| AridZone Recharge Distributed Aquifer Recharge Enhancements inArid ZonesEnrique R. VivoniRalph M. Parsons LaboratoryMassachusetts Institute of Technology[Abstract] [Introduction][Recharge] [BARS] [MODFLOW][Application] [SensitivityTests] [Results] [Conclusion][References]AbstractEnhanced aquifer recharge is one alternative to the water sustainabilitycrisis occurring in many arid regions. Intermittent and intense rainfallevents over an arid watershed can lead to short term surface water availability.Without the proper management of this water resource, the excess precipitationcan be quickly lost to the high evaporative environment or lost from thewatershed via runoff. By ensuring that the available surface water remainswithin the catchment in the form of stored groundwater, a sustainable fluxof water is obtained for the region (Eltahir, 1996). Sustainability, definedin this perspective, allows a water resources manager to focus on ensuringa consistent yield corresponding to the climatically variable input.This paper explores the possibility of enhancing natural aquifer rechargeby implementing a variant to time tested hydrologic technologies used forcenturies in various arid watersheds. The groundwater model, MODFLOW,is used to simulate an idealized catchment where a component of the BranchedAquifer Recharge System (BARS) have been suitably modeled. Comparisonsto an identical catchment without the proposed system reveal the effectof the enhanced recharge to the aquifer levels. The efficiency of the aquiferrecharge is measured as the sustainable supply of water at a supply wellfor various climatic conditions.Modeling results suggest that BARS is superior to the homogeneouscontrol case in distributing recharged water to the extraction sites. Asthe intermittency of the incident rainfall increases, so does the BARSperformanceduring storm events, a feature that makes the branched recharge systeman attractive alternative for arid and semiarid catchments. The simplifiedmodeling of the recharge system discussed here is a first step towardscombining our current hydrological understanding with hydrologic engineeringtechnology to achieve a sustainable management of water resources in aridenvironments.Keywords: Arid zone hydrology, aquifer recharge, water harvesting,sustainability, BARS, groundwater modeling, MODFLOW.IntroductionThe hydrology of arid regions has not received as much attention asother climatic regions (Scanlon et al., 1997). Ironically, the focusof hydrologic research has not been the highly water stressed regions ofthe world where the importance of sustainable water resource managementis greatest. It is only recently that the hydrologic community has attemptedto understand the variability in fluxes and processes that occur withinarid and semi-arid basins (Nash, 1999) through intensive field campaignssuch as HAPEX-Sahel (e.g. Desconnets et al., 1997) and SALSA(e.g. Goodrich et al., 1998). In addition to smaller scalefield experiments (e.g. Abu-Awwad and Shatanawi, 1997, Cattle, 1997),numerical modeling studies (e.g. Gore et al., 1998, Giao, 1999)and theoretical developments, more attention is recently being placed onthe field of arid zone hydrology. A recent review by Scanlonet al.(1997)highlights some of the major issues in flow and transport within arid,unsaturated systems. A common thread in each new discovery has been therealization that hydrologic processes in an arid or semiarid watershedare distinctly different from their humid climate counterparts.By definition, arid watersheds differ from humid watershed in theirrelative amount of precipitation as compared to the evaporative flux. Followingthe classification of Potter (1992), arid and semiarid regions are thosesubject to precipitation to evaporation ratios (P/E ratio) smaller than0.5, and between 0.5 and 1.0, respectively. In addition to a lower netprecipitation rate, rainfall events occur as infrequent, short duration,high intensity storms that bring a major portion of the annual rainfallto the surface during a very short period of time. Flash flood events maybe a direct result of this type of storm over an arid or semiarid watershed. Under these conditions, the surface layer is unable to infiltrate the incidentrainfall, resulting in precipitation excess surface flows that propagaterapidly through the watershed. Even for low intensity rainfall events,the surface crust that develops on arid watersheds can lead to significantsurface runoff (Abu-Awwad and Shatanawi, 1997). Once on the surface,water in an arid region is subject to a high evaporative demand from thelow humidity and high temperature environment. In many cases, thesurface flows never reach the valley bottom (e.g. Lavee et al.,1997).The thickness of the unsaturated zone is another distinguishing featurein arid watersheds (Scanlon et al., 1997). The presence of a deepwater table allows for unsaturated water or vapor flux to occur in eithervertical direction depending on the surface conditions. This may have theeffect of preventing infiltrated water from percolating into the underlyingaquifer. Coupled to the reduced availability of surface water, the variousloss mechanisms in the unsaturated zone (evaporation from the surface layer,uptake by plant roots, upwards vertical vapor and water flux, etc.) reducethe possibility of water reaching the underlying deep aquifer. Thus, thenatural aquifer recharge in an arid environment hinges on the downwardsvertical flow of water in the unsaturated zone and an excess availabilityof water above the evapotranspiration demand.Water availability in arid regions is both sporadic and highly variablein intensity. Consider a flash flood in desert setting. The input intothe system is extremely small and infrequent, possibly a thunderstorm ofhigh intensity lasting only a few hours, while its response, a flood wavepropagating over a crusted surface, occurs rapidly and may be of greatmagnitude. In order to manage this water resource sustainably (i.e.maintaina constant flux of water into the system), one must exert some degree ofcontrol over the system by altering its response time and storage capacity.While the precipitation inputs are not alterable, many possibilities existin modifying the watershed surface, aquifer properties and storage capacity.One alternative explored in this paper through a numerical modeling studyis to enhance the natural recharge to the underlying aquifer.Recharge Enhancements through Hydrologic TechnologyEnhanced natural recharge is defined here as the flux of water intogroundwater storage under conditions exceeding those imposed by the localhydrologic flux through the undisturbed unsaturated zone. In this regard,an enhancement to the natural recharge consists of providing mechanismsfor an increased vertical flow across the unsaturated zone and into long-termaquifer storage. Increasing the recharge rate can be achieved eitherthrough a decrease in the water loss in the upper soil layers or an increasein vertical flux through the surface or within the unsaturated zone. Theshort and long term increase in aquifer storage