Iris van Zelst , Leonhard Rannabauer, Alice-Agnes Gabriel , Ylona van Dinther

Detailed imaging of accretionary wedges reveals splay fault networks that could pose a significant tsunami hazard. However, the dynamics of multiple splay fault activation during megathrust earthquakes and the consequent effects on tsunami generation are not well understood.
We use a 2-D dynamic rupture model with complex topo-bathymetry and six curved splay fault geometries constrained from realistic tectonic loading modelled by a geodynamic seismic cycle model with consistent initial stress and strength conditions.
We find that all splay faults rupture coseismically. While the largest splay fault slips due to a complex rupture branching process from the megathrust, all other splay faults are activated either top down or bottom up by dynamic stress transfer induced by trapped seismic waves. We ascribe these differences to local non-optimal fault orientations and variable along-dip strength excess.
Generally, rupture on splay faults is facilitated by their favourable stress orientations and low strength excess as a result of high pore-fluid pressures.
The ensuing tsunami modelled with non-linear 1-D shallow water equations consists of one high-amplitude crest related to rupture on the longest splay fault and a second broader wave packet resulting from slip on the other faults. This results in two episodes of flooding and a larger run-up distance than the single long-wavelength (300 km) tsunami sourced by the megathrust-only rupture. Since splay fault activation is determined by both variable stress and strength conditions and dynamic activation, considering both tectonic and earthquake processes is relevant for understanding tsunamigenesis.