water

Article Hydrological Modeling to Assess the Efficiency of Groundwater Replenishment through Natural Reservoirs in the Hungarian Drava River Floodplain

Ali Salem 1,2,* ,József Dezs˝o 3, Mustafa El-Rawy 2,4 and Dénes Lóczy 3 1 Doctoral School of Earth Sciences, University of Pécs, Ifjúság útja 6, H-7624 Pécs, Hungary 2 Civil Engineering Department, Faculty of Engineering, Minia University, Minia 61111, Egypt; [email protected] 3 Institute of Geography and Earth Sciences, Faculty of Sciences, University of Pécs, Ifjúság útja 6, H-7624 Pécs, Hungary; [email protected] (J.D.); [email protected] (D.L.) 4 Civil Engineering Department, College of Engineering, Shaqra University, Dawadmi, Ar Riyadh 11911, Saudi Arabia * Correspondence: [email protected]; Tel.: +36-705590080



Received: 5 December 2019; Accepted: 13 January 2020; Published: 16 January 2020


Abstract: Growing drought hazard and water demand for agriculture, ecosystem conservation, and tourism in the Hungarian Drava river floodplain call for novel approaches to maintain wetland habitats and enhance agricultural productivity. Floodplain rehabilitation should be viewed as a complex landscape ecological issue which, beyond water management goals to relieve water deficit, ensures a high level of provision for a broad range of ecosystem services. This paper explores the hydrological feasibility of alternative water management, i.e., the restoration of natural reservoirs (abandoned paleochannels) to mitigate water shortage problems. To predict the efficiency of the project, an integrated surface water (Wetspass-M) and groundwater model (MODFLOW-NWT) was developed and calibrated with an eight-year data series. Different management scenarios for two 3 1 3 1 natural reservoirs were simulated with filling rates ranging from 0.5 m s− to 1.5 m s− . In both 3 1 instances, a natural reservoir with a feeding rate of 1 m s− was found to be the best scenario. In this case 14 days of filling are required to reach the possible maximum reservoir stage of +2 m. The first meter rise increases the saturation of soil pores and the second creates an open surface water body. Two filling periods per year, each lasting for around 180 days, are required. The simulated water balance shows that reservoir–groundwater interactions are mainly governed by the inflow into and outflow from the reservoir. Such an integrated management scheme is applicable for floodplain rehabilitation in other regions with similar hydromorphological conditions and hazards, too.

Keywords: floodplain rehabilitation; hydrological modeling; reservoirs; surface water-groundwater interactions; Drava river floodplain

1. Introduction Water availability is of primary importance for landscape functions. Over the last centuries, half of European wetlands and more than 95% of riverine floodplains were converted to agricultural and urban lands [1]. The human intervention has fundamentally altered the water availability of floodplain landscapes [2]. River floodplains provide various ecosystem services [3,4], including groundwater replenishment, water purification, flood protection, biodiversity/habitat, sediment and nutrient retention, recreation and tourism, cultural values, buffering capacity of climate change, and various floodplain products (fish, wood, reed, game) [5]. Changes in the water regimes of rivers are predicted to be damaging to these ecosystems under extensive human pressure, resources exploitation,

Water 2020, 12, 250; doi:10.3390/w12010250 www.mdpi.com/journal/water Water 2020, 12, 250 2 of 20 and pollution. Abandoned channels and oxbow lakes, the most common landforms of floodplains, are regarded particularly sensitive to such pressures [6,7]. In the Drava floodplain of SW Hungary drought and flood periods alternate rapidly, occasionally even within a single year [8]. With groundwater tables dropping by 1.5 to 2.5 m, the entrenchment of the river leads to intensified drought hazard and water stress in the floodplain [9–11]. Water shortages, particularly manifested in decreased river discharge and wetlands desiccation, are aggravated by the aridification trend induced by global climate change [12]. As a consequence, the ecological functioning of the natural systems and physical habitats are deteriorated. Moreover, water shortages have adverse effects on agricultural productivity, related socio-economic conditions, and the life quality of local population [13]. The optimal design of groundwater table in the floodplain is a complex task because of the conflicts between agriculture, forestry, flood control, and natural conservation demands [8]. To amend the situation, a large-scale landscape rehabilitation project, the Old Drava Programme, was launched in 2013 [14]. It is intended to solve interrelated problems of water governance, land management, and human employment. In the Drava basin alternative water resources have to be explored. International examples show that the integrated (conjunctive) use of surface water and groundwater within a proper management system increases the efficiency of water utilization [15,16], provides sufficient irrigation water for agriculture [17–19], and improves the environmental conditions of irrigated lands [20]. Construction of recharge dams and controlling the rate of groundwater abstraction are considered common practices [21,22]. Managed Aquifer Recharge (MAR) using treated wastewater is a feasible option for non-conventional water resources in arid and semiarid regions [18,23–29]. The interaction between lakes (subsurface reservoir) and groundwater is the most important part of the lake water budget and it substantially affects lake water quality [30–32]. This interaction has been studied through different approaches. The exchange between reservoir/lake water and groundwater is controlled by the texture, stratification, and isotropy of sediments in the shoreline zone [33,34]. Lee [35], Winter et al. [36], and Ong and Zlotnik [37] used point measurement tools such as seepage meters, piezometers, dye tracing, and potentio-manometers to estimate the exchange of seepage from the lake to groundwater. The difficulty of such methods lies in the spatial and temporal integration of fluxes. Geochemical analyses and mass-balance methods quantify groundwater inflow, which determines isotope or solute concentration in the lake [38,39]. The latter methods are complex, as they need a conservative tracer unaffected by land use [40–43]. Isotope-balance approaches [44] using δ2H or δ18O measurements of water were used in several hydrological studies to assess groundwater inflow to lakes, with varying degrees of success [40,45–48]. Hydrological models are the most reliable tools for the assessment of lake water/groundwater interactions [49]. A Modular Three-Dimensional Finite-Difference Ground-Water Flow Model (MODFLOW) [50] is applied widely to assess this interaction from different approaches. Leblanc et al. [51] simulated the lake as a constant stage source or sink through river package. Another approach to simulate the water balance of lakes uses the General-Head Boundary (GHB) Package [52]. The main disadvantage of these approaches is that the lake stage is assumed to be constant during each stress period. Fenske et al. [53] used the Reservoir Package to simulate the lake. This approach requires previous knowledge of lakebed seepage to assess the lake stages [54]. Another approach that represents the lake volume with high hydraulic conductivity cells [55] is called the “high-K” technique. Yihdego and Becht [56] applied a 3D flow model with the high-K technique to simulate lake/groundwater interaction in Naivasha, Kenya. In this approach, however, it is difficult to simulate the connection between the lakes and streams [54]. Cheng and Anderson [57] developed the MODFLOW lake package (LAK1) to solve the mentioned problems. The main advantage of the lake package is that it allows the fluctuation of lake stage. Also, the exchange of volumetric water between the lake and the aquifer is determined [49]. The LAK2 package is capable of simulating more than one lake simultaneously. The simulation of solute transport is also provided in LAK. Kidmose et al. [58] validated the LAK3 package through measurements by seepage meter. The advantages and disadvantages of Water 2020, 12, 250 3 of 20 three approaches for simulating lake–groundwater interactions (namely, fixed lake stages, high-K nodes, and LAK3 package) were compared by Hunt [59]. Water balance is simulated more realistically in the latest update LAK7 package [49]. The experience gathered from the MODFLOW lake packages is utilized in the modeling of water balance for the reservoir planned in the Drava floodplain. ThisWater paper 2020, evaluates12, x FOR PEERthe REVIEW feasibility of water replenishment through natural reservoirs3 of 21 using a hydrological model (Wetspass-M) and a Newton-Raphson formulation for MODFLOW-2005 more realistically in the latest update LAK7 package [49]. The experience gathered from the (MODFLOW-NWT) with ModelMuse as a graphical interface. The consequences of applying a natural MODFLOW lake packages is utilized in the modeling of water balance for the reservoir planned in reservoirthe in augmentingDrava floodplain. surface water storage are investigated under variable management scenarios of recharging orThis feeding paper the evaluates reservoir. the Thefeasibility assessment of water of replenishment such reservoir through/groundwater natural reservoirs interactions using a according to differenthydrological scenarios model of reservoir (Wetspass-M) recharge and is a preconditionNewton-Raphson to establishingformulation for a newMODFLOW-2005 water governance in the region,(MODFLOW-NWT) which can, in turn, with be ModelMuse the basis of as exploiting a graphical economic interface. opportunities.The consequences Sustainable of applying ecotourisma natural reservoir in augmenting surface water storage are investigated under variable management and relatedscenarios development of recharging would or feeding ensure the a safe reservoir. livelihood The assessment for the local of such population. reservoir/groundwater interactions according to different scenarios of reservoir recharge is a precondition to establishing a 2. Materialsnew andwater Methods governance in the region, which can, in turn, be the basis of exploiting economic opportunities. Sustainable ecotourism and related development would ensure a safe livelihood for 2.1. Studythe Area local population.

The2. 15 Materials km-wide and DravaMethods river floodplain is located on the national border between Hungary and Croatia. The Hungarian section of the floodplain is traditionally affected by floods prior to river regulation.2.1. Study FloodsArea mostly take place during June–July due to an Atlantic influence and during October–NovemberThe 15 km-wide due to Drava a Mediterranean river floodplain influence. is located on Drava the national river border channelization between Hungary began and in 1750 and reduced theCroatia. river The length Hungarian by 50% section [60]. of The the floodplainfloodplain is has traditionally a total area affected of 500 by kmfloods2. There prior to are river 13 tributary streams, 20regulation. major sideFloods channels, mostly take hundreds place during of meander June–July scars,due to andan Atlantic 18 oxbow influence lakes and in theduring Hungarian October–November due to a Mediterranean influence. Drava river channelization began in 1750 and section ofreduced the Drava the river River length (Figure by 50%1 [60].)[ 61 The]. Thefloodplain long-term has a total average area of 500 discharge km2. There of are the 13 Drava tributary River was 3 1 595 m s−streams,for the 20 periodmajor side 1896–2018 channels, hundreds [62]. Maximum of meander discharge scars, and of18 oxbow 0.1% probabilitylakes in the Hungarian is assumed to be 3 1 3 1 3 1 3070 m ssection− (reached of the Drava during River the (Figure flood 1) of[61]. 1827), The lo whileng-term 170 average m sdischarge− and 900of the m Dravas− Riverare accounted was to 3 −1 baseflow595 and m bankfull s for the discharge, period 1896–2018 respectively [62]. Maximum [62]. Asdischa a possiblerge of 0.1% source probability of water is assumed replenishment, to be the 3 −1 3 −1 3 −1 3070 m s (reached during the flood of 1827), while 170 m s and 900 m 3s are1 accounted to Fekete-vízbaseflow stream and is abankfull tributary discharge, withan respectively average [62]. discharge As a possible of Qav source= 2–2.5 of water m sreplenishment,− . The main the features of the floodplainFekete-víz are stream oxbows, is a scrolltributary bar with systems, an average abandoned discharge of channels Qav = 2–2.5 with m3 s natural−1. The main levees, features backswamps, of and in thethe north, floodplain elongated are oxbows, blown scroll sand bar systems, of east aban to westdoned strike channels [63 ].with The natu digitalral levees, elevation backswamps, model (DEM) of the studyand areain the wasnorth, obtained elongated fromblown thesand South-Transdanubian of east to west strike [63]. The Water digital Management elevation model Directorate, (DEM) Pécs, at 10 m resolution.of the study area The was Drava obtained floodplain from the South-Tr is apparentlyansdanubian flat Water at 83–133 Management m elevation, Directorate, with Pécs, an average at 10 m resolution. The Drava floodplain is apparently flat at 83–133 m elevation, with an average 2 relative reliefrelative 2 mrelief km 2− m km(Figure−2 (Figure1). Settlements1). Settlementswere were built built on on elevated elevated terrain. terrain.