can be provided eitherby a rising water table or engineered storage structures in the unsaturatedzone. The enhanced aquifer recharge system proposed here is a low technologyand low energy consumption alternative to the artificial recharge systemspresently implemented in many arid regions of the world.Artificial aquifer recharge has become a popular alternative to manywater supply and wastewater disposal problems. Bouwer (1998) and Kimrey(1989) review the current state-of-the-art in engineered recharge systemsin the United States, while Raju (1998) reviews the artificial rechargesystems implemented in India. In addition, hydrologic technologies forwater harvesting have been implemented in various forms in Europe, theMiddle East, Northern Africa, and Eurasia. Water harvesting systems differin the hydrologic approach taken to the issue of capturing and storingrain water. Some attempt to increase surface runoff into storage or collectionareas by decreasing infiltration (e.g. van Wesemaelet al.,1998, Ciuff, 1989), while others depend on increasing infiltration intothe aquifer along the route of travel of surface runoff (Raju, 1998). Still others, rely on engineered structures to collect and pump water intothe deep aquifer (e.g. Verma and Sarma, 1990). A review of the existingapproaches reveals one major hydrologic deficiency, the exposure of thewater to a high evaporative environment via surface runoff collection andponding. The conceptual hydrologic plan proposed here relies on minimizingthe exposure time to surface conditions of an incident rainfall by providingfor an efficient mechanism of transporting precipitation into the deepaquifer.Artificial recharge systems are not a new solution in arid regions.In fact, various hydrologic technologies were implemented hundreds of yearsago in portions of Iran, Israel, Spain and India (see Fitzmaurice, 2000).Despite the lack of knowledge regarding hydrologic processes, earlier civilizationsestablished in arid regions were able to modify the natural environmentto provide a source of water for the population. Many of these systemsevolved as a result of initial empirical observations and years of trialand error until an efficient engineering design was developed. Some ofthese systems are still in use today, while others are quickly being replacedby modern technology.Very few attempts have been made to study these ancient systems usingour present hydrological understanding of rainfall-runoff processes, infiltration,saturated and unsaturated flow and evapotranspiration. In one of the fewstudies of its kind, van Wesemael et al. (1998) describe the relationshipbetween geomorphic location of a hydrologic engineering system and thewater yield using a simplified curve number (Soil Conservation Service,1986) approach to surface runoff. With the advanced knowledge concerninghydrological fluxes presently available, a critical look at these ancienthydrologic engineering systems should lead to: a) a theoretical understandingof the empirically arrived methods and b) improvements to these systemsbased on current hydrologic understanding.Branched Aquifer Recharge System (BARS)The proposed hydrologic engineering system is designed to enhance thenatural recharge into a deep aquifer in an arid region by taking advantageof the time tested ideas implemented in ancient systems and incorporatingmodern-day hydrologic understanding. Hydrologic engineering systems canbenefit greatly from the low-energy, sustainable systems developed in thepast as a means of providing a long-term supply of high-quality water withoutthe need for modern technology. The various elements of the proposed systemare shown schematically in Figure 1and consists of the following elements:Hillslope surface runoff collectors with high infiltration capacities.Branching network of distribution lines for unsaturated zone transmissionand storage.Topographic convergence zone collectors of surface and subsurface runoff.Transmission line to supply well for sustainable flux of groundwater.The Branched Aquifer Recharge System (BARS) is an engineered hydrologicsystem that relies on various principles to enhance the natural rechargeto an arid zone aquifer. The four principle components outlined aboveimplement the following hydrologic concepts:Hillslope collectionOrographic effects on precipitation tend to create different hydrometeorologicalconditions on hillslopes (Bras, 1990). The windward side of a mountainrange may receive significantly more rainfall than the leeward side orregions at lower altitude. In an arid environment, this effect may be evenmore significant (van Wesemael et al., 1998). BARS takesadvantage of the increased short-term availability of water on hilltopsby placing hillslope collectors in high altitude locations within a watershed.These collectors are themselves placed in local depressions, assuring thatsurrounding flow converges into the sinkhole opening. The collector headsare designed to have a high infiltration capacity by placing a highly porousand conductive media. One potential material is a large-diameter rubbleor gravel media. The gravel media also has a high heat absorbance whichreduces the available surface energy for evaporation. Since the media isa poor substrate for vegetation growth, the possibility of transpirationlosses are also reduced.In order to maximize the infiltration into the hillslope collectors,these must be placed optimally to capture the maximum amount of hillsloperunoff while reducing the exposure time for the surface flow. This mayoccur naturally along a surface-sealed hillslope, or may require surfacetreatment upstream of the hillslope collector. Regardless, the collectorwill effectively serve as a sink for runoff, increasing the infiltrationcapacity through the additional macropore space in the collector whileassuring that ponding conditions on the surface do not occur. Themacropore system enhances the vertical flux of water by providing for shortterm storage of the rainfall pulse and circumventing the tightly packedsoil matrix. Time to ponding has been shown to be increased as a surface-sealeddesert soil is perforated and its macroporosity increased (Cattle, 1999).Branched networkWith a substantial amount of surface flow reaching the hillslope collectors,the distribution lines must be sized to allow for a proper storage capacitywithout incurring in surface ponding. The distribution lines consist ofan underground system of tunnels constructed between the hillslope andconvergence collectors. The intended purpose is to provide regions of preferentialflow and storage for runoff in the unsaturated zone. In an effort to useBARSas a low technology system, the distribution lines may be simple tunnelslined with an impermeable layer and filled with a permeable, highly conductiveand porous media. One simple solution is a clay lined tunnel consistingof sand or gravel media. Similar systems called ghanats have beenbuilt in the arid regions of Iran to transport aquifer water from a hillslopeto a population center by gravity flow (Farshad and Zinck, 1998). Thesesystems consisted of