Application of principal component analysis to geo-electrical recordings

Abstract

Nowadays, electrical resistivity tomography (ERT) is able to furnish a global reconstruction of the main features of buried structures at a shallow scale, in a faster and cost-effective manner. The use of a controlled source makes ERT less sensitive to noise with respect to other geophysical methodologies based on natural sources, such as magnetotelluric. During field practice, the depth of exploration is limited by the relative distances between the source and receiver dipoles. The latest generation resistivimeters favor the simultaneous recording of signals in correspondence with multiple and physically separated receivers. These characteristics concur towards a more profitable application of ERT imaging in the case of deeper targets distributed within harsh or densely inhabited large areas. At the same time, the ease of placing a good number of receivers far from the electric sources enhances the need of tools apt for extracting weaker signals from background noise of both natural and anthropic nature. For this purpose, the use of the well-known method, i.e., principal component analysis (PCA) as a filtering tool is tested on the geoelectrical time series acquired in the Mt. Vesuvius area (Naples, Italy). A field and a test dataset have been derived from such recordings and subsequently processed through an original procedure based on PCA. Subsequently, the results have been compared with the ones obtained using more typical estimators, such as stacking or frequency filtering, thus evaluating the usefulness of PCA in the processing of the geoelectrical time series. A good estimate of the said parameters can also be obtained in the case of source-to-receiver distance of more than 5 km.