1 23

Hydrogeology JournalOfficial Journal of theInternational Association ofHydrogeologists ISSN 1431-2174 Hydrogeol JDOI 10.1007/s10040-011-0757-7

Using three-dimensional geologicalmapping methods to inform sustainablegroundwater development in a volcaniclandscape, Victoria, Australia

Bruce Gill, Don Cherry, Michael Adelana,Xiang Cheng & Mark Reid

1 23

Your article is protected by copyright and

all rights are held exclusively by Springer-

Verlag. This e-offprint is for personal use only

and shall not be self-archived in electronic

repositories. If you wish to self-archive your

work, please use the accepted author’s

version for posting to your own website or

your institution’s repository. You may further

deposit the accepted author’s version on a

funder’s repository at a funder’s request,

provided it is not made publicly available until

12 months after publication.

Using three-dimensional geological mapping methods to informsustainable groundwater development in a volcanic landscape,Victoria, Australia

Bruce Gill & Don Cherry & Michael Adelana &

Xiang Cheng & Mark Reid

Abstract This study investigated the use of three-dimen-sional (3D) geological methods to provide better groundwaterresource estimates for the Spring Hill area in central Victoria,Australia. Geological data were gathered in 3D geologicalsoftware, which was utilised to derive fundamental dimen-sional parameters of the groundwater system in the study area.Mining industry software and hydrogeological methods werecombined to give volumetric determinations of the basaltaquifer that were used to improve estimates of the ground-water resource. The methods reduce uncertainty about thephysical attributes of the aquifer systems and greatly improveconceptual understanding of their behaviour. A simplenumerical water-balance model was developed to refine theestimates of aquifer volume and fluxes to approximateobserved water-level behaviour in the area. This enabled amuch better comparison of groundwater resource use to thenatural inputs and outputs for the area. A key conclusion wasthat the main issues for sustainable development and use inthe study area are more to do with the physical aspects of theaquifer system, rather than simply the volume of waterpumped. Visualisations of the area’s hydrogeology also

provide improved hydrogeological understanding and com-munication for groundwater users and administrators.

Keywords Australia . Groundwater management . Three-dimensional geological mapping . Sustainable yield .Volcanic aquifer

Introduction

Three-dimensional (3D) geological mapping methods forhydrogeological purposes are now becoming more widelyused. A growing global community of users is exploringhow to exploit 3D computational and visualisationmethods used by the minerals and hydrocarbon geologistsfor groundwater purposes. The evolution of 3D visual-isation of geological information has been worked on bymany groups around the world for many years (e.g. Turner1991; Berg et al. 2009). Key information, including acomprehensive coverage of the history and currentactivities in this field, is provided by the Illinois StateGeological Survey (2011) website (McKay 2011).

Robins et al. (2004) discuss the role of 3D visualisationof groundwater systems as an analytical tool preparatory tonumerical modelling. Through a couple of case studies, theydemonstrate its role and value. A key assertion they make isthat it is in the initial conceptualisation stage that mostproblems arise, so if the conceptual model of the physicalconfiguration and behaviour of the aquifer system is wrong,subsequent numerical modelling will also be incorrect. Theyalso found that a 3D approach could be of assistance inconceptual model preparation by providing a check on thelogic of the hydrogeological conceptualisation. This isespecially the case if the 3D conceptual model is coupledwith a simple water-balance model. Bredehoeft (2002)likewise makes a similar finding that the conceptual modelis the foundation of any numerical groundwater model.

With advances in digital technology and a growing baseof users/scientists wanting to see their subsurface project in a3D portrayal, improved conceptualisation of complex geo-logical settings can aid in better understandings of anotherwise invisible realm. Other key references that describecase studies on how 3D mapping and visualisations have

Received: 29 October 2010 /Accepted: 17 June 2011

* Springer-Verlag 2011

B. Gill ())Future Farming Systems Research,Department of Primary Industries,Ferguson Road, Tatura, Victoria, Australia 3616e-mail: bruce.gill@dpi.vic.gov.au

D. Cherry :M. Adelana :M. ReidFuture Farming Systems Research,Department of Primary Industries,Epsom, Victoria, Australia 3554

X. ChengFuture Farming Systems Research,Department of Primary Industries, PO Box 4166, Parkville, Victoria,Australia 3052

B. GillEnvironmental Geoscience,La Trobe University, Bundoora, Australia 3086

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

been used to improve hydrogeological characterisationinclude: Artimo et al. (2003), Ross et al. (2005) and Nuryet al. (2010). This study has sought to further explore howthese methods can be used in a practically focussed studythat aims to improve the fundamental groundwater resourceassessment of the study area, support subsequent technicalanalysis and provide relevant information for groundwaterresource-management processes.

The proposition this study tests is that 3D hydrogeologicalmethods provide significant advantages when conductinggroundwater resource assessments, which lead to improvedresource-management outcomes. By applying these newtools and methods to a case study area, this report describeswhat was found in regard to the aforementioned proposition.

The report explores the use of the 3D geological datasets to build visualisations and virtual constructs of theaquifers in the study area. The 3D software is then used tocalculate the size of the aquifers and the area of criticalcross sections to estimate the quantities of water withinand moving through the study area. The size of the aquiferprovides the core component of a water-balance model, towhich the recharge and loss volumes then relate. Thisprovides a better context in which to understand thebehaviour of the aquifer against which the sustainability ofgroundwater usage in the study area can be considered.

Background

Water resource pressures facing south-eastAustraliaA key reason for exploring the potential of 3D geologicalgroundwater-resource-assessment methods is the increas-ingly water-stressed situation that affects much of Aus-tralia (NWC 2006; Kiem and Verdon-Kidd 2010). There isstrong interest in testing whether 3D-based hydrogeologymethods can yield more robust, re-useable and evolvablehydrogeological results that enable groundwater resourcesto be better managed, especially during water-stressedtimes. There is also an ever-pressing need to improveunderstanding of surface water and groundwater interactionsto support a more precise total water-resource-managementframework. Further interest lies in the ability to integrate 3Dgeological and hydrogeological conceptual-model develop-ment with numerical modelling packages to improvepredictive capability to respond to future climate variabilitychallenges. Artimo et al. (2003) recognised that developingphysically based and numerical models in parallel provides aprocess of feedback and incremental improvement as the twoevolve in tandem. Improved analysis of landscape- andcatchment-scale water-movement processes will also givebetter understanding of other hydrologically driven issuessuch as land-salinity occurrence, groundwater base flow tostreams or protection of groundwater-dependent ecosystems.

Climatic influence on groundwater levelsMost of southern Australia has experienced a decline inrainfall over the past decade. Kiem and Verdon-Kidd

(2010) undertook an extensive review of stream flow andrainfall data across Victoria in order improve under-standing of the hydroclimatic change that has occurredin the past decade. They concluded that a step change inannual rainfall was apparent from 1994 onwards. This wascaused by a reduced reliability of late autumn rains arisingfrom changes in the frequency and timing of the synopticweather patterns that drive the Victorian climate. Theimpact of this step change in rainfall has also manifestedin groundwater observation bore records throughoutVictoria. Particularly seen in bores installed during the1980s for land salinity monitoring, the vast majority haverecorded a decline in groundwater level corresponding tothe reduced rainfall and recharge over the past decade.

State groundwater-resource management processesIn Victoria, the state government established there was aneed for groundwater management plans in the 1990s,when declining groundwater levels and high developmentdensities became apparent in some aquifers around thestate. Recognising that a trigger was required to ascertainwhere and when management plans were needed, a simple,rapid empirical approach was developed to estimate theavailable annual resource (DNRE 1996). The term initiallyused to describe the annual available resource was ‘permis-sible annual volume’ (PAV), which was essentially analo-gous to ‘safe groundwater yield’. It was used as the basis forsetting initial groundwater pumping limits in Victoriangroundwater management areas (GMAs) and identifyingwhere allocations may be in, or approaching, excess.

The approach developed to derive PAVs involvedseveral calculations that were based upon the readilyavailable data and geological interpretations of eachgroundwater management area. As a matter of someurgency, the initial PAV assessments were carried outrapidly, did not seek new data and were relatively simpleassessments. An audit (Reid and Cherry 2004) of the PAVvalues derived for 35 of the 60 GMAs in Victoria,identified a range of issues with both the approach usedand the PAVestimates proposed, although it was acknowl-edged that it provided a starting point for implementinggroundwater-management plans around Victoria.