vertical, open shafts connected by a underground tunnelalong the hillslope gradient. Although effectively used for centuries,the incorporation of various elements in BARS should improve theefficiency of the underground tunnel system:Porous media filled tunnels provide for decreased air entrance and reducedevaporation.Ambient temperatures in tunnels are maintained similar to surrounding soilmatrix.Surface exposure at outlet eliminated through recharge to aquifer.Branching network designed for increased storage and system robustness.A potential solution to the need for short term storage within reasonablesystem dimensions is the use of a branching network design for the BARSdistribution lines. Within this network, each hillslope collectoris connected to each other and to the convergence collector located intopographic depressions. The branching network provides for an increasein unsaturated zone storage as well as adds robustness to the system designby distributing an input from any node to the entire system. Network designis an efficient method for collecting the water from a distributed sourceand transporting it to a single outlet. Significant advances regardingthe behavior and properties of network systems have been made in the lasttwo decades, in particular as related to river networks (Rinaldo etal., 1998), which may serve as a benchmark for studying the hydrologicbehavior of this unsaturated zone network.Topographic convergenceIn a study of the relationship between topography and unsaturated flowin an arid region, Scanlon et al. (1999) showed that water fluxeswere highest beneath topographic depressions that periodically flood asa response to local, intense, short duration storms. This is due to thefact that topographic convergence zones in a watershed provide a mechanismfor surface and subsurface flow concentration. The relationship betweensurface topography and hydrologic flux in a watershed is a well understoodand widely used concept in hydrological modeling. The TOPMODEL approach(Beven and Kirby, 1979) is a popular statistic-dynamic model for subsurfaceflow based on the distribution of topography in watershed. It predictsthat flat depressions with large contributing areas will behave as convergencezones, in accord with the variable source area concept.BARS takes advantage of the topographic gradient by placing largercollectors in topographic depression where subsurface flow from the distributionand storage lines is concentrated into one location with a minimal amountof energy expenditure. In addition, surface flow not captured by the hillslopecollectors naturally flows into these convergence collectors whose surfaceinfiltration conditions are designed similar to their hillslope counterparts. Flow concentration within the unsaturated zone assures a decrease in evaporationdue to a lower surface area to volume ratio, the decrease in exposure tothe high evaporative demand atmosphere and a reduced evaporation due toa lower temperature and increased moisture content.The flow concentration within the unsaturated zone also assures thatlarge rainfall pulses are stored underground, thus avoiding the evaporativedemand of the surface. The entire branched network serves as a temporarystorage of harvested water, while the convergence collector has the dualpurpose of transmitting this water into the underlying aquifer via a deeppercolation tunnel. It is in the convergence zone that the underlyingaquifer will receive the collected runoff from the catchment. A groundwatermound will form underneath the convergence collector as the water is slowlytransmitted through the phreatic aquifer. As the static water level inthe aquifer increases, any overflow from the deep percolation tunnel maytransported through a transmission line to lower parts of the watershed.This is illustrated in a schematic picture in Figure2.Transmission line and supply wellThe BARS system is envisioned as a potential source of short-termand long-term water supply needs. This is achieved by using the collectorcomponents as long-term recharge units to the aquifer and short-term reservoirsfor water. If the amount of stored water in the collector exceedsa particular water level, an overflow structure redirects the excess waterinto a transmission line that feeds a supply well. During periodof water excess, the system can provide water quickly without the needof an energy consuming, deep extraction well. During period of watershortage, BARS is used exclusively to recharge the underlying aquifer,and extraction is performed using a conventional well system. Overflowstructures regulate the short or long-term water storage and provide asustainable water supply at various time scales.A preliminary evaluation of the Branched Aquifer Recharge System isimplemented in a numerical study using a groundwater model for a simplesystem. The major design criteria are incorporated into the groundwatermodel and a sensitivity test is performed to assess the effectivity ofBARSinrecharging an idealized arid zone aquifer. A more complete design of theBARSsystem would include a hydrologic analysis to determine the probabilityof rainfall occurrence and quantity in a region, modeling the rainfall-runoffresponse given the appropriate characterization of the surface conditions,a hydraulic design to determine the physical dimensions of the system componentsand the behavior of the system to different forcings, and modeling themovement of water within the BARS system and the effect on the aquiferlevel. In this preliminary study, however, the implementation ofBARSwithin MODFLOW is limited to a simple system consisting of one hillslopecollector with a branching network of storage and distribution lines intoan underlying aquifer, as illustrated in Figures5 and 6.MODFLOW Groundwater ModelThe MODFLOW groundwater flow model is applied in this study toan idealized arid watershed in order to evaluate the impact of the BARShydrologicengineering system on the aquifer recharge. MODFLOW (USGS ModularThree-Dimensional Ground-Water Flow Model) is a recognized standard groundwatermodel that has been in development and use for the past three decades.It is capable of simulating a wide variety of groundwater flow and transportproblems and its modular design has permitted developers to add capabilitiesas the need has arisen. For example, MODFLOWhas been modified foraquifer remediation and biological transport, uses that were not envisionedduring its initial conception. MODFLOW has been subject to rigorouspeer review and scrutiny in courts, regulatory agencies, universities andwithin the consultant community, making it the most well known and testedgroundwater model, and a suitable benchmark for this study.For this study, a user interface program, VisualMODFLOW (Waterloo Hydrogeologic,Inc.) has been used to simplify the creation of input files, modeloperations and results. While Visual MODFLOW is a useful utility for settingup and running the model, it has very limited capabilities for presentingresults. MODFLOW output was linked in this study to a Geographic InformationSystem (ArcView