Imaging and forecasting rupture dynamics of induced microearthquakes

Understanding the faulting dynamics of natural earthquakes is fundamentally limited by the scarcity of near-source observations and the incompleteness of knowledge about in situ conditions at depth. The Bedretto Underground Laboratory for Geosciences and Geoenergies (BedrettoLab) in Switzerland addresses these limitations through controlled hydraulic stimulation experiments that generate seismicity beneath more than 1 km of rock overburden, thereby bridging the scale gap between laboratory studies and observed natural earthquakes. A key advantage of the BedrettoLab is the ability to characterize in situ conditions prior to seismicity induction. This includes imaging the geometry of the target fault, estimating the local stress state and pore fluid pressure, and examining host and fault rock properties.

Seismicity induced by controlled hydraulic stimulation is recorded by a comprehensive suite of near-source instrumentation, including strong-motion seismometers, borehole geophones and accelerometers, high-frequency acoustic emission sensors, and fibre-optic cables enabling Distributed Acoustic Sensing (DAS) and Fibre Bragg Grating (FBG) measurements. Past experiments have successfully generated seismicity sequences with mainshock magnitudes between Mw −0.5 and 0.0. We construct dynamic rupture models for one such mainshock, constrained by the available near-source observations, to image slip distribution, rupture directivity, and rupture velocity at meter-scale resolution. We find that rupture directivity has a substantially stronger impact on spectral amplitudes than average stress drop. The inferred stress and friction drops are interpreted in terms of the maximum possible confining pressure, providing insights into dynamic weakening processes during earthquake rupture.

The next experiment aims to induce Mw 1.0 earthquakes along a selected fault zone. Using constraints from hydraulic fracture tests, fault geometry imaging, and injection protocols, we seek to forecast the potential rupture dynamics of the induced mainshock by generating a suite of dynamic rupture models representing plausible rupture scenarios, against which the observed mainshock dynamics can be evaluated. In particular, we assess how reliably pre-experiment slip tendency analyses translate into the actual rupture behavior under these controlled conditions. Ultimately, this research will advance our understanding of earthquake source physics and contribute to improved forecasting and mitigation of worst-case scenarios associated with hydraulic stimulation.