GNGTS 2023 - Atti del 41° Convegno Nazionale

Session 2.2 GNGTS 2023 Predicting shear wave velocity from standard stratigraphic logs: the contribution of the Italian Seismic Microzonation database and the machine learning approach F. Mori, A. Mendicelli, C. Varone, G. Naso, M. Moscatelli Abstract From the intersection of DH-MASW surveys and boreholes in the Italian Seismic Microzonation database, an association dataset was derived between the 32 geological-technical (GT) groups at various depths up to 30 meters (with the step of 1 meter) of the Italian standards for seismic microzonation and the natural logarithm of Vs, for a total of about 58 kilometers of Vs-lithology pairs. A preliminary classification analysis with Machine Learning shows 95% accuracy between GT classification and features used (i.e., latitude, longitude, elevation, natural logarithm Vs, depth). The machine learning regression algorithm "Ensemble Bagged Tree" allows us to predict the value of the natural logarithm of Vs from 5 features (latitude, longitude, elevation, depth, GT class) with an R2 of 0.8 and an average RMSE of 0.23. The goal of our work is to translate the knowledge of stratigraphic logs into Vs profiles to improve spatializations of synthetic Vs parameters (e.g., Vs30) at the national scale and consequently improve the estimation of GMM models.