GIS) for storage, manipulation and presentation of thespatially variable input and output data. The enhancement of MODFLOW througha GIS interface has recently been developed at the USGS(1998).The preparation of the MODFLOW groundwater model consists of the followingbasic tasks:Creation of three-dimensional model grid (x,y,z) cells of (x, y,z)size.Creation of surface elevation model from DEM or interpolation between knownpoints.Creation of underlying layer elevations.Description of default soil hydraulic parameters: Hydraulic conductivity,Porosity, Specific yield, Specific storageDescription of flow boundary conditions: Constant head, River head, Drainagehead, among others.Placement of pumping, recharge or observation wells. Definition of pumpingrates and time-series.Description of flux boundary conditions: Recharge and Evaporation at toplayer, No flux boundaries.Specification of zones for water balance calculations.After setting up the model in its most basic form, a set of experimentsis designed to model the behavior of the system under alternative conditions,depending on the purpose of the modeling exercise. This may involvechanges in pumping rates and strategies, variable climatic inputs, soilhydraulic behavior variations within the model layers, among many otherpossibilities. In this study, a Homogeneous case is designed by going throughthe basic tasks outlined above. The major variation to the model is theimplementation of the Branched Aquifer Recharge System. A comparisonbetween the Homogeneous case and the BARS case under various climaticconditions will reveal the efficiency of the proposed hydrologic engineeringsolution to the water crisis in arid zones.Model ApplicationGroundwater modeling studies of artificial recharge to aquifers in aridcatchments have been performed by several researchers. The studies of Helleretal. (1999), Tompson et al. (1999), Bekesi and McConchie (1999),Munevar and Marino (1999) are some good examples of the various techniquesand models used for this purpose. Modeling of the nature attempted in thisstudy has not been performed specifically for the purpose of assessingrecharge to an underlying aquifer. In particular, the implementation ofthe Branched Aquifer Recharge System in MODFLOW is based on theidea that the tunnel network can be modelled as zones of high hydraulicconductivity contained within layers or walls of low permeability. Sincethis is rarely seen in nature and has not been considered as an alternativefor artificial recharge, modeling studies have not been performed. In thissense, this study is a first of its kind.One potentially analogous system to BARS are the undergroundkarstic caverns common to many regions of the world. Given their hydrologicalimportance in karstic areas, modeling efforts of the large subterraneouscaves and rivers have been made. A literature review on the subject hasrevealed that the scientific community is just beginning to understandthe interaction of a groundwater system with these large, open, highlypermeable cavities. Halihan and Wick (1998) present a simple reservoirmodel for the flow within the karstic aquifer based on some earlier fieldwork by Halihan et al. (1998). A careful assessment of the similaritiesbetween both flow systems may lead to gaining some insight regarding thehydrologic behavior of BARS. Nevertheless, the use of a system suchas BARS as an alternative for recharging the aquifer is a novelapproach.The implementation of BARS for enhanced natural recharge of anaquifer in MODFLOW consisted of creating a simple, idealized landscapeand modeling the impact of the recharge system on the aquifer levels throughvarious case studies and a control case. The idealized system is a threedimensional grid containing 50 cells by 50 cells in the horizontal planeand 10 layers in the vertical direction. The grid cell size is a constantspacing of 50 meters such that the simulation area is 2.5 kilometer by2.5 kilometer in extent. The layers vary in depth from 1 meter spacingnear the low-lying regions to tens of meters in the underlying aquifer.An idealized topography was created for the small arid watershed to simulatethe effect of surface slope on the groundwater flow based on a simulateddigital elevation model (DEM). The surface topography was implemented byspecifying the elevation of a group of selected points and interpolatingthe grid cell elevations using a weighted inverse distance scheme to thenearest ten neighbors. The elevations in the DEM range from 50 meters inthe corners of the idealized hillslope to 250 meters in the central portion.Using a hillslope length scale of approximately 1800 meters, the DEM hasa nominal surface slope of 0.1 or 10%. A contour map of the topographyis shown in Figure 3 with the locationof the specified elevation points denoted by crosses. For a more realisticview of the surface topography, a three dimensional rendering of the idealizedcatchment is shown in Figure 4.The hydrologic engineering system was implemented by varying the hydraulicconductivity and storage capacity within the various soil layers in thecatchment to accommodate the design of a simple branched distribution rechargenetwork, as shown in Figures 5 and 6.As can be seen from the top and side views of the BARS model, thispreliminary study is considering an extremely idealized case. In an attemptto establish a benchmark for future work, the BARS system consistsof one collector on a hillslope leading to four branched storage and distributionlines. This is thus a model of one possible BARS node within onehillslope receiving rain at a higher rate than its surroundings, eitherdue to an orographic rain effect or to topographical convergence. An attemptwas not made to model the interconnection of multiple nodes, the surfacewater concentration into a hillslope collector or an overflow structureor transmission line leading to a supply well.Two cases were initially considered, a control case with homogeneoussoil conditions over the arid watershed and a BARS base case withthe implemented changes in soil parameters designed to model the rechargesystem. To properly set up the control case and the BARS base casein MODFLOW, the following parameters were specified based on literaturevalues for various soil textures and climatic forcings to an arid zonewatershed (Freeze and Cherry, 1979, Eagleson, 1970):Isotropic hydraulic conductivity values (Kx = Ky =Kz= K):K = 1e-5 m/s Catchment default (silty sand)K = 0.01 m/s BARS interiorcells (gravel)K = 1e-11 m/s BARS exterior cells (clay)Specific storage, specific yield, porosity values:Ss = 3e-6 (1/m), Sy = 0.1, n = 0.5 Catchment default and BARS exterior cellsSs = 3e-6 (1/m), Sy = 0.1, n = 0.7 BARS interior cellsNo flux boundary condition along edges of the simulation domain.Constant initial head in all grid cells: Hi = 45 meters. For low-lyingareas, this implies an initial depth to water table of approximately 5meters. For the higher hillslope areas, the water table is extremely deep,approaching 245 meters at the highest DEM elevation. These initial conditionsare considered representative of an arid