You are on page 1of 20

Journal of Hydrology: Regional Studies 35 (2021) 100822

Contents lists available at ScienceDirect

Journal of Hydrology: Regional Studies


journal homepage: www.elsevier.com/locate/ejrh

Fully integrated numerical simulation of surface


water-groundwater interactions using SWAT-MODFLOW with an
improved calibration tool
Tina Jafari a, *, Anthony S. Kiem a, Saman Javadi b, *, Takashi Nakamura c,
Kei Nishida c
a
Centre for Water, Climate and Land (CWCL), School of Environmental and Life Sciences, College of Engineering, Science and Environment (CESE),
University of Newcastle, Callaghan, NSW 2308, Australia
b
Department of Irrigation and Drainage, College of Aburaihan, University of Tehran, Tehran, Iran
c
Interdisciplinary Centre for River Basin Environment, University of Yamanashi, Yamanashi 400-8511, Japan

A R T I C L E I N F O A B S T R A C T

Keywords: Long-term sustainable water resources management requires comprehensive assessment of all
Surface water-groundwater modeling water cycle components. The combined calibration of surface water (SW) and groundwater (GW)
SWAT-MODFLOW calibration parameters is one of the biggest challenges in simulating SW-GW interactions. SWAT-MODFLOW
Sensitivity analysis
has improved our ability to simulate SW and GW interactions, however there remain challenges
Regional modeling
associated with (i) understanding what parameters most influence model outputs and (ii) prop­
erly calibrating the GW related components of the water cycle (as represented within SWAT-
MODFLOW). In this paper, we applied SWAT-MODFLOW to the 1452 km2 Shiraz catchment,
located in southwestern Iran, for the period 2003–2019. A comprehensive sensitivity analysis
(SA) of both SW and GW parameters was conducted to determine what controls runoff and GW
level in the study catchment. The SA reveals that the Curve Number (CN2) and GW hydraulic
conductivity (K) parameters have the most significant impact on runoff and GW level simulation.
This work also presents a “GW Calibrator” (GWCal) tool along with a demonstration of how
GWCal improves the performance of SWAT-MODFLOW in the Shiraz catchment as a result of the
insights gained from the SA.

1. Introduction

Hydrological models have been used to evaluate water availability and determine appropriate water management policies for a
long time. These models are divided into three main categories: (i) hydrological models which simulate runoff such as HBV (Bergstrom
and Forsman, 1973; Bergstrom, 1992), SAC-SMA (Burnash et al., 1973), TOPMODEL (Beven and Kirkby, 1979), VIC (Liang et al.,
1994), HEC-HMS (William et al., 1995; U.S. Army Corps of Engineers, 2016), SWAT (Arnold et al., 1998), SWIM (Krysanova et al.,
2000), and WaterGAP (Verzano, 2009); (ii) hydrogeological models which simulate GW systems such as MODFLOW (Harbaugh,
2005), MicroFEM (Diodato, 2000), ZOOMQ3D (Jackson, 2001), SVFLUX (Thode and Fredlund, 2013), FEHM (Zyvoloski et al., 2015);
and (iii) physically based integrated surface–subsurface hydrological models (ISSHM) which solve the governing surface and

* Corresponding authors.
E-mail addresses: Tina.Jafari@uon.edu.au (T. Jafari), javadis@ut.ac.ir (S. Javadi).

https://doi.org/10.1016/j.ejrh.2021.100822
Received 21 February 2020; Received in revised form 12 April 2021; Accepted 12 April 2021
Available online 23 April 2021
2214-5818/© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Fig. 1. Study area location in southwest of Iran, aquifer boundary and river network.

subsurface flow equations simultaneously (Furman, 2008). Popular ISSHM codes include Integrated Hydrology Model (InHM; Van­
derKwaak, 1999), MODHMS (HydroGeoLogic Inc, 2003), ParFlow (Kollet and Maxwell, 2006) and HydroGeoSphere (Therrien et al.,
2010). Detailed studies of ISSHM internal flow dynamics are usually undertaken at scales of less than 0.5 km2 (e.g. Ebel and Loague,
2006(Ebel and Loague, 2006)). Regional-scale, surface-subsurface applications of ISSHMs (e.g. (Li et al., 2008)) are less common
because of the large volume of data required to parameterize and evaluate the models (Sebben et al., 2013). Paniconi and Putti, 2015
reviewed developments of physically based hydrological models over the past 50 years. Maxwell et al. (2014) has provided a com­
parison between different coupled hydrological models.
The Soil Water Assessment Tool (SWAT) is a semi distributed catchment scale SW model that can simulate surface runoff. SWAT has
been shown to perform well in streamflow simulation in a wide range of watersheds and has been tested in many regions, conditions
and time scales (Gassman et al., 2007). Although SWAT has been one of the most widely used hydrological models, it has limited
capability in simulating the GW system (Neitsch et al., 2011). In SWAT, GW is represented by a lumped module which is divided into
shallow and deep GW. Both the shallow and the deep aquifer may contribute to streamflow as baseflow through a linear reservoir
approximation, ignoring distributed parameters such as hydraulic conductivity (K) and storage coefficients (S) (Kim et al., 2008). This
simplification of GW dynamics results in a mischaracterisation of GW resources especially when streamflow is strongly dependent on
GW discharge (Gassman et al., 2007). In fact, SWAT compares shallow aquifer depth with a threshold to estimate the river recharge
and it is not capable of considering the river bottom elevation and aquifer depth to properly simulate recharge (Gassman et al., 2007;
Kim et al., 2008; Arnold et al., 2013; Gayathri et al., 2015, Aliyari et al., 2019).
MODFLOW is a fully distributed three-dimensional GW standard model which simulates GW flow. MODFLOW is the most popular
GW model and has been successfully used in many global and regional studies (Harbaugh et al., 2000; Harbaugh, 2005; Jafari et al.,
2016a, 2016b, de Graaf et al., 2017; Deb et al., 2019b). Although MODFLOW is the most popular GW model, it has considerable
limitations in simulating surface and subsurface processes of land-atmosphere interactions, agricultural management practices, Un­
saturated Zone Flow (UZF), and surface runoff and does not directly account for hydrologic processes that occur in the root zone
(Lachaal et al., 2012; Surinaidu et al., 2014). Consequently, MODFLOW lacks the ability to consider the difference in recharge rates as

2
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Fig. 2. Shiraz catchment river network and location of sub-basins, stream gauges and weather stations.

a result of changes in precipitation and infiltration from the river. Infiltration varies due to land use, irrigation and agronomic practices
implemented on the surface domain (Guzman et al., 2015).
Several attempts have been made to address the above-mentioned shortcomings by linking SWAT with MODFLOW to simulate SW-
GW interactions (e.g. Kim et al., 2008; Guzman et al., 2015; Izady et al., 2015; Bailey et al., 2016; Deb et al., 2019a, 2019b, Abbas et al.,
2018; Gao et al., 2019; Chunn et al., 2019; Molina-Navarro et al., 2019; Sith et al., 2019; Aliyari et al., 2019). Bailey et al. (2016)
developed a new coupled SWAT-MODFLOW model that geographically locates and integrates Hydrologic Response Units (HRUs) in
SWAT sub-basin with the spatially gridded MODFLOW model. This model has been tested against observed stream flow and water
table elevation in a catchment area in the United State of America (USA). In SWAT-MODFLOW, deep percolation from SWAT is passed
to the grid cells of MODFLOW as recharge, and MODFLOW-calculated SW-GW interaction fluxes are passed to the stream channels of
SWAT. Hence, the model accounts for two-way interactions between SW and GW which enables better representation and under­
standing of the spatial-temporal patterns of SW-GW interactions (Bailey et al., 2016). SWAT-MODFLOW has been applied to different

