Some reasoning on the improvement of the ETAS modeling at the occurrence of the 2016 Central Italy seismic sequence

Abstract

This study presents an application of the ETAS model to the first 20 days of the 2016 Central Italy sequence. Despite of the provisional nature of data, the model is able to describe the occurrence rate, but for the first hours after the mainshock occurrence. A sensitivity analysis of the model to two uncertainty sources, the model parameters and the occurrence history, shows that the second has a main role in controlling the performance of the ETAS model. Previous results, together with the clear inability of ETAS to forecast the occurrence of a sequence before its starting time, give important suggestions about possible improvements.