catchment.Rainfall simulated as a recharge rate applied exclusively to the top activesoil layer:R = 0 for all time within the majority of the catchment.R = Ri at the BARS recharge heads for the BARScase.R = Ri at the BARS collector head for the homogeneouscase.Ri varies as a step function from 0 m/yr during interstorm periodsto 2m/yr during storm periods.Evapotranspiration values applied to top layer: E = 2 m/yr for alltime. Evapotranspiration flux, however, only occurs if the water tablesaturates the ground surface or is within the extinction depth, specifiedtobe one meter. For arid conditions with deep water tables, this flux isnegligible.Four supply wells operating at constant pumping rate of 250 cubic metersper day each during the simulation period.A transient simulation was performed for a total time of 180 days duringwhich alternating storm and interstorm periods were simulated. In evaluatingthe performance of the model to various simulation periods, initial conditions,pumping rates and climatic conditions, the parameters specified above demonstratedthe most interesting behavior. Whether to use a steady-state or a transientsimulation, for example, was tested under various conditions. The transientsimulation case proved to be far more interesting. A steady state simulationresults in a low-lying groundwater table greatly impacted by the constantpumping rate. The initial and boundary conditions, pumping rates and climaticconditions were chosen such that:The groundwater flow in the arid catchment is a closed system where allthe inputs and outputs are accounted for.The rate of extraction and recharge into the system is such that the dynamicimpact of the recharge is observable. Higher pumping rates create conditionsunder which the recharge effect on the groundwater table are negligible.A spatially-uniform initial water table was specified so that the effectsof recharge and withdrawal from the aquifer could be evaluated independentlyof the regional or large scale changes occurring if a spatially-variablegroundwater table were initially imposed. Water table head differencescause horizontal and vertical flows unrelated to recharge.The intermittency of the recharge rate is used as a parameter to evaluatethe effect of BARS under various climatic forcings.The incident rainfall is modeled in MODFLOW as as recharge rateonto the either the top active soil layer or the surface. Since MODFLOWdoes not simulate unsaturated groundwater flow, specifying a recharge orevapotranspiration onto a dry surface does not result in the anticipatedimpact to the deep groundwater aquifer. For this reason, this simulationspecified the recharge rate onto the top active soil layer, that whichcontains saturated conditions for the two cases considered. For BARS,this recharge is specified at the recharge heads as shown in Figure5 by the four dark rectangles. For the homogeneous case, recharge isspecified at the hillslope collector head, illustrated in Figure5 by the central hatched square. Due to the modeling limitations ofMODFLOWforimplementing the conceptual BARS design, this preliminary studyassumes the following:The rainfall rate onto the hillslope collector is transmitted instantaneouslyto the lower recharge zones. In BARS, these consist of the fourrecharge heads, while in the homogeneous case this is at a single centrallocation.The homogeneous case acts as an unbranched distributor of rainfall to theaquifer, while BARS acts as a branched recharge distributor to theaquifer. Conceptually, the impact of BARS should be to smoothenthe spatial variability in aquifer levels and provide a higher head tothe locations of water extraction.Identical total recharge amounts are maintained between the BARSand homogeneous cases by doubling the recharge rate for the homogeneouscase since it consists of half the number of recharge grid cells. The massbalance calculations performed substantiate that both recharge quantitiesare identical.The dynamics of the flow and storage within the BARS tunnels orbetween the hilltop and the water table are not modeled. It is assumedthat these are full and have inflows identical to outflows. In other words,the rainfall incident on the hillslope collector is modeled as being exactlyrepresented by a discharge (recharge) from the systems.More importantly, it is assumed that the homogeneous case is equally efficientin transporting infiltrated rainfall into the underlying aquifer. Thiswill not necessarily be true since BARS is designed to enhance thevertical transport and storage of high intensity stormwater by eliminatingthe various loss mechanisms expected in the homogeneous case. This preliminarystudy represents an extremely good performance of the control case.For this simulation, the recharge conditions were constant over each stormperiod and negligible during interstorm periods. The only spatial variationallowed on the rainfall or recharge rate was the orographic effect ontothe hillslope collector. The evapotranspiration rate is modelled inMODFLOWby assigning a spatially and temporally constant rate to the top soil layer. Variations to the climatic forcing are considered in the sensitivity studiesdescribed in the following section.Sensitivity TestsThe MODFLOW groundwater model was utilized to evaluate the effectof the Branched Aquifer Recharge System on aquifer levels during differentclimatic forcings. In order to assess how BARS impacts aquifer levels,a base case comparison was made between a homogeneous soil case and theBARSbase case. The homogeneous case simulates the centralized or unbrancheddistribution of recharge to an aquifer while the BARS case simulatesthe effect of distributing the recharge over a larger region of the aquifer.The effect of the branching network is explicitly tested by comparing thesetwo cases for a variety of climatic forcings. The input parameters forthe comparisons are those outlined in the Model Application section. Themodels were run for the length of the simulation time and the results evaluatedbased on four criteria:Contour maps of water table depth at the end of the simulation time (t=180 days).Time series of aquifer levels at three locations:BARS Recharge head (x=625m,y=1250m).BARS Hillslope collector (x = 1250m, y=1250m).Supply Well (x=y=625m).Water balance computations for the simulation period in accumulated volumes.Water exchange rates between three zones:Zone 1: Aquifer zone : Catchment grid cells.Zone 2: Extraction zone : Supply Well grid cells.Zone 3: Recharge zone : Recharge zone grid cells.In addition, contour maps of the groundwater flow direction and magnitude,as well as the estimated recharge into the aquifer were selectively usedto evaluate the performance of the BARS case over the homogeneouscase. Given the three dimensional nature of the groundwater flow in thebasin and the transient nature of the simulation, it is difficult to graphicallyshow all the model results. For this reason, an attempt is made to conciselydescribe the effect of the BARS system on the arid zone aquiferrecharge by utilizing horizontal and vertical cross sectional views, aquiferlevel