Figure 1. Location of the Drava River floodplain, SW Hungary, with water-courses, existing (Cún-Szaporca) lake, and topography. Water 2020, 12, x FOR PEER REVIEW 4 of 21

Figure 1. Location of the Drava River floodplain, SW Hungary, with water-courses, existing (Cún- Water 2020, 12, 250 4 of 20 Szaporca) lake, and topography.

2.1.1.2.1.1. Meteorological Meteorological Data Data TheThe meteorological meteorological data data for the for period the period 2010–2018 2010–2018 (time selected (time asselected this study as this period) study were period) obtained were fromobtained the South-Transdanubian from the South-Transdanubian Water Management Water Directorate,Management P éDirectorate,cs. The climate Pécs. of The the Dravaclimate Plain of the is moderatelyDrava Plain wet is moderately and warm wet in the and west warm under in the Atlantic west under influence, Atlantic and influe moderatelynce, and dry moderately and warm dry in theand east warm under in the Mediterranean east under Mediterranean influence [64]. influence The long-term [64]. The mean long-term annual mean temperature annual istemperature 10.8 ◦C in theis 10.8 east °C and in the 10.2 east◦C inand the 10.2 west. °C in Absolute the west maximum. Absolute andmaximum minimum and temperaturesminimum temperatures are 35 ◦C and are 35 17°CC, and respectively. −17 °C, respectively. The number The of number cold days of cold (mean days daily (mean temperature daily temperature below 10 belowC) is − 10,10 °C) while is 10, − ◦ − ◦ thewhile number the of number hot days of (meanhot days temperature (mean temperature above 30 ◦ aboveC) is 21. 30 The°C) long-termis 21. The long-term average annual average rainfall annual 1 −1 1 −1 1 −1 rangesrainfall from ranges 383 mm from y− 383to 1057 mm mmy to y− 1057, with mm a meany , with value a mean of 682 value mm y −of .682 The mm rainy y season. The rainy (summer season and(summer spring) hasand a spring) share of has 60% a ofshare the annualof 60% rainfall,of the annual while rainfall, the remaining while 40%the remaining falls in the 40% drier falls season in the (winterdrier andseason autumn) (winter [65 and]. Daily autumn) evaporation [65]. Daily from evapor the lakeation surface from the was lake calculated surface usingwas calculated the Penman using open-waterthe Penman equation open-water [66] based equation on wind [66] speed,based on hourly wind incoming speed, hourly solar radiation,incoming airsolar temperature, radiation, air relativetemperature, humidity, relative air temperature, humidity, andair temperature, wind-speed. Evaporationand wind-speed. from Evaporation the lake is concentrated from the lake on is 1 1 −1 theconcentrated summer months on the and summer shows amonths large variation and shows between a large 0.0 variation mm d− betweenand 6.8 mm 0.0 mm d− , d with and a mean 6.8 mm −1 1 −1 valued , of with 1.85 a mmmean d− value. Daily of 1.85 potential mm d evapotranspiration. Daily potential evapotranspiration (PET) was computed (PET) from was meteorological computed from datameteorological using the Food data and using Agriculture the Food Organization and Agricult ofure the UnitedOrganization Nations of (FAO) the United Penman–Monteith Nations (FAO) methodPenman–Monteith [67] and varies method from 0.0 [67] mm and to varies 46 mm, from with 0.0 an mm average to 46 mm, of 19.8 with mm. an Aboutaverage 88% of 19.8 of the mm. PET About is attributed88% of the to summer PET is attributed and spring. to summer and spring.

2.1.2.2.1.2. Soil Soil Data Data Altogether,Altogether, 72 72 soil soil samples samples were were taken taken from from the the floodplain floodplain for for particle particle size size analysis analysis and and the the identificationidentification of texturalof textural classes classes (Figure (Figure2a). 2a). Over Over around around 50% 50% of theof the floodplain floodplain soils soils are are ofsandy of sandy texture,texture, in 12% in 12% clay, clay, in 8% in loam, 8% whileloam, inwhile over in 30% over of the 30% area of sandythe area loam sandy and clayloam loam and occur. clay loam Fluvisols occur. formedFluvisols under formed periodical under water periodical surplus and water forest surplus vegetation and dominateforest vegetation the area. Soildominate thickness the is area. highly Soil variable,thickness from is ahighly few centimeters variable, from to a meter,a few thecentimeters top 0.5 m to has a moremeter, organic the top material. 0.5 m has In additionmore organic to particlematerial. size distribution,In addition to soils particle were size studied distribution, using GPR soils (ground were penetratingstudied using radar) GPR and (ground satellite penetrating image analysis,radar) inand combination satellite image with analysis, the survey in combination of landforms with [68,69 the]. survey of landforms [68,69].

(a) (b)

FigureFigure 2. (2.a) ( Spatiala) Spatial distribution distribution of soilof soil textures textures in thein the Drava Drava floodplain, floodplain, (b) ( verticalb) vertical distribution distribution of of sedimentsediment textures textures in the in top the 100 top m. 100 ClMu: m. ClMu: clayey clayey mud; FiS:mud; fine FiS: sand; fine COMedSa: sand; COMedSa: coarse to coarse medium to sand.medium sand. Figure2b shows the vertical distribution and frequency of the di fferent textural types and their thicknessesFigure by 2b a boxshows diagram the vertical for the distribution top 100 m. and The frequency CMu (clayey of the mud) different layers textural mostly types occurred and attheir aroundthicknesses 25–30 mby depths a box diagram (ranging for from the 20 top to 100 40 m) m. andThe theirCMu average (clayey thicknessmud) layers was mostly 5 m (2 occurred to 8 m). at Allaround texture 25–30 classes m turned depths out (ranging in the topfrom 20 20 m to of 40 the m) modeling and their space. average These thickness sediments was were5 m (2 deposited to 8 m). All in pointtexture bars, classes channels, turned and out oxbows in the [top69]. 20 During m of the the modeling late Pleistocene–Holocene space. These sediments in the top were 100 deposited m fluvial in coarse-sand–heavypoint bars, channels, clay seriesand oxbows deposited [69]. in During the basin the (Figure late Pleistocene–Holocene2b). Subsurface water in flow the top and 100 capillary m fluvial rise are influenced by the above-mentioned landforms and deposits.

Water 2020, 12, x FOR PEER REVIEW 5 of 21 coarse-sand–heavy clay series deposited in the basin (Figure 2b). Subsurface water flow and capillary riseWater are2020 influenced, 12, 250 by the above-mentioned landforms and deposits. 5 of 20

2.1.3. Land Use and Groundwater Depth 2.1.3. Land Use and Groundwater Depth Land use data for the Drava floodplain in 2018 were obtained from the Coordination of Land use data for the Drava floodplain in 2018 were obtained from the Coordination of Information Information on the Environment (CORINE) data base (https://land.copernicus.eu/pan- on the Environment (CORINE) data base (https://land.copernicus.eu/pan-european/corine-land-cover/ european/corine-land-cover/clc2018). Land use was composed of 13 classes, with a dominance of clc2018). Land use was composed of 13 classes, with a dominance of agriculture (45%), followed by agriculture (45%), followed by deciduous forests (23.83%), whereas shrub areas, sealed surfaces, and deciduous forests (23.83%), whereas shrub areas, sealed surfaces, and water bodies took up 11%, 4%, water bodies took up 11%, 4%, and 1.1%, respectively (Figure 3a). For daily groundwater depth, 18 and 1.1%, respectively (Figure3a). For daily groundwater depth, 18 observation wells were used to observation wells were used to depict the spatial distribution of groundwater level for the Drava depict the spatial distribution of groundwater level for the Drava floodplain (Figure3b). The average floodplain (Figure 3b). The average depth to groundwater table during the period from 2000 to 2018 depth to groundwater table during the period from 2000 to 2018 ranged from 91.3 m to 117.8 m, with ranged from 91.3 m to 117.8 m, with an average of 99.3 m [70]. an average of 99.3 m [70].

(a) (b)

Figure 3. (a(a) )Land Land use use classes classes of of the the Drava Drava floodplain, floodplain, ( (bb)) contour contour line line of of the the average average depth depth to to groundwater level.

2.2. Groundwater Groundwater Recharge Recharge and and Evapotranspiration Evapotranspiration The spatial spatial distribution distribution of of groundwater groundwater recharge recharge and evapotranspiration evapotranspiration were estimated by the the WetSpass-M modelmodel [[71].71]. TheThe WetSpassWetSpass model model has has been been applied applied to to assess assess water water balance balance components components in indi ffdifferenterent regions regions around around the the world world [72 [72–77].–77]. The The WetSpass WetSpass model model considers considers the the spatial spatial distribution distribution of ofthe the soil, soil, land land use, use, slope, slope, groundwater groundwater depth, depth, and theand meteorological the meteorological data of data every of gridevery cell grid are dividedcell are dividedinto four into fractions four fractions (vegetated, (vegetated, bare soil, bare open soil, water open surface).water surface). The seasonal The seasonal water water balance balance for each for eachfraction fraction is calculated is calculated as: as: P = S + ET + R, (1) P = S + ET + R, (1) where P is precipitation, S is surface runoff, ET is evapotranspiration, and R is groundwater recharge. whereThe actual P is precipitation, evapotranspiration S is surface is calculated runoff, ET asis evapotranspiration, a function of potential and R evapotranspiration is groundwater recharge. and a Thevegetation actual coeevapotranspirationfficient [78]. R is is determined calculated asas a a residual function part of ofpotential the water evapotranspiration balance. The simulated and a vegetation coefficient [78]. R is determined as a residual part of the water1 balance. The1 simulated annual groundwater recharge of the Drava floodplain ranged from 0 mm y− to 360 mm y− as the lowest annual groundwater recharge of the Drava floodplain1 ranged from 0 mm y−1 to 360 mm y−1 as the and highest values, with an average of 241 mm y− (Figure4a). The estimated evapotranspiration lowest and highest values,1 with an average of 2411 mm y−1 (Figure 4a). The estimated1 varied from 211 mm y− to a maximum of 1394 mm y− , with an average value of 310 mm y− −1 −1 evapotranspiration(Figure4b). varied from 211 mm y to a maximum of 1394 mm y , with an average value of 310 mm y−1 (Figure 4b).

