of the study area, respectively. Using the unit editing function in the GMS model, the 3D numerical network was limited to the range of the study area (Fig.4(d)). Then the 3D geological model and the ground elevation were transformed into a 3D numerical network based on the linear interpolation concept and the

inverse distance weighted method. Figure 4(e) presents the final 3D hydrogeological conceptual model.

Fig.4. Work sequence of establishing the 3D hydrogeological conceptual model, (a) The 3D geological model, (b) The initial 3D numerical network analysis, (c) Range of the study area, (d) Modification of the 3D numerical network boundry, (e) Transforming the 3D geological model into a 3D numerical network and modification of the 3D numerical network elevation.

Boundary Condition and Initial Condition

JOUR.GEOL.SOC.INDIA, VOL.83, JAN. 2014

Conceptual Model

The accuracy of a hydrogeological model depends on precise definition of the aquifer physical boundaries and on a proper estimation of water mass balance (Laronne BenItzhak and Gvirtzman, 2005). The data extracted from both spatially and point source defined measurements were combined into coverage (conceptual model). This model allows a much better understanding of site conditions to define the groundwater problem for development of a numerical model and to aid in selecting a suitable numerical model (Spitez, 1996). The conceptual model includes the potentiometric surface, hydraulic properties, and recharge and discharge components. Developing the conceptual model is the most important part of the modeling process. It simplifies the field situation and organizes associated field data for easy analysis of the system (Kushwaha et al. 2009). In the present study the input for conceptual model comes from three coverage layers. The first coverage was used to define source and sinks. Base on observations achieved from the region, the Gamasib river is recharged by groundwater (Fig.5). For this reason this river was determined as drain with specified conductance. The second coverage was used to define the areal attributes such as recharge zones and recharge values. This value was assigned to the model based on NRCS-CN calculations. For this purpose, average of high and low level of calculations (0.000301m/day) was entered asssuming that this value has even distribution in the region. The third coverage was used to define groundwater table measured at 18 observation wells. Once the set of observation wells is entered and a solution to groundwater simulation is imported, GMS automatically interpolates the computed solution to observation wells and the error (computed – observed) will be displayed graphically based on the confidence interval or standard error (Froukh, 2002). After defining the coverages, the conceptual model converts to 3D numerical network. This conversion allows the modeling engine to do the mathematical calculations, and to produce the results in the form of groundwater heads, water flow budget, and water flow velocity and flow directions.

Hydrological features adjacent to and within the model domain must be represented in the model by mathematical boundary conditions. In MODFLOW, cells used to simulate boundary conditions, are grouped into two categories “constant head” cells and “no-flow” cells. Constant head

88


Fig.5. Gamasiab river, (a) recharging of Gamasiab river from groundwater, (b) flow in Gamasiab river.

cells are those for which the head is specified for each time, and the head value does not change as a result of solving the flow equations. No-flow cells are those for which no flow into or out of the cell is permitted. The remaining cells of the grid, termed “variable-head” cells. Equation 2 must be formulated for each variable-head cell in the grid (Harbaugh et al. 2000). The border of the region was considered as a boundary with no flow, while river was considered as constant head boundaries (Fig.6). The spatial distribution of the groundwater table in the study area, for initial solution of the model, was defined from the digital terrain information because according to the concept proposed by Hubbert (1940), a water table is a subdued replica of the ground surface. After determing the boundary conditions, the Layer Property Flow (LPF) in MODFLOW code was used by the finite difference method to calculate groundwater level and flow. This method directly simulates each individual hydrological stratum and independently calculates the flow in each inter-cell regardless of the dimensions of hydrological stratum units (Yang et al. 2009). Initial value of hydraulic conductivity according to numbers proposed

by Todd (2005) was entered 1m/day and 50 m/day for clay, silt and sand, and gravel respectively. RESULT AND DISCUSSION Model Calibration

Calibration is the process of adjusting model inputs until the resultant predictions give a reasonably good fit to the observed data. It is an iterative and very time consuming process (Chen et al. 1998). Calibration is needed in order to account for unmeasured, unknown, or unrepresented conditions or processes and uncertainty in measured input data. Traditionally, models are calibrated by trial-and-error processes in which model parameters are adjusted within reasonable limits from one simulation to the next to achieve the best model fit. Model fit is commonly evaluated by visual comparison of simulated and measured heads and flows or by comparing root mean square (RMS) errors of heads and flows between simulations (Yaouti et al. 2008). Models can also be calibrated using inverse methods, in which the optimal parameter values for a given parameter structure are determined using a mathematical technique, such as nonlinear regression (Cooley and Naff, 1990). This technique is sometimes referred to as parameter estimation. In this study the parameter estimation program PEST was used for parameter optimization because of its ability to limit parameter value ranges and parallel process utilities. The model was calibrated by steady state calibration of average conditions during 1997. Calibration in Steady State

