Hybrid Discrete Fracture Network Inversion of Hydraulic Tomography Data from a Fractured-Porous Field Site

The accurate characterization of hydraulic conductivity heterogeneities in an aquifer is crucial for predicting flow and transport processes correctly. Hydraulic tomography (HT) experiments are often used to infer the hydraulically relevant features, but the correct inversion of the data remains a challenging task. We implemented a discrete fracture network (DFN) inversion approach that is expanded by considering a nonzero matrix permeability. The hybrid model allows the accurate characterization of fractured-porous sites by taking into account both matrix and fracture flow. This novel inversion algorithm is successfully applied to HT data acquired at a field site in Goettingen (Germany), and the results are compared with those of a standard travel time inversion. Furthermore, we validate the inversion results by using them as the underlying material parameters for simulating heat tracer experiments and comparing the modeled temperature responses with those of heat tracer tests actually conducted at the site. It is shown that the DFN ensemble predicts the thermal response of the experiments correctly for the two major fractures in terms of location, amplitude, and time-dependent behavior of the temperature anomaly, as long as the stochastic nature of the results is taken into account. We conclude that considering both matrix and fracture flow in a hybrid DFN inversion approach can lead to significant improvements in flow and transport modeling at fractured-porous sites. For understanding groundwater processes correctly, precise knowledge about the relevant geological structures in the subsurface is crucial. To infer this information, hydraulic tomography (HT) experiments are often used, which are based on sequential pumping tests in boreholes. The acquired data are usually processed by inversion techniques that allow for computing a subsurface model from the field observations. There are two different conceptual approaches: (a) continuum models, which assume a smooth distribution of the hydraulic parameters, and (b) discrete fracture network (DFN) models, which are based on an impermeable rock matrix, crossed by a limited number of fractures exclusively responsible for groundwater flow. However, certain sites show characteristics of both models, meaning that both the rock matrix and individual fractures are relevant. Therefore, we developed a new inversion technique (hybrid DFN inversion) that combines the two approaches. The method is tested on HT data acquired at a field site in Goettingen (Germany), and the results are validated against data from independent experiments (thermal tracer tests). It is shown that the subsurface models from the new hybrid DFN inversion are more accurate and reliable. We conclude that groundwater modeling at those sites can be improved significantly when using the new inversion method. A novel inversion approach for hydraulic tomography (HT) data considering flow in both discrete fractures and porous matrix is introducedThe method is applied to HT data from a fractured-porous field site in Gottingen (Germany)The results are successfully validated against independent thermal tracer test data