Modeling interbasin groundwater flow in karst areas: Model development, application, and calibration strategy
Introduction
The term “karst” refers to a region with distinct landscape features (e.g., sinking streams, sinkholes, and springs) and underground features (e.g., underground conduits and caves). In some karst regions, the karst landscape features could be absent or subtle, but their aquifers could be heavily karstified (Ford and Williams, 2007). Karst aquifers are developed as a result of dissolution of karstifiable rocks (e.g., limestone, dolomite, gypsum, and rock salt), the so-called karstification (Ford and Williams, 2007, Bögli, 1980, Howard, 1963). Karst aquifers account for about 10% to 15% of the continental area and karst groundwater is one of the sources of drinking water for approximately a quarter of the world’s population (Ford and Williams, 2007). However, karst groundwater is particularly vulnerable to contamination due to their distinct hydrogeologic characteristics (Goldscheider, 2005, Doerfliger et al., 1999, Drew and Hötzl, 1999). Therefore, understanding the hydrogeologic characteristics of karst aquifers plays an important role in water resources management in karst regions.
Hydrogeologic characteristics of karst aquifers are different from other aquifers (Bakalowicz, 2005). Karst aquifers often exhibit a duality of recharge, infiltration, porosity, flow and storage (Goldscheider and Drew, 2007, White, 2002, Gun, 1986). Karst aquifers also show a high degree of spatial heterogeneity in hydraulic properties (Bonacci et al., 2006). Especially, the surface drainage basin in karst aquifers usually do not coincide with the groundwater basin (Spangler, 2001, Dar et al., 2014). Karstification is considered as one of the most common causes of interbasin groundwater flow (IGF) (Le Moine et al., 2007). Water recharged to karst aquifers could flow through an underground conduit system spanning over several basins and emerge at springs located at distant sites (e.g., Anderson et al., 2006, Belcher et al., 2009, Le Moine et al., 2008). It should be noted that IGF could also occur in porous aquifer in form of regional groundwater flow (Tóth, 1963, Nguyen and Dietrich, 2018, Danapour et al., 2019), however, in this study we focus on IGF in karst areas. The term IGF in this study could be also understood as regional groundwater flow across surface topographic divides. IGF in karst areas could significantly alter the water budget of a basin (e.g., Anderson et al., 2006, Le Moine et al., 2008). Considering the aforementioned facts, IGF in karst areas should be accounted for in hydrological modeling, especially in the context of transboundary or interbasin groundwater management.
Various models have been used to simulate IGF in karst aquifers with varying model complexity, ranging from physically based distributed to conceptual lumped models. Physically based distributed models simulate groundwater flow based on hydraulic head gradient, therefore, groundwater could flow across topographic divide units, which are normally considered as isolated groundwater units in surface hydrology. Conceptual models can simulate IGF by allowing the simulation (or routing) of groundwater flow between topographical basins. Some models of these types are the Modular Three-Dimensional Finite-Difference Ground-Water Flow Model (MODFLOW, Scanlon et al., 2003), the modified WetSpa model (Liu et al., 2005), the modified Soil and Water Assessment Tool (SWAT, Arnold et al., 1998, Nerantzaki et al., 2015, Malagó et al., 2016, Palanisamy and Workman, 2014), modèle du Génie Rural à 4 paramètres Journalier (GR4J, Perrin et al., 2003, Le Moine et al., 2007, Le Moine et al., 2008), the tank model (Anaya and Wanakule, 1993), and the multi-cell aquifer model (Rozos and Koutsoyiannis, 2006, Barrett and Charbeneau, 1997). SWAT is one of the most widely-used models to simulate the effect of land use, agricultural management practices and climate change on water and chemical yields in non-karst areas (Arnold and Fohrer, 2005, Gassman et al., 2007, Krysanova and White, 2007, Molina-Navarro et al., 2017). Therefore, the modified SWAT versions which account for IGF in karst areas could potentially help to explore these effects in karst regions.
