SUMMARYHow does megathrust earthquake rupture govern tsunami behaviour? Recent modelling advances permit evaluation of the influence of 3-D earthquake dynamics on tsunami genesis, propagation, and coastal inundation. Here, we present and explore a virtual laboratory in which the tsunami source arises from 3-D coseismic seafloor displacements generated by a dynamic earthquake rupture model. This is achieved by linking open-source earthquake and tsunami computational models that follow discontinuous Galerkin schemes and are facilitated by highly optimized parallel algorithms and software. We present three scenarios demonstrating the flexibility and capabilities of linked modelling. In the first two scenarios, we use a dynamic earthquake source including time-dependent spontaneous failure along a 3-D planar fault surrounded by homogeneous rock and depth-dependent, near-lithostatic stresses. We investigate how slip to the trench influences tsunami behaviour by simulating one blind and one surface-breaching rupture. The blind rupture scenario exhibits distinct earthquake characteristics (lower slip, shorter rupture duration, lower stress drop, lower rupture speed), but the tsunami is similar to that from the surface-breaching rupture in run-up and length of impacted coastline. The higher tsunami-generating efficiency of the blind rupture may explain how there are differences in earthquake characteristics between the scenarios, but similarities in tsunami inundation patterns. However, the lower seafloor displacements in the blind rupture result in a smaller displaced volume of water leading to a narrower inundation corridor inland from the coast and a 15 per cent smaller inundation area overall. In the third scenario, the 3-D earthquake model is initialized using a seismo-thermo-mechanical geodynamic model simulating both subduction dynamics and seismic cycles. This ensures that the curved fault geometry, heterogeneous stresses and strength and material structure are consistent with each other and with millions of years of modelled deformation in the subduction channel. These conditions lead to a realistic rupture in terms of velocity and stress drop that is blind, but efficiently generates a tsunami. In all scenarios, comparison with the tsunamis sourced by the time-dependent seafloor displacements, using only the time-independent displacements alters tsunami temporal behaviour, resulting in later tsunami arrival at the coast, but faster coastal inundation. In the scenarios with the surface-breaching and subduction-initialized earthquakes, using the time-independent displacements also overpredicts run-up. In the future, the here presented scenarios may be useful for comparison of alternative dynamic earthquake-tsunami modelling approaches or linking choices, and can be readily developed into more complex applications to study how earthquake source dynamics influence tsunami genesis, propagation and inundation.