Numerical versus statistical modelling of natural response of a karst hydrogeological system

Abstract Structural and hydrodynamic characteristics of karst aquifers are mostly deduced from studies of global responses of karst springs (hydrographs, chemical or isotopic composition). In this case, global response is often used to make inferences with respect to infiltration and ground water flow processes as well as on the hydrodynamic parameters. Obviously, the direct verification of these inferences is very difficult. We have used an indirect method of verification, introducing well defined theoretical karst structures into a finite element model and then analysing the simulated global response according to the currently accepted interpretation schemes. As we know what we are introducing into the numeric model, the consistency of the interpretation may be checked immediately. The results obtained in the hydrogeological study of two karst basins in the Swiss Jura and from 2-D and 3-D numerical simulations show the difficulty of finding structural parameters and hydrodynamic behaviour from statistical methods alone, i.e. correlation analyses discharge–discharge and precipitation–discharge. In effect, our first results show that the form of the correlograms depends on several factors besides the structure of the karst aquifer: (i) on the form of the floods, in other words the contrast between quick flow and base flow, (ii) on the frequency of hydrological events during the period analysed and (iii) on the type of infiltration processes, in other words the ratio of diffuse infiltration to concentrated infiltration. Obviously, the variability of a karst hydrograph is a result of a combination of these factors. Distinction between them is not always possible on hydrographs, and therefore on correlations (discharge–discharge and precipitation–discharge).