Water 2020, 12, 250 6 of 20 Water 2020, 12, x FOR PEER REVIEW 6 of 21

(a) (b)

Figure 4.4.Spatial Spatial distribution distribution of simulatedof simulated mean mean water water balance balance components: components: (a) Groundwater (a) Groundwater recharge; (recharge;b) actual ( evapotranspiration.b) actual evapotranspiration.

2.3. Numerical Groundwater Modeling MODFLOW-NWT [[79]79] with ModelMuse [[80]80] as a graphicalgraphical user interface was used to simulate groundwater flow.flow. ItIt waswas expresslyexpressly developed developed to to solve solve problems problems which which are are caused caused by by nonlinearities nonlinearities of dryingof drying and and rewetting rewetting of the of unconfinedthe unconfined groundwater groundwater flow, flow, similar similar to that to treated that treated in this in paper. this Thispaper. is inThis contrast is in contrast with other with versions other versions of MODFLOW, of MODFLOW, which excludewhich exclude dry cells dry from cells calculation. from calculation. The LAK7 The packageLAK7 package was developed was developed to simulate to lake simulate/groundwater lake/g interactions.roundwater Theinteractions. lake stage isThe determined lake stage based is ondetermined the exchanges based of on volumetric the exchanges water of in volumetric and out of wa theter lake in and and out the overallof the lake water and balance the overall of the water lake. Thebalance seepage of the between lake. The a lakeseepage and between its groundwater a lake an systemd its groundwater depends on system lake stage, depends lakebed on lake leakage stage, in gridlakebed cell leakage dimensions, in grid the cell hydraulic dimensions, conductivity the hydraulic of the lakebed conductivity material, of the the lakebed adjacent material, aquifer, andthe hydraulicadjacent aquifer, heads in and the groundwaterhydraulic heads system. in the The groundwater seepage is calculated system. byThe Darcy’s seepage law is [ 54calculated], as follows: by Darcy’s law [54], as follows: q = K (h ha)/d, (2) 1 − q = K (ℎ − ℎ)/d, (2) 1 1 where q is the seepage rate (L T− ); K is the hydraulic conductivity (L T− ) of materials between the where q is the seepage rate (L T−1); K is the hydraulic conductivity (L T−1) of materials between the lake lake and the aquifer; h1 is the stage of the lake (L); ha is the aquifer head (L); d is the distance (L) and the aquifer; ℎ is the stage of the lake (L); ℎ is the aquifer head (L); d is the distance3 1 (L) between between the points at which h1 and ha are measured; and the volumetric flux Q (L T− ) is established the points at which ℎ and ℎ are measured; and the volumetric flux Q (L3 T−1) is established by the by the following equation: following equation:

QQ= =q q A A= =(K (K ( h(h1 1 − hhaa))/d)/d) AA == CC( (h1 − hha),a), (3) (3) − − where c = KA/d is the conductance (L2 2T−1) 1and K/d is the leakage (T–1). 1 where c = KA/d is the conductance (L T− ) and K/d is the leakage (T− ). Inputs toto thethe groundwater groundwater system system are are groundwater groundwate recharger recharge from from precipitation, precipitation, river river seepage seepage and lakeand seepage,lake seepage, lateral lateral groundwater groundwater inflow inflow from a from part ofa part the northern of the northern boundary, boundary, while output while from output the groundwaterfrom the groundwater system includes system includes groundwater groundwater seepage toseepage rivers to and rivers lakes, and lateral lakes, groundwater lateral groundwater outflow fromoutflow the from southwestern the southwestern boundary, boundary and evaporation, and evaporation from groundwater. from groundwater.

2.3.1. Groundwater Flow Model Setup According to evidence from geologicalgeological boreholes,boreholes, pumping tests and GPR surveys,surveys, the groundwater flow flow system consists of four layers. Since the sediments in the area range from coarse sand to clay,clay, thethe fourfour layers layers are are highly highly heterogeneous heterogeneous in in space. space. The The model model was was discretized discretized by by a 100a 100 m bym by 100 100 m m grid grid and and the the model model was was refined refined for for lake lake area area with with 30 30 m m by by 30 30 m,m, resultingresulting inin aa domaindomain of 361 rows and 742 columns, and four layers with a total number of 1,339,310 cells, 655,390 of which were active cells. The records of GPR, supplemente supplementedd with auger holes, presented high heterogeneity in the hydraulic proprieties of the sediment depending on particle size (ranging from coarse sand to clay). The The fall fall head head method was was applied to to calcul calculateate the hydraulic conductivity for for each each borehole. According to the boreholes data, the firstfirst layerlayer consistedconsisted of fourfour K-zonesK-zones of hydraulichydraulic conductivity −1 (Figure5 5),), varyingvarying fromfrom 375 375 to to 0.15 0.15 m m d d− ,, while the second layer was characterized by three zoneszones with a range of 375 and 75 m d−1. The third and fourth layers were described by two zones, with a range from between 375 and 75 m d−1 and 0.15 to 3 m d−1, respectively (Figure 5). Although all the

Water 2020, 12, 250 7 of 20

Water 2020, 12, x FOR PEER REVIEW 1 7 of 21 with a range of 375 and 75 m d− . The third and fourth layers were described by two zones, with a range from between 375 and 75 m d 1 and 0.15 to 3 m d 1, respectively (Figure5). Although all hydraulic parameters were determined− in the field, because −of limited spatial representativeness and the hydraulic parameters were determined in the field, because of limited spatial representativeness the influence of modeling scale in accordance with [81–83], they still had to be adjusted in the and the influence of modeling scale in accordance with [81–83], they still had to be adjusted in the calibration process. Fortunately, the majority of the units had natural boundaries. calibration process. Fortunately, the majority of the units had natural boundaries.

(a) (b)

(c) (d)

Figure 5. HydraulicHydraulic conductivity conductivity zones for ( a) layer 1; ( b) layer 2; (c) layer 3; (d) layer 4.

The southern boundary is represented byby the DravaDrava River and assigned as a river package. Based on data available available from from observatio observationn wells, wells, the the general general direction direction of of groundwater groundwater flow flow is from is from north north to south.to south. The The Fekete-víz Fekete-v íandz and Okor Okor streams streams comprise comprise the the northern, northern, eastern, eastern, and and part part of of the western boundary and they are assigned by a riverriver packagepackage (Figure(Figure6 6a).a). AllAll streamsstreams inin thethe floodplainfloodplain areare 2 1 assigned by a river package.package. The conductanceconductance of thethe riverbedriverbed waswas initiallyinitially setset toto 55 mm2 dd−−1,, and then adjusted during the calibration process. MODFLOWMODFLOW General-Head Boundary (GHB) was assigned to the southwestern and part of the northernnorthern boundaryboundary to simulate lateral groundwater inflowinflow and outflowoutflow of the system based on the time series of groundwater level from the availableavailable observation 2 1 wells. Initially, the conductanceconductance of GHBGHB waswas setset toto bebe 2020 mm2 dd−−1,, based on geological boreholes and adjusted during the calibration process.process. Cún-Szaporca Cún-Szaporca Lake was a assignedssigned in the model through the LAK7 package (Figure 66a).a). TheThe lakelake simulationsimulation waswas performedperformed usingusing aa separateseparate additionaladditional inactiveinactive layer [54] [54] with with variable variable thickness. thickness. The The top top of ofinactive inactive layer layer had had a 1.0 a 1.0m thickness m thickness outside outside the lake the lakeand theand top the topequaled equaled the themaximum maximum lake lake stage stage within within the the lake lake area. area. The The bottom bottom was was equal equal to to the bathymetrically defineddefined lake bottom.bottom. Lakebed leakage was assigned from field field measurements for the 1 vertical hydraulic conductivity andand thethe thicknessthickness ofof thethe lakebed.lakebed. LakebedLakebed leakageleakage waswas setset toto 0.050.05 dd−−1..

Water 2020, 12, 250 8 of 20 Water 2020, 12, x FOR PEER REVIEW 8 of 21

(a)

(b) (c)

FigureFigure 6. ( a6.) Boundary(a) Boundary conditions, conditions, observation observation wells, and and planned planned reservoirs; reservoirs; (b) ( bKorcsina) Korcsina reservoir; reservoir; (c) Okor(c) Okor reservoir. reservoir.

2.3.2.2.3.2. Calibration Calibration and and Sensitivity Sensitivity Analysis Analysis A steady-stateA steady-state model model was was developed developed to represent to represent an initial an initial case forcase the for transient the transient model model by applying by long-termapplying average long-term parameters, average i.e.,parameters, mean precipitation, i.e., mean precipitation, mean groundwater mean groundwater levels, mean levels, PET, andmean mean PET, and mean groundwater recharge for the period from 1 January 2010 to 20 September 2018. The groundwater recharge for the period from 1 January 2010 to 20 September 2018. The steady-state model steady-state model was calibrated using average groundwater levels for 16 observation wells. was calibrated using average groundwater levels for 16 observation wells. Calibrated results showed Calibrated results showed a good agreement between observed and simulated groundwater heads, a good agreement between observed and simulated groundwater heads, with R2 = 0.98. The mean with R2 = 0.98. The mean error was −0.016 m, and the mean absolute error was 0.09 m. However, error was 0.016 m, and the mean absolute error was 0.09 m. However, starting the transient model by starting− the transient model by steady-state model as an initial condition proved to be unsuccessful steady-statedue to a large model difference as an initial between condition the steady-state proved to parameters be unsuccessful and the due measured to a large daily diff valueserence at between the the steady-statestart-up of the parameters transient model and on the 1 measuredJanuary 2010. daily values at the start-up of the transient model on 1 JanuaryThe 2010. first three hydrological years from which data were available, i.e., from 1 January 2010 until 31The December first three 2012, hydrological were used as years a warm-up from which period, data after were which available, the model i.e., was from run 1(calibrated) January 2010 untilthroughout 31 December the 2012,following were six used hydrological as a warm-up years period,from 1 January after which 2013 theto 20 model September was run2018 (calibrated) with a throughoutdaily time the step. following The calibrated six hydrological results of years the transient from 1 Januarymodel against 2013 to the 20 time September series of 2018 observation with a daily timewell step. heads The and calibrated the stages results of the of Cún-Szaporca the transient modelLake are against shown the in Figure time series 7. The of calibration observation process well of heads andgroundwater the stages of head the C wasún-Szaporca carried out Lake using are 16 shown time seri ines Figure of daily7. Thegroundwater calibration levels process from of observation groundwater wells extending over 10 years. Calibrated results showed a good match between observed and head was carried out using 16 time series of daily groundwater levels from observation wells extending simulated groundwater heads, with R2 = 0.98. The mean error was −0.08 m, and the root mean square over 10 years. Calibrated results showed a good match between observed and simulated groundwater error was 0.4 m (Figure 7a). Additionally, the calibration of lake stages showed a good agreement heads, with R2 = 0.98. The mean error was 0.08 m, and the root mean square error was 0.4 m (Figure7a). between simulated and observed stages− with a correlation coefficient of 0.90, a mean error of −0.07 Additionally,m, mean absolute the calibration error of of 0.15 lake m, stages and root showed mean a square good agreementerror of 0.2 between m (Figure simulated 7b). The andcalibrated observed stages with a correlation coefficient of 0.90, a mean error of 0.07 m, mean absolute error of 0.15 m, and − root mean square error of 0.2 m (Figure7b). The calibrated parameters of hydraulic conductivity, river conductance, general head conductance, and groundwater recharge are shown in Table1. Water 2020, 12, x FOR PEER REVIEW 9 of 21

Waterparameters2020, 12, 250of hydraulic conductivity, river conductance, general head conductance, 9 ofand 20 groundwater recharge are shown in Table 1.

