Appendix A: Conference Paper 1 (supplement to Chapter 2)

Quantifying biogeochemical changes during ASR of reclaimed water at Bolivar, South Australia

Abstract

A modelling study was carried out to provide a process-based quantitative interpretation of the biogeochemical changes that were observed during an ASR experiment in which reclaimed water was injected into a limestone aquifer at a field-site near Bolivar, South Australia. A site-specific conceptual model for the interacting hydrodynamic and biogeochemical processes that occur during reclaimed water ASR was developed and incorporated into an existing reactive multi-component transport model. The major reactive processes considered in the model were microbially mediated redox reactions, driven by the mineralisation of organic carbon, mineral precipitation/dissolution and ion exchange. The study showed that the geochemical changes observed in the vicinity of the ASR well could only be adequately described by a model that explicitly considers microbial growth and decay processes, while an alternative, simpler model formulation based on the assumption of steady state biomass concentration failed to reproduce the observed hydrochemical changes. However, both, the simpler and the more complex model approach were able to reproduce the geochemical changes further away from the injection/extraction well. These changes were interpretated as a result of the combined effect of ion exchange, calcite dissolution and mineralisation of dissolved organic carbon.

Introduction

Aquifer Storage and Recovery (ASR, Pyne, 1995) is an increasingly popular technique to augment groundwater resources and secure and enhance water supplies. During ASR, physical, chemical and biogeochemical processes modify the water quality within the target aquifer. For instance, the injection of both oxic potable water and oxic nutrient-rich reclaimed water into an anaerobic aquifer can lead to a number of microbially mediated redox reactions (Stuyfzand, 1998; Vanderzalm et al., 2002), which in turn may trigger further geochemical reactions that have a considerable effect on the water quality and the composition of the aquifer matrix. (Eckert and Appelo, 2002). In order to design and operate efficient, sustainable and safe ASR schemes, a qualitative and quantitative understanding of those processes is important. Therefore, several field-scale investigations have been carried out over the past years, some of them specifically investigating the geochemical response to the injection of high quality water (e.g., Stuyfzand, 1998; Mirecki et al., 1998) but also of reclaimed wastewater (e.g., Valocci et al., 1981; Vanderzalm et al., 2002). However, numerical models which investigate the interacting hydrodynamic and biogeochemical processes during ASR (e.g., Valocchi et al., 1981, Saaltink et al., 2003; Prommer and Stuyfzand, 2005) and assist in the analysis and interpretation of observed field data are rather scarce to date.

As part of a larger-scale study (Dillon et al., 2005), Greskowiak et al., (2005, Chapter 2) developed a site-specific modelling framework that provides a process-based interpretation of the biogeochemical changes that occurred during a field experiment in which pre-treated, nutrient-rich reclaimed water was injected into a limestone aquifer at Bolivar, South Australia.

The present paper addresses the ubiquitous question of selecting an appropriate level of model complexity under such circumstances. To illustrate this issue we compare and discuss two alternative conceptual biogeochemical models and their numerical implementation in relation to their respective capability of describing the observed field data.

Background

The Bolivar ASR site is located in the Northern Adelaide Plains, South Australia and used to investigate the viability of storage and recovery of reclaimed water intended to compensate the greater demand of irrigation water during summer (Vanderzalm et al., 2002). During the experiment, the reclaimed water was injected into a brackish limestone aquifer, which is separated from the overlying fresh water aquifer by a 7.5 m thick confining clay layer. Discontinuous injection of 250 ML took place between October 1999 and April 2000, followed by a storage period of 110 days. Subsequently, about 150 ML were recovered within the following 130 days (Figure A.1). The injection took place over the entire depths of the target aquifer, i.e., from 103 m to 160 m below the ground surface. Several multilevel wells have been installed at various radial distances from the ASR well, monitoring the geochemical evolution of the groundwater during injection, storage and recovery.

