HR: 1340h
AN: H23E-1470    [Abstracts]
TI: Identification of the monitoring point density needed to reliably estimate contaminant mass fluxes
AU: Liedl, R
EM: rudolf.liedl@uni-tuebingen.de
AF: Center for Applied Geoscience, University of Tuebingen, Sigwartstrasse 10, D-72076, Tuebingen, 72076 Germany
AU: * Liu, S
H23E-1470 AF: Center for Applied Geoscience, University of Tuebingen, Sigwartstrasse 10, D-72076, Tuebingen, 72076 Germany
AU: Fraser, M
H23E-1470 AF: Department of Earth Sciences, University of Waterloo, 200 University Ave. W., Waterloo, ON N2L 3G1 Canada
AU: Barker, J
H23E-1470 AF: Department of Earth Sciences, University of Waterloo, 200 University Ave. W., Waterloo, ON N2L 3G1 Canada
AB: Plume monitoring frequently relies on the evaluation of point-scale measurements of concentration at observation wells which are located at control planes or `fences' perpendicular to groundwater flow. Depth-specific concentration values are used to estimate the total mass flux of individual contaminants through the fence. Results of this approach, which is based on spatial interpolation, obviously depend on the density of the measurement points. Our contribution relates the accurracy of mass flux estimation to the point density and, in particular, allows to identify a minimum point density needed to achieve a specified accurracy. In order to establish this relationship, concentration data from fences installed in the coal tar creosote plume at the Borden site are used. These fences are characterized by a rather high density of about 7 points/m² and it is reasonable to assume that the true mass flux is obtained with this point density. This mass flux is then compared with results for less dense grids down to about 0.1points/m². Mass flux estimates obtained for this range of point densities are analyzed by the moving window method in order to reduce purely random fluctuations. For each position of the moving window the mass flux is estimated and the coefficient of variation (CV) is calculated to quantify variablity of the results. Thus, the CV provides a relative measure of accurracy in the estimated fluxes. By applying this approach to the Borden naphthalene plume at different times, it is found that the point density changes from sufficient to insufficient due to the temporally decreasing mass flux. By comparing the results of naphthalene and phenol at the same fence and at the same time, we can see that the same grid density might be sufficient for one compound but not for another. If a rather strict CV criterion of 5% is used, a grid of 7 points/m² is shown to allow for reliable estimates of the true mass fluxes only in the beginning of plume development when mass fluxes are high. Long-term data exhibit a very high variation being attributed to the decreasing flux and a much denser grid would be required to reflect the decreasing mass flux with the same high accurracy. However, a less strict CV criterion of 50% may be acceptable due to uncertainties generally associated with other hydrogeologic parameters. In this case, a point density between 1 and 2 points/m² is found to be sufficient for a set of five tested chemicals.
DE: 1848 Monitoring networks
DE: 1873 Uncertainty assessment (3275)
DE: 1895 Instruments and techniques: monitoring
SC: Hydrology [H]
MN: Fall Meeting 2005