(a) (b)

(c)

FigureFigure 7.7. ModelModel calibration.calibration. (a) ScatterScatter plotplot forfor simulatedsimulated versusversus observedobserved groundwatergroundwater piezometricpiezometric heads;heads; ((bb)) simulatedsimulated andand observedobserved dailydaily variabilityvariability ofof lakelake stages;stages; ((cc)) compositecomposite scaledscaled sensitivitiessensitivities calculatedcalculated withwith thethe parameterparameter values.values.

TheThe sensitivitysensitivity analysisanalysis waswas achievedachieved usingusing computercomputer codecode forfor universaluniversal sensitivitysensitivity analysis, calibration,calibration, andand uncertaintyuncertainty evaluationevaluation (UCODE-2005)(UCODE-2005) [[84]84] withwith thethe helphelp ofof ModelMateModelMate [[85].85]. ItIt waswas studiedstudied forfor thethe groundwatergroundwater rechargerecharge (RCH),(RCH), hydraulichydraulic conductivityconductivity (HK),(HK), whichwhich waswas divideddivided intointo fivefive zones,zones, evapotranspirationevapotranspiration (EVT),(EVT), riverriver conductanceconductance (RIVC), and general head boundary (GHB) parameters.parameters. TheThe parameters of the groundwater recharge (RCH), hydraulic conductivity (HKZ1),(HKZ1), andand evapotranspirationevapotranspiration (EVT)(EVT) hadhad thethe highesthighest compositecomposite scaledscaled sensitivitysensitivity valuesvalues andand werewere thereforetherefore thethe mostmost sensitivesensitive parametersparameters (Figure(Figure7 7c).c).

TableTable 1.1. CalibratedCalibrated parametersparameters forfor thethe calibratedcalibrated model.model. RR representsrepresents thethe spatialspatial distributiondistribution mapmap ofof averageaverage groundwatergroundwater recharge.recharge.

ParameterParameter Value Value Unit Unit Description Description HKZ1 360 m d−1 Hydraulic conductivity of zone 1 HKZ1 360 m d− Hydraulic conductivity of zone 1 −1 HKZ2 HKZ225 25 m dd− Hydraulic Hydraulic conductivity conductivity of zone of 2 zone 2 1 HKZ3 HKZ30.10 0.10 m dd−−1 Hydraulic Hydraulic conductivity conductivity of zone of 3 zone 3 1 HKZ4 HKZ45 5 m dd−−1 Hydraulic Hydraulic conductivity conductivity of zone of 4 zone 4 HKZ5 60 1 Hydraulic conductivity of zone 5 HKZ5 60 m dd−−1 Hydraulic conductivity of zone 5 RIVC1 300 m2 d 1 River conductance of streams 1 2 −−1 RIVC1 300 m2 d 1 River conductance of streams 1 RIVC2 5 m d− River conductance of streams 2 RIVC2 5 m22 d−1 River conductance of streams 2 GHB 40 m d− Conductance of general head boundary 2 −11 GHB RCH40 0.95R mmm d d− ConductanceGroundwater of general recharge head boundary

Water 2020, 12, 250 10 of 20

3. Results and Discussion

3.1. Water Budget of the Groundwater Flow Model The water balance at the end stress of the calibrated transient model is shown in Table2. The inflow rate from GHB boundary to the aquifer represents 42.25% of the total inflow to the aquifer through 3 1 627,789.00 m d− . Water seepage from streams and the Cún-Szaporca Lake represents 4.68% of 3 1 the total inflow. Groundwater discharge from the aquifer to streams is 1,275,244.38 m d− , which amounts to 85.82% of the total outflow from groundwater system as the area characterized by shallow groundwater table, while evapotranspiration represents 10% of total outflow.

Table 2. Water balance at the end stress of the calibrated transient model.

Input Output Input Output − 3 1 3 1 3 1 (m day− ) % (m day− ) % (m day− ) Recharge 546,403 36.8 457 0.0 545,946 Evapotranspiration 0 0.0 149,992 10.1 149,992 − GHB boundary 627,789 42.3 10,949 0.7 616,840 River 62,766 4.2 1,275,244 85.8 1,212,478 − Lake 6868 0.5 4967 0.3 1901 Storage 242,055 16.3 44,265 3.0 197,783 Total 1,485,881 100.0 1,485,874 100.0 0

3.2. Simulated Scenarios The natural water reservoirs are the former oxbows, paleomeanders, and deeper-lying floodplain sections, which are under excess water effect. These areas may be suitable for the storage of surplus water provided by the adjacent watercourses. The model space covered two subareas: 1, the Korcsina oxbow system (Figure6b); and 2, the Okor–Fekete-v íz backswamps (Figure6c). The exact location of the potential reservoir was selected from the analysis of the topography and hydraulic soil properties explored by auger samples. The selected subareas were simulated as natural reservoirs with filling and discharge conditions based on +2 m vertical water depth increasing theoretically with different stream 3 1 discharges of 0.5, 0.75, 1, and 1.5 m s− . (The excess water hazard on agricultural land excludes water level rise above +2 m.) Based on measurements of stream discharge, to obtain feeding with stream 3 1 discharge larger than 1 m d− , constructing a dam is required. Evaporation from the open water surface of the reservoir is a main challenge of the natural reservoir solution. The forested environs of the reservoir, however, reduce evaporation. This model is the first attempt to investigate whether water management through establishing a natural reservoir can be a major contribution to floodplain rehabilitation. Results of the management scenarios are discussed below.

3.2.1. Korcsina Subarea The Korcsina paleochannel system is a potential water reservoir in the western central part of the 3 1 floodplain. With streams feeding 0.5 m s− , the total upfilling (+2 m relative water level from 98 to 100 m) in the Korcsina reservoir would take 55 days, with 2,376,000 m3 water demand. The discharge (seepage) period lasts 11 days from a level of 100 m to 98.75 m. By the end of the filling period, seepage from the reservoir to groundwater reaches 35,815 m3. Recharge from groundwater to the reservoir in the filling period from the water level stage of 98 to 99 m and over this level becomes zero (Figure8a). 3 1 For filling at a rate of 0.75 m s− , the reservoir reaches the stage 100 m after 27 days with reservoir seepage of 38,900 m3, while the recharge from groundwater to the reservoir takes place in the first five days of the filling process at the level of 99 m (Figure8b). In the filling with a discharge of 1 3 1 and 1.5 m s− , 18 and 11 days are required to achieve +2 m relative water level rise in the reservoir (Figure8c,d). For all scenarios with di fferent discharge to the reservoir, the reservoir stage did not reach 98 m again because of the shallow groundwater table, which allows groundwater recharge larger Water 2020, 12, x FOR PEER REVIEW 11 of 21

Water 2020, 12, 250 11 of 20 (Figure 8c,d). For all scenarios with different discharge to the reservoir, the reservoir stage did not reach 98 m again because of the shallow groundwater table, which allows groundwater recharge largerthan the than seepage the seepage when reachingwhen reaching a level ofa level 98.83 of m 98.83 (i.e., them (i.e., inflow the from inflow groundwater from groundwater to the reservoir, to the reservoir,which is equal which to is the equal seepage to the from seepage the reservoir from the atreservoir this stage) at this (Figure stage)8). (Figure 8).

(a) (b)

(c) (d)

FigureFigure 8. DailyDaily variability variability of of Lake Lake Korcsina Korcsina stage, stage, groundwater groundwater seepage seepage into into the the lake lake (GW-Inflow), (GW-Inflow), seepageseepage from from the the lake lake into into groundwater groundwater (GW-Outf (GW-Outflow),low), and and net net lake lake seepag seepagee (LAKnet) (LAKnet) during during the the simulationsimulation period period of of one one year. year. Note Note that that the thefluxes fluxes are arereferenced referenced to the to temporally the temporally variable variable lake with lake 3 −1 3 1 3 −1 3 1 differentwith diff erentfeeding feeding discharge discharge from fromthe stream. the stream. (a) Feeding (a) Feeding at Q = at 0.5 Q m= 0.5 s , m (b)s −Q ,(= b0.75)Q m= 0.75 s , ( mc) Qs− = , 3 −1 3 1 3 −1 3 1 1.0(c)Q m = s1.0, (d m) Qs −= 1.5,(d m)Q s= .1.5 m s− .

3.2.2.3.2.2. Simulation Simulation of of the the Water Water Storage Storage O Opportunitiespportunities in the Okor–Fekete-víz Okor–Fekete-víz Area TheseThese randomly distributed andand irregular-shapedirregular-shaped paleochannels paleochannels and and backswamps backswamps are are located located at atthe the confluence confluence of of the the Okor Okor and and Fekete-v Fekete-vízíz streams. streams. The The mainmain texturaltextural typestypes in the west are medium toto coarse coarse sands, sands, in in the the middle middle part part fine fine sands, sands, an andd in in the the southeastern southeastern part part loamy loamy sands sands (Figure (Figure 2a).2a). TheThe bed bed level level of of the the Okor Okor natural natural reservoir reservoir is is at at 95.95 95.95 m. m. Feeding Feeding is is from from the the two two mentioned mentioned streams. streams. 3 1 With streams feeding atat aa rate ofof 0.50.5 mm3 s−1, ,the the reservoir reservoir stage stage does does not not reach reach +2+2 m m relative relative water water level level (from(from 95.95 95.95 to to 97.95 97.95 m), m), but but it it reaches reaches 97.7 97.7 m m within within 61 61 days days (Figure (Figure 99a),a), while with a stream discharge 3 1 3 ofof 0.75 0.75 m 3 ss−−1 thethe reservoir reservoir stage stage is is at at 97.95 97.95 m m after after 22 22 days, days, with with seepage seepage of of 49,733 49,733 m m3 (Figure(Figure 9b).9b). 3 1 RegardingRegarding the scenario of stream feeding at 1 m3 ss−1, ,13 13 days days are are needed needed to to reach reach the the stage stage of of 97.95 97.95 m, m, whilewhile the the discharge (seepage) period period lasts lasts for for 180 180 da daysys (the reservoir dries out) (Figure 99c).c). With aa 3 1 3 streamstream discharge discharge of 1.5 m3 s−1, ,the the reservoir reservoir achieves achieves the the stage stage of of 97.95 97.95 within within one one week week (62,215 m 3 seepage)seepage) (Figure (Figure 99d).d). There is no recharge from groundwater to the OkorOkor reservoirreservoir atat thethe stagestage ofof 97.3597.35 m m during during the the filling filling and discharge period.

