Rough or smooth plate interface? It doesn’t matter when it comes to great earthquakes

Subduction zones host the largest seismogenic zones on earth and hence the largest earthquakes. However, although some subduction margins generate some of the most destructive earthquakes (e.g., Japan, Sumatra), others appear to slip less dramatically in slow slip events, by aseismic creep or in small-moderate earthquakes (e.g., north Hikurangi). Subduction interface topography (‘roughness’) has emerged as a leading parameter in controlling the seismicity at subduction zones, although there is strong debate as to whether rough patches are asperities or barriers to large rupture. This issue persists because observational studies are limited to individual margins, seafloor bathymetry used as proxy for plate interface topography is not direct measurement, and historical earthquake record is short, which together make the precise assessment of earthquake potential in a subduction margin challenging. Here we test whether geodetic interplate coupling is an indicator of earthquake potential in lieu of a longer historical records. We then use direct seismic reflection observations from 35 plate boundary faults to quantify three types of roughness at 1–10 km length scales. We find a strong and positive relationship between maximum magnitude and interplate coupling. Strikingly, no relationship is observed between any of the roughness parameters and maximum earthquake magnitudes/interplate coupling. This result challenges the long-standing paradigm that the plate interface roughness is a pivotal factor in governing seismogenic behaviour. We suggest that short-wavelength (£10 km) roughness has different effects on earthquake nucleation depending on the prevailing stage of earthquake cycle. We conclude that plate roughness alone is not a good proxy to assess a margin’s seismic potential. Instead, interplate coupling provides a better indicator of seismic potential, highlighting the need for enhanced marine geodetic observations.