Zoe K Mildon , Gerald Roberts, Joanna Faure Walker, Francesco Iezzi

In order to investigate the importance of including strike-variable geometry and the knowledge of historical and palaeoseismic earthquakes when modelling static Coulomb stress transfer and rupture propagation, we have examined the August–October 2016 A.D. and January 2017 A.D. central Apennines seismic sequence (Mw 6.0, 5.9, 6.5 in 2016 A.D. (INGV) and Mw 5.1, 5.5, 5.4, 5.0 in 2017 A.D. (INGV)).We model both the coseismic loading (from historical and palaeoseismic earthquakes) and interseismic loading (derived from Holocene fault slip-rates) using strike-variable fault geometries constrained by fieldwork. The inclusion of the elapsed times from available historical and palaeoseismological earthquakes and on faults enables us to calculate the stress on the faults prior to the beginning of the seismic sequence. We take account the 1316–4155 yr elapsed time on the Mt. Vettore fault (that ruptured during the 2016 A.D. seismic sequence) implied by palaeoseismology, and the 377 and 313 yr elapsed times on the neighbouring Laga and Norcia faults respectively, indicated by the historical record. The stress changes through time are summed to show the state of stress on the Mt. Vettore, Laga and surrounding faults prior to and during the 2016–2017 A.D. sequence.We show that the build up of stress prior to 2016 A.D. on strike-variable fault geometries generated stress heterogeneities that correlate with the limits of the main-shock ruptures. Hence, we suggest that stress barriers appear to have control on the propagation and therefore the magnitudes of the