Presentation on theme: "Assessing Mining Impacts Groundwater Modeling in Karst Aquifers"— Presentation transcript:
1Assessing Mining Impacts Groundwater Modeling in Karst Aquifers Todd R. Kincaid, Ph.D.Hazlett-Kincaid, Inc.H2H Associates, LLCTimothy J. Hazlett, Ph.D.Ardaman Associates
2Problem / Purpose“Limestone areas are bad places for limerock quarries” Environmental Consultant PennsylvaniaIf our society needs quarry products, then the challenge is to locate them and manage them such that environmental impacts are minimized and repaired to every extent possible after the quarry closes.Impact assessments are often obscure and not accessibleDescribe how modeling contributes to impact assessmentsDescribe how I think it SHOULD be used to assess impacts
3What is a Groundwater Model? Groundwater models are tools most often used to make predictions about future impacts to groundwater resources associated with proposed activities or existing problems.They can be:analytical (meaning mathematically solvable),stochastic (meaning statistically solvable),numerical (meaning solvable through iteration),or some combination of the three.They are often required by State and Federal regulatory agencies for impact and contamination assessments specifically because of their ability to predict the future.The problem is that most of these models are based on simplifying assumptions that render the resulting predictions dangerously inaccurate particularly when applied to karst aquifers.
4Basic Conceptualizations Most commonly assumedMost commonly true
10Why Does it Matter? Wakulla Springs Basin – North Florida Simple World Simple models produce simple simulationsProduce very generalized estimations of groundwater flow directionsUnder-estimate velocitiesMisrepresent connectionsReal World
12Water Budget / Flow Chart Models produce predictions of groundwater levels from which all other things are calculatedCalibration is the process of matching model predicted conditions to observed conditionsTypically only done for groundwater levelsIn Karst it is critical that models are also calibrated tospring flows (regional)observable conduit velocities (tracer tests, observed responses to storms or collapse, etc)
13Sinkhole Overview 19 sinkholes in current model 15 known to exist4 assumedSinkhole systems can be grouped by three dominate sinksSanta Fe River Sink (largest)Capacity estimated at 1000’s of cfsLarge volume conduit network connects to Santa Fe RiseMill and Rose Creek sinksCapacities estimated up to 100 cfsLarge volume conduits extend downstream from sinksAll other sinks estimated to have capacities > 30 cfsAlligator and Waters Lakes known to drain through low capacity sinks during drought yearsWatermelon Pond, Burnetts Lake, Lake Wilson, and Lake Lona assumed to have similar sinks
14Springs Overview 18 Springs in current model Santa Fe River Rise is Largest SpringDischarge in 1000’s of cfsDirect flow from Santa Fe SinkHornsby second largest springDischarge up to 250 cfsMajority of discharged believed to be sinking surface waterCan go dry during droughts (fall, 2002)Other springs part of three groupsIchnetuckeneeGinnieDevil’s Ear
15Defining Conduits Incorporating Cave and Karst Pathways 1 2 Hand-contoured potentiometric surfaceComposite data - May 2005lines colored by elevationdepressions assumed to reflect presence of conduits/caves12Projected & known caves & pot surfaceOld Bellamy only cave shownTraced pathways based on KES data 1) Rose Sink cave – Ichetucknee group 2) Mill Creek cave – Hornsby springProjected caves - follow potentiometric lows - connect swallets to springs or to established pathways1. Data_potmap.jpg - shows hand drawn pot-map based on May 2005 static well datatalk: lines colored by elevationtalk: areas of head depression assumed to be due to presence of condiut2. Data_potmap_caves.jpg - shows projected and known caves over pot maptalk: Old Bellemy cave only surveyed cave shown (not shown; Rose Sink, mill creek, Ginnie, Hornsby)talk: Tracer caves are drawn based on pot-map connecting Pete Butts sinks and springstalk: projected caves are based only on pot-map3. Data_potmap_caves_streams.jpg - shows relationship between caves and sinking streamstalk: pot lines greyed outtalk: caves connect sinking creeks to Hornsby spring, Ichetucknee springs, and Santa Fe riseGinnie, Twin, Dogwood, Gilchrist Blue, Lilly, Poe not connected to sinksTalk: July and Devil's Eye springs not yet in modelConnectionssinking streams to Hornsby spring, Ichetucknee springs, and Santa Fe riseGinnie, Twin, Dogwood, Gilchrist Blue, Lilly, Poe not connected to sinksJuly & Devil’s Eye not yet in model
16Cave vs No Cave Results Evaluation Approach – Preliminary Comparisons Develop independent modelsOne with no cavesOne with cavesCompare results using calibration as the measuring stickNo-cave model similar to standard approachCave model significantly more thorough
17Cave vs No Cave Results Cave Model No-Cave Model 1. Flordan_nocave_head_all.jpg - shows modeled flordan head with no caves addedtalk: developed for calibration comparison purposes. Could a good flordan model be built without caves?talk: in general heads are higher across the whole modeltalk: heads are esecially high in south central part of model, suggests that condiuts are needed
18Model Comparison: Southern Area Cave ModelNo-Cave ModelQuarry Area1. Flordan_head_calsouth.jpg - shows flordan head in southern part of model with calibration points overlaidtalk: this is model with cavestalk: calibration is good everywhere (within +- 1 meter). Model needs minor calibration adjustments2. Flordan_nocave_head_calsouth.jpg - shows flordan head in southern part of model with calibration points overlaidtalk: this is model without cavestalk: calibration is good closest to Bell Ridge and Sanchez Prarietalk: calibration is poor in central part of model (heads off by 3 to 5 meters)talk: model head in general too high over entire southern half of model due to lack of condiutsmodels not very comparableno-cave model shows large area of anomalously elevated head (+ 3-5 meters) / okay in confined areascave model is well calibrated everywherecave model shows quarries in springshed
19Example: Corkscrew Mine Lee County, Florida – 2003 Quarry Permit ApplicationDefine impacts to the water table near the mine due to the removal of mined materials – operational stageHow will they change under proposed dewatering?What is the magnitude of these impacts and how long might they last?How might karst features impact the groundwater flow field?Dewatered QuarrySchmidt PropertyN11/13/03Copyright Hazlett-Kincaid, Inc All rights reserved.