Water 2020, 12, 250 12 of 20 Water 2020, 12, x FOR PEER REVIEW 12 of 21

(a) (b)

(c) (d)

FigureFigure 9.9. DailyDaily variabilityvariability of of Okor Okor lake lake stage, stage, groundwater groundwater seepage seepage into the into lake the (GW-Inflow), lake (GW-Inflow), seepage seepagefrom the lakefrom into the groundwater lake into groundwater (GW-Outflow), (GW-Ou and nettflow), lake seepage and net (LLKnet) lake seepage during simulation(LLKnet) during period simulationof one year. period Note thatof one the year. fluxes Note are that referenced the fluxes to theare temporallyreferenced to variable the temporally lake with variable different lake feeding with 3 1 3 3−1 1 3 3−1 1 3 −1 3 1 differentdischarges. feeding (a) Feeding discharges. by Q =(a0.5) Feeding m s− ,(byb )QQ == 0.50.75 m m s s, −(b,() Qc)Q = 0.75= 1.0 m m s s,− (c)and Q = ( d1.0)Q m= s1.5 and m s(−d). Q = 1.5 m3 s−1. The entrenchment and recent incision of the Drava River resulted in a dropping of groundwater 3 1 tablesThe by entrenchment 2.5 m. For both and reservoirs, recent incision and with of the a feeding Drava River discharge resulted of 0.5 in anda dropping 0.75 m ofs− groundwater, the feeding tablesstream by capacity 2.5 m. For does both not reservoirs, allow the rechargeand with ofa feeding the reservoirs discharge in 60of days0.5 and and 0.75 25 m days,3 s−1, respectively,the feeding streambased oncapacity the observed does not discharge allow the of recharge the feeding of th stream.e reservoirs However, in 60 feedingdays and the 25 reservoir days, respectively, at a rate of 3 1 based1.5 m ons− the, a +observed2 m rise todischarge be achieved of the in feeding five days stre requiresam. However, the construction feeding the of areservoir dam on theat a feedingrate of 1.5stream. m3 s− Therefore,1, a +2 m rise this to scenario be achieved would in notfive be days economically requires the feasible. construction Thus, theof a best dam scenario on the forfeeding both 3 1 stream.reservoirs Therefore, is a feeding this scenario rate of 1.0 would m s −not, as be 18 econom and 13ically days feasible. are needed Thus, for the filling best thescenario Korcsina for both and reservoirsOkor reservoirs, is a feeding respectively, rate of 1.0 and m water3 s−1, as recharge 18 and 13 to thedays reservoirs are needed could for befilling implemented the Korcsina by twoand fillingOkor reservoirs,periods per respectively, year, each providing and water water recharge storage to th fore reservoirs six months. could The be drop implemented of the water by leveltwo filling in the periodsreservoir per after year, the each filling providing period is water attributed storage to the for pressure six months. differences The drop between of the thewater reservoir level in water the reservoirand the adjacent after the groundwater. filling period The is attributed reservoir bedto the seepage pressure analysis differences was acrucial between step the to reservoir understand water the andinteractions the adjacent between groundwater. the reservoir The andreservoir groundwater bed seepage system. analysis The seepagewas a crucial from thestep reservoir to understand to the 3 1 thegroundwater interactions system, between up the to 25,000 reservoir m dand− , wasgroundwater dominant system. during stagesThe seepage higher from than ~99the reservoir m, whereas to 3 −1 3 1 thethe groundwater recharge from system, groundwater up to 25,000 to the m reservoir, d , was up dominant to 5000 m duringd− , wasstages dominant higher than during ~99 stages m, whereas below the99.00 recharge m. Water from levels groundwater in the reservoirs to the reservoir, are primarily up to e 5000ffected m3 by d−1 the, was surrounding dominant during groundwater stages below levels, 99.00regulated m. Water by the levels hyporheic in the reservoirs flow of the are streams. primar Theily effected first 1 m by of the filling surrounding increases thegroundwater saturation levels, of soil regulatedpores, which by the raises hyporheic soil moisture flow of content the streams. in the rootThe zonefirst 1 and m of the filling second increases meter creates the saturation an open of water soil pores,body, whichwhich israises useful soil to moisture create wetland content habitats in the ro inot the zone floodplain. and the second meter creates an open water body, which is useful to create wetland habitats in the floodplain.

Water 2020, 12, 250 13 of 20

3.3. Spatio-Temporal Extent of Reservoir Impact The assessment of the spatio-temporal impact of a lake (reservoir) on groundwater conditions is required to manage the groundwater systems adjacent to artificial lakes. The developed groundwater 3 1 head of lakes Korcsina and Okor for a filling scenario with a discharge of 1 m s− at different filling and discharge periods were plotted against distances at a cross-section which passes through the lakes (Figure6b,c) for the selected stress periods (Figures 10 and 11). The influence of the lake on groundwater levels decreases with distance (Figures 10 and 11). However, there are differences between the trends of the two sections due to the variance in hydrogeological and hydraulic conditions for the twoWater locations.2020, 12, x FOR The PEER results REVIEW clearly indicate that after 10 and 20 days of filling the Korcsina reservoir,13 of 21 the average groundwater head increased by 1 m and 1.6 m, respectively, with respect to the base case 3.3. Spatio-Temporal Extent of Reservoir Impact along the cross section. For the Okor reservoir, the average groundwater head increased by 0.7 m and 1.3The m assessment after filling of periodsthe spatio-temporal of 10 days andimpact 20 days,of a lake respectively, (reservoir) withon groundwater respect to the conditions base case, is whilerequired the to average manage groundwater the groundwater table decreasedsystems adjacent by 0.9 mto andartificial 1.0 m lakes. after The discharge developed periods groundwater of 70 days andhead 170 of lakes days, Korcsina respectively, and comparedOkor for a to filling the base scenario case (Figurewith a discharge11). As depicted of 1 m3 ins−1 the at different gradual risefilling in groundwaterand discharge table periods after were successive plotted filling against (Figures distan 10ces and at 11a ),cross-section the natural reservoirswhich passes improve through aquifer the storagelakes (Figure and hence 6b,c) maintainfor the selected the level stress of ecosystem periods (F servicesigures and10 and increase 11). The agricultural influence productivity of the lake on in thegroundwater Drava basin. levels decreases with distance (Figures 10 and 11). However, there are differences betweenLake–aquifer the trends exchange of the fluxestwo sections in the Korcsina due to and the Okor variance reservoirs in hydrogeological are shown in Figures and 12hydraulic and 13, identifyingconditions for the the losing two andlocations. gaining The zones results of theclearly planned indicate reservoirs that after with 10 theand help 20 days of the of cell-to-cellfilling the fluxKorcsina terms reservoir, in the model the average output files.groundwater Exchange head fluxes increased between by the 1 reservoirsm and 1.6 andm, respectively, the groundwater with systemrespect forto the diff baseerent case periods along of the filling cross and section. discharge For the were Okor analyzed. reservoir, The the signaverage on the groundwater graph indicates head theincreased direction by 0.7 of fluxesm and between 1.3 m after the filling reservoirs periods and of the 10 aquifer.days and A 20 positive days, respectively, sign represents with upward respect reservoirto the base seepage case, while gain the and average negative groundwater represents downward table decreased reservoir by 0.9 seepage m and loss. 1.0 m For after the discharge Korcsina reservoirperiods of with 70 days 20 days and of 170 filling days, (+ 2respectively, m), the seepage compared from reservoir to the base to groundwater case (Figure system 11). As ranged depicted from in 3 1 3 1 0the to 245gradual m d −risewith in angroundwater average value table of 55after m successived− , with higher filling values (Figures in the 10 northeastand 11), (Figurethe natural 12a). 3 1 Thereservoirs average improve exchange aquifer between storage reservoir and hence and aquifer mainta wasin the 1.4 level m d −of afterecosystem 40 days services of discharge and increase period, theagricultural majority productivity of which the reservoirin the Drava has basin. gained from the aquifer (Figure 12b).

Figure 10.10. SimulatedSimulated groundwatergroundwater levels levels at at a linea line crossing crossing the the Lake Lake Korcsina Korcsina at 10 at and 10 20 and days 20of days filling of andfilling at and 70 and at 70 170 and days 170 of days discharge of discharge period period with respect with respect to the base to the case. base case.

Water 2020, 12, x FOR PEER REVIEW 14 of 21

Water 2020, 12, 250 14 of 20 Water 2020, 12, x FOR PEER REVIEW 14 of 21

Figure 11. Simulated groundwater levels at a line crossing Lake Okor at 10 and 20 days of the filling period and at 70 and 170 days of the discharge period with respect to the base case.

Lake–aquifer exchange fluxes in the Korcsina and Okor reservoirs are shown in Figures 12 and 13, identifying the losing and gaining zones of the planned reservoirs with the help of the cell-to-cell flux terms in the model output files. Exchange fluxes between the reservoirs and the groundwater system for different periods of filling and discharge were analyzed. The sign on the graph indicates the direction of fluxes between the reservoirs and the aquifer. A positive sign represents upward reservoir seepage gain and negative represents downward reservoir seepage loss. For the Korcsina reservoir with 20 days of filling (+2 m), the seepage from reservoir to groundwater system ranged 3 −1 3 −1 from 0 to 245 m d with an average value of 55 m d , with higher values in the northeast (Figure 12a). The average exchange between reservoir and aquifer was 1.4 m3 d−1 after 40 days of discharge period,FigureFigure 11.the Simulatedmajority 11. Simulated of groundwater which groundwater the reservoir levels levels at hasat a a line gainedline crossingcrossing from Lake Lakethe aquiferOkor Okor atat (Figure10 10 and and 20 12b). 20days days of the of filling the filling periodperiod and atand 70 at and 70 and 170 170 days days of theof the discharge discharge period period withwith respect to to the the base base case. case.