Fig.6. determining the boundary conditions of study area

Model was calibrated in steady state conditions, assuming constant rain, steady discharge and fixed groundwater table. Also, it was supposed that horizontal hydraulic conductivity and recharge are unknown parameters and PEST method was applied to estimate these parameters. JOUR.GEOL.SOC.INDIA, VOL.83, JAN. 2014


89

Fig.7. Comparison value of the computed head and observed head, (a) the Scatter plot of observed versus computed values of head for calibration flow model, (b) the error between observed and computed value described as the length of the histogram, the error margin was ±20 m

Calibration was conducted to optimize the match between the steady state measured and simulated hydraulic head in 1997. In this area no information existed about pumping rate. Accordingly, the steady state simulation was done without integrating the water volumes extracted from the wells. After calibration, recharge value obtained was 0.0008211 m/day, which is bigger than the initial value that was entered in the model (the ratio of 2.72). Horizontal hydraulic conductivity values obtained were 5m/day and 26.52 m/day for clay, silt and sand, and gravel, respectively. In the second step, by trial-and-error calibration, the Gamasib river conductance was adjusted until the best match between the observed and simulated heads was obtained. After trial-and error the value of conductance obtained was 5m2/d/m. A plot of weighted observations versus weighted simulation equivalents of the water table (Fig.7a, b) shows the calibration fit. This plot displayed a correlation coefficient of 0.91, indicating a good overall fit. In Fig.7 (b), the upper and lower limits of the calibration indicator represent plus and minus which is permissible error value (defined as 20 m). The solid portion of the indicator is the error between calculated and observed values; errors within the permissible range are green (Yang et al. 2009). To provide information on the overall match of all the monitoring wells, the performance statistics of the model were examined. The overall root mean square error (RMSE) of the model is 10.44 m and the mean error (ME) is 2.86 m for calibrated wells, which indicates a tendency towards underestimating head values. These errors show that a good JOUR.GEOL.SOC.INDIA, VOL.83, JAN. 2014

estimation was obtained. Table 1 shows the computed errors between mean groundwater elevation and simulated groundwater elevation in 1997. The water budget of the entire aquifer obtained from the groundwater flow model is presented in Table 2. Budget terms are expressed in m3/d and are positive when entering and negative when leaving the groundwater system. The balance between inflows and outflows is consistent with the steady state modeling hypothesis. The main source of water Table 1. Comparison between the observed and computed groundwater heads in wells. Well Noa 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 a

Observed head (m)

Computed head (m)

Residual (m)

1503 1497 1519 1597 1479 1505 1467 1501 1513 1465 1451 1493 1543 1444 1473 1470 1477 1485

1487 1498 1508 1591 1486 1515 1472 1492 1495 1451 1466 1474 1537 1448 1482 1474 1473 1482

16 1 11 6 7 10 5 9 18 14 15 19 6 4 9 4 4 3

see fig.7(b) for location on map

90


Table 2. Water budget of the aquifer for steady state condition (1997) 1997

input(m3/d)

output(m3/d)

constant heads drain recharge total input-output (input-output)%

171063.69 0 233274.12 404353.43 -3104.88 -0.76

0 -407458.32 0 -407458.32

is the recharge, which is due to precipitation. The recharge amount is 233274.12 m3/d, which represents 57.69% of the total inflow of water in the aquifer. The inputs of water from the constant head are about 171063.69 m3/d. The main outputs of water from the aquifer drain into the Gamasiab river. Other outputs such as evaporation were negligible. Figure 8a,b shows the simulated aquifer head for layers 1 and 2 respectively. According to this figure the unsaturated zones led to dry cells in the model layers. Also, according to Fig.8 flow direction follows the general slope of the area (SE to NW direction). Simulation of Contaminant Transport with Particle Tracking

In particle tracking, a hypothetical particle is placed at a particular location and tracked through the velocity flow field at successive intervals of flow time. With the particle tracking model, it is possible to track a particle forward in time as it moves down the gradient from a given location or backward in time from where it originates (Barry et al. 2009).

