Fuzzy Mapping of Integrated Multiparameter Tomographic Data to Detect and Localize Geophysical Targets

In this work, we present an integrated and quantitative approach to detect and localize geophysical targets associated with both geological and anthropogenic complex scenarios. We complement electrical and seismic tomographic techniques and a machine-learning (ML) unsupervised algorithm [fuzzy C-means (FCM)] to get a final clustered section validated by borehole and well data, where the reliability is evaluated by the membership function. This is a good estimator of the reliability of the proposed procedure, as it ranges from 0 to 1, with one reflecting a high reliability of the clustering analysis. This method is applied to two case studies, related to the detection of leachate in a municipal solid waste landfill and the detection of the bedrock surface in a site prone to instability. For both cases, we also set up synthetic simulations by reproducing realistic models with similar layering and ranges of geophysical parameters as per the field cases. The results show the effectiveness of the method in providing a unique detection of the cluster associated with the desired targets, as highlighted by the matching with direct information. However, the accuracy of the reconstruction is reduced in areas where the resolution of geophysical methods is lower. We demonstrate that the proposed method can achieve a reliable reconstruction if it focuses on searching for one desired target only, rather than a comprehensive reconstruction of the whole layering at the site. This approach can be a valuable automatic tool for improving the benefit-to-cost ratio of projects, where new constructions or remediation interventions have to be planned.