3
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Fig. 3. Study area landuse.

catchments in the USA (Abbas et al., 2018; Gao et al., 2019), Canada (Chunn et al., 2019), Denmark (Molina-Navarro et al., 2019), Iran
(Semiromi and Koch, 2019), and Japan (Sith et al., 2019). SWATMOD-Prep was introduced by Bailey et al., 2017 as a graphical user
interface for coupled SWAT-MODFLOW and it was applied in Little River Experimental Watershed, Georgia.
In order to apply SWAT-MODFLOW in large scale catchments, Aliyari et al., 2019 improved the linkage between MODFLOW
pumping cells and SWAT HRUs for irrigation. This modification accounts for spatially variable recharge which in turn is affected by
dynamic land use and land management practices.
When calibrating SWAT-MODFLOW, it is essential to know which GW parameters could impact runoff simulation results from
SWAT and which SW parameters could impact GW level results from MODFLOW. Previous research has shown that a sensitivity
analysis (SA) is required to determine which parameters control runoff and GW level, and how SW and GW parameters impact each
other (Guo and Su, 2019; Jiang et al., 2018). In order to obtain meaningful results from the SA, a well calibrated SWAT-MODFLOW
model is required. However, it is challenging, especially with manual calibration, to obtain optimal parameter values when concur­
rently simulating SW and GW after linking in SWAT-MODFLOW.
The ultimate aim of this research is to obtain insights about what controls runoff and GW level in a semi-arid area. This is done by
conducting a comprehensive SA to determine the effect of both SWAT parameters and MODFLOW parameters on the results of the
SWAT-MODFLOW simulation. A new “GW Calibrator tool” (GWCal) is developed to improve the efficiency and effectiveness of the

4
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Table 1
Data sources with date and frequency used for the SWAT-MODFLOW model.
Model Data type Date Frequency/resolution Source

Soil map 2009 1000 × 1000 m FAO soil database, 2020


DEM 2015 15 × 15 m IR Ministry of energy, 2016
Land use 2012 1000 × 1000 m USGS global land use, 2020
SWAT Meteorological data (rain, temperature, Iran Water Resources Management,
2003− 2019 Daily
evaporation) 2020
Meteorological organization of Iran,
River runoff 2003− 2019 Daily
2020
Observation bores/ Provisional water authority of Fars,
GW levels 2003− 2019
Monthly 2021
2003, 2008 and Provisional water authority of Fars,
Production bores 5 years
2013 2016
MODFLOW Aquifer thickness 1998 250 × 250 m Geological survey of Iran, 2014
Provisional water authority of Fars,
Aquifer transmissivity 1998− 2015 250 × 250 m
2016
Provisional water authority of Fars,
Aquifer specific yield 1998− 2015 250 × 250 m
2016

manual calibration process and to obtain more realistic results from SWAT-MODFLOW and the SA. GWCal is applied to the latest
version of SWAT-MODFLOW for a catchment in Shiraz, Iran, where SW-GW interaction is high, to demonstrate how GWCal improves
the efficiency/effectiveness of the calibration process.

2. Materials and methods

2.1. Study area

2.1.1. Geography and climate


The Shiraz watershed covers an area of 1452 km2 in southwestern Iran (Fig. 1). The elevation in Shiraz catchment ranges from 1436
to 2991 m above mean sea level with an average of 1830 m (Fig. 1). The Shiraz catchment has mountains covering approximately 800
km2 of the total catchment area. Annual long-term average temperature in the study area is 17.9⁰C with 2550 mm average potential
evaporation and 361.5 mm annual long-term average precipitation. In this area precipitation mainly occurs during December to March
(Fars Regional Water Authority, 2015).

2.1.2. Hydrology
The Khoshk River is the major stream in Shiraz catchment, originating in the northwest of the basin it crosses the alluvial aquifer
and Shiraz city, discharges into Mahrloo lake. The river discharge is heavily affected by the wet season. Stream gauges are located at
the lower end of Azam and Khoshk sub-watershed (Eini et al., 2019). Fig. 2 shows the river network and location of sub-basins, stream
gauges and weather stations in the study area.
The land cover in the Shiraz catchment is dominated by intensive irrigated agriculture and other land use includes wetland, pasture
and urban areas (Fig. 3). The main crops include spring and winter wheat, almond and pomegranate gardens (Fars Regional Water
Authority, 2015).
Soils are heterogeneous and erosive with fine sandy loam in the eastern part, fine sandy loams and loamy soils in the north central
and south central portions of the watershed, and silt loams in the western portion (Fars Regional Water Authority, 2015).

2.1.3. Hydrogeology
The Shiraz aquifer is mainly an unconfined aquifer constituted of fine sandy clay and loamy soils in the north west and central
portions of the watershed. Shiraz aquifer average thickness is approximately 30 m. Based on results of pumping tests, the transmissivity
range varies between 100 to 200 m2/d, whereas the average specific yield is around 0.05 unit.
In the Shiraz aquifer, GW depth varies from between 3–35 metres below ground level (mbgl). The maximum GW depth is in west
and north of aquifer where GW depth level is 35 m. The minimum GW depth is towards the east and south part of the aquifer, where
GW depth reaches to less than 3 m depth. General GW flow direction is from the north west toward the south east (Fars Regional Water
Authority, 2015). Available information from groundwater and stream observation networks indicate there is interaction between the
aquifer and river mainly in the south-east part of aquifer (Fars Regional Water Authority, 2016).
Geologically, the study area is comprised of stratigraphic units of Precambrian rocks to Quaternary deposits, and tectonically,
northern altitudes of Shiraz Plain are placed in the Zagros tectonic zone. Moreover, the tectonic structure of the aquifer, including folds
and faults, are mainly extended from north to south and from north west to south east.

5
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Table 2
Land use type and soil type distribution in the study area.
Data set Class Description % Area

1 Loam 18.4
Soil 2 Clay loam 49.2
3 Sandy clay 32.4
1 0 %–1/5 % 46.5
Slope 2 1/5 %–4 % 38.3
3 >4% 15.2
1 Residential 3.6
2 Forest 1.7
3 Agriculture 8.5
Land use
4 Orchard 4.9
5 Wetland 1.4
6 Pasture 79.9

This aquifer is considered as the main source of water for irrigation, domestic, and public supply because of streamflow depletion
and reduced base flow. According to the latest water budget reports in 2019, 1802 production bores are extracting 189 million cubic
meter (MCM) of water from the aquifer. In addition, 122 drinking water, 48 industrial usage and 1632 irrigation pumping bores are
placed on agricultural land.