A number of over-arching groundwater-managementconsiderations were also identified. These included suchthings as aquifer definition or inclusiveness, adequacy ofGMA boundaries, aquifer storage, groundwater declines,groundwater quality risks, groundwater monitoring andinteraction with surface-water bodies and the sea. Overall,the audit identified that the major deficiency was ingeological and hydrogeological information upon whichto base these calculations. This meant only very lowconfidence ratings could be applied to the PAV values.

The PAV calculations were based on several differentmethods, but primarily relied upon a determination of therecharge as an analogue for sustainable yield. Many notedpractitioners have demonstrated that this principle isincorrect (e.g., Theis 1940; Brown 1963; Bredehoeft etal. 1982; Alley et al. 1999; Sophocleous 2000; Bredehoeft

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

2002; Kalf and Woolley 2005). They argue that the size ofthe sustainable groundwater development should insteadbe based upon how much groundwater can be safelycaptured by the development. As such, this ‘safe capture’is not dependent on recharge, but rather on how theaquifer system responds to the development, how long ittakes and where the impacts occur (Bredehoeft 2002). It istherefore important to closely monitor and model theaquifer’s response to development and manage theresource in a more flexible manner in tune with theaquifer dynamics. Development can and does in manycases safely exceed average annual recharge by virtue oflarge aquifer storages, disconnection from streams and acombination of aquifer characteristics and strategic boredevelopment that helps to minimise detrimental impactson the environment and groundwater resources.

The amount of groundwater that can be safelycaptured, or intercepted, in terms of aquifer and environ-mental sustainability is influenced by a number of factorsto do with aquifer extent, storage and other hydraulicproperties. This amount may vary from year to year withclimate and the dynamic status of the groundwater system.Although it is considered that ‘safe capture’ should be thekey concept upon which to ultimately base groundwatermanagement, it is recognised that a whole-of-aquifer-system water balance is an essential component in under-standing sustainable groundwater resource use and thedevelopment of a management strategy.

A detailed water-balance analysis that assesses themajor inputs and outputs to the aquifer storage isnecessary not only to gauge the size and annualvolumetric potential of the extractable resource but alsoto help understand the nature of the aquifer, its rechargeand discharge processes, its inertia and its connection tosensitive environmental areas and surface-water resources.

The PAV audit (Reid and Cherry 2004) revealed anoften-incomplete analysis of the water balance, includinga deficiency in accounting for hydraulic connection tostreams and estimates of through-flow. Most of theattempted through-flow analyses were incomplete in thatthey did not account for all flow into, or in some caseseven within, the GMA aquifer system. These deficienciesunderscored the need to build, as complete as possible,renditions of the available geological data for each areaupon which the hydrogeological conceptual model can bebased. It is against this background that the need for amore complete and reliable method for estimating thegroundwater resources of an area was realised.

In some parts of the state, local opposition and challengeto the PAVs occurred, with significant criticism being levelledat the technical veracity of the approach employed. Afterfurther independent technical analysis, one groundwatermanagement area was successful in gaining an increase of7,000 million litres (ML)/year in the PAV, whereupon asubsequent auction of the increased volume was held.

While the process set in motion by the initial PAVestimation approach was a successful catalyst for estab-lishing groundwater resource-management plans, there hassince been little improvement in the ability to define or

estimate the permissible yield, or as it is termed in Victorianow, the Permissible Consumptive Volume (PCV). It iswith this in mind that 3D hydrogeology methods havebeen developed and tested in the Spring Hill area.

Study area

The Spring Hill Groundwater Supply Protection Area(GSPA) is 253 km2 in area and lies at the southern end ofthe Loddon River catchment in central Victoria, Australia.Figure 1 shows its location in relation to the nearest largetown of Creswick, the State of Victoria and the Australiancontinent.

GeologyThe geology of the study area comprises an uplifted anddissected Palaeozoic bedrock landscape with the palaeo-valleys in-filled with Tertiary-aged sediments overlain orinterrupted by a thick sequence of Neogene-aged basaltflows and volcanic eruption points. Figure 2 shows thegeology of the study area, with a number of other featuresalso shown such as the observation bore locations (redstars), the four water trading zones within the groundwatermanagement plan area (delineated by the green lines) andthe cross section A–A’ (pink line).

In more detail, the geology of the study area comprisesfolded Ordovician age (435–500 million years, Ma)sediments comprising thinly bedded shale and sandstoneunits that underlie most of the study area (Douglas andFerguson 1988). These low-grade meta-sedimentary rocksform a poorly transmissive, fractured rock aquifer fromwhich only very small yields can be obtained for stockand domestic purposes. Water salinity is also generally toohigh for most uses and drilling data nearly always identifya meter or more of clay immediately above bedrock.Evidence indicates that the aquifer has only a minoroverall influence on the groundwater in the overlyingvolcanic sequence and essentially forms the basal confin-ing layer of the GSPA groundwater system.

Major orogeny occurred during the Devonian period(345–395 Ma) in the vicinity of the study area, resulting information of gold-bearing quartz veins, locally known as“reefs”. Subsequent erosion during the Oligocene–Plio-cene period (26–2 Ma) produced gold-bearing alluvialdeposits that in-filled the palaeo-valleys of the upperLoddon River catchment.

A period of volcanism from around 2 million years toless than 30,000 years ago filled the valleys and coveredthese alluvial deposits. These deposits of gravel, sand, silt,clay and minor lignite along the valleys were extensivelymined for alluvial gold during the 1850s to 1900, wherethey outcrop in the Creswick area. The deposits werefollowed under the basalt by the early miners, who coinedthe term ‘deep leads’ for them. At least 15 eruption pointsformed small cones in and adjacent to the study area.Scoria, ash and tuff deposits formed in the vicinity ofcones, and weathering periods between eruptions are

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

identified as clay layers between basalt flows in manydrilling logs. The basalt sequence is up to 100 m thickover the palaeo-valley and the volcanic cones are up to200 m above the present valley (DME 1983). Thevolcanic materials form the major aquifer of the studyarea. Well developed and productive soils cover thebasalts, especially around the volcanic cones, where thedominant crop is potatoes.

Drilling dataGold mining commenced in the 1850s, with initial surfaceworking of regolith and alluvial deposits proceeding todeeper reef mining and deep lead mines until the turn ofthe century. Specifically in the study area, extensive boring(or drilling) through the basalt to locate the deep leaddeposits was well documented and of great value inbuilding the 3D geology data set for this study. Thesehistorical records, often overlooked in groundwaterstudies, provide a greater density of bore data than wouldotherwise be available if the area had not had the goldmineral wealth. While 720 bores have been recorded inthe study area, only 200 mining bores and 20 more recentgroundwater bores were greater than 50 m deep and had

adequate records suitable for building the 3D geologymodel of the study area. This equates to a bore density ofapproximately two bores per square kilometre.

Groundwater-resource development in the Spring HillareaExpansion of farming followed the decline of gold miningand forest clearing in the district in the 1880s. Potatocropping became a local specialisation due to the presenceof suitable soils on the volcanic parent material. Waterboring and irrigation-pump-technology developmentsafter the Second World War enabled groundwater to beobtained in sufficient volumes for irrigation use and thisenabled the area to become a significant potato farmingarea (LCC 1980).

During the late 1990s, groundwater development hadreached sufficient levels to start causing localised seasonaldeclines in groundwater depth (Fig. 3). Part of this develop-ment included development of bores for urban supply,

Fig. 1 The Spring Hill Groundwater Supply Protection Area in the state of Victoria, Australia, which lies on the southern margin of theMurray-Darling Basin watershed

Fig. 2 Map of the study area, showing surface geology, placenames, cross section A–A’, the Spring Hill Groundwater SupplyProtection Area (GSPA) boundary and the four named zones withinthe GSPA

�

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

especially because of reduced surface-water reliability in therecent decade. This has resulted in some tension betweenirrigation and urban use and puts increased pressure on thewater authorities to define sustainable development.