20Results: Dewatering Scenario Drawdown indicated from unperturbed condition to 36 months of dewatering at 25’ b.l.s.Recharge trench includedPerimeter of mineFlux limitedModel was designed such that effect of trenches could be assessedLarger scale with less assumptions yields different predictionsMine ModelModel shows no offsite impactsMost water infiltrated through holding ponds, but not shown in modelLeakage between holding pondsTrench is built up with berms (horiz. flow)Our ModelTrench is strictly 10 ft wideHead is strictly 2 ft deepRecharge flux is limited to ~72 mgd (from permit)Trench is part of top layer of sandSand and limestone layers use thickness from GFMClay is bottom of modelHeads are steady state flow from north to southIn pit, heads are set at 25’ b.l.s., as specified in permit applicationAll hydraulic parameters taken from permit app.Pit modeled as a hole in the ground11/13/03Copyright Hazlett-Kincaid, Inc All rights reserved.
21Potential Impact of Karst Hypothetical conduit extending from wetland discharges to quarryConduit assumed to be in the limestone beneath the quarry floorFlow concentrated along conduitDrains the water table in a localized fashionLower heads in conduitsHigher flow rate in hotter colorsSame calibration but different prediction of where the impacts will be feltN11/13/03Copyright Hazlett-Kincaid, Inc All rights reserved.
22Corkscrew Modeling Results The modeled excavation of the pit demonstrates a drawdown in the surficial aquifer.The maximum drawdown is about 10 cm (~1 ft) under the Schmidt property, at the northeast corner.The water table rebounds to near equilibrium conditions in about 150 days.The water table will not reach full equilibrium over the long term, due to increased loss from the aquifer from evaporative flux from the pit.Substantially greater groundwater drawdown will occur offsite of the mine property if mechanical dewatering is implemented as described in the permit application (recently withdrawn)The recharge scenario as described in the permit application will not significantly mitigate the drawdown effectModeling results are dependent on the initial conceptualization and model designShould be designed to match real-world as closely as possible11/13/03Copyright Hazlett-Kincaid, Inc All rights reserved.
23Example: Hercules Quarry Preliminary No Karst ModelKarst ModelLocalized Cone of DepressionNo account for exchange with nearby creekLarger Cone of DepressionConduit exchange with creek included
24Summary / ConclusionsThe quality of an impact assessment is heavily determined by the accuracy and scale of the conceptual model on which they’re basedFlorida limestones are predominantly karstConceptual models therefore need to be large enough to account for local and regional hydrologic features (recharge & discharge)Modeling can be an effective tool for predicting impacts IF they’re based on reasonably accurate conceptualizations
25Summary / Conclusions Need for basin-wide characterization models Continuously updated with new dataGoal: provide ongoing background levels – ie what is the pre-mining condition from which impacts can be addressedIncorporate high frequency dataPublically available & transparentRequired foundation for all local-scale investigations & assessmentsNeed for comprehensive look at water use throughout the life of mineConsumptive use, recharge, discharge, reclaimed use (Water Balance Evolution)What is the post-operation mine going to be used for?
26Summary / Conclusions Quarries are (most likely) an unwanted given Must look at how we manage them to ensure the best possible outcome for the environmentWhere they go, how big they are, how long they operate All these factors can and probably will impact groundwater resources and should therefore be at least partly determined by comprehensive and consistent regional characterizationsHow can we get these comprehensive characterization models done?Public private partnershipsPartially publically funded / partially funded by quarries