In the present study, MODPATH, a particle tracking computer code documented by Pollock (1994), was used to calculate travel times and travel paths for advective transport of particles in groundwater flow. This program represents groundwater travel times and path lines for advective transport only and in order to calculate successive particle positions estimates of aquifer porosity are required. The particle tracking used by MODPATH can be implemented for linear velocity and finite difference methods and may not be able to calculate path line for finite element (Pollock, 1994). In this study, based on numbers proposed by Todd (2005), the porosity of clay, silt, sand and gravel was entered 30-70% and 20-35%, respectively. The particle tracking simulation, neglecting seasonal fluctuations and relatively dry or wet years, was done for 1997 steady state conditions and two particle tracking simulations were made. The first simulation was tracked from the border of the aquifer, and the second simulation was tracked from the five selected wells in the study area. In the first simulation, one particle was located in the center of each cell at both side of the plain and the forward movement of the particle was visualized (Fig.9). Travel times ranged from 30 to 1000 years and simulation was done in layer 2. Movement of particles is based on groundwater gradient and contaminant moves from the both sides towards the center of the plain. Elevation difference between center and border of the plain is the main reason for this movement. The longest flow paths of particles between 30 to 1000 years, is 12232 m.

Fig.8. Calculated water table,(a) layer 1, (b) layer 2 (dry cells are shown with white color). JOUR.GEOL.SOC.INDIA, VOL.83, JAN. 2014


91

Fig.9. simulated flow paths of particles at both side of the plain for steady state condition (1997) for forward tracking particles. Travel time (a) 30 years, (b) 100 years, (c) 1000 years.

A reverse particle tracking analysis was conducted to examine the potential origin of contaminants entering the pumped well (Fig. 10). For this purpose, twenty four particles were placed systematically around the five selected extraction wells and tracked backward in time corresponding to 10, 20 and 100 years of travel time. The particle positions after these years are shown in Fig.10. Simulation was done in layer 2. Nineteen particles moved approximately 8366 m from the center of the cell towards the sources of contamination, over 100 years. For tracing, chloride can be used because this ion is of primary concern in a geochemical analysis and it is highly nonreactive tracer (Yaouti et al., 2008). Results show that the migration of pollutants from both sides of the plain is very slow and sources of any pollutant in selected wells are locations which are placed around the plain. By understanding the source of pollutant and its path line,

contaminant transport to other places of the plain can be avoided. CONCLUSIONS

This work presents a conceptual and numerical analysis of groundwater flow and advective contaminant transport in the unconfined aquifer of Nahavand plain. For this purpose, a groundwater modeling package, the GMS supporting groundwater numerical codes MODFLOW and MODPATH were employed to simulate groundwater flow and groundwater travel times and path lines for advective transport. For the development of the numerical model all available data were integrated and processed. The model was calibrated for steady state conditions in 1997 and the results of the model calibration showed reasonable agreement between observed and calculated

Fig.10. simulate flow paths of selected particles placed at five selected wells cells for steady state conditions for backward particle tracking travel time of (a) 10 years, (b) 20 years, (c) 100 years. JOUR.GEOL.SOC.INDIA, VOL.83, JAN. 2014

92


water levels for the observation wells. The calibration was examined with statistical analysis and results show that the estimation obtained was good. The amount of recharge obtained was 0.0008211 m/day, which is more than the initial value that was entered in the model (the ratio of 2.72). The horizontal hydraulic conductivity, obtained from the model calibration, for clay, silt, sand and gravel are 5m/day and 26.52 m/day, respectively. According to computed water level, flow follow the general slope of the area (SE to NW direction). The values of

computed water level for southeast and northwest are 1564m and 1475m respectively, which with an approximately 28 km of distance, the hydraulic gradient is 0.3% in the plain. The MODPATH results show that the pollutants move from both sides towards the center of the plain. But this movement is very slow (1000 years). Backward simulation of particles from selected wells shows that this groundwater flow model can be used for investigation of times of travel and flow paths of contaminants.