The aforementioned modified SWAT models, the so-calledKarstSWAT (Palanisamy and Workman, 2014) and KSWAT (Nerantzaki et al., 2015, Malagó et al., 2016), simulate IGF in karst regions. The KarstSWAT model was specifically developed for watersheds dominated by sinkholes and springflow, which is mainly fed by the water from sinkholes (Palanisamy and Workman, 2014). The KSWAT model combines the adapted SWAT model (Fig. 3, Malagó et al., 2016) and the karst-flow model (Nikolaidis et al., 2013). The adapted SWAT model assumes that all water entering the soil profile is karst groundwater recharge (Fig. 3, Malagó et al., 2016). However, part of the infiltrated water could contribute to the streamflow as lateral flow and baseflow if the underlying aquifer of a subbasin is not entirely a karst aquifer (e.g., Palanisamy and Workman, 2014). The adapted SWAT model does not differentiate between concentrated recharge and diffuse recharge. The karst-flow model is the two-linear-storage reservoir model, which receives the recharge simulated from the adapted SWAT model (or from the original SWAT model, Nikolaidis et al., 2013) and routes it to spring. Outflows from the two reservoirs of the karst-flow model represent flow from wide conduits and narrow fractures (Kourgialas et al., 2010, Malagó et al., 2016). Because of the lumped feature of deep recharge from the adapted SWAT model, the KSWAT model does not explicitly differentiate between (1) the diffuse recharge and concentrated recharge, (2) between matrix storage and conduit storage. This is important because these recharges and storages are different in term of travel time and storage. In addition to the aforementioned disadvantages, the recharge area of the karst aquifer in the KarstSWAT and KSWAT models follows the subbasin delineation of SWAT.
In addition to the model development, parameter identification in karst regions is also subject to higher uncertainty compared to other regions (Brenner et al., 2018, Hartmann et al., 2017, Hartmann et al., 2013). This is because the karst aquifer is highly heterogeneous and the upper flux (actual evapotranspiration, ETa) and the lower flux (karst groundwater recharge) are usually unknown. In order to develop a robust model and to minimize the parameter uncertainty, especially in karst regions, multi-variable calibration is suggested. ETa is one of the main components of the hydrologic cycle. About 60% of the annual precipitation on the global land surface returns to the atmosphere as evapotranspiration (Jung et al., 2010, Oki and Kanae, 2006). Considering the aforementioned facts, observed ETa should be used for calibrating the model. However, a direct observation of ETa is very scarce.
In non-karst areas, many studies have used satellite-derived ETa for model calibration (e.g., Rajib et al., 2018, Franco and Bonumá, 2017, Vervoort et al., 2014, Rientjes et al., 2013, Droogers et al., 2010, Zhang et al., 2009, Muthuwatta et al., 2009, Immerzeel and Droogers, 2008). In these studies, satellite-derived ETa was either used as an independent calibration data set or as input data. Results showed that the model performance for streamflow could decrease when constraining model calibration with satellite-derived ETa as an additional variable (Vervoort et al., 2014). However, the above-mentioned studies showed that using satellite-derived ETa in combination with observed streamflow for calibrating a hydrologic model could (1) better reproduce the catchment’s water balance, (2) reduce the parameter uncertainty, (3) increase the model robustness, and (4) detect the structural model issues. In karst areas, the use of satellite-derived ETa as an additional calibration variable has not been given enough attention.
In this study, we developed a conceptual model which is able to (1) simulate surface and subsurface flows in both karst and non-karst areas, (2) apply for a region where the karst aquifer boundaries do not coincide with the surface subbasin boundaries, and (3) represent different recharges (diffuse recharge and concentrated recharge) and storages (matrix storage and conduit storage) in karst areas. The proposed concept was implemented into the SWAT model. The modified SWAT model was tested in the karst-dominated area in Lower Saxony, Germany. The effects of using satellite-derived ETa for model calibration on the model performance was examined in detail. The Moderate Resolution Imaging Spectroradiometer (MOD16 ETa, Mu et al., 2013) was used for the model calibration.
