Journal article

Three-dimensional statistical moment analysis of the Stanford/Waterloo Borden tracer data

We estimate the mass, mean position vector, and covariance tensor of the bromide plume investigated at the Borden site (Ontario, Canada) by researchers from Stanford University and the University of Waterloo. A number of interpretive steps was necessary to compute these plume moments. First, the spatially scattered data were interpolated onto a regular grid, with the interpolator selected on the basis of a cross-validation criterion. The moment estimates then could be calculated readily, although it was recognized that the interpolation procedure can lead to bias in the moment estimates. For this reason, reduced-bias, jackknifed estimates of the moments, as well as confidence intervals for these estimates, were derived. The total plume variance also was estimated with the jackknife method. The latter results were compared with the prediction of a recent stochastic model developed by Dagan (1988). When this model was applied in a completely predictive mode, a remarkably good estimate of the total plume variance resulted.


Related material