2.2. Data collection

In order to apply the SWAT-MODFLOW model in the study area a range of meteorological, hydrological and hydrogeological data
were collected and interpreted from various sources (Table 1). Rainfall, temperature and evaporation data was collected from four
stations in the catchment (Fig. 2). Data from two runoff gauges of Khosk river, and Azam river have been used. Rainfall data was
available for the entire simulation period of 2003− 2019 (Iran Water Resources Management, 2020). GW level data were available
from an existing network of GW monitoring bores. In the study area GW is used for agriculture, potable and industrial use (Fars
Regional Water Authority, 2015).

2.3. Model setup

2.3.1. Surface water simulation


To simulate SW system in the study area, SWAT was used. SWAT was built for the entire watershed shown in Fig. 2. SWAT
simulation is based on conceptual units of homogeneous land use, management, slope, and soil characteristics that extend below the
surface to a soil profile depth called HRUs (Arnold et al., 1998). ArcSWAT was used as the computer interface to develop the SWAT
model in the study area.
The main input data for SWAT includes slope, soil data, and land use. The National Elevation Dataset from the Iranian Ministry of
Energy have been used to develop the slope map with grid size 15 × 15 m. The Food and Agriculture Organization (FAO) soil dataset
and United State Geological Survey (USGS) global land use with 1 × 1 km raster layer were used to obtain soil and land use types in the
study area respectively (Table 2).
Required meteorological inputs, such as precipitation, wind, relative humidity, maximum and minimum air temperatures and solar
radiation from the Meteorological Organization of Iran for four weather stations have been employed as model inputs (Fig. 2). The
spatial interpolation of the meteorological data was done via Simple Kriging (SK) method. SK is a linear unbiased method to estimate
the value of regionalized variables at an unsampled location based on the available data (Sun et al., 2009).
The SWAT model was built with the total of 32 sub-basins and 188 HRUs. Sub-basin areas ranged from approximately 1.2–7.1 km2.

2.3.2. Groundwater simulation


Although SWAT is commonly used in simulating watershed hydrological systems it has limited capability in simulating GW sys­
tems. In this study, the latest three-dimensional GW standard model, MODFLOW_Newton Raphson formulation (MODFLOW), was used
to simulate GW flow. MODFLOW is a physically based and fully distributed model that solves the GW flow differential equation in 3D
using the finite difference approach (Niswonger et al., 2011). MODFLOW was built and applied only for the aquifer subsection (see
Fig. 1), and GW modelling area was gridded to 183 × 171 cells with 300 m × 300 m dimension. In order to link SWAT and MODFLOW,
the following packages were used: NAM (input and output files of model); DIS (boundary condition); BAS (initial head condition); RCH
(GW recharge); GHB (general head boundary); OBS (observation bores); UPW (aquifer hydrodynamic parameters including K and S);
NWT (solves the GW flow); and WEL (pumping/production bores).

6
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Fig. 4. SWAT dHRUs and MODFLOW grids in SWAT-MODFLOW.

2.3.3. SW-GW interaction simulation


SWAT has limited capability in simulating the heterogeneity of the GW system and applying the distribution of K, S and GW level
(Sun et al., 2016; Nguyen and Dietrich, 2018). In this study, the interaction between river and aquifer is simulated with the revised
version of integrated SWAT-MODFLOW code developed by Aliyari et al., 2019. The SWAT-MODFLOW code used in this study was first
developed by Bailey et al. (2016) by combining the updated SWAT 2012 model (Revision 591) with MODFLOW. This code spatially
represents feedback fluxes within the surface and GW domains. In this code, SWAT simulates land surface processes, crop growth,
in-stream processes and soil zone processes, whereas MODFLOW simulates three-dimensional GW flow and all associated sources and
sinks (e.g. GW recharge, pumping, discharge to tile drains and interaction with stream network). Both modelling codes are combined
into a single FORTRAN code as an executable. This code pass, deep percolation from HRUs into the grid cells of MODFLOW as GW
recharge. Then it passes SW-GW interaction fluxes from MODFLOW into the stream channels in SWAT.
In order to simulate the interaction between SW-GW the package of “River” in MODFLOW is used. This package is capable of
calculating the volumetric flow of exchanged water Qleak [L3/T] between river and aquifer based on Darcy’s equation as shown in Eq.
(1);
( )
hstr − hgw
Qleak = Kbed (Lstr Pstr ) (1)
Zbed

Where Kbed is the river bed hydraulic conductivity [L/T], Lstr is river length [L], Pstr is the wetted perimeter of the river [L], hstr is river
stage [L], hgw is the hydraulic head of GW [L], and zbed is the river bed thickness [L]. If GW recharges the river (i.e. GW hydraulic head
hgw is above the river stage hstr), Qleak will be negative and if river recharges the aquifer Qleak will be positive. The required input for
recharge package in MODFLOW is the SWAT output of deep percolation from HRUs. Data from SWAT is passed to MODFLOW grids,
according to the percent area of the DHRU (Disaggregated HRU) contained within the grid cell for use by the recharge package in
MODFLOW (Bailey et al., 2016).

7
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Fig. 5. Applied simulation strategy in this research.

In the version of SWAT-MODFLOW applied in this research, MODFLOW pumping cells are linked to SWAT HRUs which allow for
the application to basin-scale agro-urban watershed systems like Shiraz. This version is capable of simulating SW-GW interactions as
well as management schemes that transfer water between the domains in semi-arid, heavily managed agro-urban large river basins.
Fig. 4 shows the river cells (blue cells) and the aquifer cells simulated with MODFLOW (green cells).
The simulation strategy is represented in Fig. 5. In the first step, the required information is collected from different sources to setup
the SW and GW model and then the two codes are joined using Aliyari et al. (2019) linkage codes. In the second step, the
SWAT-MODFLOW model is calibrated using GWCal.

8
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Table 3
SWAT parameters used for calibration process before coupling.
SWAT Units Parameter Description Calibrated range for study
Parameter area

ALPHA_BF.gw Days Baseflow alpha factor 0− 1


PLAPS.sub mm H2O/km Precipitation lapse rate − 100 to 100
TLAPS.sub ◦
C/km Temperature lapse rate − 10 to 10
CH_N(2).rte – Manning’s "n" value for the main channel − 0.01 to 0.03
HRU_SLP.hru m/m Average slope steepness 0− 1
LAT_TTIME.hru Days Lateral flow travel time 0− 100
CH_K(2).rte mm/hr Effective hydraulic conductivity in main channel alluvium − 0.01 to 30
CN(2).mgt – Initial Soil Conservation Service (SCS) runoff curve number for moisture condition II 50− 70
ESCO.hru – Soil evaporation compensation factor 0− 1
GWQMN.gw mm Threshold depth of water in the shallow aquifer required for return flow to occur 0− 500
GW_DELAY.gw days GW delay time 0− 500
GW_REVAP.gw – GW revap. coefficienta 0.02–0.2
OV_N.hru – Manning’s roughness coefficient (n value) for overland flow 0.01–0.3
REVAPMN.gw mm Threshold depth of water in the shallow aquifer for revap. or percolation to the deep 0− 500
aquifer to occur
SOL_AWC.sol mm H2O mm/ Available water capacity of the soil layer 0− 1
soil
SOL_BD.sol g/cm3 Moist bulk density 0.9 – 2.5
SOL_K.sol mm/hr Saturated hydraulic conductivity 0− 200
SOL_CRK.sol – Crack volume potential of soil 0− 1
RCHRG_DP.gw – Deep aquifer percolation fraction 0− 1
a
Revap; movement of water from the shallow aquifer into the overlying unsaturated zone (Arnold et al., 2013).