time series at various points and mass balance computations for thesimulation period.The scientific questions that this modeling study attempts to answercan be expressed as follows:How does the distributed aquifer rechargeprovided by BARS impact aquifer levels under a variety of climaticforcings? What effect does rainfall intermittency have on its performanceas compared to an undistributed recharge?To provide some insight into these two questions, four climatic scenariosare considered for the BARS and homogeneous cases. A sensitivitystudy was performed to the precipitation forcing by altering the rechargetime series. Four cases were considered:Case 1: Constant recharge rate R = Ri = 2m/yr for BARS andR= Ri = 4m/yr for control case.Case 2: Two recharge periods with R = Ri m/yr during stormsand R = 0 during interstorms.Case 3: Six recharge periods with R = Ri m/yr during stormsand R = 0 during interstorms.Case 4: Three recharge periods with R = 2Ri m/yr during stormsand R = 0 during interstorms.The climatic conditions for the arid catchment are shown in Figure7. The reasoning behind the choices of these recharge rates are straightforward.These four cases are designed to observe the following effects in an aridwatershed characterized by a deep water table:Orographic rainfall effect on an arid mountain. (Cases 1-4)Constant rainfall for an extended period. This case simulates the wet ormonsoonal season and represents the most humid watershed conditions. (Case1)Alternating wet and dry periods. These cases simulated varying degreesof rainfall intermittency observed in most arid and semi-arid watersheds.(Cases 2-4)Rainfall intensity versus rainfall intermittency effect. Case 3 and 4 havethe same volume of recharge but with different storm intensity and recurrenceintervals.Constant water extraction from the four wells at 250 cubic meters per dayto simulated demand on aquifer levels and address the issue of a sustainablegroundwater supply. (Cases 1-4)Due to the short simulation period, the deep initial groundwater tableand the constant withdrawal rate from the aquifer, it is difficult to describeeach case as representative of arid, semiarid or humid climatic forcings.Though large (on the order of 2-4 meters/yr), each rainfall input occursover a limited portion of the catchment, is intermittent in nature andforces a system that initially has low storage. For these reasons, it canbe argued that all the cases represent rainfall over an arid climatic condition.The Results section will discuss the rainfall effect on these initial conditionsover the 180 day simulation period.It is also interesting to evaluate how varying the spatial distributionof the aquifer recharge affects the available supply of water to the extractionwells. This evaluation is thus performed in tandem with the comparisonof the two cases for the different climatic conditions. In this manner,the efficiency of BARS in increasing the available flux of waterout of the aquifer, the sustainability condition defined previously, canbe evaluated.Results and DiscussionThe efficiency of the aquifer recharge system can be measured by thesustainable supply of water provided to a supply well. One way of evaluatingthis effect in this modeling study is by comparing the water table levelswithin various regions of the aquifer. The spatial and temporal variabilityof the water table level can be analyzed through contour maps for a specifiedtime instance or by a time series plot of aquifer level at a specifiedlocation. Alternatively, the amount of water transported between variouspredefined zones can give an indication of the origin of the extractedwell water. This technique can supplement a mass balance calculation forthe closed system and is performed in MODFLOW through the ZONEBUDGET package. Tables 1 and 2 presents the computed mass balancefor each climatic case and recharge system. In addition, the computed volumetricflowrate between the three water budget zones (Catchment: Zone 1; RechargeZone: Zone 2; Supply Well: Zone 3) is shown. Mass Balance Term(cubic meters)Case 1Case 2Case 3Case 4 Storage Change49672.159544.859549.659540.6Recharge19726.29863.19863.19863.1Well Extraction69419.569419.569419.569419.5Transport Term (cubic meters/day)Case 1Case 2Case 3Case 4 Zone 1 - 3385.2385385.1385.2Zone 2 - 11100.891.772.6Table 1. Homogeneous, unbranched recharge system Mass Balance Term(cubic meters) Case 1 Case 2 Case 3Case 4 Storage Change49669.55953859544.159523Recharge19726.29863.19863.19863.1Well Extraction69419.569419.569419.569419.5Transport Term(cubic meters/day)Case 1 Case 2 Case 3 Case 4 Zone 1 - 3385.2385385.2385.1Zone 2 - 1110.53.283.94.46Table 2. Branched Aquifer Recharge SystemThe MODFLOW model runs for the climatic forcing cases in eachrecharge system are summarized here by using the techniques discussed inthe Sensitivity Tests section. Figure 8and Figure 10 show plan views of thewater table depth contours for the Homogeneous (unbranched) andBARS(branched) recharge systems. The water table level is measured from a datumof z = 0 at the lowest soil layer. Unfortunately, the transfer ofthe VISUAL MODFLOW AutoCAD graphics to ArcView GIS resultsin the loss of the contour labels. For this reason, the purpose of thesegraphics is to qualitatively show the impact of rainfall variability onthe simulated water table level for the two systems. More quantitativemeasures will be provided for specific observation wells within the catchmentdomain.Figure 9 and Figure11 show side views of the water table depth and groundwater flow directionfor a specific cross section through the BARS Recharge Head andthe BARS Collector Head. These simulation results only representthe system behavior at a specific time, t = 180 days, the end ofthe simulation period. Changes in the groundwater table elevations, flowdirections and magnitudes occur throughout the transient simulation. Amore thorough analysis would be required to determine if case comparisonsat other time periods lead to extremely different behaviors. The aquiferlevel time series at the three specific points suggest that this is notthe case.The aquifer level was monitored for all time periods at three locations:one supply well, one recharge head and the hilltop. Due to the radial symmetryinherent in the system, these results are representative of conditionselsewhere. The time series of aquifer level at these three locations areshown in Figure 12 for the Supply Well,Figure13 for the Recharge Well Head and Figure14 for the Hilltop. Each figure shows the results from the BARSand Homogeneous cases for the various climatic conditions discussed inthe Sensitivity Test section. As previously discussed, the water levelvalues represent the depth from the datum z = 0 at the bottom layer.In inspecting these figures, it is important to remember that an initialcondition of Hi = 45 meters was uniformly applied to the catchment.In addition, the constant pumping rate in the four supply wells imposesa general decrease in aquifer levels throughout the catchment. This effectis only counteracted by the recharge supplied by the two