In the study area, the State Minister responsible forwater legislation established a locally based consultativecommittee in 1999. This committee, in consultation withthe responsible water authority, prepared a groundwatermanagement plan for the Spring Hill Groundwater SupplyProtection Area (GMW 2001). Although there was noevidence of severe stress on the groundwater resources ofthe area in total, it was recognised that some parts wouldbe stressed if all existing water licenses were fully utilisedon a regular basis. There was limited monitoring of theaquifer up until then, so priority was given to establishinga monitoring program. No new licenses were issued,temporary trading rules were established, meters wereinstalled on irrigation bores and small volume boresrequired registration under the plan.

Existing groundwater resource estimates in the SpringHill areaThe permissible annual volume adopted by governmentfor the Spring Hill Groundwater Management Plan wascalculated using the methods described in DNRE (1996).In this case, it was primarily based upon an estimate of thepotential annual recharge over the whole area, as apercentage of rainfall. It assumed 7% of the averageannual rainfall of 625 mm recharged the aquifer over thevolcanic cone areas and 2% of rainfall over the plainsareas. Over the total area of 253 km2, this equated to5,139 ML/year. An alternative estimate, using a hydro-graph fluctuation method, gave an annual estimate of10,140 ML/year. An estimate of the total aquifer storagewas also made. It assumed an average aquifer thickness of20 m and a porosity of 5% for the whole of the 253 km2

area, giving a volume of 253,500 ML.

The current groundwater situationThe majority of active groundwater usage occurs on theslopes around Spring and Forest Hills (the two scoriacones in the Forest Hill Zone, see Fig. 2). These are wherethe most suitable soils for cultivation and low salinity,high yielding aquifer locations coincide. Groundwaterbehaviour in the study area is variable. The greatestdecline in groundwater level coincides with the area ofgreatest groundwater extraction intensity. Groundwaterlevels in some areas recover to pre-pumping levelsfollowing winter while some areas are showing long-termdecline (Fig. 3). Notable among the areas that recover isthe Blampied Zone, whereas the central area aroundSpring Hill and Forest Hill shows the greatest fall inlevel. The Forest Hill Zone has the greatest licenceentitlement with a total of 2,887 ML/year–more thandouble that of Blampied Zone (1,130 ML/year) which isthe next largest zone by licence entitlement. In the 2008/09 season, reduced allocations of 50 and 80% wereimposed on the two zones by the groundwater manage-ment plan in response to the falling groundwater levels.All licensed bores now have flow meters installed,enabling accurate usage figures to be recorded. Totalusage from the 75 metered bores for 2008/2009 seasonwas 2,041 ML, which is 63% of the restricted licensedentitlement in the area (GMW 2009).

Topography and rainfallThe study area is at the uppermost end of the LoddonRiver catchment, which drains northwards into the MurrayRiver. The highest elevations in the area are volcaniccones with elevations ranging from 650 to 700 m aboveAustralian Height Datum (AHD). The land slopes gentlyto the north, with the north-west corner being about 380 mabove AHD. Figure 4 shows annual rainfall variabilitysince 1889, the long-term mean rainfall of 575 mm/yearand cumulative deviation from the mean. The latter showsa rising trend from 1946 to 1997, and a steeply fallingtrend since.

520

525

530

535

540

545

550

1994

1996

1998

2000

2002

2004

2006

2008

2010

Year

RW

L (m

AH

D)

B116382 B138658 B141228 B141296

Fig. 3 Time series plot of rest groundwater levels (RWL) abovethe Australian height datum (AHD) for four observation bores in theSpring Hill area. Note that only bore B116382 is continuing todecline, whereas the other three appear reasonably stable over thelast 3 years. Data sourced from the Victorian Water data warehouse(DSE 2011)

-2000

-1500

-1000

-500

0

500

1000

1500

1889

1897

1905

1913

1921

1929

1937

1945

1953

1961

1969

1977

1985

1993

2001

2009

Year

Cu

mm

ula

tive

dev

iati

on

fro

m m

ean

rain

fall

(mm

)

0

200

400

600

800

1000

1200

An

nu

al r

ain

fall

(mm

)Annual rainfall

Cummulative departure from mean

Long-term mean rainfall (1889-2009)

Fig. 4 The annual rainfall and cumulative deviation from meanannual rainfall for the Spring Hill area. Data Sourced from SILO,Enhanced Meteorological data, Creswick station (SILO 2009)

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

3D geological model development

This research has been undertaken using ArcGIS (ESRI)and the GOCAD suite, a software product of ParadigmGeophysical Pty Ltd. GOCAD is a software solutiondeveloped specifically by the petroleum industry to allowgeoscientists to construct 3D models of the subsurfacefrom a range of available data sets.

To develop the 3D geological model for the Spring Hillarea, detailed surface-geology geographic informationsystem (GIS) polygon layers, bore logs, digital elevationmodel of the surface topography and geophysical datawere collated. Bore logs were closely examined foraccuracy and detail to help understand the stratigraphy,then interpretations were added to the logs to markspecific and important horizons. Logs were omitted ifthey (1) had spurious data, (2) were poorly or incorrectlyrecorded, or (3) were too shallow or simply not useful inadding to the geological knowledge base. This left 220bores for use in constructing the sub-basaltic surfaces inthe 3D model. As this was a pilot study, simplistic rastermodelling was undertaken using the Inverse DistanceWeighting interpolation tool in ArcGIS 9.2. To keepprocessing straightforward, the default options wereaccepted; however, a cell size of 200 × 200 m was chosento smooth the variable density of boreholes. This was thenexported to ASCII format, which was then read directlyinto GOCAD. While GOCAD provides many tools tomodel such surfaces and volumes (e.g. DSI), the researchteam’s limited experience in this software restricted its useto simpler tasks.

Surfaces were corrected for cross-overs and anomalies,and modified if other data (e.g. the surface geology orgeophysics) suggested another interpretation was morelikely. Once the surface topography, bedrock palaeo-topography and any intervening layers were constructed,GOCAD modelling tools were used to construct voxelmodels (3D blocks) of the respective elements (geologicalunits or aquifers). The volume and cross-sectional area ofthese units were then calculated using tools in GOCAD,upon which calculations of the groundwater resource werecarried out. These shapes were also able to be used forvisualisation purposes, for example, they can be colouredaccording to groundwater salinity (Fig. 5b) or cut by watertables (Fig. 6) or structural controls.

Hydrogeology conceptual-model developmentThe size and shape of the aquifers within the bedrockpalaeo-valleys can be visualised easily in GOCAD,especially using appropriate vertical exaggeration. Thisshows that the main aquifer unit comprises the volcaniceruption points (or scoria cones) and the extensive basaltsthat infill the bedrock valleys and gently slope towards thenorth-west corner (Fig. 6). Note that this image is a voxet(a regular 3D grid constructed from voxels) of the mainbasalt aquifer, enabling the volume to be calculated in the3D software.

The boundaries of the area are the catchment dividearound the south-east perimeter and the bedrock hills onthe east and south. No significant groundwater flow islikely across the southern and eastern boundaries due tothe low permeability of the bedrock and lack of pressure

Fig. 5 A stack of 3D block diagrams beneath a geology drapedover a digital elevation model (DEM) surface, viewed from thesouth-west corner of the study area. a Surface topography withmapped geological units; see legend on Fig. 2. Vertical exaggerationof DEM is 10 times. b Voxet model of whole area, showinggroundwater salinity in the basalt aquifer (bedrock shown inpurple); yellow, total dissolved solids (TDS), where 1,000 < TDS<3,500 mg/L; pale blue, 500 < TDS < 1,000 mg/L; darker blue,TDS < 500 mg/L. Salinity segments from EPA 1997. c Voxet modelof bedrock (basalt removed) with the water table shown. The colourgradation represents the change in water level for the period 2000–2010 (blue, no change; orange, up to 20 m fall). The GOCAD-calculated volume difference between the two water tables is0.37189 km3

Fig. 6 Voxet representation of the basalt aquifer (orange) withintersecting groundwater surface (blue). (10 times vertical exaggeration)

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

gradient across the catchment divide. The basalt aquifer islargely unconfined and assumed to be so for groundwaterstorage calculations, but semi-confined conditions occur insome places due to the presence of palaeosols, oftenlogged as clay layers between basalts, as well as due tocontrasts between more fractured, vesicular basalt flowsand more dense layers. However, no attempt to portraysuch features has been made at this stage, given that theobjective is to assess the total groundwater resource in thestudy area, not investigate local-scale behaviour.