After calibrating SWAT-MODFLOW, a comprehensive SA is conducted to identify the most important parameters.

2.3.4. Calibration of SWAT-MODFLOW


Calibration of SWAT is performed at two outlets associated with the two branches of rivers for the periods of 2003–2019 on a daily
time step, with three years for a warm-up period that allows the model to stabilize for further simulations. To perform successful
calibration the Sequential Uncertainty Fitting 2 (SUFI-2) approach (Abbaspour et al., 2015) with SWAT Calibration Uncertainty
Procedure (SWAT-CUP) was used. The SUFI-2 is a semi-automated approach that makes the calibration process easier to carry out
within the reliable time bounds (Semiromi and Koch, 2019). The calibration and validation of the model within SWAT-CUP was
conducted considering 24 hydrologic key parameters based on detailed previous literature sources (Abbaspour, 2015). Each SWAT
parameter was set to default lower and upper values (Table 3) as suggested by the SWAT expert group and the best fitted parameter
values obtained from SWAT-CUP were incorporated into the database for stream discharge simulations. SWAT model performance was
evaluated previously using a range of different parameters including P-Factor, R-Factor, Nash-Sutcliffe efficiency (NSE), (Nash and
Sutcliffe, 1970), coefficient of determination (R2), (Nagelkerke, 1991), and percentage of bias (PBIAS) (Moriasi et al., 2007). Root
mean square method (RMSE) was only used to evaluate MODFLOW.
GW level was manually calibrated with GWCal and five different groups of K and S values were generated in the modeling area. The
model was manually calibrated based on GW level data from observation bores. The GW monitoring network and pumping bore lo­
cations are shown in Fig. 6.
The GWCal code developed in this research was used to calibrate GW parameters in SWAT-MODFLOW. GWCal is written in Visual
Basic for Applications (VBA) and linked to the SWAT-MODFLOW model to allow calibration of GW parameters in SWAT-MODFLOW.
The main process of the GWCal is represented in Fig. 7 and the GWCal interface is illustrated in Fig. 8. Each time that GW parameters
are calibrated in this interface the corresponding files in SWAT-MODFLOW are replaced with the new input data.
In the first step, the results of the coupled SWAT-MODFLOW model are considered as initial values in various zones of the aquifer
(multiple polygons). In the next step, the calibrated values from polygons in ArcGIS format were changed to raster. Following that, the
raster files of K and S were converted to ASCII file as MODFLOW input (UPW package) in SWAT-MODFLOW. Finally, the model was run
using the calibrated GW values. After calibration, the model output is represented in the GW head in each observation bore. The GWCal
code is capable of presenting the results of GW model performances from SWAT-MODFLOW after each run. By comparing the model
performance (RMSE) from the last run with the previous run the user would be able to track changes in the model performance result
and improve the result. The GWCal main process commands are as below (see Fig. 7 for further details);

1 ZONE COMMAND: When the user changes MODFLOW parameters such as K and S in each zone in the main window, GWCal
creates corresponding K and S in that particular zone in the text format.

9
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

(caption on next page)

10
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Fig. 6. Aquifer boundary, GW level monitoring network, and pumping bore locations.

2 RASTER COMMAND: Create raster data from new S and K in ArcGIS.


3 ASCII COMMAND: Convert raster data from new raster data to ASCII.
4 UPW COMMAND: Set up UPW package and create SWAT-MODFLOW input file in UPW format from ASCII data.
5 RUN COMMAND: Run SWAT-MODFLOW automatically with new parameters.
6 CLEAR DATA COMMAND: Remove all previous run data from corresponding SWAT-MODFLOW folder.
7 OBS COMMAND: Import simulated data at observed location from MODFLOW package.
8 MON COMMAND: Convert daily data to monthly average as the observed values are monthly.
9 RESULTS COMMAND: 1) Refresh the result and represent GW level in each bore against the previous run which is similar to what
PMWIN and GMS do; 2) calculate RMSE and R2 for each observation bore as well as overall and represented simulated result of GW
level as “underestimated” (negative) and “overestimated” (positive) for each bore (red circles in Fig. 8) and whole aquifer as total;
and 3) create plots from simulated and observed GW level automatically over the simulation period.

The SWAT result is then extracted and represented alongside the MODFLOW result to give the modeler a good understanding of
whether the model performance is improving or not for both SW and GW.

2.4. Local sensitivity analysis (SA)

SA determines the relative change in model output values given changes in model input values (Cacuci et al., 2005). Local SA of
SWAT-MODFLOW was conducted by manually adjusting model parameter values to test their impact on model outputs of “river
runoff” and “GW level”. To determine the nature of different parameters contribution to the model result, model parameters were
manually changed and river runoff and GW level result changes were assessed.
This SA shows whether an increase in parameter values results in an increase or decrease in river runoff and GW level. Table 4
shows the parameters chosen for SA.
The key steps involved in the SA were:

1) Choosing parameters and output variables

In order to conduct SA, parameters related to SW, unsaturated zone and saturated zone were selected as follows:

• Group 1: SW related parameters including CN, LAT_TIME, ALPHA_BF, SLSUBBSN


• Group 2: Unsaturated zone related parameters including SOL_CRK, SOL_AWC, SOL_BD;
• Group 3: Saturated zone (GW) related parameters including GW delay, S and K.
1) Determination of the value ranges

The default values of each selected parameters were changed manually by 25 %, 50 %, 75 %, 150 %, 200 % and 400 %.

3. Result

3.1. SWAT-MODFLOW performance for surface water

The SWAT-MODFLOW results for runoff simulation at two points (Chenar Sokhteh (outlet 19) and Chenar Khoshk (outlet 16)) are
presented in Fig. 9. Table 5 shows that the SWAT-MODFLOW model performed well in simulating runoff in both outlets based on the
guidelines in Moriasi et al. (2007). While the results generally show good agreement, some differences are also evident (e.g. the 2008
flood is underestimated). The differences between simulated and observed values mostly occur at times of peak runoff and this is due to
SWAT’s inherent limitation in simulating floods. Based on Yu et al., 2018, this is due to the fact that SWAT is a long-term yield model
and daily time steps limit SWAT model application in event-based flood simulation. The calibrated SWAT-MODFLOW values are listed
in Table 6. It is interesting to note that the sub-basin CN2 parameters which are related to SW, are close to their original reference
values in SWAT, while parameters related to saturated zone and GW are not close to their original values in SWAT. This highlights a
limitation of SWAT in simulating the GW-related components in the water cycle.

3.2. SWAT-MODFLOW performance for groundwater

As outlined in Section 2.3.2, the MODFLOW model was calibrated using observation bores, water budget results and two stream

11
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

(caption on next page)

12
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Fig. 7. GWCal main process of calibrating SWAT-MODFLOW.

