Probabilistic Evaluation of a Seismic Site Class from Electrical Resistivity Test Data

Uncertainty invoked deviations are an emerging aspect in the domain of numerous geotechnical as well as geophysical applications. The implications of existing variabilities and variabilities surfacing due to transformation models have always shown a disparity in the final outcome. In this study, variabilities emerging in the process of asserting seismic site class as per IBC (International Building Code, International Code Council, Country Club Hills, IL, 2009) specifications by transforming electrical resistivity test data of soil have been addressed, which had not been explored yet. This study has embraced all possibilities of uncertainty entanglement in the estimation of time average shear wave velocity of 30-m soil deposit such as randomness in ERT data, mode of assessment, deployment of transformation models and extrapolation techniques. Two different assessment modes have been utilised for procuring shear wave velocity (V ) profiles from ERT data; consequently, V parameter has been ascertained via two extrapolation techniques. The aforementioned process resulted in 686 V estimates comprising all extents of uncertainty invoked deviations. The implications of such deviations in seismic site class assessment have been averted by deploying probabilistic approaches, i.e., first order reliability method and Monte Carlo simulation. These approaches have been implemented on V estimates acquired through different processes such as individual extrapolation technique, transformation models correlating ERT with V and both extrapolation techniques. This study clearly illustrated the variation in probabilistic output as per the aforementioned choice of acquiring V estimates and conveyed this particular site under site class D with a good level of achieved confidence. This novel approach to probabilistically evaluating a seismic site class from ERT data while comprehending the full extent of uncertainty entanglement paves a new way to handle uncertainty invoked deviations.