The underlying Tertiary gravels, or ‘deep leads’, form adendritic system beneath the basalt, but are limited inextent. This hydrogeological unit outcrops near themargins of the basalt, where it was mined for alluvialgold (Neogene Shepparton formation, Fig. 2). As itplunges below the basalt, it becomes saturated, anddepending upon the occurrence of clay layers associatedwith the volcanics or with pre-basaltic alluvial conditions,it varies from unconfined to semi-confined conditions(Beatty 2003; Fawcett et al. 2006). To the north west ofthe study area, it joins with similar deposits emanatingfrom similar basalt covered palaeo-valleys to the west. Itshydraulic conductivity is known to be significantly higherthan the basalt aquifer and it is likely to be draining waterfrom the overlying basalts and transporting it from thestudy area. However, mining records indicate it istypically confined to narrow trenches and is semi-continuous due to disruption by fault movement sincedeposition (Hunter 1909; Holdgate et al. 2006). Whilethere is little mining or drilling data on the Kingston toSmeaton area to confirm its presence or absence,continuations of the Bullarook, Smeaton Reserve and un-named lead parallel to Birch Creek are inferred (DME1983). Drilling completed in 2009 about 4 km to the southof the cross section intersected approximately 30 m ofalluvial materials beneath the basalt. The implications forthe hydrogeological conceptual model are significant, withits assumed presence or absence being a significant sourceof uncertainty for the model estimates.

Given the topography and the structure of the bedrockand younger aquifers, there is interpreted to be littleopportunity for significant movement of groundwater acrossany of the study area boundaries except for the north-westcorner. The small area of basalt on the northeast boundary islargely isolated by a bedrock divide from the main aquiferand is excluded from the calculations.

As the basalt aquifer in the Spring Hill area forms theuppermost end of a groundwater flow system, there is acontinuous natural loss of groundwater occurring all yearround. This loss is controlled by the flow rate through theaquifer, and an estimate of this can be more readily madewhere the bedrock structure confines the basalt, as incross-section A–A’ (Fig. 7).

Recharge in excess of the flow losses from the studyarea will cause a rising groundwater level in the upperreaches of the basalt aquifer. When recharge (rainfall)declines, the water level in the aquifer up gradient ofcross-section A–A’ is likely to fall at a rate controlled bythe flow rate through the cross section and upstream base-

flow losses to Birch Creek. This conceptual understandingof the groundwater flow system suggests that the keyneeds are to quantify the flux of water leaving the areathrough this bedrock structural restriction (cross sectionA–A’), base flow, and the amount of recharge. This thenallows comparison of the relative contribution of pumpingon groundwater levels, therefore informing permissiblegroundwater consumption.

Estimating groundwater resources from 3Dgeology

On the basis of the 3D geological mapping of the studyarea, estimates of the groundwater resources have beenmade. The estimates include:

1. A calculation of the aquifer volumes and the maximumpossible volumes of groundwater resource in the area,based upon estimates of porosity

2. Annual recharge amounts based upon the outcropareas, rainfall data and hydrographs

3. Flow of groundwater leaving the area via the basalt anddeep lead aquifers

4. Loss of groundwater, as base flow, to the main drainageline (Birch Creek)

Aquifer volumes and cross-sectional areas derivedfrom the 3D software were combined with groundwatergradients and estimates of aquifer properties usingfundamental hydrogeological principles (e.g. Fetter 1988)using an Excel spreadsheet to build a comprehensiveunderstanding of the likely volumes of groundwater in thestudy area. The aforementioned four groundwater volumeestimates are considered to be the key elements thatgovern the behaviour of the groundwater resource of theSpring Hill area. The initial water volume and fluxescalculated from the 3D data were refined using a spread-sheet-based mass-balance model. The annual extractionscan then be seen in the context of the annual recharge andnatural loss estimates from the aquifer system.

An estimate of the total groundwater resourceIn order to establish the total size of the groundwaterresource (R), volumetric calculations of the two main

Fig. 7 Cross section A–A’. Bedrock (lavender), deep lead (green)and basalt (orange) profile with 2010 groundwater level. See Fig. 2for cross-section location. The basalt aquifer cross-sectional areabelow the water table is 79,200 m2. The presence of the Tertiarygravels aquifer under the basalt is assumed from mining records thatsurmise two ‘leads’ either side of the centre of the cross sections(DME 1983). It is estimated to have a cross-sectional area of 12,000m2 in this location

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

hydrogeological units in the study area (the volcanic andthe underlying Tertiary sedimentary aquifers, V) wereobtained from GOCAD. These were then multiplied byestimates of effective porosity (or specific yield, S) to giveestimates of the total groundwater volume using thesimple equation R ¼ V � S.

The value used for specific yield of basalt is a majordeterminant of this potential maximum volume. Refer-ences to basaltic and volcanic aquifer hydraulic parame-ters in the literature highlight a wide range of possiblevalues. Fetter (1988) quotes Schoeller (1962) as reportingbasalt generally ranging from 1 to 12%, with coolingfractures, collapse rubble and gas vesicles contributing tothe wide range. Pyroclastic materials such as pumice,scoria and ash can have much higher porosities, but wouldmost likely be found only in and around the eruptionpoints. Other papers (Saha and Agrawal 2005) report asimilarly wide range for Deccan Trap basalt in India of 0.1to 1% for massive basalt and 5–11% for vesicular basalts.A deep road cutting through basalt near Smeaton showssolid basalt flows with a low gas content and smallnumbers of unevenly distributed fractures (Fig. 8). Themajority of the water capacity in the whole of the basaltaquifer volume is likely to be in fractures, cooling jointsand interflow features and so a lower porosity isconsidered more likely. Given the inhomogeneous natureof volcanic aquifers, selecting parameters for large scalecalculations provides a significant source of uncertainty.

From GOCAD, the total volume of volcanic aquifer inthe central area above cross section A–A’ was determinedto be 7.75 km3, with the saturated volume below theSpring 2007 water table being 4.63 km3. Assuming an

average effective porosity of 2% results in a groundwatervolume of 4:63� 0:02 ¼ 0:0926 km3, or 92,580 ML. Thevolume of the underlying deep lead, or sub-basalticTertiary age alluvial aquifer was calculated as 0.689 km3.Assuming a conservative effective porosity of 10% gives acontained water volume of 0.0689 km3, or 69,000 ML,totalling nearly 162,000 ML.

Annual recharge estimatesThe original PAV recharge estimate was calculated usingthe whole area of 253 km2, but the surface area of basaltin the Spring Hill area totals only 147 km2, considerablyless than the total Spring Hill Groundwater Managementarea. Recalculating for the basalt area only, using the samepercentage rainfall amounts as the original PAV estimatedid (2% of rainfall for the plains areas and 7% of rainfallon the 3% of area that is the volcanic cones), gives a totalannual recharge of 2,100 ML/year. However, this flat rateof recharge used for the PAVestimate appears to be too lowand does not take into account the seasonal and annualrainfall variability, which is substantial (Fig. 4). Basedupon the method described in Healy and Cook (2002) forfractured rock recharge, a threshold-based analysis ofdaily rainfall and evaporation records with correlation tolocal groundwater hydrograph response has been used toestimate annual recharge amounts. Recharge usuallyoccurs between May and September each year, whenwinter rainfall predominates and evapotranspiration ratesare low. Based upon 1975–2009 climate data, the analysisindicates that annual recharge ranged from 8 mm in verydry years (1982) to as much as 120 mm in the wettestyears (1981). Annual recharge estimates for the 107 km2

surface area of basalt aquifer up gradient from crosssection A–A’ provides a bar graph of total annual recharge(Fig. 9).

The 35-year-average-annual recharge is 6,680 ML/year,while the reduced rainfall of the recent decade gives anaverage that is approximately one third lower at 4,960

Fig. 8 Basalt exposed in a road cutting just south of Smeaton,showing a range of fracture densities. Each view is approximately 3m high

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

2007

2009

Year

To

tal

Rec

har

ge

(ML

/yea

r)

ML/year

Fig. 9 Annual recharge estimates to the basalt aquifer for 1975–2008 calculated from rainfall-evaporation surpluses (as mm effec-tive rainfall) multiplied by the (107 km2) basalt aquifer surface areaup gradient from the cross section. Note the extreme variability inpotential recharge by comparing the wet winter in 1981 with thefollowing very dry year and the lower average recharge since 2001

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

ML/year (1997–2008). While the averages are similar tothe PAVestimate, the annual estimates have the advantageof reflecting the dynamics of the past 35 years of rainfallvariability and hence show more realistically the impact ofclimate variability on the groundwater resource.