Fig. 8. GWCal interface.

Table 4
Selected parameters for SA.
Parameter Unit Parameter Description

CN2 – Initial SCS runoff curve number for moisture condition II


LAT_TIME Days Lateral flow travel time
Group 1: Parameters related to SW
ALPHA_BF Days Base flow alpha factor for shallow aquifer
SLSUBBSN M Average slope length
SOL_CRK – Crack volume potential of soil
Group 2: Parameters related to unsaturated zone SOL_AWC mm H2O mm/soil Available water capacity of the soil layer
SOL_BD g/cm3 Moist bulk density
K m/day GW hydraulic conductivity
Group 3: Parameters related to saturated zone
S – Specific Yield
GW_DELAY Days GW delay time

13
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Fig. 9. Observed and simulated river discharge using the SWAT-MODFLOW model during calibration and validation periods.

Table 5
Values for objective functions for evaluating the performance of SWAT at each outlet.
Outlet Drainage Area (km2) p-Factor r-Factor R2 Initial R2 bR2 Initial bR2 NSE

Outlet 16 513.1 0.80 0.78 0.70 0.01 0.16 0.001 0.66


Outlet 19 140.9 0.78 0.84 0.72 0.02 0.34 0.002 0.68
Average 0.79 0.81 0.71 0.015 0.25 0.001 0.67

gauges. After linking MODFLOW with SWAT the GW parameters were calibrated using GWCal. The GW level calculated by SWAT-
MODFLOW and observed GW level values in some piezometers are shown in Fig. 10. Table 7 shows the SWAT-MODFLOW perfor­
mance at GW bores.
Table 8 outlines the SWAT-MODFLOW water balance result. Based on the SWAT-MODFLOW result, a considerable percent (65 %)
of estimated rainfall (225.5 MCM) is evaporated. Of the remaining water (80 MCM), around 30 % (24.7 MCM) contributes to runoff
and 70 % (58.3 MCM) goes into GW system.
The initial and calibrated distribution maps for K and S (all extracted from GWCal) are presented in Fig. 11. These maps provide
insight about the aquifer hydraulic parameters all around the model domain. Based on the result, interaction between river and aquifer
in the north of the Shiraz basin is lower than it is in the south.

3.3. Local sensitivity analysis (SA)

The results of the SA in group 1 (parameters related to SW: CN2, LAT_TIME, ALPHA_BF, SLSUBBS), group 2 (parameters related to
unsaturated zone: SOL_CRK, SOL_AWC, SOL_BD) and group 3 (parameters related to saturated zone: K, S, GW_DELAY) on river runoff
and GW level are represented in Fig. 12.

14
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Table 6
Calibrated parameters for SWAT-MODFLOW.
Calibrated values
Sub-basin
Units Parameter Description Outlet Outlet
parameter
16 19

CN(2).mgt – Initial SCS runoff curve number for moisture condition II 57.5 55.3
Alpha_BF.gw days Baseflow alpha factor for shallow aquifer 0.2 0.12
REVAPMN.gw mm Threshold depth of water in the shallow aquifer for revap. or percolation to the deep aquifer to 185 205
occur
GWQMN.gw mm Threshold depth of water in the shallow aquifer required for return flow to occur 4565 4855
GW_REVAP.gw – GW revap. Coefficient 0.087 0.018
GW_DELAY.gw days GW delay time 13 15
HRU_SLP.hru m/m Average slope steepness 0.02 0.02
ESCO.hru – Soil evaporation compensation factor 0.5 0.51
SLSUBBSN.hru m Average slope length 115 121
OV_N.hru – Manning’s “n” value for overland flow 30 30
PLAPS.sub mm H2O/ Precipitation lapse rate 12 30
km
TLAPS.sub ◦
C/km Temperature lapse rate 3.5 1.5
CH_N(2).mgt – Manning’s "n" value for the main channel 0.25 0.26
LAT_TTIME.hru Days Lateral flow travel time 85 92
CH_K(2).rte mm/hr Effective hydraulic conductivity in main channel alluvium 185 240
SOL_AWC.sol mm H2O/ Available water capacity of the soil layer 0.6 0.9
soil
SOL_BD.sol g/cm3 Moist bulk density 2.5 2
RCHRG_DP.gw – Deep aquifer percolation fraction 0.75 0.9

Based on SA result, K and S parameters had the most significant influence on the model outputs in GW level while CN2, SOL_BD and
SOL_CRK parameters had the biggest impact on river runoff. ALPHA_BF, LAT_TIME, GW_DELAY and SLSUBBSN parameters have no
significant effect in the river runoff and the GW level result of SWAT-MODFLOW.
The largest changes in river runoff were associated with changes in the CN2 parameter. A CN2 decline of 25 % resulted in a 15 %
change in river runoff and a CN2 increase of 400 % resulted in 24 % increase in runoff. However, the 25 %–400 % CN2 changes will
only change GW level up to 5 %. LAT_TIME from SW parameters had the least impact on the model result in both river runoff and GW
level (Fig. 12). SLSBBSN and ALPHABF change by 25 % and 400 % had less than 2 % effect on the model results.
Parameters from group 2 (related to unsaturated zone) have more impact on river runoff than GW level. Based on Fig. 12, decrease
and increase in SOL_BD, result in less than 5 % changes in GW level and 10 %–18 % change in river runoff.
Changes in parameters related to the saturated zone (K and S), had the most impact on GW level (Fig. 12). The 25 % decrease and
400 % increase in K parameter, resulted in 15 % and 20 % change in GW level respectively and up to 10 % in river runoff. The changes
in S had similar impact on GW level, the 25 % decline and 400 % increase will cause 10 % and 15 % changes in GW level and up to 7 %
on river runoff.

4. Discussion

This research provides insights into the parameters that control runoff and GW level in a semi-arid area. These insights were gained
via a comprehensive SA that explored the sensitivity of SWAT-MODFLOW SW and GW parameters. The results of the SA deomstrate
that GW related parameters hydraulic conductivity (K) and storage coefficients (S) have impact on GW level as well as runoff. This
shows the importance of considering GW in hydrological simulations. Moreover, the results show except CN, other parameters related
to the surface water did not have a significant impact on GW level or runoff. However, curve number (CN2) from SW parameters and
moist bulk density (SOL_BD) and crack volume potential of soil (SOL_CRK) from unsaturated zone had the most impact on runoff. This
is an important finding because in the standalone SWAT model, unsaturated zone parameters have been found to have considerable
impact on runoff (e.g. Abbaspour, 2015). Overall, the results of the SA conducted in this research shows that CN2 from SWAT, K and S
from MODFLOW have the largest impact on both runoff and GW level and control the model result.
In this study the calibration of SWAT and MODFLOW parameters were conducted using GWCal, which is a tool developed to
facilitate and visualize the manual calibration of MODFLOW parameters along with SWATCUP for SWAT parameters calibration after
linking. In summary, the advantage of using the GWCal tool are: i) facilitation of the calibration process of MODFLOW parameters; ii)
improved efficiency of the process of SWAT-MODFLOW calibration; and iii) improvement in the performance of the SWAT-MODFLOW
model in GW simulation. The results demonstrate that the coupled SWAT-MODFLOW performance improves when GWCal is used. The
calibrated values of parameters related to SW-GW interaction with GWCal were considerably different to their default values in

15
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Fig. 10. Observed and simulated GW level using SWAT-MODFLOW model in observation bores.

