Including Multiple Types of Information Measured at Non-Colocated Locations for Simulation of One Spatially Distributed Variable - Hydraulic Conductivity at the MADE Site

Real world data-sets typically exhibit non-Gaussian spatial dependence structures. Simulation algorithms that take the structures encountered in the real world into account is critical for the simulation of groundwater water flow and solute transport.

The conception of the presented methodology is to combine different kinds and levels of the available information for the estimation of a real scale non-Gaussian stochastic model of the primary parameter (K) at the highly heterogeneous Macrodispersion Experiment (MADE) site:

• The degree of the non-symmetric dependence (\non-Gausianness") is quantified empirically via the measure of asymmetry, that is one part of the multi-objective simulation algorithm.

• Two independent data sets of observations of the primary parameter (K) are available without an overlap on locations (non-colocated): 31123 direct push injection logging based measurements and 2611 flowmeter based measurements.

• Different qualities of the measurements are considered: The order of one variable and/or the measurement value can be used in the conditioning. Additionally, ȩnsored measurements" (e.g., measurements below detection limit) are included in the simulation. This is important in areas where K is so large that the pressure during DPIL dissipates too quickly into the formation such that the method's sensitivity is decreased.

• The whole domain is simulated in three dimensions. Due to the necessary fine resolution in vertical direction, the full three-dimensional simulation is computationally demanding. A more efficient algorithm is implemented where the vertical dimension is included via an AR(MA) approach.

All of the above information is integrated together and simulated by using the phase annealing method. This improved dependence structure of K is expected to improve the modelling of the solute concentrations at the site.