Integrating faults and past earthquakes into a probabilistic seismic hazard model for peninsular Italy

Italy is one of the most seismically active countries in Europe. Moderate to strong earthquakes, with magnitudes of up to ~ 7, have been recorded on many of active faults in historical times. Currently, probabilistic seismic hazard assessments in Italy are mainly based on area source models, in which the seismicity is modelled on a number of seismotectonic zones and the occurrence of earthquakes is assumed to be uniform. However, in the last decade, efforts have increasingly been directed towards using fault sources in seismic hazard models to obtain more detailed and possibly more realistic patterns of ground motion. In our model, we used two categories of earthquake sources. The first involves active faults, and fault slip rates were used to quantify the seismic activity rate. We produced an inventory of all fault sources, with details on their geometric, kinematic and energetic properties. The parameters are used to compute the total seismic moment rate for each fault. We evaluated the magnitude-frequency distributions of each fault source using two models, a characteristic Gaussian model centred on the maximum magnitude and a Truncated Gutenberg-Richter model. The second earthquake source category involves distributed seismicity, and a fixed-radius smoothed approach and a historical catalogue were used to evaluate seismic activity. Under the assumption that deformation is concentrated along faults, we combined the earthquakes derived from the geometry and slip rates of active faults with the earthquakes from the spatially smoothed earthquake sources and assumed that the smoothed seismic activity in the vicinity of an active fault gradually decreases by a fault-size driven factor. We computed horizontal peak ground acceleration maps for return periods of 475 and 2,475 yr. Although the range and gross spatial distribution of the expected accelerations obtained here are comparable to those obtained through methods involving seismic catalogues and classical zonation models, the spatial pattern of our model is far more detailed. Our model is characterized by areas that are more hazardous and that correspond to mapped active faults, while the previous models yield expected accelerations that are almost uniformly distributed across large regions. In addition, we conducted sensitivity tests to determine the impact on the hazard results of the earthquake rates derived from two magnitude-frequency distribution models for faults and to determine the relative contributions of faults versus distributed seismic activity. We think our model represents an advance for Italy in terms of input data (quantity and quality) and methodology in the field of the fault-based regional seismic hazard modelling.