Table 7
RMSE and MAE values in SWAT-MODFLOW.
Bore Number RMSE MAE Bore number RMSE MAE Bore number RMSE MAE

W1 1.21 1.18 W8 1.2 1.18 W15 1.50 1.29


W2 0.59 0.46 W9 0.93 0.86 W16 0.88 0.83
W3 2.45 2.28 W10 0.68 0.51 W17 1.21 1.17
W4 1.37 1.04 W11 1.05 0.89 W18 1.03 0.88
W5 2.43 2.3 W12 1.08 0.95 W19 2.45 2.29
W6 1.77 1.64 W13 1.04 0.88 W20 2.96 2.38
W7 0.92 0.87 W14 1.78 1.63 W21 3.13 2.48
Mean RMSE : 1.51 Mean MAE : 1.33

Table 8
SWAT-MODFLOW water balance.
Model Rainfall (MCM/Year) Evapotranspiration (MCM/Year) Infiltration (MCM/Year) Runoff (MCM/Year)

SWAT-MODFLOW 225.5 142.5 58.3 24.7

16
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Fig. 11. a) Initial K and b) Initial S distribution map and c) Calibrated K and d) Calibrated S obtained from SWAT-MODFLOW using GWCal.

standalone SWAT which represent an inherent limitation of SWAT in simulating GW. Based on this research, R2 and NSE for SWAT and
RMSE for MODFLOW change considerably after linking the two together in SWAT-MODFLOW. Indeed, calibrated values in the
standalone SWAT and MODFLOW, models are totally different to the calibrated values in SWAT-MODFLOW.
The results are consistent with previous research such as the SWAT-MODFLOW application in North Denmark in the Odderbak
basin (Molina-Navarro et al., 2019). Molina-Navarro et al. (2019) also stated that SWAT-MODFLOW demonstrated better results
compared to SWAT. Semiromi and Koch, 2019 calibrated SWAT and MODFLOW separately and linked together for the Gharehso Basin
in Iran and again found that the linked version performed better than SWAT in isolation. It should further be noted that, coupled
models like SWAT-MODFLOW have an increased number of parameters. Therefore, applying coupled models like SWAT-MODFLOW
requires the relevant input data and this will limit the application of SWAT-MODFLOW in areas that lack SW or GW data.

5. Conclusion

SW-GW interactions were studied for Shiraz basin in Iran using SWAT-MODFLOW. The SW and GW parameters have been cali­
brated and validated using two variables of river runoff and GW level.
The SWAT-MODFLOW calibrated values are considerably different to standalone SWAT in the SW parameters and standalone
MODFLOW in the GW parameters. This difference is the result of the exchange happening between SW and GW. Based on the result, the
calibration of SW parameters along with GW parameters in SWAT-MODFLOW significantly influenced the outputs of the model. These
results also demonstrate that the GWCal tool, developed to enable the manual calibration of MODFLOW and SWAT, improves the
efficiency and effectiveness of the calibration process which ultimately leads to improved performance of SWAT-MODFLOW.
This sensitivity analysis showed that curve number (CN2), GW hydraulic conductivity (K) and storage coefficients (S) are the most
influential parameters on GW level and runoff in the study catchment. This highlights that it is critical to consider and account for GW
when conducting regional scale hydrological simulations. The results also demonstrate that properly calibrating both SW and GW
components of the water cycle is critical.
The approach developed in this research, could be applied in other basins. Future work should be conducted to determine if
automatic calibration of GW parameters can further accelerate and/or improve the calibration of SWAT-MODFLOW.

Author statement

Tina Jafari: Conceptualization, Methodology, Analysis, Software, Visualization, Writing- Original draft preparation.
Anthony S. Kiem: Supervision, Conceptualization, Methodology, Reviewing and Editing.

17
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

(caption on next page)

18
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Fig. 12. SWAT-MODFLOW result of SA from SW parameters (CN2, LAT_TIME, ALPHA_BF, SLSUBBS), unsaturated zone parameters (SOL_CRK,
SOL_AWC, SOL_BD) and GW parameters (K, S, GW_DELAY) on two variables of river runoff and GW level.

Saman Javadi: Supervision, Methodology, Conceptualization, Data curation, Software, Validation, Reviewing and Editing.
Takashi Nakamura: Reviewing and Editing.
Kei Nishida: Reviewing and Editing.

Declaration of Competing Interest

The authors report no declarations of interest.

Acknowledgements

The authors would like to express their gratitude to Dr. Ryan T. Bailey from Colorado State University for the collaboration and for
providing the SWAT-MODFLOW linkage code.

Appendix A. Supplementary data

Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.ejrh.2021.
100822.

