Impact of two geostatistical hydro-facies simulation strategies on head statistics under non-uniform groundwater flow

We present a numerical study keyed to the analysis of the impact on hydraulic head statistics of two selected methodologies for the stochastic simulation of hydro-facies spatial arrangement. We analyze the distribution of hydraulic heads in a confined aquifer under steady-state convergent three-dimensional flow to a fully penetrating well, superimposed to a mean uniform regional gradient. The heterogeneous structure of the system is modeled on the basis of available field information comprising detailed lithological data collected within an aquifer system located in northern Italy. These data are grouped into five litho-type categories and the aquifer system is modeled as a random composite medium. Monte Carlo realizations of the three-dimensional geo-material distributions are generated through the Sequential Indicator and the Truncated Plurigaussian Simulation methods. The latter enables one to integrate geological conceptual information in the simulation procedure, while the former relies mainly on a variogram-based analysis. Point and vertically averaged hydraulic heads, corresponding to typical observations collected within screened boreholes, are analyzed by evaluating the dependence of their sample probability distributions on (i) the hydro-facies generation scheme, (ii) the extent of the vertical averaging interval and (iii) the relative distance between the location of observation boreholes, hydrological boundaries and the source term. Theoretical probability density function models are fitted against numerically simulated distributions within a maximum likelihood context. Our results indicate that hydraulic heads associated with the Truncated Plurigaussian Simulation method exhibit increased variability when compared to their counterparts evaluated upon relying on a Sequential Indicator based modeling strategy of the system heterogeneity. Covariance matrices and probability distributions of point and vertically averaged hydraulic heads display similar key representative features and patterns. This suggests that typical measurements collected in screened boreholes can be used to infer qualitative information about the correlation structure and the statistical properties of heads.