Paper No. 7
Presentation Time: 3:20 PM

BIAS OF ENVIRONMENTAL TRACER METHODS IN HETEROGENEOUS AQUIFERS


MCCALLUM, James, School of the Environemt, National Centre for Groundwater Research and Training, Flinders University, GPO Box 2100, Adelaide, 5001, Australia, COOK, Peter G., National Centre for Groundwater Research and Training/CSIRO Land and Water, Flinders University, Ring Road, Bedford Park, 5042, Australia, SIMMONS, Craig T., College of Science and Engineering & National Centre for Groundwater Research and Training (Australia), Flinders University, GPO Box 2100, Adelaide, 5001, Australia and WERNER, Adrian D., National Centre for Groundwater Research and Training, Flinders University, GPO Box 2100, Adelaide, 5000, Australia, mcca0206@flinders.edu.au

Knowledge of groundwater age allows for the direct determination of the timescales of groundwater flow. This can be important in determining hydraulic fluxes (recharge rates, flow rates) or arrival times of solutes. The use of environmental tracers (compounds with known historical atmospheric concentrations and/ or decay rates) to determine apparent age has been demonstrated to be inaccurate in heterogeneous environments. In this study, we use numerical simulations to examine the bias of CFC’s, SF6, 85Kr and 39Ar in heterogeneous fields. Three types of hydraulic conductivity (K) field are simulated – multi-Gaussian, connected and structured fields, with increasing levels of ln(K) variance. Groundwater flow was simulated using the finite element model HydroGeoSphere. Environmental tracer concentrations and residence time distributions were simulated using a random walk particle tracking model. We demonstrate that when using environmental tracers in heterogeneous environments, a bias occurs when the distribution of water ages covers a range of ages corresponding to times when atmospheric concentrations were non-linear. The direction of the bias (younger/older) is dependent on the shape of the input concentration and the magnitude is dependent on the width of the age mixture. This width is determined by both variance and structure of the K field. As the structure of the K field is generally unknown in practice this bias cannot be corrected for. However, over age ranges where different tracers show different levels of bias, the relationship between two apparent ages may be useful in determining mean ages. We present a case where the use of CFC-12 and CFC-113 data can be used conjunctively to determine mean ages.