References

Abbas, S., Xuan, Y., Bailey, R., 2018. Improving river flow simulation using a coupled surface-groundwater model for integrated water resources management. EPiC
Ser. Eng. 3, 1–9.
Abbaspour, K., 2015. User Manual for SWAT-CUP: SWAT Calibration and Uncertainty Analysis Programs. Available:. Swiss Fed. Inst. of Aquatic Science and
Technology, Dübendorf, Switzerland http://www.eawag.ch/organization/abteilungen/siam/software/swat/index_EN/.
Abbaspour, K.C., Rouholahnejad, E., Vaghefi, S., Srinivasan, R., Yang, H., Klove, B., 2015. A continental-scale hydrology and water quality model for Europe:
calibration and uncertainty of a high-resolution large-scale SWAT model. J. Hydrol. 524, 733–752.
Aliyari, F., Bailey, R.T., Tasdighi, A., Dozier, A., Arabi, M., Zeiler, K., 2019. Coupled SWAT-MODFLOW model for large-scale mixed agro-urban river basins. Environ.
Model. Softw. 115, 200–210.
Arnold, J.G., Srinivasan, R., Muttiah, R.S., William, J.R., 1998. Large area hydrologic modeling and assessment part i model development. JAWRA 34 (1), 73–89.
https://doi.org/10.1111/j.1752-1688.1998.tb05961.x.
Arnold, J.G., Kiniry, J.R., Srinivasan, R., Williams, J.R., Haney, E.B., Neitsch, S.L., 2013. SWAT 2012 Input/Output Documentation. Texas Water Resources Institute.
Bailey, R., Rathjens, H., Bieger, K., Chaubey, I., Arnold, J., 2017. SWATMOD-Prep: graphical user interface for preparing coupled SWAT-MODFLOW simulations.
JAWRA 53 (2), 400–410.
Bergstrom, S., 1992. The HBV Model-Its Structure and Applications. Swedish Meteorological and Hydrological Institute Reports Hydrology.
Bergstrom, S., Forsman, A., 1973. Development of a conceptual deterministic rainfallrunoff model. Nordic Hydrol. 4, 147–170 (Munksgaard, Copenhagen, Denmark).
Beven, K.J., Kirkby, M.J., 1979. A physically based variable contributing area model of basin hydrology. Hydrol. Sci. Bull. 24 (1), 43–69.
Burnash, R.J.C., Ferral, R.L., McGuire, R.A., 1973. A Generalized Streamflow Simulation System Conceptual Modeling for Digital Computers. U.S. Department of
Commerce, National Weather Service, and State of California, Department of Water Resources.
Cacuci, D.G., Ionescu-Bujor, M., Navon, I.M., 2005. Sensitivity and Uncertainty Analysis, Volume II: Applications to Large-Scale Systems, Vol. 2. CRC press.
Chunn, D., Faramarzi, M., Smerdon, B., Alessi, D.S., 2019. Application of an integrated SWAT–MODFLOW model to evaluate potential impacts of climate change and
water withdrawals on groundwater–Surface water interactions in West-Central Alberta. Water 11 (1), 110.
Deb, P., Kiem, A.S., Willgoose, G.R., 2019a. Mechanisms influencing non-stationarity in rainfall-runoff relationships in southeast Australia. J. Hydrol. 571, 749–764.
https://doi.org/10.1016/j.jhydrol.2019.02.025.
Deb, P., Kiem, A.S., Willgoose, G.R., 2019b. A linked SW-GW modelling approach to more realistically simulate rainfall-runoff non-stationarity in semi-arid regions.
J. Hydrol. 575, 273–291. https://doi.org/10.1016/j.jhydrol.2019.05.039.
Diodato, D.M., 2000. Software spotlight (MicroFEM). Gr. Water 38 (5), 649–650.
Ebel, Loague, 2006. Near-surface hydrologic response for a steep, unchanneled catchment near Coos Bay, Oregon: 2. Comprehensive physics-based simulations.
American Journal of Science 307, 709–748.
Eini, M.R., Javadi, S., Delavar, M., Monteiro, J.A.F., Darand, M., 2019. High accuracy of precipitation reanalyses resulted in good river discharge simulations in a
semi-arid basin. Ecol. Eng. 131, 107–119.
Fars Regional Water Authority, 2015. Report of Water Budget in Shiraz Plain. Fars Provincial Water Authority, No 1, 22-23. Written in Persian. Only available via
formal written request from University of Tehran, Tehran. Iran to Fars Provincial Water Authority, Fars, Iran.
Fars Regional Water Authority, 2016. Report of GW and Geology in Maharloo Basin, Geological Survey of Iran, No 4, 122-231. Written in Persian. Only available via
formal written request from University of Tehran, Tehran. Iran to Fars Provincial Water Authority, Fars, Iran.
Food and agriculture organization of the United Nations (FAO), 2020, http://www.fao.org/home/en/.
Furman, 2008. Modeling coupled surface–subsurface flow processes: A review. Vadose Zone Journal 7, 741–756.
Gao, F., Feng, G., Han, M., Dash, P., Jenkins, J., Liu, C., 2019. Assessment of surface water resources in the big sunflower river watershed using coupled
SWAT–MODFLOW model. Water 11 (3), 528.
Gassman, P.W., Reyes, M.R., Green, C.H., Arnold, J.G., 2007. The soil and water assessment tool: historical development, applications, and future Research directions.
Trans. ASABE 50 (4), 1211–1250.
Gayathri, K. Devi, Ganasri, B.P., Dwarakish, G.S., 2015. A review on hydrological models. ICWRCOE Conference 1001–1007. https://doi.org/10.1016/j.
aqpro.2015.02.126. Aquatic Procedia 4.
Geological survey and mineral exploration of Iran, 2014, https://gsi.ir/en/map/1183/hydrogeology-and-environmental-geology-shiraz.
Graaf, I.E., van Beek, R.L., Gleeson, T., Moosdorf, N., Schmitz, O., Sutanudjaja, E.H., Bierkens, M.F., 2017. A global-scale two-layer transient groundwater model:
development and application to groundwater depletion. Adv. Water Resour. 102, 53–67.
Guo, J., Su, X., 2019. Parameter sensitivity analysis of SWAT model for streamflow simulation with multisource precipitation datasets. Nord. Hydrol. 50 (3), 861–877.

