Abstract
Analyses of abundant small earthquakes have the potential to map the Earth's small-scale stress and frictional properties. However, standard seismological characterizations lack the resolution to capture the physical complexity of earthquake rupture, particularly its spatial and temporal heterogeneity. Advanced dynamic rupture models, which integrate elastodynamic simulations with frictional laws, have so far been applied to large earthquakes. Here, we develop a Bayesian dynamic inversion from station-specific (apparent) source spectra, extending up to 25 Hz, for slip-weakening friction, assuming heterogeneous parameterization along a finite-extent planar fault with resolution down to ~100-m scale. The approach is demonstrated on two Mw~4 earthquakes in Central Italy with distinct spectral behavior, one directive and one nondirective. Results show that the inversion resolves key mean source parameters, pinpoints possible pitfalls in standard estimates, and infers a power-law decay of stress and friction heterogeneity spectra down to the smallest scales. Such advanced studies promise to unravel so-far elusive small-scale characteristics of earthquake ruptures.