systems. Is thedistributed recharge provided by BARS more effective than the concentratedrecharge occurring for the homogeneous case? Figure15 and the results in Tables 1 and 2 should help in answeringthe first of our scientific questions.Distributed Recharge EffectA prominent effect of recharge distribution is observed due to the branchingnetwork of the Branched Aquifer Recharge System. The branching networktransports a concentrated rainfall input occurring at a hilltop due toan orographic effect or in a depression due to topographic convergenceto distant locations within the catchment. This recharge dispersal effectincreases the aquifer level near the supply wells and enhances the amountof available of water for extraction. This can be clearly seen in Figure15 where a comparison between the concentrated and the distributedrecharge systems is made. The important conclusion provided by the headratio is that the branching network increases the aquifer levels slightlyfor all times at the supply well, where it is most desired, at the expenseof decreasing the levels more significantly in the recharge zone, whereit is constantly being replenished. The calculated transport between therecharge zones and the aquifer as shown above in Tables 1 and 2 furthersupport this conclusion. The volumetric flowrate from the recharge zoneto the aquifer is consistently higher for the BARS case (Zone 2- 1) for identical mass balances. This implies that BARS is moreefficient in supplying recharge water to the extraction wells. By doingso, BARS increases the sustainable water supply available from thecatchment, as implied by the increased groundwater mound areal coverageshown in the contour maps in Figures 8 and 10. Not only is the spatialextent of the groundwater mound increased due to the distributed recharge,but the temporal distribution of the peak head in the recharge well varies.This is most clearly seen for Case 1 in Figure 13 for the BARScase and Figure 14 for the Homogeneous case. The peak water tablelevel occurs earlier in the simulation period at the recharge zone forthe BARS case, suggesting that the system is capable of fulfillingshort term water needs more quickly due to its distributed nature. Indirectly,the spatial and temporal variations suggests that the distribution of rechargeover a branched network can lead to higher sustainable flux into the aquifer,even for the adverse conditions considered here. It must be noted thata major assumption has been made in that the homogeneous case is able totransport the same amount of water through the unsaturated zone as theengineered hydrological system. Due to the elements of theBARSdesign, this is highly unlikely. Finding that BARS is more efficientin transporting water to a supply well, even under the assumption of equalperformance, is significant.Climatic Forcing EffectThe temporal distribution of the incident rainfall has a profound effecton the spatial and temporal distribution of aquifer levels within the idealizedarid catchment for both recharge systems. The first order effect is thatan increase in the rainfall intermittency, keeping a constant rainfallvolume and pumping rate, reduces the aquifer level at the supply well,as clearly demonstrated in Figure 12. Given that Cases 2, 3 and4 only differ in the degree of intermittency, the clear decrease in aquiferlevels in the supply well indicate the importance of the ratio of interstormperiod length to storm period length. For a constant pumping rate, an increasein the rainfall intensity does compensate for a proportional reductionin storm duration. This is a significant finding that suggests that anarid watershed managed for water supply is more sensitive to the arrivalfrequency of storms than to the amount of rainfall incident upon its rechargezones. A second order effect of the rainfall intermittency can be observedin relation to Figure 15, the comparison of the system rechargeperformances. As compared to the constant recharge rate, intermittencyreduces the effectiveness of BARS in providing an increased aquiferlevel to the supply well at the expense of maintaining higher levels inthe recharge zone. As intermittency is increased among Cases 2, 3 and 4,however, the BARS efficiency increases during pulsed recharge periodsof higher intensity. This is also evidenced by the increase in the volumetricflowrate from the recharge zone to the aquifer in Tables 1 and 2.The first order and second order effects mentioned above work in oppositedirections. The overall decrease in aquifer levels at the supply well isdue to the first order effect of having longer interstorm periods witha constant pumping rate. The magnitude of this effects hides the true sensitivityof the BARS system to rainfall intermittency, that expressed bythe second order effect. As a rainfall pulse becomes more intense and lessfrequent, the recharge zones are more efficient in transmitting the pulsedrecharge. Given the nature of storm arrival in arid watersheds, we expectthat BARS would outperform the homogeneous case during the rechargeevents. Thus, discovering that the performance of BARS as a distributedrecharge system is highly dependent on the temporal distribution of theincident rainfall is a significant finding in this study.ConclusionsCurrently, the use of artificial recharge to an aquifer is seen as apotential solution to the sustainable water supply needs of various aridand semiarid regions. One needs only to inspect the rising number of projectsbeing designed and constructed in the semiarid southwest of the UnitedStates to appreciate the growing popularity of engineered recharge systems(ENR, 1999a,b; WATER/Engineering and Management, 1995, 1999). Other typesof aquifer recharge and water harvesting systems have been implementedin many arid region over the past hundreds of years. Despite this wealthof engineering knowledge, a quantitative hydrological analysis of an artificialrecharge system that considers the impact of the competing hydrologicalprocesses in arid regions has not been performed. This preliminary studytakes a small step in that direction by presenting the conceptual designand initial modeling results of the Branched Aquifer Recharge System (BARS).The Branched Aquifer Recharge System takes advantage of the hydrologicprocesses that favor the increase of vertical transport of water throughthe unsaturated zone, the concentration of flow by topographical constraintsand the reduction of evaporative loss due to the high atmospheric evaporativedemand. Four basic elements comprise the hydrologic engineering technology:hillslope collectors, a branched network of underground tunnels for storageand distribution, convergence zone collectors with an overflow structureto a transmission line and supply well. BARS is designed to combinethe time-tested elements of ancient water harvesting systems (i.e.ghanats)with advances achievable due to our current understanding of hydrologicprocesses in arid catchments.This modeling study concentrates on performing an initial investigationinto the efficiency of the branched distribution system proposed