What Do Early Source Time Functions Reveal About Rupture Dynamics in Laboratory Experiments?

The predictability of the final size of an ongoing rupture remains a fundamental open question in earthquake physics. If small and large earthquakes differ during their earliest stages, then information contained in the initial part of the source time function (STF) could provide a clue about the final rupture size, representing a major potential improvement for early warning systems. Testing this hypothesis using natural earthquakes is challenging because STFs are indirectly inferred, rupture dimensions are uncertain, and observational catalogs are incomplete.

To overcome these limitations, we built a new experimental catalog of laboratory earthquakes with a wide range of contained rupture lengths. Laboratory earthquakes were conducted using a biaxial apparatus holding PMMA plates, allowing controlled conditions and direct observations of rupture processes. We compute STFs using three independent approaches: (1) true STFs obtained via digital image correlation using a high-speed camera, (2) inferred STFs from quasi-static spatio-temporal slip inversions using 20 near-field accelerometers, and (3) approximate STFs estimated using a far-field acoustic sensor.

Our results show no robust relationship between final rupture length and the initial slope of the STF. Instead, we observe a relationship between the initial slope of the STF and the rupture velocity: events with higher initial rupture velocities exhibit steeper initial slope of the STF. Our results are supported by analytical results. These observations suggest that the initial growth of the STF mainly reflects rupture dynamics and nucleation processes rather than the ultimate size of the event. In this laboratory system, any information about final rupture length appears to emerge in the STF only once the rupture has grown to a length that is significant compared to its final size.