Groundwater leaving the area (outflow)The Spring Hill area lies at the top of the Loddoncatchment with its southern boundary lying along thecatchment divide. The geology, landscape and hydraulicgradients govern movement of groundwater in the area.Estimation of the groundwater balance requires calcula-tion of the amount of water leaving via the basalt and deeplead aquifers as outflow, via the major streams as baseflow, and via extraction wells.

An estimate of groundwater outflow from the basaltand deep lead aquifers can be made at the cross-section(A–A’, shown in Fig. 7) that is parallel to the water-tablecontours (Fig. 10) and lies between the two bedrock highsat Allendale (A’) and Smeaton (A). GOCAD was used tocalculate the cross-sectional areas of the basalt and thedeep lead aquifers at this section and Darcy’s law appliedto derive flow rates (Fetter, 1988) using Q ¼ Hc � I �Awhere Q is the quantity of flow (m3), Hc is the hydraulicconductivity (m/day), I is the hydraulic gradient and A isthe cross sectional area of the aquifer (m2).

Estimates of hydraulic conductivity (Hc) for calculatingthroughflow in the relevant aquifers were derived fromearlier work in similar hydrogeological terrain. Firstly, inregard to the basalt aquifer, Ife (1979) compiled pumpingtest results for bores in Pliocene Basalts of South-westVictoria. Transmissivity values obtained from 17 testsranged from 1 to 746 m2/day, with an average of200 m2/day. Similar transmissivity values are reportedfor basalt in Moghaddam and Fijani (2009) of 24–870 m2/day and Saha and Agrawal (2006) of 10–700 m2/day. Pumping test results from nine bores in avolcanic aquifer in or near the study area (B. Cossens,Goulburn-Murray Water, personal communication, 2011)list transmissivity values between 30 and 172 m2/day.Based upon the reported saturated thickness, theaverage hydraulic conductivity from these tests is2.3 m/day. An average hydraulic conductivity for basaltof 4 m/day is derived based on these references.

Secondly, in regard to the Tertiary alluvial aquifer, twosites north of Spring Hill are listed as recording hydraulicconductivities of 130 and 25 m/day (Tickell andHumphrys 1986). However, as these are derived frompumping test data from private production bores, they areconsidered likely to be higher than the average for thewhole formation thickness, especially given that themining and drilling records indicated the presence oflower permeability materials in the sequence. A hydraulicconductivity of 15 m/day may be more representative andwas used in the mass balance calculations.

GOCAD calculations of cross-sectional area of thebasalt aquifer in cross-section A–A’ (Fig. 7) is 79,000 m2.Using a hydraulic conductivity of 4 m/day and a hydraulic

gradient of 0.016 gives an annual flow rate of79; 000� 4� 0:016 ¼ 5056m3=day, or 1,845 ML/yearthrough the basalt. Assuming a hydraulic conductivity of15 m/day and a hydraulic gradient of 0.016 through anestimated 12,000 m2 of Tertiary aquifer gives a flow of2,950 m3/day, or 1,078 ML/year.

Groundwater discharge to surfaceGroundwater loss to Birch Creek is likely, as the creek lineis incised by up to 30 m into the basalt plains and springactivity is observed along the creek bed. Hydrograph datafrom the Smeaton stream gauge on Birch Creek (stationNo. 407227, DSE 2011) were analysed to ascertainpossible baseflow amounts (Fig. 11). In earlier years (priorto 1997) baseflow appears to be approximately 5 ML/day,totalling approximately 1,800 ML/year. During 2008 and2009, flow ceased in February and March, suggesting therecent decline in groundwater level was sufficient toreduce baseflow to the stream. This is consistent withmany smaller streams in central Victoria that ceasedflowing during the last 10 years. Assuming aquiferdischarge does provide some baseflow to sustain the creeksystem during summer, it needs to be taken into accountwhen considering the water budget for the area.

A water budget for Spring HillThe calculation of groundwater fluxes leaving the SpringHill GSPA and calculation of its water budget areimportant requirements in understanding the sustainabilityof the groundwater resource and, indeed, in testing thesoundness of the conceptual hydrogeological model.Given the location of the area near a catchment divideand assuming unconfined conditions in the basalt aquifer(that contains the primary groundwater resource), a naturaldecline in groundwater level occurs as water drains fromthe higher parts of the aquifer during dry periods. Whenirrigation pumping exacerbates the influence of climate onthe natural drainage, a more significant decline in ground-water level is inevitable. Additionally, the volcanic conesare the highest points in the landscape, so the naturaldrainage rate within the cone area is even greater due tothe steeper groundwater gradients.

A simple mass-balance modelA simple spreadsheet-based mass-balance model was builtusing the aquifer volumes and fluxes illustrated in Fig. 12and is shown in Table 1. Run for the period 1975–2009,the model is instructive in determining volume estimatesfor the main Spring Hill aquifer. The model used a totalinitial aquifer storage volume of 162,000 ML (basalt anddeep lead), the calculated annual recharge (Fig. 9), annuallosses from the aquifer to the north-west (through crosssection A–A’) of 1,900 ML in the basalt and 1,080 ML inthe deep lead, and a stream flow loss of 2,000 ML/year inwetter years, declining to 1,500 ML/year in the recent

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

decade. Groundwater abstraction in the model is based onmetering data for recent years, but was assumed to be1,000 ML/year prior to 1983 to reflect the lower levels ofearlier development. The output of this model is shown aschange in aquifer storage volume in Fig. 13.

The value of this simple model was in being able totest a range of possible recharge and discharge values toarrive at the ones given in Fig. 13. For example, the massbalance model output (Fig. 14) shows an increase in totalvolume when lower fluxes of 1,000 ML/year for the basaltand 750 ML/year for the deep lead are allowed throughcross-section A–A’. This does not accord as well with themonitoring data. Likewise, the mass balance is strongly

influenced by the recharge inputs. If a flat 5000 ML peryear recharge rate is used, the model shows a steadydecline in total resource from 160,000 to 115,000 ML,whereas an average recharge of 6,400 ML/year keeps thestart and finish volumes the same.

The mass-balance model was further checked byadjusting the cross-section fluxes until the decrease involume from 2000– 2010 approximated the total loss ofstorage derived from the difference between the 2000 and2010 water tables. The volume of water lost from storageover the 2000–2010 period (illustrated in Fig. 5c) wasestimated to be 14,870 ML. The sum of recharge minusthe sum of system losses (including groundwater pumping

Fig. 10 Water-table contours for Spring 2007 expressed as metres Australian Height Datum (m AHD). Contour spacing is 10 m and thesurface is derived from 23 bore data points and allowed to a maximum of 2 m below surface away from data points. This surface has beenvisualised in GOCAD and is shown in Fig. 6. For base map colour key, please refer to Fig. 2

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

and Birch Creek base flow) for the same period in themodel totals 14,246 ML (using the fluxes shown inFig. 13). This comparison gives additional confidence thatthe numbers estimated are reasonable, despite the uncer-tainty inherent in the selection of aquifer parameters andrecharge rates.