Figure A.1: Cumulative volume of injected water

The hydrogeochemistry of the target aquifer is characterised by anoxic conditions and the aquifer matrix is composed of about 74 % calcite, 18 % quartz and small amounts of ankerite and hematite (Vanderzalm et al., 2002). The average total cation exchange capacity (CEC) is 20 meq/kg.

Pumping tests, flowmeter, temperature and anisotropy measurements revealed that the target aquifer could be classified into four stratified zones of similar thickness but of distinctly different permeability with an average horizontal hydraulic conductivity of about 3 m/d (Pavelic et al., 2001). Thereby, lateral flow occurs preferentially in two layers, referred to as Layer 1 and Layer 3.

Throughout the trial period, the quality of the injectant varied significantly with time. For example, oxygen and nitrate concentrations ranged between <0.02 and 0.32 mmol/L and <0.0004 and 0.34 mmol/L, respectively. Correspondingly, ammonium concentrations varied between <0.02 and 2 mmol/L. However, dissolved organic carbon concentrations (DOC) were relatively constant with an average concentration of 1.40 mmol/L. The average total organic carbon concentration (TOC) was slightly higher (~1.51 mmol/L) and it is assumed that organic matter in particulate form (POC) compensated the concentration difference between DOC and TOC. The injection water is generally under-saturated with respect to calcite (SICalcite = -1.44 to 0.13). As the chloride concentration of the injectant is approximately 50 % of the ambient groundwater concentration, chloride was thought to be a suitable tracer for the identification of physical transport. More details on the monitored hydrochemistry are given in Vanderzalm et al. (2002).

Groundwater flow and reactive transport model

In a first step a three-dimensional flow and conservative transport model of the ASR trial was set up and calibrated. In a second step the results were used to determine the groundwater flow within the most permeable layer (Layer 3) and to construct a computationally more efficient quasi-radial flow and transport model for this particular layer. The quasi-radial flow model formed the basis for the subsequent simulations with the reactive multi-component transport model PHT3D (Prommer et al., 2003). The reaction network considered in those simulations included all major ions, oxygen, one type of ion exchanger site, two forms of mobile organic carbon (i.e., DOC, POC), immobile organic carbon, four minerals (i.e., calcite, hematite, siderite and amorphous iron sulphide) and two microbial groups. The two microbial groups were defined as facultative aerobic/denitrifying bacteria and facultative iron/sulphate-reducing bacteria, respectively. The reaction stoichiometry of the redox reactions, which incorporate microbial growth and decay were adapted from Prommer et al. 2002 and linked to a standard Monod-type microbial growth model (e.g., Barry et al., 2002). The POC contained in the injectant was expected to rapidly become immobile in the close vicinity of the ASR well due to filtration (Skjemstad et al., 2002) and attachment was simulated using first order kinetics. From there it was assumed to solubilise and to form a continuous source of DOC. The solubilisation was simulated by a kinetic approach adapted from Kinzelbach et al., 1991. More details are given in Greskowiak et al. (2005, Chapter 2). For the present study two alternative conceptual biogeochemical models of differing complexity were investigated:

In alternative (A), the simplifying assumption of a steady state microbial concentration was incorporated and the mineralisation of DOC releases ammonia (Jacobs et al., 1988)

In alternative (B), microbial growth and decay were explicitly modelled and the release of ammonium was assumed not to be associated with the solubilisation of the filtered POC. Instead, ammonium was cycled through the occurrence of biomass growth and decay. That is, ammonium was included as a nitrogen source during biomass formation while biomass decay was assumed to release ammonium back into the aqueous phase.

For both alternatives, adjustable model parameters, i.e., the rate constants of the kinetic reactions were fitted such that the residual between simulated and observed concentrations was minimised. For this process the model-independent nonlinear parameter estimation program PEST (Doherty, 2002) was coupled to PHT3D.