19
T. Jafari et al. Journal of Hydrology: Regional Studies 35 (2021) 100822

Guzman, J.A., Moriasi, D.N., Gowda, P.H., Steiner, J.L., Starks, P.J., Arnold, J.G., Srinivasan, R., 2015. A model integration framework for linking SWAT and
MODFLOW. J. Environ. Model. Softw. 73, 103–116. https://doi.org/10.1016/j. envsoft.2015.08.011.
Harbaugh, A.W., 2005. MODFLOW-2005, The US Geological Survey Modular Ground-Water Model: the Ground-Water Flow Process. US Department of the Interior,
US Geological Survey, Reston, VA, pp. 6–A16.
Harbaugh, A.W., Banta, E.R., Hill, M.C., McDonald, M.G., 2000. MODFLOW-2000, The U. S. geological survey modular ground-water model-user guide to
modularization concepts and the ground-water flow process. Open-file Report. U. S. Geological Survey (92), p.134.
Iran Ministry of Energy, 2016, http://www.greengoalsco.com/services/gis/2020-04-02-12-31-23.
Iran Water Resources Management, 2020, http://wrs.wrm.ir/m3/istgah_baranlist.asp?t=istgah_baran&RecPerPage=ALL.
Izady, A., Davary, K., Alizadeh, A., Ziaei, A.N., Akhavan, S., Alipoor, A., Joodavi, A., Brusseau, M.L., 2015. Groundwater conceptualization and modeling using
distributed SWAT-based recharge for the semi-arid agricultural Neishaboor plain. Iran. Hydrogeology journal 23 (1), 47–68.
Jackson, C.R., 2001. The development and validation of the object-oriented quasi three-dimensional regional groundwater flow model ZOOMQ3D. British Geological
Survey Internal Rep, pp. 57 IR/01/144.
Jafari, F., Javadi, S., Golmohammadi, G., Karimi, N., Mohammadi, K., 2016a. Numerical simulation of groundwater flow and aquifer-system compaction using
simulation and InSAR technique: Saveh basin, Iran. Environ. Earth Sci. 75 (9), 833.
Jafari, F., Javadi, S., Golmohammadi, G., Mohammadi, K., Khodadadi, A., Mohammadzadeh, M., 2016b. Groundwater risk mapping prediction using mathematical
modeling and the Monte Carlo technique. Environ. Earth Sci. 75 (6), 491.
Jiang, S., Ren, L., Xu, C.Y., Liu, S., Yuan, F., Yang, X., 2018. Quantifying multi-source uncertainties in multi-model predictions using the Bayesian model averaging
scheme. Nord. Hydrol. 49 (3), 954–970.
Kim, N.W., Chung, I.M., Won, Y.S., Arnold, J.G., 2008. Development and application of the integrated SWAT–MODFLOW model. J. Hydrol. 356, 1–16. https://doi.
org/10.1016/j.jhydrol.2008.02.024.
Kollet, Maxwell, 2006. Integrated surface–groundwater flow modeling: A free-surface overland flow boundary condition in a parallel groundwater flow model.
Advances in Water Resources 29, 945–958.
Krysanova, V., Wechsung, F., Arnold, J., Srinivasan, R., Williams, J., 2000. SWIM (Soil and Water Integrated Model) User’s Manual. Ver. SWIM-8. Potsdam Institute
for Climate Impact Research, Potsdam, Germany.
Lachaal, F., Mlayah, A., Bédir, M., Tarhouni, J., Leduc, C., 2012. Implementation of a 3-D groundwater flow model in a semi-arid region using MODFLOW and GIS
tools: The Zéramdine–Béni Hassen Miocene aquifer system (east-central Tunisia). Comput. Geosci. 48, 187–198.
Li, Kinzelbach, Brunner, Li, Dong, 2008. Topography representation methods for improving evaporation simulation in groundwater modeling. Journal of Hydrology
356, 199–208.
Liang, X., Lettenmaier, D.P., Wood, E.F., Burges, S.J., 1994. A simple hydrologically based model of land surface water and energy fluxes for general circulation
model. J. Geophys. Res. 99 (D7), 14,415–14,428. https://doi.org/10.1029/94JD00483.
Maxwell, R.M., Putti, M., Meyerhoff, S., Delfs, J.O., Ferguson, I.M., Ivanov, V., Kim, J., Kolditz, O., Kollet, S.J., Kumar, M., Lopez, S., 2014. Surface-subsurface model
intercomparison: A first set of benchmark results to diagnose integrated hydrology and feedbacks. Water Resour. Res. 50 (2), 1531–1549.
Meteorological organization of Iran, 2020, http://www.irimo.ir/far/wd/720-%D9%85%D8%AD%D8%B5%D9%88%D9%84%D8%A7%D8%AA-%D8%AE%D8%AF
%D9%85%D8%A7%D8%AA.html.
Molina-Navarro, E., Bailey, R.T., Andersen, H.E., Thodsen, H., Nielsen, A., Park, S., Jensen, J.S., Jensen, J.B., Trolle, D., 2019. Comparison of abstraction scenarios
simulated by SWAT and SWAT-MODFLOW. Hydrol. Sci. J. 64 (4), 434–454.
Moriasi, D.N., Arnold, J.G., Van Liew, M.W., Bingner, R.L., Harmel, R.D., Veith, T.L., 2007. Model evaluation guidelines for systematic quantification of accuracy in
watershed simulations. Trans. ASABE 50 (3), 885–900.
Nagelkerke, N.J., 1991. A note on a general definition of the coefficient of determination. Biometrika 78 (3), 691–692.
Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models part I — a discussion of principles. J. Hydrol. 10, 282–290.
Neitsch, S.L., Arnold, J.G., Kiniry, J.R., Williams, J.R., 2011. Soil and water assessment tool theoretical documentation version 2009. Texas Water Resources Institute
647.
Nguyen, V.T., Dietrich, J., 2018. Modification of the SWAT model to simulate regional groundwater flow using a multicell aquifer. Hydrol. Process. 32, 939–953.
Niswonger, R.G., Panday, S., Ibaraki, M., 2011. MODFLOW-NWT, a Newton formulation for MODFLOW-2005. US Geol. Surv. Techn. Methods 6 (A37), 44.
Paniconi, C., Putti, M., 2015. Physically based modeling in catchment hydrology at 50: survey and outlook. Water Resour. Res. 51 (9), 7090–7129.
Provisional water authority of Fars, 2016, https://www.frrw.ir/?l=EN.
Sebben, M.L., Werner, A.D., Liggett, J.E., Partington, D., Simmons, C.T., 2013. On the testing of fully integrated surface–subsurface hydrological models. Hydrolog.
Process. 27 (8), 1276–1285.
Semiromi, M.T., Koch, M., 2019. Analysis of spatio-temporal variability of surface–groundwater interactions in the Gharehsoo river basin, Iran, using a coupled
SWAT-MODFLOW model. Environ. Earth Sci. 78 (6), 201.
Sith, R., Watanabe, A., Nakamura, T., Yamamoto, T., Nadaoka, K., 2019. Assessment of water quality and evaluation of best management practices in a small
agricultural watershed adjacent to Coral Reef area in Japan. Agric. Water Manag. 213, 659–673.
Sun, Y., Kang, S., Li, F., Zhang, L., 2009. Comparison of interpolation methods for depth to groundwater and its temporal and spatial variations in the Minqin oasis of
northwest China. Environ. Model. Softw. 24 (10), 1163–1170.
Sun, X., Bernard-Jannin, L., Garneau, C., Volk, M., Arnold, J.G., Srinivasan, R., Sauvage, S., Sánchez-Pérez, J.M., 2016. Improved simulation of river water and
groundwater exchange in an alluvial plain using the SWAT model. Hydrol. Process. 30 (2), 187–202.
Surinaidu, L., Rao, V.G., Rao, N.S., Srinu, S., 2014. Hydrogeological and groundwater modeling studies to estimate the groundwater inflows into the coal Mines at
different mine development stages using MODFLOW, Andhra Pradesh, India. Water Resour. Ind. 7, 49–65.
Therrien, McLaren, Sudicky, Panday, 2010. A three-dimensional numerical model describing fully-integrated subsurface and surface flow and solute transport. user
guide.
Thode, R., Fredlund, M., 2013. 2D/3D Seepage Modeling Software (SVFLUX) Tutorial Manual. SoilVision Systems Ltd., Saskatoon, Saskatchewan, Canada.
U.S. Army Corps of Engineers, 2016. HEC-HMS Hydrologic Modeling System User’s Manual. Ver. 4.2. Hydrol. Eng. Center, Davis, CA.
United State Geological Survey (USGS), 2020, https://www.usgs.gov/centers/eros/science/usgs-eros-archive-land-cover-products-global-land-cover-
characterization-glcc?qt-science_center_objects=0#qt-science_center_objects.
VanderKwaak, J.E., 1999. Numerical simulation of flow and chemical transport in integrated surface-subsurface hydrologic systems. https://uwspace.uwaterloo.ca/
bitstream/handle/10012/412/NQ38276.pdf?sequence=1.
Verzano, K., 2009. Climate Change Impacts on Flood Related Hydrological Processes: Further Development and Application of a Global Scale Hydrological Model. Ph.
D. diss. International Max Planck Research School on Earth System Modelling. University of Kassel, Germany.
William, C., Pabst, A., Peters, J., 1995. The Hydrologic Modeling System (HEC-HMS): Design and Development Issues. Institute for Water Resources, Hydrol. Eng.
Center, Davis, CA.
Yu, D., Xie, P., Dong, X., Hu, X., Liu, J., Li, Y., Peng, T., Ma, H., Wang, K., Xu, S., 2018. Improvement of the SWAT model for event-based flood simulation on a sub-
daily timescale. Hydrol. Earth Syst. Sci. 22 (9), 5001–5019.
Zyvoloski, G.A., Robinson, B.A., Dash, Z.V., Kelkar, S., Viswanathan, H.S., Pawar, R.J., Stauffer, P.H., Miller, T.A., Chu, Sh, 2015. Software User’s Manual (UM) for the
FEHM Application Version 3.1-3.X. Los Alamos National Laboratory, United States.

20

You might also like