Discussion

The original appraisal of groundwater resources in theSpring Hill GSPA in 1998 applied a basic method thattook no account of the detailed geology of the area. Thisinitial water-resource appraisal (DNRE 1998) was suitablefor the immediate need, which was to determine a volumeof water that defined the ‘safe yield’ for the area in order

0

50

100

150

200

250

300

350

400

450

500

01-Jan-95 01-May-95 01-Sep-95

Date

Str

eam

flo

w (

ML

/d)

0

5

10

15

20

25

30

Rai

nfa

ll (m

m/d

ay)

Daily Rainfall (mm)Daily Flow (ML/Day)

Fig. 11 One year (1995) of stream hydrograph data for BirchCreek (station No. 407227, Smeaton Stream Gauge, DSE 2011)plotted with daily rainfall data. Baseflow during summer months istypically 5 ML/day during wetter years

Table 1 Simple water-balance model for Spring Hill groundwater resource from 1975 to present (2009). All values in ML (million litres).The long-term average recharge is 6,667 ML/year, while 1999–2009 was 4,957 ML/year. For the last 10 years, total recharge was 47,978ML and losses total 62,223 ML, resulting in a deficit of 14,246 ML. In comparison, the volume of aquifer that drained (Fig. 5c, 0.37189km2) times specific yield of 0.04 equals 14,875 ML

Year Startingvolumea

Annualrechargeb

Tertiary aquifer DSlossc

Basalt DSlossc

Estimatedpumpingd

Streamdischargee

Newvolumef

1975 161,537 8,544 1,000 1,650 1,000 2,000 164,4191976 164,419 3,638 1,000 1,650 1,000 2,000 162,3951977 162,395 6,794 1,000 1,650 1,000 2,000 163,5261978 163,526 9,096 1,000 1,650 1,000 2,000 166,9601979 166,960 7,875 1,000 1,650 1,000 2,000 169,1731980 169,173 8,229 1,000 1,650 1,000 2,000 171,7401981 171,740 12,845 1,000 1,650 1,000 2,000 178,9221982 178,922 814 1,000 1,650 1,500 2,000 173,5741983 173,574 10,547 1,000 1,650 1,200 2,000 178,2591984 178,259 5,922 1,000 1,650 1,200 2,000 178,3191985 178,319 5,815 1,000 1,650 1,200 2,000 178,2721986 178,272 9,305 1,000 1,650 1,200 2,000 181,7151987 181,715 7,138 1,000 1,650 1,500 2,000 182,6911988 182,691 8,685 1,000 1,650 1,500 2,000 185,2131989 185,213 8,243 1,000 1,650 1,500 2,000 187,2931990 187,293 5,199 1,000 1,650 1,500 2,000 186,3301991 186,330 8,571 1,000 1,650 1,500 2,000 188,7391992 188,739 11,082 1,000 1,650 1,500 2,000 193,6591993 193,659 8,226 1,000 1,650 1,500 2,000 195,7221994 195,722 3,569 1,000 1,650 1,500 2,000 193,1291995 193,129 8,635 1,000 1,650 1,500 1,800 195,8011996 195,801 8,975 1,000 1,650 1,800 1,800 198,5141997 198,514 5,060 1,000 1,650 1,800 1,800 197,3121998 197,312 6,006 1,000 1,650 1,800 1,800 197,0551999 197,055 6,548 1,000 1,650 1,800 1,800 197,3412000 197,341 7,801 1,000 1,650 1,800 1,800 198,8792001 198,879 4,698 1,000 1,650 1,800 1,800 197,3142002 197,314 2,721 1,000 1,650 2,000 1,800 193,5732003 193,573 5,137 1,000 1,650 2,000 1,800 192,2482004 192,248 4,977 1,000 1,650 2,000 1,500 191,0632005 191,063 5,451 1,000 1,650 2,000 1,500 190,3512006 190,351 2,162 1,000 1,650 2,200 1,500 186,1522007 186,151 5,134 1,000 1,650 2,200 1,200 185,2232008 185,222 3,797 1,000 1,650 2,200 1,200 182,9572009 182,957 6,100 1,000 1,650 2,100 1,200 183,095

a Start of year aquifer volume, 1975 value is the sum of basalt and Tertiary aquifers derived from GOCAD volumes and porosities of 2%and 10%. In subsequent years, it is the last column value from the previous yearb Annual recharge to 107 km2 surface area of the basalt aquifer (Fig. 6) multiplied by annual recharge (Fig. 9)c Annual flows calculated through cross section A–A’. DS downstreamd Pumping estimated prior to metering records, which started in 2005e Stream discharge estimated from Birch Creek hydrograph data (Fig. 11) decreased due to drought in recent decadef End of year volume calculated thus: g ¼ aþ b� cþ dþ eþ fð Þ. This data shown in Fig. 13; other scenarios in Figs. 14 and 15

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

to precipitate initial management and investigative actions,and to develop a groundwater-management plan.

This study has taken a fresh look at the process ofdefining the groundwater resources and defining what thesustainable yield should be. The approach developedutilises 3D geological software and all available geo-logical data to give a higher level of confidence to thevolumetric estimates of groundwater resources in the area.The aquifer volumes calculated from the 3D model arephysically based and provide a sound baseline from whichto gain a better appreciation of the actual size of thegroundwater resource. Nury et al. (2010) used ArcGIS andGOCAD to explore the geometry and settings of theEastern View aquifer system in the Barwon Downs area ofVictoria. They presented visualisations of the geologywith calculations of water volume in the main aquiferbased upon the calculated formation volumes. While theydid not use the method to directly consider sustainablegroundwater usage (instead focussing on improvingunderstanding of surface-water interactions), their workdid suggest the 3D geology-derived volumetric-basedcalculations of water availability would be valuable inconsiderations about usage of the resource.

Other 3D mapping case studies in the literature thataddress groundwater needs include Artimo et al. (2003), Rosset al. (2005), MacCormack et al. (2005) and Wycisk et al.(2009). These studies have developed a range of sophisticated3Dmapping approaches and products. They have not directlyaddressed groundwater usage for resource sustainabilityneeds, instead focussing mostly on the potential use of the3D mapping for consideration of groundwater contaminationprotection or remediation (e.g. MacCormack et al. 2005.Kessler et al. (2010) indicate that their mapping work isintended to be used by the responsible water authority to

quantify the sustainable resources of the aquifer through thedevelopment of a conceptual model of the aquifer that willprovide the framework for future resource management.Given the interest in Victoria and elsewhere (both nationallyand internationally) to undertake 3D geological mapping forgroundwater management needs, development of methodsand procedures for quantifying storage volumes and flowsthrough groundwater systems appears to be an area ripe forinvestigation.

While a total aquifer storage volume alone may give afalse sense of how much water resource is available (Alley2007), putting it in context with the critical values ofrecharge, groundwater outflow rates, surface discharge and

Basalt aquifer95,280 ML storage

Deep Lead Aquifer69,000 ML storage

1900 ML/yr

1080 ML/yr

Recharge (rain)

Average6,680 ML/yr

Usage2100 ML/yr

Outflow from area

Irrigation and town

extractions

?

Birch Ck 1800 ML/yr Basalt aquifer95,280 ML storage

Deep Lead Aquifer69,000 ML storage

1900 ML/yr

1080 ML/yr

Recharge (rain)

Average6,680 ML/yr

Usage2100 ML/yr

Outflow from area

Irrigation and town

extractions

?

Birch Ck 1800 ML/yr

Fig. 12 Estimated groundwater storage and annual average flux volumes for the study area. The outflow volumes shown here are thosederived from the mass-balance model (output shown in Fig. 13). They were adjusted from the initial cross section estimates in order toallow the resource volumes to approximate the observed groundwater levels (Fig. 3)

140,000

150,000

160,000

170,000

180,000

190,000

200,000

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

2007

2009

Year

Vo

lum

e (M

L)

Storage Volume

Fig. 13 Spring Hill groundwater resource—a simple mass balancemodel of annual residual volume based upon numbers shown in Fig.12. Assuming a starting volume of 180,000 ML (basalt and deeplead), the estimated recharge rates and losses produce a steady risein storage volume up until 1995, with a decline since. Thespreadsheet model that produced this plot is shown in Table

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

pumping volumes in a mass balance model gives a muchbetter indication of the dynamics of the resource.Sophocleous (1997) discusses how the construct of ‘safeyield’ is a flawed concept if it is equated only with theamount of recharge, and with the expectation that itprovides a single product exploitation goal that defines thevolume of water that can be exploited every year withoutdestroying the resource base. Maimone (2004) recognisedthat explicitly defining sustainable yield is becoming morecommonplace, but that the true complexity of the conceptof sustainable yield is usually avoided and consensus on afirm number is unlikely to ever be reached. He then goeson to suggest that discussion of this is a critical need, andan explicit process of bounding the problem, analysingtradeoffs and interacting with stakeholders is a better wayof developing a workable definition of sustainable yield.

The physical and mass balance models developed in thisstudy help identify these other important components of thewater balance, hence it allows the actual pumping to be seenin context of the total resource base. The 3D-based methodstested in this study provide tools that can more explicitlymeasure and visualise the groundwater systems in an area.The uncertainties can be readily demonstrated and tradeoffscan be simply modelled. Interacting with stakeholders isaided through the use of visualisations. The simple water-balance model approach can also play a major role in thedevelopment of adaptive groundwater management plan-ning by allowing stakeholders to explore the limits anddynamics of the system. In this way, many of the needsidentified by Maimone (2004) can be addressed.