Results and discussion

The simulation results of both calibrated alternatives (A) and (B) were compared with the hydrogeochemical data collected at the ASR well and the 50m well. The simulation results indicated that both alternatives were capable of reproducing the key features of the hydrochemical changes which occurred at the 50 m well during injection, storage and recovery. In the model, the degradation of DOC is accompanied by the consumption of oxygen (not shown), nitrate (Figure A.2) and sulphate (Figure A.2), as measured in the field. Furthermore, the observed retarded ammonium breakthrough at the 50m well can be decribed accurately by the simulated ion exchange reactions (Figure A.2), while the increased concentrations of calcium (relative to nonreactive transport, see Figure A.2) result from the dissolution of calcite.

However, alternative (A) was unable to account for some of the highly dynamic changes of the local geochemistry observed in the close vicinity of the ASR well during the storage period. While the observed rapid increase of DOC was well matched by the simulations (Figure A.2), the increase of ammonia (Figure A.2), alkalinity (Figure A.2), calcium (Figure A.2) and dissolved iron (not shown) as well as the drastic decrease of sulphate (Figure A.2) and pH (Figure A.2) was not reproduced by alternative (A).

On the other hand, all of the geochemical changes observed at the ASR well could be adequately simulated by alternative (B), providing evidence that the observed hydrochemistry is strongly affected by the dynamics of microbial growth and decay. The model results suggest that in particular during the storage period, microbial decay, i.e., the mineralisation of highly degradable biomass, plays a key role and consumes considerable amounts of oxidation capacity, as discussed in Prommer et al. (2002). Correspondingly, inorganic carbon and protons are produced during decay, as was observed in the field. In the model this effect was caused by the decay of facultative aerobic/denitrifying bacteria (Figure A.2), which reduced sulphate (Figure A.2) and hematite (not shown) during the storage period.

Figure A.2: Concentrations of DOC, ammonia/ammonium, nitrate, sulphate, calcium, alkalinity, pH and bacteria at the ASR and the 50m well. Concentrations of the injectant (crosses), observations (circles), alternative A (dash-dot lines), alternative B (solid lines) and non-reactive simulations (dotted lines) are shown.

Conclusions

The present study investigated two alternative conceptual biogeochemical models with respect to their capability of providing a quantitative and consistent description of the biogeochemical processes at a reclaimed water ASR research site in South Australia. The results indicate that an adequate simulation of the geochemical changes in the vicinity of the ASR well could be achieved by a more complex conceptual biogeochemical model, while geochemical changes at larger distances were well-described by both models. This suggests that dynamic changes in bacterial mass may be important in interpreting near-well hydrochemical data from reclaimed water ASR schemes and should at least be considered during formulation of conceptual and numerical models. However, the identified processes and their respective parameters are highly nonlinear and may be subject to non-uniqueness. Thus, the possibility cannot be excluded that other conceptual models might describe the data equally well (Reichert and Omlin, 1997). Generally, the benefit of mechanistic multi-component reactive transport models is seen primarily in their capacity to constrain or reject hypotheses on interactions of physical, chemical and biological processes and to a lesser extent in their predictive capabilities.

Literature

Barry, D. A., Prommer H., Miller, C. T., Engesgaard, P., and Zheng, C. (2002). Modelling the fate of oxidisable organic contaminants in groundwater. Adv. Water Resour., 25, 945-983.

Dillon, P., Toze, S., Pavelic, P., Vanderzalm, J., Barry, K., Ying, G-G., Kookana, R., Skjemstad, J., Nicholson, B., Miller, R., Correll, R., Prommer, H., Greskowiak, J. and Stuyfzand, P. (2005). Water quality improvements during aquifer storage and recovery at ten sites, Proceedings of 5th International Symposium on Management of Aquifer Recharge, Berlin, June 2005, in press.

Doherty, J. (2002). PEST - Model-Independent Parameter Estimation. User’s manual, Fifth Edition. Watermark Numerical Computing, Brisbane, Australia.

