Local‐ and Plume‐Scale Dispersion in the Twin Lake 40‐ and 260‐m Natural‐Gradient Tracer Tests

We present analysis and interpretation of experimental data collected in two natural-gradient dispersion tests over distances of 40 and 260 m, respectively. In the 40-m test, the use of a gamma-emitting tracer (131I) and a dry-access-well monitoring technique provided 1-cm resolution of the vertical concentration distribution. This in turn provided an opportunity to calculate dispersion coefficients at the local-scale for individual layers. In the 260-m test, the use of a beta-emitting tracer (3H) and multilevel sampling devices with 1-m vertical spacing did not allow the identification of individual layers. The local-scale longitudinal dispersivities, which may be only calculated at large distances from the injection well at monitors located next to the trajectory of the plume centroid, were similar to those calculated in the 40-m test for individual layers. Tritium data were also averaged over the cross section perpendicular to the mean direction of flow and the obtained plume-scale concentration time data were used for calculating the coefficient of longitudinal dispersion. There is an order of magnitude difference between plume-scale and local-scale longitudinal dispersivities. The local-scale velocities were used to perform transformation of observed concentration time data into concentration distance data. The latter were used to calculate plume-scale dispersivities using moments analysis. The plume-scale dispersivities calculated for the vertically averaged data were used for predictive purposes. It was found that the vertically averaged data of tracer distribution are poorly described by the two-dimensional solution to the advection-dispersion equation with constant coefficients determined from the moments analysis.