Implications for groundwater managementin the study areaThe current metered groundwater usage in the Spring Hillarea is 2,100 ML/year out of a maximum licensed

entitlement volume of 4,900 ML/year. The annual permis-sible consumptive volume (PCV) allowed under thegroundwater management plan is 5,130 ML. Under thelower recharge conditions of the past 10 years, ground-water levels have declined by up to 20 m in the moreintensively pumped, higher landscape positions. This hasbeen a direct result of the combination of pumping and thenatural groundwater flux draining from the more elevatedparts of the aquifer. The modelling suggests that duringwetter years, there is sufficient groundwater to allowpumping up to 2,500 ML/year, but running the model atthe total licensed volume of 4,900 ML/year for the35 years shows a severe decline in total groundwaterresource from 160,000 to 64,000 ML (see Fig. 15).Allowance for flows through cross-section A–A’ and baseflow to Birch Creek needs to be allowed for; therefore, aPCVof 5,130 ML is too high. Sustainable development isalso highly dependent upon future climate and thelikelihood of rainfall being sufficient to maintain rechargeat the current long-term average level.

In respect to the concepts of ‘safe yield’ and ‘safecapture’ (Bredehoeft 2002), for Spring Hill groundwaterusers, the relatively small volume of their basalt aquifer‘reservoir’ means there is insufficient capacity to allowpumping to exceed annual recharge for more than a fewyears at a time. Future management, therefore, needs tocontinue to respond to groundwater level and rainfall data,and appropriately restrict groundwater usage during dryyears.

The current metered groundwater usage in the SpringHill GSPA is about 50% of the total natural flux estimatedin the model, so consideration of the sustainability of thecurrent usage can be made based on the understanding ofthe groundwater resource developed in this study. Itappears that the Spring Hill groundwater system iscontrolled by two key factors. The first is the significanceof the natural drainage of groundwater away from theelevated, high recharge parts of the system. The second isthe constriction at cross-section A–A’ (Figs. 2 and 7) that

140,000

150,000

160,000

170,000

180,000

190,000

200,000

210,000

220,000

230,000

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

2007

2009

Year

Vo

lum

e (M

L)

Storage Volume

Fig. 14 Spring Hill groundwater resource—a simple mass balancemodel run with lower cross section A–A’ flows of 1,000 ML/yearand 750 ML/year through the basalt and deep lead aquifers,respectively. Here, the model output suggests these fluxes are toolow for the chosen recharge rates, as a 50,000 ML increase in totalresource since 1975 does not seem plausible, nor does the smalldecline in resource since 2000 accord with the observed declines ingroundwater levels (Fig. 3)

60,000

70,000

80,000

90,000

100,000

110,000

120,000

130,000

140,000

150,000

160,000

170,000

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

2007

2009

Year

Vo

lum

e (M

L)

Storage Volume

Fig. 15 Spring Hill groundwater resource—a simple mass-balancemodel output with 4,900 ML/year total pumping rate over 35 years.In reality, the rate of decline is likely to reduce as the gradientsdriving the natural losses decline and pumping becomes lessefficient as wells fail due to continued declining groundwater levels

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

limits the flow of water out of the aquifer. The numbersestimated in the analysis suggest that the current rate ofirrigation and urban use is sustainable, provided rechargerates do not continue at the average of the last decade, ordecline further. It is also apparent that the most elevatedparts of the aquifer are most at risk due to the combinationof pumping and natural drainage if recharge rates remainlow.

Conclusions

The methods developed in this study significantly improvegroundwater-resource assessments by reducing uncer-tainty about the physical dimensions and configurationof the aquifers systems. They also improve the ability toconceptualise the aquifer system and identify key features.Three-dimensional geology software such as GOCADallows reliable and repeatable calculation of aquifervolumes, cross-sectional areas, aquifer surface areas, andvolumes of drawdown. The ability to visualise the aquifersystem and build a physically based, simple mass-balancemodel makes the analysis more transparent and open tochallenge, modification and refinement than traditionalgroundwater resource assessments.

Limitations such as the uneven and low density of dataand the uncertainty of hydrogeological parameters used tocalculate groundwater volume, flows and recharge areacknowledged, but exist whatever approach is taken. Themain advantage of the 3D-modelling methods applied inthis study is that construction of a high quality, dimen-sionally sound virtual model of the hydrogeology of theSpring Hill study area allows more accurate resourceestimates to be made from the available data than has beenpossible in the past. The generated visualisations andgroundwater volumes show that much more can belearned from the available geological information thanhas been achieved previously. As new data is gathered, orparameters are revised, the ready ability to evolve themodel without having to redo fundamental mapping willbe a great time saving in the future.

For groundwater managers and resource users, themethods used in this study offer a considerable advanceon resource-estimation methods previously available. Asmore groundwater systems become mapped in comingyears, and as improved rigor in assessing water resourcesis increasingly sought due to dry conditions and increasingcompetition for the resource, the methods provide a morerigorous and transparent approach to defining sustainabledevelopment. While the inputs to simple mass-balancemodels tend to have large ranges and uncertainties, puttingall the estimates together in the model and relating theoutput to the observed situation (in the case of the SpringHill GSPA, the measured water-table decline of the past10 years) allows the groundwater system to be concep-tualised, simulated and tested. Overall, the estimatedvolumes and annual changes seem reasonable and accordwith the seasonal recharge and natural losses from theaquifer.

Despite some recognised data shortfalls, the 3D geo-logical modelling approach and mass-balance modellingmakes better use of the available data to quantify thegroundwater resource in the area in a transparent andrepeatable way. A range of parameters can be used in themodel to test the validity of key assumptions and giveconfidence to the results. The method has also generated alegacy of digital, 3D geological and hydrogeological datathat can be stored in the public data library to be availablefor subsequent users and updating.

The study has developed and tested new combinationsof methods that go some way to addressing issuesidentified in Bredehoeft and Durbin (2009) aroundgroundwater development impacts and sustainability.They note, in their discussion of uncertainty associatedwith predictive models, that no one argues that modelpredictions are not useful, provided they are used in fullawareness of the difficulties and resulting uncertainties.The results presented here recognise the significantuncertainty associated with predicting groundwater quan-tities. However, by mapping the physical dimensions andconfiguration of the aquifer system of interest, a signifi-cant amount of the conceptual and mass-balance modeluncertainty has been reduced. Further work to explore thepotential benefits of 3D hydrogeology methods willinclude testing the methods on some larger groundwaterflow systems. This will be challenging, especially wheresuch systems are not confined to a single basin.

For the Spring Hill area itself, and the regionallysignificant potato industry that is reliant upon its ground-water resource, the study results support the current limitson usage as being necessary under the prevailing reducedrecharge period. Continued adaptive management andseasonal restriction on usage will remain necessary unlessrainfall significantly increases above the average of thepast 10 years.

When working with the groundwater user communityin the ongoing development of their groundwater manage-ment plan, the 3D visualisations and quantificationsprovide new means to explain what is known about theresource. A broader audience will be more easily able tounderstand the visualisations of the main aquifers underrecognisable surface features. Information on bedrockboundaries, salinity constraints and changes in the watertable can also be seen in their proper landscape context.The key cross-sections, flow directions, hydraulic gra-dients and their implications on groundwater flow ratesand recharge are readily illustrated.

Most usefully for water-resource management, themain fluxes of groundwater into and out of the area havebeen quantified, allowing usage to be seen in this context.This is beneficial in being more open about how theresource usage compares to the natural (uncontrolled)system and also supports the case for adaptive ground-water management (Maimone 2004) by providing sup-porting evidence. In the case of Spring Hill GSPA, theneed to manage impact and equity at the most vulnerableparts of the aquifer system is evident. It is also evident thatin the area above the flow constriction (cross-section A–A’),

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

usage has already reached a sustainable level ofdevelopment, particularly in the most elevated parts ofthe aquifer.

Acknowledgements This study was made possible by the fundingmade available from the Australian Government via the NationalWater Commission as well as the Victorian Government through theVictorian Water Trust. Thanks are also extended to Dr P. Dahlhaus ofBallarat University and Dr J. Webb of La Trobe University for theirconstructive criticisms of earlier drafts of the paper. We are alsograteful for the experienced and considered comments of reviewersG. D. Lecain and M. Ross that led to significant improvements inthe manuscript.