Section snippets
The original SWAT model
In SWAT, a basin can be divided into subbasins, which are further divided into Hydrologic Response Units (HRUs). HRUs are created by lumping all areas having the same combination of land use, soil type and slope within a subbasin. The HRU concept is computationally efficient while incorporating the aforementioned landscape properties. SWAT simulates two phases of the hydrologic cycle, the land phase and the routing phase. The land phase includes HRU-related processes such as surface processes
Study area and data
The study area is located in the southwest Harz Mountains (non-karst area) and the southern Harz rim (karst-dominated area) in Northern Germany with a drainage basin of about 384 km (Fig. 2). The study area has two outlets located at the Rhume spring and Lindau gauging stations. The study area receives inflow from the Odertalsperre reservoir. The Digital Elevation Model (DEM) obtained from the Niedersächsische Landesbetrieb für Wasserwirtschaft, Küsten-und Naturschutz (NLWKN) shows that the
Model setup
The study area was divided into 26 subbasins and 1094 HRUs based on LULC, soil, DEM and aquifer map (Fig. 3). The thresholds for defining HRUs were set to zero to include all of the basin landscape. The SWAT_IGF model uses the conceptual model presented in Fig. 1A for the southwest Harz Mountains and the conceptual model presented in Fig. 1B for the southern Harz rim (Fig. 3C). Infiltration losses () and river transmission losses () from the karst area located outside the recharge
Sensitivity analysis and best calibrated parameter set
Table 1 shows the results of global sensitivity analysis for 21 model parameters. Parameter sensitivity ranking was based on the values of t-stat and p-value. It is seen that CN2 is the most sensitive parameter. This indicates that streamflow, karst groundwater recharge, and evapotranspiration are strongly affected by the surface runoff generation process. The parameter CH_K2 (riverbed hydraulic conductivity) is listed among the most sensitive parameters. This is because river transmission
Conclusions and recommendations
Interbasin groundwater flow (IGF), especially in karst areas, could significantly alter the water budget of a region. In this study, the original SWAT model was modified for simulating IGF in karst areas, resulting in the SWAT_IGF model. A two-linear-reservoir model was proposed to represent the duality of recharge, infiltration, storage, and discharge in the karst area. The study area is located in a karst-dominated region in the southwest Harz Mountains, Germany. The model was successfully
Acknowledgments
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. We thank the Editor and three anonymous reviewers for their constructive comments, which helped to improve the quality of the manuscript significantly. We also thank the DWD, HWW, BGR, and NLWKN for providing data.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in
References (95)
- et al.
Modelling hydrology and water quality in the pre-alpine/alpine Thur watershed using SWAT
J. Hydrol.
(2007) - et al.
Interbasin flow revisited: the contribution of local recharge to high-discharge springs, Death Valley, CA
J. Hydrol.
(2006) - et al.
A parsimonious model for simulating flow in a karst aquifer
J. Hydrol.
(1997) - et al.
Interbasin flow in the great basin with special reference to the southern Funeral Mountains and the source of Furnace Creek springs, Death Valley, California, US
J. Hydrol.
(2009) - et al.
Estimating actual irrigation application by remotely sensed evapotranspiration observations
Agricult. Water Manag.
(2010) - et al.
Decomposition of the mean squared error and NSE performance criteria: Implications for improving hydrological modelling
J. Hydrol.
(2009) - et al.
Calibration of a distributed hydrological model based on satellite evapotranspiration
J. Hydrol.
(2008) - et al.
An integrated framework for the hydrologic simulation of a complex geomorphological river basin
J. Hydrol.
(2010) - et al.
The isotopic composition of strontium and sulfur in seawater of Late Permian (Zechstein) age
Chem. Geol.
(1991) - et al.
Assessing the spatial and temporal variations of water quality in lowland areas, northern Germany
J. Hydrol.
(2012)
Regional scale hydrologic modeling of a karst-dominant geomorphology: the case study of the Island of Crete
J. Hydrol.
The impact of the objective function in multi-site and multi-variable calibration of the SWAT model
Environ. Modell. Softw.
River flow forecasting through conceptual models. Part 1: a discussion of principles
J. Hydrol.
Modeling suspended sediment transport and assessing the impacts of climate change in a karstic Mediterranean watershed
Sci. Total Environ.
Hydrologic and geochemical modeling of a karstic Mediterranean watershed
J. Hydrol.
Improvement of a parsimonious model for streamflow simulation
J. Hydrol.
Hydrologic model predictability improves with spatially explicit calibration using remotely sensed evapotranspiration and biophysical parameters
J. Hydrol.
Multi-variable calibration of a semi-distributed hydrological model using streamflow data and satellite-based evapotranspiration
J. Hydrol.
A multicell karstic aquifer model with alternative flow equations
J. Hydrol.
Can we simulate regional groundwater flow in a karst system using equivalent porous media models? Case study, Barton Springs Edwards Aquifer, USA
J. Hydrol.
Evaluation of SWAT simulated soil moisture at catchment scale byfield measurements and landsat derived indices
Agricult. Water Manag.
Karst hydrology: recent developments and open questions
Eng. Geol.
SWAT-CUP 2012. SWAT Calibration and Uncertainty Program—A User Manual
Estimating uncertain flow and transport parameters using a sequential uncertainty fitting procedure
Vadose Zone J.
A guideline for successful calibration and uncertainty analysis for soil and water assessment: A review of papers from the 2016 international SWAT conference
Water
A penman for all seasons
J. Irrig. Drain Eng. ASCE
Operational estimates of reference evapotranspiration
Agron. J.
A Lumped Parameter Model for the Edwards Aquifer
SWAT2000: Current capabilities and research opportunities in applied watershed modelling
Hydrol. Process.