Lake–aquifer exchange fluxes in the Korcsina and Okor reservoirs are shown in Figures 12 and 13, identifying the losing and gaining zones of the planned reservoirs with the help of the cell-to-cell flux terms in the model output files. Exchange fluxes between the reservoirs and the groundwater system for different periods of filling and discharge were analyzed. The sign on the graph indicates the direction of fluxes between the reservoirs and the aquifer. A positive sign represents upward reservoir seepage gain and negative represents downward reservoir seepage loss. For the Korcsina reservoir with 20 days of filling (+2 m), the seepage from reservoir to groundwater system ranged from 0 to 245 m3 d−1 with an average value of 55 m3 d−1, with higher values in the northeast (Figure 12a). The average exchange between reservoir and aquifer was 1.4 m3 d−1 after 40 days of discharge period, the majority of which the reservoir has gained from the aquifer (Figure 12b).

Water 2020, 12, x FOR PEER REVIEW 15 of 21 (a)

(a)

(b)

FigureFigure 12. Lake–aquifer 12. Lake–aquifer exchange exchange fluxes fluxes for for thethe KorcsinaKorcsina reservoir: reservoir: (a) ( aat) atthe the end end of filling of filling period period of of 14 days14 days (+2 m(+2 reservoir m reservoir stage), stage), (b ()b for) for 40 40 days days ofof dischargedischarge period period (+1 (+ m)1 m) (positive (positive represents represents upward upward reservoirreservoir seepage seepage gain gain and and negative negative represents represents downward downward reservoir reservoir seepage seepage loss). loss).

Similar seepage patterns can be observed for the Okor reservoir (Figure 13a,b), with average values of seepage of 25.3 m3 d−1 and 0.7 m3 d−1 after 14 days of filling (+2 m) and 40 days of discharge (+1 m), respectively. The patterns of spatial seepage at various reservoir stages through filling and discharge periods are different, attributed to variations in reservoir-bed leakage and reservoir stages. The magnitude and direction of seepage between the reservoir and underlying groundwater system relies on the reservoir-bed leakage and the difference between reservoir stage and hydraulic head of an underlying aquifer. In the Drava Plain, of highly variable alluvial deposits and soil textural types, the modeling of reservoir/groundwater interchanges is central to the planning of rehabilitation measures.

(a)

Water 2020, 12, x FOR PEER REVIEW 15 of 21

(b)

Figure 12. Lake–aquifer exchange fluxes for the Korcsina reservoir: (a) at the end of filling period of 14 days (+2 m reservoir stage), (b) for 40 days of discharge period (+1 m) (positive represents upward Water 2020,reservoir12, 250 seepage gain and negative represents downward reservoir seepage loss). 15 of 20

Similar seepage patterns can be observed for the Okor reservoir (Figure 13a,b), with average valuesSimilar of seepageseepage of patterns 25.3 m3 d can−1 and be 0.7 observed m3 d−1 after for 14 the days Okor of filling reservoir (+2 m) (Figure and 40 13 daysa,b), of with discharge average 3 1 3 1 values(+1 of m), seepage respectively. of 25.3 The m patternsd− and of 0.7 spatial m d− seepagaftere 14 atdays various of fillingreservoir (+2 stages m) and through 40 days filling of discharge and (+1 m),discharge respectively. periods are The different, patterns attributed of spatial to seepagevariations at in various reservoir-bed reservoir leakage stages and through reservoirfilling stages.and dischargeThe magnitude periods areand di directionfferent, attributedof seepage between to variations the reservoir in reservoir-bed and underlying leakage groundwater and reservoir system stages. The magnituderelies on the andreservoir-bed direction leakage of seepage and the between differen thece reservoirbetween reservoir and underlying stage and groundwater hydraulic head system of reliesan on underlying the reservoir-bed aquifer. In leakage the Drava and Plain, the di offf erencehighly betweenvariable alluvial reservoir deposits stage and soil hydraulic textural head types, of an underlyingthe modeling aquifer. of Inreservoir/groundwater the Drava Plain, of highly interchanges variable is alluvialcentral depositsto the planning and soil of textural rehabilitation types, the measures. modeling of reservoir/groundwater interchanges is central to the planning of rehabilitation measures.

Water 2020, 12, x FOR PEER REVIEW 16 of 21 (a)

(b)

FigureFigure 13. 13.Lake–aquifer Lake–aquifer exchange exchange fluxes fluxes of of the the Okor Okor reservoir: reservoir: (a ()a at) at the the end end of of filling filling period period of of 10 10 days (+2days m reservoir (+2 m reservoir stage), stage), (b) for (b 40) for days 40 days of theof the discharge discharge period period ((+1+1 m) m) (positive (positive represents represents upward upward reservoirreservoir seepage seepage gain gain and and negative negative represents represents downwarddownward reservoir reservoir seepage seepage loss). loss).

The findings point out that several factors influence the interactions of reservoir with the groundwater system. The regulated recharge to the reservoir and the large spatio-temporal variability of interactions indicate that sophisticated and complex modeling methods are required to depict such a complex hydrological system. Therefore, the integration of a spatially-distributed water balance model (WetSpass-M), and a groundwater flow model (MODFLOW-NWT) to represent the volumetric exchanges between reservoir and groundwater was applied for case studies in the Drava floodplain.

4. Conclusions The regulation works and incision of Drava River have changed the water regime and water budget of the floodplain, and, as a consequence, the water stress has increased and threatens the sustainable development of different sectors in the Drava basin. This study assessed the feasibility of the natural reservoir in augmenting the water resources of the floodplain through surface water storage and to mitigate water shortage problems caused by drought and human interventions using MODFLOW-NWT with ModelMuse as a graphical interface. The transient groundwater flow model was calibrated with a correlation coefficient of 0.96, a mean error of −0.016 m, and a mean absolute error of 0.4 m. A digital elevation model and hydraulic soil properties revealed by auger samples were used to identify possible allocations of a natural reservoir. The two selected subareas were selected as potential natural reservoir sites. Simulated reservoirs had filling and discharge conditions based on +2 m vertical water depth with different discharges from adjacent streams at 0.5, 0.75, 1, and 1.5 m3 s−1 rates. For both reservoirs, around 60 days and 25 days of stream feeding of 0.5 and 0.75 m3 s−1, respectively, are required to achieve the target water depth for the reservoir (+2 m), the optimal and economic scenario of filling the reservoirs is with a stream discharge of 1 m3 s−1 for 14 days, based on the capacity of the feeding stream. Feeding at a rate of 1.5 m3 s−1 would need around just six days, but this is not an economic scenario as it necessitates impoundment and dam construction. With feeding at 1 m3 s−1, filling would be repeated twice per year, as the reservoir lasts for 180 days. The effect of the reservoir on the groundwater table in the adjacent area depends on reservoir stage,

Water 2020, 12, 250 16 of 20

The findings point out that several factors influence the interactions of reservoir with the groundwater system. The regulated recharge to the reservoir and the large spatio-temporal variability of interactions indicate that sophisticated and complex modeling methods are required to depict such a complex hydrological system. Therefore, the integration of a spatially-distributed water balance model (WetSpass-M), and a groundwater flow model (MODFLOW-NWT) to represent the volumetric exchanges between reservoir and groundwater was applied for case studies in the Drava floodplain.

4. Conclusions The regulation works and incision of Drava River have changed the water regime and water budget of the floodplain, and, as a consequence, the water stress has increased and threatens the sustainable development of different sectors in the Drava basin. This study assessed the feasibility of the natural reservoir in augmenting the water resources of the floodplain through surface water storage and to mitigate water shortage problems caused by drought and human interventions using MODFLOW-NWT with ModelMuse as a graphical interface. The transient groundwater flow model was calibrated with a correlation coefficient of 0.96, a mean error of 0.016 m, and a mean absolute − error of 0.4 m. A digital elevation model and hydraulic soil properties revealed by auger samples were used to identify possible allocations of a natural reservoir. The two selected subareas were selected as potential natural reservoir sites. Simulated reservoirs had filling and discharge conditions based on 3 1 +2 m vertical water depth with different discharges from adjacent streams at 0.5, 0.75, 1, and 1.5 m s− 3 1 rates. For both reservoirs, around 60 days and 25 days of stream feeding of 0.5 and 0.75 m s− , respectively, are required to achieve the target water depth for the reservoir (+2 m), the optimal and 3 1 economic scenario of filling the reservoirs is with a stream discharge of 1 m s− for 14 days, based on 3 1 the capacity of the feeding stream. Feeding at a rate of 1.5 m s− would need around just six days, but this is not an economic scenario as it necessitates impoundment and dam construction. With feeding 3 1 at 1 m s− , filling would be repeated twice per year, as the reservoir lasts for 180 days. The effect of the reservoir on the groundwater table in the adjacent area depends on reservoir stage, leakage, and the distribution of the groundwater head using the reservoir results in raising the groundwater table by 0.8 m at the end of the filling period. Natural reservoirs can be regarded as an economically feasible management practice of water resources retention and can help protect ecological values, the restoration of wetlands, and enhance agriculture productivity on the Hungarian section of the Drava floodplain. Moreover, the findings of this research can be utilized in planning rehabilitation measures in lowlands with water shortage.

Author Contributions: Conceptualization, A.S.; Data curation, J.D.; Formal analysis, A.S. and M.E.-R.; Methodology, A.S. and M.E.-R.; Project administration, J.D. and D.L.; Software, A.S. and M.E.-R.; Supervision, J.D. and D.L.; Validation, A.S.; Visualization, M.E.-R.; Writing—original draft, A.S.; Writing—review and editing, J.D., M.E.-R. and D.L. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by the project that supported by the European Union, and co-financed by the European Social Fund: Comprehensive Development for Implementing Smart Specialization Strategies at the University of Pécs (EFOP-3.6.1.-16-2016- 00004). Acknowledgments: The present scientific contribution is dedicated to the 650th anniversary of the foundation of the University of Pécs, Hungary. The first author would like to thank the Egyptian Ministry of Higher Education (MoHE) and Tempus Public Foundation for providing him the Stipendium Hungaricum Scholarship. Also, the authors are grateful for financial support from the project that supported by the European Union, and co-financed by the European Social Fund: Comprehensive Development for Implementing Smart Specialization Strategies at the University of Pécs (EFOP-3.6.1.-16-2016- 00004), the Hungarian National Science Foundation (OTKA, contract no K 104552) and by FEKUTSTRAT (No. 20765-3/2018) “Innovation for a Sustainable, Healthy Life and Environment” grant and to the South-Transdanubian Water Management Directorate for providing access to the necessary data. Conflicts of Interest: The authors declare no conflict of interest. Water 2020, 12, 250 17 of 20