References

Alley WM (2007) Another water budget myth: the significance ofrecoverable groundwater in storage. Groundwater 45(3):251

Alley WM, Reilly TE, Frank OL (1999) Sustainability of ground-water resources. US Geol Surv Circ 1186

Artimo A, Makinen J, Berg RC, Abert CC, Salonen VP (2003)Three-dimensional geologic modelling and visualisation of theVirttaankangus aquifer, southwestern Finland. Hydrogeol J11:378–386

Beatty B (2003) The hydrology and water chemistry of the upperLoddon catchment, Clunes, Victoria. Honours Thesis, La TrobeUniversity, Australia

Berg RC, Russel HAJ, Thorleifson LH (2009) Introduction: threedimensional geologic mapping—an international perspective.Three dimensional Geologic Mapping Workshop GSA PortlandOregon. http://www.isgs.uiuc.edu/research/3DWorkshop/2009/workshop.shtml). Cited May 2011

Bredehoeft JD (2002) The water budget myth revisited: whyhydrogeologists model. Ground Water 40(4):340–345

Bredehoeft JD, Durbin TJ (2009) Ground water development: thetime to full capture problem. Ground water-2009. NationalGround Water Association, Westerville, OH

Bredehoeft JD, Papadopulos SS, Cooper HH Jr (1982) The waterbudget myth. In: Scientific basis of water resource management:studies in geophysics. National Academy Press, Washington,DC, pp 51–57

Brown RH (1963) The cone of depression and the area of diversionaround a discharging well in an infinite strip aquifer subject touniform recharge. US Geol Surv Water Suppl Pap 1545c, ppC69–C85

DME (1983) Castlemaine and creswick deep leads maps. BulletinNo. 62, deep lead gold deposits in Victoria. Geological Survey,within the Department of Minerals and Energy, Melbourne,Australia

DNRE (1996) Summary report on PAV methodology, draft B.Prepared by SK Merz for the Department of Natural Resourcesand Environment, Melbourne, Australia

DNRE (1998) Permissible annual volume project for the Spring HillGMA. Monograph series ISSN 1328–4495, Groundwater reportno. 38, Prepared by SK Merz for the Department of NaturalResources and Environment, Melbourne, Australia

Douglas JG, Ferguson JA (eds) (1988) Geology of Victoria.Geological Society of Australia, Melbourne, 664 pp

DSE (2011) Victorian Water data Warehouse. Online access toVictorian surface and groundwater monitoring data. http://www.vicwaterdata.net/vicwaterdata/home.aspx. Cited May 2011

EPA (Environment Protection Authority) (1997) State environmentprotection policy: groundwaters of Victoria—beneficial useclasses (salinity segments). Victorian Government Gazette, S160, Melbourne, Australia

Fawcett J, Cherry D, Reid M (2006) Hydrogeological units of theUpper Loddon and Spring Hill Groundwater Management units.Department of Primary Industries, Melbourne, Australia

Fetter CW (1988) Applied hydrogeology, 2nd edn. Merrill,Columbus, OH

GMW (2001) Spring Hill water supply protection area managementplan (groundwater). Developed by the Spring Hill water supplyprotection area consultative committee. Goulburn Murray Water,Tatura, Australia. http://www.g-mwater.com.au/downloads/Groundwater/Springhill_Groundwater_Mgt_Plan.pdf. CitedMay 2011

GMW (2009) Spring Hill water supply protection area managementplan (groundwater). Annual report for the year ending June2009 Document Number 2696001. Goulburn Murray Water,Tatura, Australia. http://www.g-mwater.com.au/downloads/Annual_Reports/Groundwater_Annual_Reports/2009/2009_Springhill_Groundwater_Annual_Report.pdf. Cited May2011

Healy RW, Cook PG (2002) Using groundwater levels to estimaterecharge. Hydrogeol J 10(1):91–109

Holdgate GR, Wallace MW, Gallagher SJ, Witten RB, Stats B,Wagstaff BE (2006) Cenozoic fault control on ‘deep lead’palaeoriver systems, Central Highlands, Victoria. Aust J EarthSci 53:445–468

Hunter S (1909) The deep leads of Victoria. Memoirs of theGeological Survey of Victoria, no. 7. Department of Mines,Melbourne, Australia

Ife D (1979) Groundwater manual: an introduction to practicalhydrogeology. State Rivers and Water Supply Commission ofVictoria, Melbourne, Australia

Kalf FRP, Woolley DR (2005) Applicability and methodology ofdetermining sustainable yield in groundwater systems. Hydro-geol J 13(1):295–312

Kessler H, Mathers S, Sobisch H-G (2010) The capture anddissemination of integrated 3D geospatial knowledge at theBritish Geological Survey using GSI3D software and method-ology. Comput Geosci 35:1311–1321

Kiem AS, Verdon-Kidd DC (2010) Towards understanding hydro-climatic change in Victoria, Australia – Preliminary insights intothe “Big Dry”. Hydrology and Earth Systems Sciences 14,European Geosciences Union, Munich, Germany, pp 433–445

LCC (1980) Report on the Ballarat area. Land ConservationCouncil, Melbourne, Australia

MacCormack KE, Maclachlan JC, Eyles CH (2005) Viewing thesubsurface in three dimensions: initial results of modelling theQuaternary sedimentary infill of the Dundas Valley, Hamilton,Ontario. Geosphere 1(1):23–31

Maimone M (2004) Defining and managing sustainable yield.Ground Water 42(6):809–814

McKay ED (2011) Illinois State geological Survey (29011) Three-dimensional geological mapping for groundwater applicationsworkshops. http://www.isgs.uiuc.edu/research/3DWorkshop/index.shtml. Cited May 2011

Moghaddam AA, Fijani E (2009) Hydrogeologic framework of theMaku area basalts, northwestern Iran. Hydrogeol J 17:949–959

Nury SN, Zhu X, Cartwright I, Ailleres L (2010) Aquifer visual-ization for sustainable water management. Manage EnvironQual 21(2):253–274

NWC (2006) Australia’s water supply status and seasonal outlook.National Water Commission (NWC), Canberra, Australia, 17 pp

Reid MA, Cherry D (2004) Geological base plans for the audit ofpermissible annual volumes for 35 groundwater managementareas in Victoria (this document). CD-ROM. Primary IndustriesResearch Victoria, Bendigo, Australia

Robins NS, Rutter HK, Dumpleton S, Peach DW (2004) The role of3D visualisation as an analytical tool preparatory to numericalmodelling. J Hydrol 301:287–295

Ross M, Parent M, Lefebvre R (2005) 3D geological frameworkmodels for regional hydrogeology and land-use management: acase study from a Quaternary basin of southwestern Quebec,Canada. Hydrogeol J 13:690–707

Saha D, Agrawal AK (2006) Determination of specific yield using awater balance approach: a case study of Torla Odha watershedin the Deccan Trap province, Maharastra State, India. Hydro-geol J 14:625–635

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy

SILO (2009) NRM enhanced meteorological data sets, developedby a consortium of national and state government agenciesunder the under the Climate Variability in Agriculture R&Dprogram (CVAP). http://www.longpaddock.qld.gov.au/silo/.Cited May 2011

Sophocleous M (1997) Managing water resource systems: why“safe yield” is not sustainable. Ground Water 35(4):561

Sophocleous M (2000) From safe yield to sustainable developmentof water resources: the Kansas experience. J Hydrol 235:27–43

Theis CV (1940) The source of water derived from wells: essentialfactors controlling the response of an aquifer to development.Civil Eng 10:277–280

Tickell SJ, Humphrys WG (1986) Groundwater resources andassociated salinity problems of the Victorian part of the RiverinePlain. Report no. 84, Dept. of Industry, Technology andResources, Melbourne, Australia

Turner AK [Editor] (1991) Three-dimensional modelling withgeoscientific information systems, NATO ASI series C: mathe-matical and physical sciences, vol 354, Kluwer, Dordrecht, TheNetherlands, 443 pp

Wycisk P, Hubert T, Gossel W, Neumann Ch (2009) High-resolution3D spatial modelling of complex geological structures for anenvironmental risk assessment of abundant mining and indus-trial megasites. Comput Geosci 35:165–182

Hydrogeology Journal DOI 10.1007/s10040-011-0757-7

Author's personal copy