Veronica Pazzi

Performing a reliable stability analysis of a landslide slope requires a good understanding of the internal geometries and an accurate characterisation of the geotechnical parameters of the identified strata. Geotechnical models are commonly based on geomorphological data combined with direct and intrusive geotechnical investigations. However, the existence of numerous empirical correlations between seismic parameters (e.g., S-wave velocity) and geotechnical parameters in the literature has made it possible to investigate areas that are difficult to reach with direct instrumentation. These correlations are often overlooked even though they enable a reduction in investigation costs and time. By means of geophysical tests, it is in fact possible to estimate the N-SPT value and derive the friction angle from results obtained from environmental seismic noise measurements. Despite the empirical character and a certain level of uncertainty derived from the estimation of geotechnical parameters, these are particularly useful in the preliminary stages of an emergency, when straight data are not available and on all those soils where other direct in situ tests are not reliable. These correlations were successfully applied to the Theilly landslide (Western Alps, Italy), where the geotechnical model was obtained by integrating the results of a multi-parameter geophysical survey (H/V seismic noise and ground-penetrating radar) with stratigraphic and geomorphological observations, digital terrain model and field survey data. The analysis of the triggering conditions of the landslide was conducted by means of hydrological–geotechnical modelling, evaluating the behaviour of the slope under different rainfall scenarios and considering (or not) the stabilisation interventions present on the slope. The results of the filtration analyses for all events showed a top-down saturation mechanism, which led to the formation of a saturated face with a maximum thickness of 5 m. Stability analyses conducted for the same events showed the development of a shallow landslide in the first few metres of saturated soil. The modelling results are compatible with the actual evolution of the phenomenon and allow us to understand the triggering mechanism, providing models to support future interventions.

Real-time seismic monitoring is of primary importance for rapid and targeted emergency operations after potentially destructive earthquakes. A key aspect in determining the impact of an earthquake is the reconstruction of the ground-shaking field, usually expressed as the ground-motion parameter. Traditional algorithms compute the ground-shaking field from the punctual data at the stations relying on ground-motion prediction equations computed on estimates of the earthquake location and magnitude when the instrumental data are missing. The results of such algorithms are then subordinate to the evaluation of location and magnitude, which can take several minutes. To fill the temporal gap between the arrival of the data and the estimate of these parameters, a new data-driven algorithm that exploits the information from the station data only is introduced. This algorithm, consisting of an ensemble of convolutional neural networks (CNNs) trained on a database of ground-shaking maps produced with traditional algorithms, can provide estimates of the ground-shaking maps and their associated uncertainties in real time. Because CNNs cannot handle sparse data, a Voronoi tessellation of a selected peak ground parameter recorded at the stations is computed and used as the input to the CNNs; site effects and network geometry are accounted for using a (normalized) VS30 map and a station location map, respectively. The developed method is robust to noise, can handle network geometry changes over time without the need for retraining, and can resolve multiple simultaneous events. Although having a lower resolution, the results obtained are statistically compatible with the ones from traditional methods. A fully operational version of the algorithm is running on the servers at the Department of Mathematics and Geosciences of the University of Trieste, showing real-time capabilities in handling stations from multiple Italian strong-motion networks and outputting results with a resolution of 0.05° × 0.05°.