References AYERS, R.S. and WESTCOT, D.W. (1989, 1994) Water quality for agriculture. FAO Irrigation and Drainage paper 29 Rev. 1 BARRY, F., OPHORI, D., HOFFMAN, J. and CANACE, R . (2009) Groundwater flow and capture zone analysis of the Central Passaic River Basin, New Jersey. Environ. Geol., v.56, pp.1593-1603. CHEN , D.W., C ARSEL , R.F., M OETI, L. and VONA , B. (1998) Assessment and Prediction of Contaminant Transport and Migration at a Florida Superfund Site. Environ. Monit. Assess., v.57, pp.291-299. COOLEY, R.L. and NAFF, R.L. (1990) Regression modelling of ground-water flow.U.S. Geological Survey Techniques of Water-Resources Investigations,233p. (Book 3, Chapter B4). DON, N.C., HANG, N.T.M., ARAKI, H., YAMANISHI, H. and KOGA, K. (2006) Groundwater resources management under environmental constraints. Environ. Geol., v.49, pp.601-609. EL-BIHERY, M.A. (2009) Groundwater flow modeling of Quaternary aquifer Ras Sudr, Egypt. Environmental Geology 58:1095– 1105. E NVIRONMENTAL M ODELING R ESEARCH L ABORATORY . (2004) Groundwater modeling system. Brigham Young University. FROUKH, L.J. (2002) Groundwater Modelling in Aquifers with highly Karstic and Heterogeneous Characteristics (KHC) in Palestine . Water Resources Management, v.16, pp.369-379. HAMEDAN REGIONAL WATER ORGANISATION website(http://hmrw.ir). HARBAUGH, A.W., BANTA, E.R., HILL, M.C., and MCDONALD, M.G. (2000) MODFLOW-2000, The U.S. Geological Survey Modular Ground-Water Model e User Guide to Modularization Concepts and the Groundwater Flow Process. U.S. Geological Survey, Water-Resources Investigations Report 00-92, 121p. HUBBERT, M.K. (1940) The theory of ground water motion. Jour. Geol., v.48(8), Pt.1., pp.785-944 . KUSHWAHA, R.K. PANDIT, M.K. and GOYAL, R. (2009) MODFLOW Based Groundwater Resource Evaluation and Prediction in Mendha Sub-Basin, NE Rajasthan. Jour. Geol. Soc. India, v.74, pp.449-458. LARONNE, B.I. and GVIRTZMAN, H. (2005) Groundwater flow along and across structural folding: an example from the Judean Desert, Israel. Jour. Hydrology, v.312, pp.51-69.

MANFREDA, S., FIORENTINO, M. and IACOBELLIS, V. (2005) DREAM: a distributed model for runoff, evapotranspiration, and antecedent soil moisture simulation. Adv. Geosci., v.2, pp.3139. MCDONALD, M.G. and HARBAUGH, A.W. (1988) A modular threedimensional finite difference flow model. Techniques of water resources investigations of the U.S. Geological Survey, Book 6, 586p. P ATIL, J.P., SARANGI , A., SINGH, A.K. and AHMAD, T. (2008) Evaluation of modified CN methods for watershed runoff estimation using a GIS-based interface. Bio-systems Engg., v.100, pp.137-146. POLLOCK, D.W. (1994) User’s guide for MODPATH/MODPATHPLOT, Version3: a particle tracking post-processing package for MODFLOW, the U.S. Geological Survey finitedifference groundwater flow model, USGS. Open-File Report 94-464. REGLI, C., RAUBER, M. and HUGGENBERGER, P. ( 2003) Analysis of aquifer heterogeneity within a well capture zone, comparison of model data with field experiments: a case study from the river Wiese, Switzerland. Aquat. Sci., v.65, pp.111-128. SPITEZ, K. (1996) A practical guide to groundwater and solute transport modeling. John Wiley and Sons Inc., 238p. TODD, D.K. and MAYS, L.W. (2005) Groundwater Hydrology. John Wiley& Sons. YANG, F.R. HAW, L.C. KUNG, W.J. and YEH, H.F. (2009) The impact of tunneling construction on the hydrogeological environment of “Tseng-Wen Reservoir Transbasin Diversion Project” in Taiwan. Engg. Geol., v.103, pp.39-58. YAOUTI, F.EL. MANDOUR, A.EL. KHATTACH and D. KAUFMANN, O. (2008) Modelling groundwater flow and advective contaminant transport in the Bou-Areg unconfined aquifer (NE Morocco). Jour. Hydro-environment Res., v.2, pp.192-209. YIDANA, S.M. and OPHORI, D. (2008) Groundwater Resources Management in the Afram Plains Area, Ghana. KSCE Jour. Civil Engg., v.12(5), pp.349-357. ZOU, R., TENG, Y. and WANG, J. (2009) Modeling migration of strontium in sand and gravel aquifer in the candidate VLLW disposal site. Jour. Radioanal Nucl. Chem., v.281, pp.653662.

(Received: 30 December 2011; Revised form accepted: 26 November 2012) JOUR.GEOL.SOC.INDIA, VOL.83, JAN. 2014