for BARSonthe aquifer levels near the zone of water extraction, as compared to ahomogeneous, unbranched recharge to the same idealized arid catchment.The popular groundwater model, MODFLOW, is used for this purposedespite its limitations in modeling the surface hydrology and the unsaturatedzone flow. By specifying the appropriate recharge conditions, the BARSsystemis mimicked within the MODFLOW modeling environment and tested undera variety of climatic forcings. The case studies demonstrated that theBARSsystem increases the amount of available water to the supply systemand decreases the time for recharge water to be available. These two effectssuggest that the BARS system is capable of increasing the sustainableflux of water into the aquifer even under identical system performances,an unrealistic expectation for the homogeneous, unbranched system. Thecase studies also demonstrated that the intermittency of the recharge intothe system is a crucial parameter that governs the efficiency of BARSin transporting water to the demand sites. BARS is best suited tohandle the highly intense and intermittent rainfall conditions expectedin an arid watershed.AcknowledgmentsThis study was performed to fulfill a term paper requirement for theclass 1.714 Surface Hydrology offered at the Massachusetts Institute ofTechnology, Department of Civil and Environmental Engineering, Ralph M.Parsons Laboratory by Prof. Elfatih Eltahir (Spring Term 2000). Fruitfuldiscussions with E.A.B. Eltahir and D. Collins are acknowledged.ReferencesAbu-Awwad, A.M. and Shatanawi, M.R. 1997. Water harvesting and infiltrationin arid areas affected by surface crust: examples from Jordan. Journalof Arid Environments. 37: 443-452.Bekesi, G. and McConchie, J. 1999. Groundwater recharge modelling usingthe Monte Carlo technique, Manatawu region, New Zealand. Journal ofHydrology. 224: 137-148.Bouwer, H. 1998. Issues in Artificial Recharge. Water Science andTechnology. 33(10-11): 381-390.Bras, R.L. 1990. Hydrology: An introduction to hydrologic science.Addison-Wellesly. Reading, Massachusetts.Cattle. S.R. 1999. Efficacy of perforating the soil to capture and storerain during fallow in dry regions. European Journal of Soil Science.50: 481-487.Ciuff, C.B. 1989. Water harvesting systems in arid lands. Desalination.72: 149-159.Desconnets, J.C., Taupin, J.D., Lebel, T. and Leduc, C. 1997. Hydrologyof the HAPEX-Sahel Central Super-Site: surface water drainage and aquiferrecharge through the pool system. Journal of Hydrology. 188-189:155-178.Eltahir, E.A.B. 1996. Sustainable water resources: concept,definition and example. Ralph M. Parsons Laboratory Department of Civiland Environmental Engineering. Massachusetts Institute of Technology. Unpublishedmanuscript.Engineering News Record. 1999a. Water Campus ready to begin ArizonaAquifer recharge. July. 14.Engineering News Record. 1999b. California landowner proposes aquiferstorage. September. 16-17.Farshad, A. and Zinck, J.A. 1998. Traditional irrigation water harvestingand management in semiarid western Iran: A case study of the Hamadan Region.WaterInternational. 23: 146-154.Fitzmaurice, J. 2000. Ancient hydrologic technologies. MIT Class Project:1.714 Surface Hydrology.Freeze, R. A. and Cherry, J. A. 1979. Groundwater. Prentice-Hall.Englewood Cliffs, New Jersey. 604 pp.Giao, P.H., Phien-Wej, N. and Honjo, Y. 1999. FEM quasi-3D modelingof responses to artificial recharge in the Bangkok multiaquifer system.EnvironmentalModelling and Software. 14. 141-151.Goodrich, D.C., Chehbouni, A., Goff, B., MacNish, B., Maddock, T., Moran,S., Williams, D.G., Watts, C. 1998. An Overview of the 1997 Activitiesof the Semi-Arid Land-Atmosphere (SALSA) program. American MeteorologicalSociety. Special Symposium on Hydrology. Phoenix, AZ.Gore, K. P. , Pendke, M.S., Gurunadha, V.V.S., Gupta, C.P. 1998. Groundwatermodeling to quantify the effect of water harvesting structures in Wagarwadiwatershed, Parbhani district, Maharashtra, India. Hydrological Processes.12. 1043-1052.Halihan, T., Wicks, C.M. and Engeln, J.F. 1998. Physical responses ofa karst drainage basin to flood pulses: example of the Devil's Icebox cavesystem (Missouri, USA). Journal of Hydrology. 204: 24-36.Halihan, T. and Wicks, C.M. 1998. Modeling storm responses in conduitflow aquifers with reservoirs. Journal of Hydrology. 208: 82-91.Hellers, J.A, Guertin, D.P., Miller, S.N. and Stone, J.J. 1999. GISfor watershed assessment: Integrated spatial and tabular data to deriveparameters for a hydrologic simulation model (ARDBSN). Proceedings of1999 ESRI International User Conference. July 26-30. San Diego, CA.Lavee, H., Poesen, J. and Yair, A. 1997. Evidence of high efficiencywater harvesting by ancient farmers in the Negev desert, Israel. Journalof Arid Environments. 35: 341-348.Kimrey, J.O. 1989. Artificial recharge of groundwater and its role inwater movement. Desalination. 72: 135-147.Munevar, A. and Marino, M.A. 1999. Modeling analysis of groundwaterrecharge potential on alluvial fans using limited data. Ground Water.37(5): 649-659.Nash, D.J. 1999. Arid Geomorphology. Progress in Physical Geography.23(3): 429-439.Potter, L.D. 1992. Desert characteristics as related to waste disposal.In: Deserts as dumps? The disposal of hazardous materials in arid ecosystems.C.C. Reith and B.M. Thomson, ed. Univ. of N.M. Press. Albuquerque, N.M.21-56.Raju, K.C.B. 1998. Importance of recharging depleted aquifers: Stateof the art of artificial recharge in India. Journal Geological Societyof India. 51. 429-454.Rinaldo, A., Rodriguez-Iturbe, I. and Rigon, R. 1998. Channel networks.AnnualReview of Earth Planetary Science. 26: 289-327.Scanlon, B.R., Tyler, S.W. and Wierenga, P.J. 1997. Hydrologic issuesin arid, unsaturated systems and implications for contaminant transport.Reviewsof Geophysics. 35(4): 461-490.Scanlon, B.R., Langford, R. P. and Goldsmith, R. S. 1999. Relationshipbetween geomorphic setting and unsatured flow in an arid setting. WaterResources Research. 35(4): 983-999.Soil Conservation Service. 1986. Urban hydrology for small watersheds.Technical Release 55, US Department of Agriculture, Washington, D.C.Tompson, A.F.B., Carle, S.F., Rosenberg, N.D. and Maxwell, R.M. 1999.Analysis of groundwater migration from artificial recharge in a large urbanaquifer: A simulation perspective. Water Resources Research. 35(10):2981-2998.van Wesemael, B., Poesen, J.,Sole Benet,A., Cara Barrionuevo, L. andPuigdefabregas, J. 1998. Collection and storage of runoff from hillslopesin a semi-arid environment: geomorphic and hydrologic aspects of the aljibesystem in Almeria Province, Spain. Journal of Arid Environments.40: 1-14.Verma, H.N. and Sarma, P.B.S. 1990. Design of storage tanks for waterharvesting in rainfed areas. Agricultural Water Management. 18:195-207.WATER/Engineering & Management. 1995. Aquifer recharge enhancedwith rubber dam installations. January. 37-40.WATER/Engineering & Management. 1999. Aquifer recharge: A naturalsolution. January. 30-32. Page created and maintained by Enrique R. VivoniLast modified on: May 15, 2000For comments or information contact: vivoni@mit.edu |
|