SWAT: model use, calibration, and validation
Trans. ASABE
Large area hydrologic modeling and assessment. Part I: model development
J. Am. Water Resour. Assoc.
Karst groundwater: a challenge for new resources
Hydrogeol. J.
Chloride circulation in a lowland catchment and the formulation of transport by travel time distributions
Water Resour. Res.
Karst Hydrology and Physical Speleology
Karst flash floods: an example from the dinaric karst (Croatia)
Nat. Hazards Earth Syst. Sci.
The stable isotopic geochemistry of the sulfur and carbon cycles in a modern karst environment
Isot. Environ. Health Stud.
Catchment residence and travel time distributions: The master equation
Geophys. Res. Lett.
Process-based modelling to evaluate simulated groundwater levels and frequencies in a chalk catchment in south-western England
Nat. Hazards Earth Syst. Sci.
Assessment of regional inter-basin groundwater flow using both simple and highly parameterized optimization schemes
Hydrogeol. J.
Review: carbonate aquifers and future perspectives of karst hydrogeology in India
Hydrogeol. J.
Water vulnerability assessment in karst environments: a new method of defining protection areas using a multi-attribute approach and GIS tools (EPIK method)
Environ. Geol.
Karst Hydrogeology and Geomorphology
Multi-variable SWAT model calibration with remotely sensed evapotranspiration and observed flow
Brazilian J. Water Resour
The soil and water assessment tool: historical development, applications, and future research directions
Trans. ASABE
Die Wasserbilanz der Rhumequelle. Bericht Wasserwirtschaftsamt Göttingen
Karst groundwater vulnerability mapping: application of a new method in the Swabian Alb, Germany
Hydrogeol. J.
Cited by (18)
Deciphering inter-catchment groundwater flow: A water balance perspective in the Choshui River Basin, Taiwan
2024, Journal of Hydrology: Regional StudiesAn interactive graphical interface tool for parameter calibration, sensitivity analysis, uncertainty analysis, and visualization for the Soil and Water Assessment Tool
2022, Environmental Modelling and SoftwareCitation Excerpt :Each file includes different parameters located in different lines with different number formats. In addition, new input files and new parameters can be introduced as SWAT is constantly being developed for specific purposes (e.g., Nguyen et al., 2020; Nguyen and Dietrich, 2018). Therefore, this study developed a general approach that allows modification of any parameter defined via the text file “swatParam.txt” (please see the supporting material R-SWAT.pdf).
Land-use changes and precipitation cycles to understand hydrodynamic responses in semiarid Mediterranean karstic watersheds
2022, Science of the Total EnvironmentCitation Excerpt :Recent studies adopted past and present land-use cover conditions to project extreme future scenarios, which can be used as inputs to design new hydrological models (García-Ruiz and Lana-Renault, 2011; Boongaling et al., 2018; Rafiei-Emam et al., 2016). Among different approaches, GIS-based hydrological modeling systems are becoming increasingly relevant tools because of their ability to handle spatial variations at the basin scale with a huge number of physiographic inputs (Nguyen et al., 2020; Senent-Aparicio et al., 2020; Jodar-Abellan et al., 2018). Several models have been developed such as SWAT (Arnold et al., 1998; Arnold et al., 2012), MODFLOW (Mcdonald and Harbaugh, 1984) or HEC-RAS (Brunner and Bonner, 1994; Ferreira et al., 2020).
Development of alternative SWAT-based models for simulating water budget components and streamflow for a karstic-influenced watershed
2020, CatenaCitation Excerpt :The standard SWAT model was applied in 23 studies versus the application of modified SWAT models in 11 studies (Table 1); four studies reported applications of both the standard SWAT model and a modified SWAT model. The influence of external or interbasin groudwater flows were accounted for in 11 studies, which included simulations with the standard SWAT model and/or modified SWAT models: Salerno and Tartari (2009), Jiang et al. (2011), Palazón and Navas (2013), Palanisamy and Workman (2014), Gamvroudis et al. (2015), Nerantzaki et al. (2015; 2020), Amin et al. (2017), Delavar et al. (2020), Nguyen et al. (2020) and Senent-Aparicio et al. (2020). The majority of the studies reported Nash-Sutcliffe Efficiency (NSE) statistics (Krause et al., 2005) for daily and/or aggregated monthly comparisons between simulated and measured steam flows (Table 1), which are considered satisfactory if NSE ≥ 0.50 based on suggested criteria by Moriasi et al. (2007; 2015).
- 1
Present address: Department of Hydrogeology, Helmholtz Centre for Environmental Research - UFZ, Leipzig, Germany.