References

1. Gumiero, B.; Mant, J.; Hein, T.; Elso, J.; Boz, B. Linking the restoration of rivers and riparian zones/wetlands in Europe: Sharing knowledge through case studies. Ecol. Eng. 2013, 56, 36–50. [CrossRef] 2. De Vries, A. (Ed.) Summary of Master Plans of Wetland Sub-Group Partners for Water Project; University of Debrecen, Centre for Environmental Management and Policy: Debrecen, Hungary, 2013; p. 76. 3. Mori, K. Can market-based policies accomplish the optimal floodplain management? A gap between static and dynamic models. J. Environ. Manag. 2009, 90, 1191–1194. [CrossRef][PubMed] 4. Hopkins, K.G.; Noe, G.B.; Franco, F.; Pindilli, E.J.; Gordon, S.; Metes, M.J.; Claggett, P.R.; Gellis, A.C.; Hupp, C.R.; Hogan, D.M. A method to quantify and value floodplain sediment and nutrient retention ecosystem services. J. Environ. Manag. 2018, 220, 65–76. [CrossRef][PubMed] 5. WWF International. Assessment of the Restoration Potential along the Danube and Main Tributaries. In Working Paper for the Danube River Basin; World-Wide Fund for Nature: Vienna, Austria, 2010; p. 60. 6. Anderson, J.; Arblaster, K.; Bartley, J.; Cooper, T.; Kettunen, M.; Kaphengst, T.; Leipprand, A.; Laaser, C.; Umpfenbach, K.; Kuusisto, E. Climate Change-Induced Water Stress and Its Impacts on Natural and Managed Ecosystems; European Parlament: Brussels, Belgium, 2008; p. 108. 7. Hulisz, P.; Michalski, A.; D ˛abrowski, M.; Kusza, G.; Łéczy´nski,L. Human-induced changes in the soil cover at the mouth of the Vistula River Cross-Cut (northern Poland). Soil Sci. Annu. 2015, 66, 67–74. [CrossRef] 8. Lóczy, D.; Dezs˝o,J.; Czigány, S.; Gyenizse, P.; Pirkhoffer, E. Amadé Halász Rehabilitation potential of the Drava river floodplain in Hungary. In Proceedings of the Water Resources and Wetlands, Tulcea, Romania, 11–13 September 2014; pp. 21–29. 9. Lóczy, D.; Dezs˝o,J.; Czigány, S.; Prokos, H.; Tóth, G. An environmental assessment of water replenishment to a floodplain lake. J. Environ. Manag. 2017, 202, 337–347. [CrossRef] 10. Dezs˝o,J.; Lóczy, D.; Salem, A.; Gábor, N. Floodplain connectivity. In The Drava River: Environmental Problems and Solutions; Springer Science + Media: Cham, Switzerland, 2019; pp. 215–230. 11. Burián, A.; Horváth, G.; Márk, L. Channel Incision Along the Lower Drava. In The Drava River: Environmental Problems and Solutions; Springer Science + Media: Cham, Switzerland, 2019; pp. 139–157. 12. Goler, R.A.; Frey, S.; Formayer, H.; Holzmann, H. Influence of Climate Change on River Discharge in Austria. Meteorol. Z. 2016, 25, 621–626. [CrossRef] 13. Gálosi-Kovács, B.; Császár, Z.M.; Pap, N.; Remény, P.; Elekes, T. Social problems of the most disadvantaged southern transdanubian micro-regions in Hungary. Rom. Rev. Reg. Stud. 2011, 7, 41–50. 14. AQUAPROFIT Os-Dr˝ áva Program—Összefogással az Ormánság fellendítéséért. Vezet˝oiösszefoglaló (Old Drava Programme—Cooperation for the Progress of Ormánság: Executive summary); AQUAPROFIT: Budapest, Hungary, 2010; p. 29. 15. Cheng, Y.; Lee, C.-H.; Tan, Y.-C.; Yeh, H.-F. An optimal water allocation for an irrigation district in Pingtung County, Taiwan. Irrig. Drain. 2009, 58, 287–306. [CrossRef] 16. Salem, A.; Dezs˝o,J.; El-Rawy, M.; Loczy, D. Water management and retention opportunities along the Hungarian section of the Drava River. In Proceedings of the 2nd Euro-Mediterranean Conference for Environmental Integration (EMCEI), Sousse, Tunisia, 10–13 October 2019. 17. Singh, H.; Sinha, T.; Sankarasubramanian, A. Impacts of Near-Term Climate Change and Population Growth on Within-Year Reservoir Systems. J. Water Resour. Plann. Manag. 2015, 141, 04014078. [CrossRef] 18. El-Rawy, M.; Zlotnik, V.A.; Al-Raggad, M.; Al-Maktoumi, A.; Kacimov, A.; Abdalla, O. Conjunctive use of groundwater and surface water resources with aquifer recharge by treated wastewater: Evaluation of management scenarios in the Zarqa River Basin, Jordan. Environ. Earth Sci. 2016, 75, 1146. [CrossRef] 19. Seo, S.B.; Mahinthakumar, G.; Sankarasubramanian, A.; Kumar, M. Conjunctive Management of Surface Water and Groundwater Resources under Drought Conditions Using a Fully Coupled Hydrological Model. J. Water Resour. Plann. Manag. 2018, 144, 04018060. [CrossRef] 20. Liu, L.; Cui, Y.; Luo, Y. Integrated Modeling of Conjunctive Water Use in a Canal-Well Irrigation District in the Lower Yellow River Basin, China. J. Irrig. Drain Eng. 2013, 139, 775–784. [CrossRef] 21. Kreši´c,N. Groundwater Resources: Sustainability, Management, and Restoration; McGraw-Hill: New York, NY, USA, 2009; ISBN 978-0-07-149273-7. 22. Healy, R.W.; Scanlon, B.R. Estimating Groundwater Recharge; Cambridge University Press: Cambridge, UK, 2010; ISBN 978-0-511-78074-5. Water 2020, 12, 250 18 of 20

23. Asano, T.; Cotruvo, J.A. Groundwater recharge with reclaimed municipal wastewater: Health and regulatory considerations. Water Res. 2004, 38, 1941–1951. [CrossRef][PubMed] 24. Gale, I. Strategies for Managed Aquifer Recharge (MAR) in Semi-Arid Areas; UNESCO IHP: Paris, France, 2005; p. 33. 25. Allam, M.N.; Allam, G.I. Water Resources in Egypt: Future Challeges and Opportunities. Water Int. 2007, 32, 205–218. [CrossRef] 26. Khan, S.; Mushtaq, S.; Hanjra, M.A.; Schaeffer, J. Estimating potential costs and gains from an aquifer storage and recovery program in Australia. Agric. Water Manag. 2008, 95, 477–488. [CrossRef] 27. Abiye, T.A.; Sulieman, H.; Ayalew, M. Use of treated wastewater for managed aquifer recharge in highly populated urban centers: A case study in Addis Ababa, Ethiopia. Environ. Geol. 2009, 58, 55–59. [CrossRef] 28. Al-Maktoumi, A.; El-Rawy, M.; Zekri, S. Management options for a multipurpose coastal aquifer in Oman. Arab. J. Geosci. 2016, 9, 636. [CrossRef] 29. El-Rawy, M.; Al-Maktoumi, A.; Zekri, S.; Abdalla, O.; Al-Abri, R. Hydrological and economic feasibility of mitigating a stressed coastal aquifer using managed aquifer recharge: A case study of Jamma aquifer, Oman. J. Arid Land 2019, 11, 148–159. [CrossRef] 30. Schwartz, F.W.; Gallup, D.N. Some factors controlling the major ion chemistry of small lakes: Examples from the prairie parkland of Canada. Hydrobiologia 1978, 58, 65–81. [CrossRef] 31. Stauffer, R.E. Effects of citrus agriculture on ridge lakes in Central Florida. Water Air Soil Pollut. 1991, 59, 125–144. [CrossRef] 32. Nakayama, T.; Watanabe, M. Missing role of groundwater in water and nutrient cycles in the shallow eutrophic lake Kasumigaura, Japan. Hydrol. Process. 2008, 22, 1150–1172. [CrossRef] 33. Sanford, W. Recharge and groundwater models: An overview. Hydrogeol. J. 2002, 10, 110–120. [CrossRef] 34. Salem, A.; Dezs˝o,J.; Lóczy, D.; El-Rawy, M.; Slowik, M. Modeling Surface Water-Groundwater Interaction in an Oxbow of the Drava Floodplain. In Proceedings of the 13th International Conference on Hydroinformatics HIC 2018, Palermo, Italy, 1–6 July 2018. 35. Lee, D.R. A device for measuring seepage flux in lakes and estuaries1. Limnol. Oceanogr. 1977, 22, 140–147. [CrossRef] 36. Winter, T.C.; LaBaugh, J.W.; Rosenberry, D.O. The design and use of a hydraulic potentiomanometer for direct measurement of differences in hydraulic head between groundwater and surface water. Limnol. Oceanogr. 1988, 33, 1209–1214. [CrossRef] 37. Ong, J.B.; Zlotnik, V.A. Assessing Lakebed Hydraulic Conductivity and Seepage Flux by Potentiomanometer. Ground Water 2011, 49, 270–274. [CrossRef] 38. Stauffer, R.E. Use of solute tracers released by weathering to estimate groundwater inflow to seepage lakes. Environ. Sci. Technol. 1985, 19, 405–411. [CrossRef] 39. Sacks, L.A.; Swancar, A.; Lee, T.M. Estimating Ground-Water Exchange with Lakes Using Water-Budget and Chemical Mass-Balance Approaches for Ten Lakes in Ridge Areas of Polk and Highlands Counties, Florida; U.S. Geological Survey: Reston, VA, USA, 1998; p. 52. 40. Krabbenhoft, D.P.; Bowser, C.J.; Anderson, M.P.; Valley, J.W. Estimating groundwater exchange with lakes: The stable isotope mass balance method. Water Resour. Res. 1990, 26, 2445–2453. [CrossRef] 41. Brindha, K.; Neena Vaman, K.V.; Srinivasan, K.; Sathis Babu, M.; Elango, L. Identification of surface water-groundwater interaction by hydrogeochemical indicators and assessing its suitability for drinking and irrigational purposes in Chennai, Southern India. Appl. Water Sci. 2014, 4, 159–174. [CrossRef] 42. Kanduˇc,T.; Grassa, F.; McIntosh, J.; Stibilj, V.; Ulrich-Supovec, M.; Supovec, I.; Jamnikar, S. A geochemical and stable isotope investigation of groundwater/surface-water interactions in the Velenje Basin, Slovenia. Hydrogeol. J. 2014, 22, 971–984. [CrossRef] 43. Sacks, L.A.; Lee, T.M.; Swancar, A. The suitability of a simplified isotope-balance approach to quantify transient groundwater–lake interactions over a decade with climatic extremes. J. Hydrol. 2014, 519, 3042–3053. [CrossRef] 44. Coplen, T.B. Guidelines and recommended terms for expression of stable-isotope-ratio and gas-ratio measurement results: Guidelines and recommended terms for expressing stable isotope results. Rapid Commun. Mass Spectrom. 2011, 25, 2538–2560. [CrossRef] 45. Dinçer, T. The Use of Oxygen 18 and Deuterium Concentrations in the WaterBalance of Lakes. Water Resour. Res. 1968, 4, 1289–1306. [CrossRef] Water 2020, 12, 250 19 of 20

