Joint inversion of tracer tests using reversed flow fields

Two successive dipole tracer experiments were performed in a fractured rock at the Grimsel Test Site in February/March 1993. In both experiments NaCl was used as a tracer and the extraction rate was twice the injection rate. In the second experiment injection and extraction were interchanged (‘reverse experiment’). The withdrawal rate remained twice the injection rate. Long tailing was characteristic for the breakthrough curves in both experiments. The tests were interpreted using a single fracture flow model. Tracer transport is described by advection/dispersion along the fracture allowing for diffusion into an immobile matrix. We were able to interpret the breakthrough curves for both experiments by one unique set of parameters, describing transport and baseflow. Uniqueness could only be achieved when using the information from both experiments. We conclude that performing a reverse experiment is an indispensable tool for parameter identification in dipole tracer tests. A sensitivity analysis proved that not only the exchange of tracer between the mobile zone and the immobile matrix but also transversal dispersivity is responsible for the tailing in the breakthrough curves. Further, the typical exchange time between mobile and immobile media was too small to be attributed to mere molecular diffusion into the matrix. The interpretation of the estimated parameters allows us to conclude that we did not observe matrix diffusion but a fast exchange of tracer between the mobile zone and both the adjacent rock and less pervious zones within the fracture. Analysis of the joint covariance matrices showed that the estimated transport parameters have small errors.