Eckert, P., and Appelo, C. A. J. (2002). Hydrogeochemical modeling of enhanced benzene, toluene, ethylbenzene, xylene (BTEX) remediation with nitrate. Water Resour. Res., 38(8), doi: 10.1029/2001WR000692.

Greskowiak, J., Prommer, H., Vanderzalm, J., Pavelic, P. and Dillon, P. (2005) Modelling of carbon cycling and biogeochemical changes during a wastewater injection and recovery experiment at Bolivar, South Australia. Water Resour. Res. 41(10), W10418, doi: 10.1029/2005WR004095.

Jacobs, L.A., von Gunten, H.R., Keil, R., Kuslys, M. (1988). Geochemical changes along a river-groundwater infiltration flow path: Glattfelden, Switzerland. Geochim. Cosmochim. Acta, 52, 2693-2706.

Kinzelbach, W., Schäfer, W., and Herzer, J. (1991). Numerical Modeling of Natural and Enhanced Denitrification Processes in Aquifers, Wat. Resour. Res., 27(6), 1123-1135.

Mirecki, J. E., Campell, B. G., Conlon, K. J., and Petkewich, M. D. (1998). Solute changes during Aquifer Storage and Recovery testing in a limestone/clastic aquifer. Ground Water, 36(3), 394-403.

Pavelic, P., Dillon, P. J., Martin, R. R., Traegar, B., and Simmons, C. T. (2001). Multi-scale permeability characterisation of a confined carbonate aquifer targeted for aquifer storage and recovery. In: New approaches characterising groundwater flow, Seiler, K.P., and Wohnlich, S. (Eds.), Proceedings of the IAH XXXI. Congress 2001, Munich, Germany, 10 - 14 September, pp. 859 –862.

Prommer, H., Barry, D. A., and Davis, G. B. 2002. Modelling of physical and reactive processes during biodegradation of a hydrocarbon plume under transient groundwater flow conditions. J. Cont. Hydrol., 59, 113-131.

Prommer H., Barry, D. A., and Zheng, C. (2003). MODFLOW/MT3DMS based reactive multi-component transport modelling. Ground Water, 41(2), 247-257.

Prommer, H., and Stuyfzand, P. J. (2005). Identification of temperature-dependent water quality changes during a deep well injection experiment in a pyritic aquifer. Environ. Sci. and Technol., 39, 2200 -2209.

Pyne, R.D. (1995). Groundwater Recharge and Wells: A Guide to Aquifer Storage Recovery. Lewis Publishers, Boca Raton, Florida.

Reichert, P., and Omlin, M. (1997). On the Usefulness of Overparameterized Ecological Models. Ecological Modelling, 95, 289-299.

Saaltink, M. W., Ayora, C., Stuyfzand, P. J., Timmer, H. (2003). Analysis of a deep well recharge experiment by calibrating a reactive transport model with field data. J. Cont. Hydrol., 65, 1-18.

Skjemstad, J., Hayes, M., and Swift, R. (2002). Changes in Natural Organic Matter during Aquifer Storage. In: Management of Aquifer Recharge for Sustainability, Dillon, P. J (Ed.), A. A. Balkema, 149-154.

Stuyfzand, P. J. (1998). Quality changes upon injection into anoxic aquifers in the Netherlands: Evaluation of 11 experiments. In Artificial recharge of groundwater, Peters, J. H. et al. (Ed.), A.A. Balkema, 283-291.

Valocchi, A. J., Street, R. L., and Roberts, P. V. (1981). Transport of ion-exchanging solutes in groundwater: Chromatographic theory and field simulation. Water. Resour. Res., 17(5), 1517-1527.

Vanderzalm, J. L., Le Gal La Salle, C., Hutson, J. L., and Dillon, P. J. (2002). Water quality changes during aquifer storage and recovery at Bolivar, South Australia. In: Management of Aquifer Recharge for Sustainability, Dillon, P. J (Ed.), A. A. Balkema, 83-88.


© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 4.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML-Version erstellt am:
06.07.2007