Is it possible to retrieve karst network geometry and properties using an inverse technique? Framing the inverse problem for karst simulation using a pseudo-genetic algorithm. Inverse problem and karst simulation.

@inproceedings{BorghiGM2014,
 abstract = { Karst aquifers are characterized by a very high heterogeneity, which makes them difficult to handle. The principal source of heterogeneity is the presence of high conductive features like karst conduit embedded in a very poorly conductive limestone matrix. The major issue, is the great uncertainty related to the location and size of these conduits. In many hydrogeological contexts, the application of uncertainty analysis and inverse problem techniques shows their benefits, but so far very little has been done in karst hydrogeology.
The present paper proposes an inverse approach to simulate karst aquifers, and investigate its applicability. The proposed methodology is based on the combination of: 1) many realizations of karst conduit networks are generated using a stochastic approach; 2) for each karst realization, flow and transport are simulated and the hydraulic properties are optimized with respect to observed data (field measurement); 3) finally, for the uncertainty analysis, rejection sampling s used to select only the best fit models resulting from the combination of step 1 and 2.
To investigate the applicability of this inverse approach, two numerical experiments have been made. In the first test, the flow and transport results of 18’000 simulations is used to test the robustness of rejection sampling to retrieve the parameter (both geometrical and physical) of a reference simulation. In the second test, a complex flow model (which includes non linear flow equations in conduits) is used to test the ability of gradient-based inverse techniques to estimate the parameters that influence the radius of the conduits and the hydraulic conductivity of the limestone matrix. },
 author = { Borghi, Andrea AND Renard, Philippe AND Cornaton, Fabien },
 booktitle = { Proc. 34th Gocad Meeting, Nancy },
 title = { Is it possible to retrieve karst network geometry and properties using an inverse technique? Framing the inverse problem for karst simulation using a pseudo-genetic algorithm. Inverse problem and karst simulation. },
 year = { 2014 }
}