HR: 0800h
AN: H21E-1058    [Abstracts]
TI: Kriging Plume Volumes: Sensitivity of Search Parameters
AU: * Watts, A W
EM: Awatts@unh.edu
AF: University of New Hampshire, 35 Colovos Road, Durham, NH 03824
AU: Davis, J M
EM: Matt.davis@unh.edu
AF: University of New Hampshire, 35 Colovos Road, Durham, NH 03824
AB: Geostatistical methods such as kriging are commonly used to estimate the concentration distribution of compounds in a ground water plume. Plume volume estimates can also be obtained by summing the volume of kriged cells with an estimated concentration above some threshold (e.g. MCL). Similarly, estimated concentrations can be used to estimated the total plume mass, assuming that the average estimated concentration is present in pore water within the estimated volume. Estimating the total mass with ordinary kriging has the advantage of being non-sensitive to the concentration at individual points, but careful restriction of the search parameters must be applied to accurately delineate the volume of the plume. In this study we evaluated the effect of search radii and octant criteria restriction on plume volume and mass estimates using the ok3d routine (Deutsch and Journel, 1998). Data from the Borden field tracer experiment were used, and the resulting mass estimates were compared to the initial injected mass, and to mass estimates obtained using spatial moment analysis and geometrical averaging. Ordinary kriging can be used to estimate the mass and volume of a compound in a ground water plume with results that correlate well with other techniques, but the method is highly sensitive to the octant search parameters. We found that the method should require data from at least five octants, with no more than one data point per octant. This prevents extrapolation beyond the plume boundary, while allowing sufficient coverage within the plume. The method is not highly sensitive to variogram parameters, but requires careful selection of search radii. The optimal search radius is between one and two well spacings. If the search radius is lower, plume mass is underestimated, and if the search radius is less than one well spacing the estimate may be completely invalid. A larger search radius tends to result in an overestimation of the plume mass and volume. However, when the octant criteria is adopted, the method is less sensitive to overly large search radii.
DE: 1829 Groundwater hydrology
SC: Hydrology [H]
MN: 2004 AGU Fall Meeting