Potential shallow aquifers characterization through an integrated geophysical method: multivariate approach by means of k-means algorithms

Main Article Content

Stefano Bernardinetti
Stefano Maraio
Pier Paolo Gennaro Bruno
Valentina Cicala
Serena Minucci
Miriana Giannuzzi
Marilena Trotta
Francesco Curedda
Simone Febo
Matteo Vacca
Enrico Guastaldi *
Tommaso Colonna
Filippo Bonciani
Emanuele Tufarolo
Fabio Brogna
Andrea Zirulia
Omar Milighetti
(*) Corresponding Author:
Enrico Guastaldi | guastaldi@geoexplorer.cgtgroup.org


The need to obtain a detailed hydrogeological characterization of the subsurface and its interpretation for the groundwater resources management, often requires to apply several and complementary geophysical methods. The goal of the approach in this paper is to provide a unique model of the aquifer by synthesizing and optimizing the information provided by several geophysical methods. This approach greatly reduces the degree of uncertainty and subjectivity of the interpretation by exploiting the different physical and mechanic characteristics of the aquifer. The studied area, into the municipality of Laterina (Arezzo, Italy), is a shallow basin filled by lacustrine and alluvial deposits (Pleistocene and Olocene epochs, Quaternary period), with alternated silt, sand with variable content of gravel and clay where the bottom is represented by arenaceous-pelitic rocks (Mt. Cervarola Unit, Tuscan Domain, Miocene epoch). This shallow basin constitutes the unconfined superficial aquifer to be exploited in the nearly future. To improve the geological model obtained from a detailed geological survey we performed electrical resistivity and P wave refraction tomographies along the same line in order to obtain different, independent and integrable data sets. For the seismic data also the reflected events have been processed, a remarkable contribution to draw the geologic setting. Through the k-means algorithm, we perform a cluster analysis for the bivariate data set to individuate relationships between the two sets of variables. This algorithm allows to individuate clusters with the aim of minimizing the dissimilarity within each cluster and maximizing it among different clusters of the bivariate data set. The optimal number of clusters “K”, corresponding to the individuated geophysical facies, depends to the multivariate data set distribution and in this work is estimated with the Silhouettes. The result is an integrated tomography that shows a finite number of homogeneous geophysical facies, which therefore permits to distinguish and interpret the porous aquifer in a quantitative and objective way.

Downloads month by month


Download data is not yet available.

Article Details