STOCHASTIC-ANALYSIS OF SOLUTE TRANSPORT IN PARTIALLY SATURATED HETEROGENEOUS SOIL .2. PREDICTION OF SOLUTE SPREADING AND BREAKTHROUGH

The applicability of results of Lagrangian-stochastic analyses of vado se-zone transport [Russo, 1993a, b] to realistic situations is investi gated using results of detailed numerical simulations of transport in a hypothetical, yet realistic heterogeneous, partially saturated soil, obtained in the first companion paper (Russo et al., this issue) for both quasi steady state and transient, nonmonotonic flows. For both fl ow regimes, lower mean water saturation and longer travel time are sho wn to increase solute spreading, while lower water saturation and smal ler travel distance are shown to increase the skewing of mean solute b reakthrough curves. Solute plumes associated with the latter flow regi mes, however, exhibit less spreading in the longitudinal direction and more spreading in the transverse direction, while the respective brea kthrough curves are less skewed and less erratic, as compared with sol ute spreading and breakthrough associated with the former flow regimes . Components of the time-dependent displacement covariance tensor, X a nd expected solute flux through a given horizontal control plane [s], based on the Lagrangian-stochastic analyses, compared favorably with e stimates of X and [s] obtained from the simulated transport under quas i steady state, essentially unidirectional flows, but failed to predic t estimates of X and [s] obtained from the simulated transport under t ransient, nonmonotonic, multidirectional flows. The latter can be pred icted by a particle-tracking method [Rubin, 1990] that allows for devi ation of the solute particles from their mean path, provided that the pertinent flow regime is quantifiable in terms of an appropriate veloc ity covariance function.