Testing the b-value variability in Italy and its influence on Italian PSHA

The supposed b-value spatial variability is the central topic of many scientific works dealing with forecast modeling applications or geological correlations. If used for seismicity rate determination, the b-value plays an important role in probabilistic seismic hazard assessment, but how much does it influence the results? In the logic tree approach used for the new probabilistic seismic hazard map of Italy, named MPS04, one of the sources of epistemic uncertainty considered was the procedure for computing seismicity rates. Two alternatives were adopted: 1) to compute the activity rates for each binned magnitude class and 2) to compute a Gutenberg-Richter distribution. In the logic tree branches, where the Gutenberg-Richter distribution was adopted, the corresponding b-value was evaluated for each seismogenic zone: it spans between 0.63 and 2.01. After analysing the b-value variability in the Italian region, this work evaluates the impact of setting the b-value equal to I on the results of seismic hazard assessment in terms of PGA and energy release compared to the choices adopted for MPS04.