46. Turner, J.V.; Allison, G.B.; Holmes, J.W. The water balance of a small lake using stable isotopes and tritium. J. Hydrol. 1984, 70, 199–220. [CrossRef] 47. LaBaugh, J.W.; Winter, T.C.; Rosenberry, D.O.; Schuster, P.F.; Reddy, M.M.; Aiken, G.R. Hydrological and chemical estimates of the water balance of a closed-basin lake in north central Minnesota. Water Resour. Res. 1997, 33, 2799–2812. [CrossRef] 48. Sacks, L.A. Estimating Ground-Water Inflow to Lakes in Central Florida Using the Isotope Mass-Balance Approach; Water-Resources Investigations Report 2002-4192; U.S. Dept. of the Interior, U.S. Geological Survey: Reston, VA, USA, 2002. 49. El-Zehairy, A.A.; Lubczynski, M.W.; Gurwin, J. Interactions of artificial lakes with groundwater applying an integrated MODFLOW solution. Hydrogeol. J. 2018, 26, 109–132. [CrossRef] 50. McDonald, M.G.; Harbaugh, A.W. A Modular Three-Dimensional Finite-Difference Ground-Water Flow Model; Book 6, Chapter A1; U.S. Geological Survey, Techniques of Water-Resources Investigations: Reston, VA, USA, 1988; p. 586. 51. Leblanc, M.; Favreau, G.; Tweed, S.; Leduc, C.; Razack, M.; Mofor, L. Remote sensing for groundwater modelling in large semiarid areas: Lake Chad Basin, Africa. Hydrogeol. J. 2007, 15, 97–100. [CrossRef] 52. Mylopoulos, N.; Mylopoulos, Y.; Tolikas, D.; Veranis, N. Groundwater modeling and management in a complex lake-aquifer system. Water Resour Manag. 2007, 21, 469–494. [CrossRef] 53. Fenske, J.P.; Leake, S.A.; Prudic, D.E. Documentation of A Computer Program (RES1) to Simulate Leakage from Reservoirs Using the Modular Finite-Difference Ground-Water Flow Model (MODFLOW); U.S. Geological Survey: Reston, VA, USA, 1996; p. 51. 54. Merritt, M.L.; Konikow, L.F. Documentation of A Computer Program to Simulate Lake-Aquifer Interaction Using the MODFLOW Ground Water Flow Model and the MOC3D Solute-Transport Model; Water-Resources Investigations Report 2000-4167; U.S. Dept. of the Interior, U.S. Geological Survey: Reston, VA, USA, 2000; p. 146. 55. Lee, T.M. Hydrogeologic Controls on the Groundwater Interactions with an Acidic Lake in Karst Terrain, Lake Barco, Florida. Water Resour. Res. 1996, 32, 831–844. [CrossRef] 56. Yihdego, Y.; Becht, R. Simulation of lake–aquifer interaction at Lake Naivasha, Kenya using a three-dimensional flow model with the high conductivity technique and a DEM with bathymetry. J. Hydrol. 2013, 503, 111–122. [CrossRef] 57. Cheng, X.; Anderson, M.P. Numerical Simulation of Ground-Water Interaction with Lakes Allowing for Fluctuating Lake Levels. Groundwater 1993, 31, 929–933. [CrossRef] 58. Kidmose, J.; Engesgaard, P.; Nilsson, B.; Laier, T.; Looms, M.C. Spatial Distribution of Seepage at a Flow-Through Lake: Lake Hampen, Western Denmark. Vadose Zone J. 2011, 10, 110. [CrossRef] 59. Hunt, R. Ground Water-Lake Interaction Modeling Using the LAK3 Package for MODFLOW 2000. Ground Water 2003, 41, 114–118. [CrossRef] 60. Buchberger, P. A Dráva-völgy árvédelmének története (History of flood control in the Drava Valley). Vízügyi Közlemények 1975, 75, 103–113. 61. Pálfai, I. Magyarország Holtágai (Oxbows in Hungary); Hungarian Ministry of Transport and Water Management: Budapest, Hungary, 2001; p. 231. 62. VKKI Vízgy˝ujt˝o-gazdálkodási terv. Dráva részvízgy˝ujt˝o(Water Basin Management Plan: Drava Partial Water Basin); Central Directorate for Water Management and Environmental Protection: Budapest, Hungary, 2010. 63. Lóczy, D. Geological and Geomorphological Setting. In The Drava River: Environmental Problems and Solutions; Springer Science + Media: Cham, Switzerland, 2018; pp. 5–21. 64. Lóczy, D.; Dezs˝o,J.; Gyenizse, P.; Czigány, S.; Tóth, G. Oxbow Lakes: Hydromorphology. In The Drava River: Environmental Problems and Solutions; Springer Science + Media: Cham, Switzerland, 2019; pp. 177–198. 65. Salem, A.; Dezs˝o,J.; El-Rawy, M.; Lóczy, D. Statistical analysis of precipitation trend for Drava flood plain region in Hungary. In Proceedings of the GSRD International Conference, 579th International Conferences on Engineering and Natural Science (ICENS), Istanbul, Turkey, 21 March 2019; pp. 43–47. 66. Penman, H.L. Natural evaporation from open water, bare soil and grass. Proc. R. Soc. Lond. A 1948, 193, 120–145. 67. Crop Evapotranspiration: Guidelines for Computing Crop Water Requirements; Allen, R.G.; Food and Agriculture Organization of the United Nations (Eds.) FAO Irrigation and Drainage Paper; Food and Agriculture Organization of the United Nations: Rome, Italy, 1998; ISBN 978-92-5-104219-9. Water 2020, 12, 250 20 of 20

68. Dezs˝o,J.; Salem, A.; Lóczy, D.; Slowik, M.; Dávid, P. Randomly layered fluvial sediments influenced groundwater-surface water interaction. In Proceedings of the 17th International Multidisciplinary Scientific GeoConference SGEM 2017, Vienna, Austria, 27 June–6 July 2017; Volume 17, pp. 331–338. 69. Słowik, M.; Dezs˝o,J.; Marciniak, A.; Tóth, G.; Kovács, J. Evolution of river planforms downstream of dams: Effect of dam construction or earlier human-induced changes?: Evolution of river planforms downstream of dams. Earth Surf. Process. Landf. 2018, 43, 2045–2063. [CrossRef] 70. Salem, A.; Dezs˝o,J.; El-Rawy, M.; loczy, D.; Halmai, Á. Estimation of groundwater recharge distribution using Gis based WetSpass model in the Cun-Szaporca oxbow, Hungary. In Proceedings of the 19th International Multidisciplinary Scientific GeoConference SGEM 2019, Albena, Bulgaria, 28 June–7 July 2019; Volume 19, pp. 169–176. 71. Abdollahi, K.; Bashir, I.; Verbeiren, B.; Harouna, M.R.; Van Griensven, A.; Huysmans, M.; Batelaan, O. A distributed monthly water balance model: Formulation and application on Black Volta Basin. Environ. Earth Sci. 2017, 76, 198. [CrossRef] 72. Batelaan, O.; Smedt, F.D. WetSpass: A flexible, GIS based, distributed recharge methodology for regional groundwater modelling. In Impact of Human Activity on Groundwater Dynamics; Gehrels, H., Peters, J., Leibundgut, C., Eds.; International Association of Hydrological Sciences: Wallingford, UK, 2001; pp. 11–17. 73. Abu-Saleem, A.; Al-Zubi, Y.; Rimawi, O.; Al-Zubi, J.; Alouran, N. Estimation of Water Balance Components in the Hasa Basin with GIS based WetSpass Model. J. Agron. 2010, 9, 119–125. [CrossRef] 74. Gebreyohannes, T.; De Smedt, F.; Walraevens, K.; Gebresilassie, S.; Hussien, A.; Hagos, M.; Amare, K.; Deckers, J.; Gebrehiwot, K. Application of a spatially distributed water balance model for assessing surface water and groundwater resources in the Geba basin, Tigray, Ethiopia. J. Hydrol. 2013, 499, 110–123. [CrossRef] 75. Armanuos, A.M.; Negm, A.; Yoshimura, C.; Valeriano, O.C.S. Application of WetSpass model to estimate groundwater recharge variability in the Nile Delta aquifer. Arab. J. Geosci. 2016, 9, 553. [CrossRef] 76. Zhang, Y.; Liu, S.; Cheng, F.; Shen, Z. WetSpass-Based Study of the Effects of Urbanization on the Water Balance Components at Regional and Quadrat Scales in Beijing, China. Water 2017, 10, 5. [CrossRef] 77. Salem, A.; Dezs˝o,J.; El-Rawy, M. Assessment of Groundwater Recharge, Evaporation, and Runoff in the Drava Basin in Hungary with the WetSpass Model. Hydrology 2019, 6, 23. [CrossRef] 78. Batelaan, O.; De Smedt, F. GIS-based recharge estimation by coupling surface–subsurface water balances. J. Hydrol. 2007, 337, 337–355. [CrossRef] 79. Niswonger, R.G.; Panday, S.; Ibaraki, M. MODFLOW-NWT, A Newton Formulation for MODFLOW-2005; U.S. Geological Survey: Reston, VA, USA, 2011. 80. Winston, R.B. ModelMuse-A Graphical User Interface for MODFLOW-2005 and PHAST: US Geological Survey Techniques And Methods 6-A29; U.S. Geological Survey: Reston, VA, USA, 2009. 81. Guimerà, J.; Vives, L.; Carrera, J. A discussion of scale effects on hydraulic conductivity at a granitic site (El Berrocal, Spain). Geophys. Res. Lett. 1995, 22, 1449–1452. [CrossRef] 82. Zhang, Y.; Gable, C.W.; Person, M. Equivalent hydraulic conductivity of an experimental stratigraphy: Implications for basin-scale flow simulations: Equivalent conductivity of experimental stratigraphy. Water Resour. Res. 2006, 4, W05404. [CrossRef] 83. Zhang, Y.; Person, M.; Gable, C.W. Representative hydraulic conductivity of hydrogeologic units: Insights from an experimental stratigraphy. J. Hydrol. 2007, 339, 65–78. [CrossRef] 84. Poeter, E.E.; Hill, M.C.; Banta, E.R.; Mehl, S.; Christensen, S. UCODE_2005 and Six Other Computer Codes for Universal Sensitivity Analysis, Calibration, and Uncertainty Evaluation; U.S. Geological Survey: Reston, VA, USA, 2006; p. 283. 85. Banta, E.R. ModelMate—A Graphical User Interface for Model Analysis; U.S. Geological Survey: Reston, VA, USA, 2011; p. 26.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).