Critical Views about the Interpretation of Karst Spring Response: Findings from the Milandre Test-Site (Switzerland)

The Milandre test-site is a karst system that includes a 10.5-km-long cave with an underground stream. The good access to this cave and the decision to build a freeway across its catchment area motivated a very large amount of data collection and studies. This exceptional observation network makes it possible to apply various interpretation methods and to verify in the field if inferred information depicts the true processes ongoing. This approach first was used for the analysis of hydrographs. Classical recession-curve decomposition methods were applied and results compared to reality. Interpretations derived from classical methods were often not meaningful! Using numerical models a series of inverse models have been attempted, leading to a physically based method for deriving the conduit density and hydraulic conductivity from the recession curve analysis. Concerning transport, a conceptual model has been developed for the interpretation of the behavior of several natural tracers such as TDS (or conductivity), water chemistry (major ions), stable isotopes (18O and 13C), temperature, and turbidity. Responses to discharge variations are quite different for the respective tracers, leading to different models that often are incompatible. Solving these inconsistencies requires entering the cave and making direct observations of flow in the sub-systems (e.g., epikarst, vadose zone, and conduits). This on-site approach resulted in a large number of new interpretations, leading to a general conceptual model that includes most of the variations observed. This approach provides evidence that most classical interpretation models tend to homogenise the flow behaviour that occurs between the recharge and the outlet. This is not the case in reality, and even simple processes such as the mixing of two tributaries may produce a substantial deviation from a simplistic model.