Undercover karst imaging using a Fuzzy c-means data clustering approach (Costa Brava, NE Spain)

Abstract

Identification of undercover karst areas can be essential for engineering or hydrogeological studies. Geological reconnaissance of karst areas can be challenging due to their complex and highly heterogeneous internal structure. Herein, we present a case study carried out at the carbonate coastal Montgrí Massif in Costa Brava (Spain), which, in the study area, is characterized by a well-developed karst system covered by Quaternary sediments. The objectives of the work are to delineate and characterize karst structures using geophysical techniques and to produce a subsurface geological model that goes beyond the individual physical properties determined independently by each geophysical method. We acquired electrical resistivity and seismic datasets for five collocated profiles. We analyzed the seismic datasets to retrieve P-wave (Vp) and S-wave (Vs) velocity models. Clay-filled karst formations may produce seismic and electrical signatures that are difficult to interpret due to the wide range of values that can be present depending on the properties of the karst's infill. In this type of environment, the Fuzzy c-means (FCM) data clustering approach is a way of producing an integrated model that can help to constrain the lithological interpretation. In this study, FCM is first applied to an electrical resistivity-Vs dataspace in order to characterize near surface sediments. Four different groups are recognized corresponding to different lithologies, fluid conductivities, and saturations in the unconsolidated sediments. Secondly, FCM analysis of the resistivity-Vp datasets provides integrated models that help to delineate the karst areas. The interpretive constraints used derive from the results of the first FCM application and the lithological description of a single borehole that reaches the karst below the Quaternary sediments. In this way, karst is identified in all five profiles. This study shows that the integration of seismic and electrical resistivity information using the FCM approach can extend borehole information throughout the study area and thus provide comprehensive geological models in karst environments.