Anne Lausten Hansen (alha@geus.dk) 1,2

1. A concept for estimating depths to the redox interface for catchment scale nitrate modelling in a till area Anne Lausten Hansen (alha@geus.dk)1,2 Christensen BSB3, Ernstsen V1, He X1 and Refsgaard JC1 (1) Geological Survey of Denmark and Greenland (2) Department of Geosciences and Natural Resource Management, University of Copenhagen (3) Rambøll

2. Introduction • Nitrate can be naturally transformed by reducedcompounds (OM, Fe+2,pyrite) in thesediments • Transition from oxic to reducedconditions = redox interface • Spatial variation in the redox interface and in water flow pathsleads to nitrate sensitive and nitrateroboust areas • Importantto know the location of the redox interface to delineatetheseareas Nitrate sensitive area Nitrate roboust area

3. Introduction • Location of the redox interface in till areas varies several meters within short distances • The interface can only be determined by drilling boreholes => Limited data Large uncertainty on the location of the redox interface • Interestingtodevelopmethodologiestoinferthelocationoftheredoxinterfacefromother variables Objective of this study

4. Redox interface development - Hypothesis- • The reduced compounds (redox capacity) in the sediments is depleted by oxygen and nitrate • The present location of the redox interface is the result of the cumulative flux of oxygen in recharging groundwater since the onset of Holocene (11.700 years) • Development of the interface in parts of the unsaturated zone can have happened fast due to oxygendiffusion in the air phase. In a clay till, however, this is only important in the root zone

5. Redox interface concept • Key principle: estimatethespatialpatternoftheredoxinterfacefromvariability in groundwaterrechargeandsedimentredoxcapacity • Redoxequation: the redoxdepth in grid i is estimated as: Redox depthi = fluxi · f + min. redox depth • flux: rechargefluxestimated with hydrological model • f: redox interface migration constant (m over 11.700 pr mm yearlyrecharge) • min. redoxdepth: Upper part of UZ whereredoxcapacity have beendepleted fast due to air phase diffusion Additional parameters: • Maximum redoxdepth • Lowerredoxdepth in riparianlowlands Dependent on the sediment redoxcapacity !

6. Redox interface concept Step 1: Extraction of rechargeflux from hydrological model (nodrainage and pumping) Step 2: Difference in redoxcapacitybetween sediment types applied to rechargemap Step 3: Applyredoxequation, define f for main sediment type Step 4: Run nitrate model with estimatedredox interface => Simulatednitrate arrival (% of nitrate input, NAP) at catchmentoutlet Step 5: Comparesimulated and observed NAP If sim >< obs => new constant f and min. redoxdepth

7. Application in Norsminde fjord catcment Redoxdepth observations Topography Soil type

8. Models • Geological model • 11 hydrogeological units • Based on borehole data from Jupiter and geophysical data from Mini-SkyTEM • Hydrological model • MIKE SHE/MIKE 11 • All hydrological processes • Grid scale 100x100 m • Nitrate model • Particletracking (MIKE SHE)

9. Nitrate model - particle tracking • Nitrate input: Daily N leaching from root zone • N balance methodcombined with Daisy simulations (Thirup (2013), available at www.nitrat.dk) • Redox interface implemented as registration zone => particleregistreted if crossing interface • Nitrate arrival: particlesarriving in fjord withoutcrossingredox interface • The model is run 4 years with N input (2000-2003) and thenadditional 4 years to get all nitrate out (flow recycled) Distribution of particles at different sim. time (N addedfirst 4 years)

10. Calibration target- Nitrate arrival percentage (NAP) to Norsminde fjord - 41 – 49 % of the nitrate leaching arrives in Norsminde fjord

11. Redox scenarios and calibration • Redox scenarios (based on sensitivityanalysis) • Scenario 1: Recharge flux layer 2 (3 - 4 m.b.s) Redox depth in riparian lowlands 1.5 m • Scenario 2: Recharge flux layer 1 (0 - 3 m.b.s) Redox depth in riparian lowlands 1.5 m • Scenario 3: Recharge flux layer 2 (3 - 4 m.b.s) No riparian lowlands • Calibration • All 3 scenarios wascalibrated to NAP = 45% Calibrated parameter values Norsminde redoxdata (claytill) Avg. redoxcapacity: 418 meq-e/kg O2conc.: 11.4 mg/l (10oC) => Constant f = 0.025

12. Redox interface and reduction maps

13. Estimated versus observed redox depths- point scale -

14. Estimated versus observed redox depths- catchment scale -

15. Evaluation of Results • The model is able to simulate observed nitrate arrival (NAP) to Norsminde fjord • All 3 scenarios can be cailbrated to NAP = 45% => equifinality • Redox depth observations not sufficient to choose between scenarios • Cumulative distribution of redox depths close to observed • Site-specific redox depths is not well estimated • Results okay on cathment scale, but not on small scale

16. Factors affecting the results • Rechargeflux • Constantflux • Onlyvertical component of flux • Migration constant f • Uniform migration constant f within sediment type • Variation in sediment type with depth not included • Scaleissue (Model grid scale 100x100 m) • Affectsestimatedredoxdepths due to averaging • Affectscompariosn of estimated vs. obsevedredoxdepths • Nitrate data • N leaching • N flux to Norsminde fjord • Geological and hydrological model • Flow pathscorrect ? Norsminde data Redoxcapacity (claytill) Avgerage: 418 meq-e/kg St.dev.: 150 meq-e/kg

17. Conclusions • The concept is capable of estimating the general location of the redox interface, but not at grid scale • The model is therefore not able to accurately simulate nitrate reduction at grid scale • The uncertainty on the reduction potential maps needs to be evaluated

18. Work in progress- Application of redox concept on 20 geological models - Uncertaintyon nitratereduction at differentaggregationscales

19. Thankyou for your attention!