We introduce a comprehensive methodology that combines multi-objective optimization, global sensitivity analysis (GSA) and 3D groundwater modeling to analyze subsurface flow dynamics across large-scale domains. In this way, we effectively consider the inherent uncertainty associated with subsurface system characterizations and their interactions with surface waterbodies. We demonstrate the effectiveness of our proposed approach by applying it to the largest groundwater system in Italy.

Our study shows that (i) monitoring wells installed with packers provide the (overall) best conductivity estimates; (ii) conductivity estimates anchored on information from partially and fully screened wells are of similar quality; (iii) inflation of the measurement-error covariance matrix can improve conductivity estimates when a simplified flow model is adopted; and (iv) when compared to the MC-based EnKF, the MEs-based EnKF can efficiently and accurately estimate conductivity and head fields.

Permeability of natural systems exhibits heterogeneous spatial variations linked with the size of the measurement support scale. As the latter becomes coarser, the system appearance is less heterogeneous. As such, sets of permeability data associated with differing support scales provide diverse amounts of information. In this contribution, we leverage information theory to quantify the information content of gas permeability datasets collected with four diverse measurement support scales.