A vulnerability-based approach for delineating groundwater protection zones around springs in fractured media has been developed to implement Swiss waterprotection regulations. It takes into consideration the diversity of hydrogeological conditions observed in fractured aquifers and provides individual solutions for each type of setting. A decision process allows for selecting one of three methods, depending on the spring vulnerability and the heterogeneity of the aquifer. At the first stage, an evaluation of spring vulnerability is required, which is essentially based on spring hydrographs and groundwater quality monitoring. In case of a low vulnerability of the spring, a simplified method using a fixed radius approach ("distance method") is applied. For vulnerable springs, additional investigations must be completed during a second stage to better characterize the aquifer properties, especially in terms of heterogeneity. This second stage includes a detailed hydrogeological survey and tracer testing. If the aquifer is assessed as slightly heterogeneous, the delineation of protection zones is performed using a calculated radius approach based on tracer test results ("isochrone method"). If the heterogeneity is high, a groundwater vulnerability mapping method is applied ("DISCO method"), based on evaluating discontinuities, protective cover and runoff parameters. Each method is illustrated by a case study.

Several methods have been developed for vulnerability mapping in karstic areas. These methods need additional validation by field experiments. Several tests have been carried out in the Swiss Jura with natural and artificial tracers. The protective role of some intrinsic properties of the system, such as glacial deposits covering karst, epikarst storage and system dilution effect, have been clearly demonstrated. Use of three tracers in parallel showed the reactivity of the epikarst: all tracers arrived at the same time, but their relative concentration stayed clearly different. A classification of contamination scenarios into four classes is proposed. It is shown that the relevance of some intrinsic properties depends on the considered scenario class. The hydrodynamic state of the aquifer influences greatly flow velocities and can strongly modify contaminant concentrations at the output of the system. The spatial repartition (point vs diffuse) and the quantity of contaminant entering the system will also influence the output response. Hence, results from tracing experiments cannot be used straightforward for obtaining a representative value of flow velocity, dispersion or recovery rate.

No abstract