Abstrakt
Stress drop characterizing dynamics of earthquake rupture propagation has various definitions and measures with a yet unclear mutual relationship. We aim to reconcile this discrepancy by analyzing synthetic dynamic rupture models generated by random sampling of the model space of spatially heterogeneous dynamic rupture parameters.
An ensemble of similar to 1,600 dynamic models, with waveforms statistically fitting empirical Ground Motion Prediction Equations, is treated as a synthetic event database. The events exhibit various magnitudes, degrees of complexity, and realistic omega(-2) spectral decay of their average moment rates.
While the variability of the stress drop Delta tau(e) estimated from the moment rates agrees with real data studies (sigma(ln(Delta tau e)) (similar to)1.1), the true static stress drops Delta tau(s) evaluated directly from the dynamic models exhibit larger values and lower variability (sigma(ln(Delta tau s)) (similar to)0.4). Our study suggests that the disagreement between various stress drop definitions is related to the source model approximation and not to any particular data processing.
Plain Language Summary During an earthquake, the stress accumulated on a fault suddenly releases. Earthquake stress drop is thus a significant parameter characterizing the dynamics of rupture propagation.
However, its definition and its inference from observed data is ambiguous and inferred stress drops vary among researchers both in terms of values and variability. Here we have reconciled this discrepancy by inspecting synthetic physics-based dynamic rupture models, which provide parameters that are difficult to assess in reality, in particular, the stress change before and after the earthquake.
We randomly generate synthetic events that are compatible with observed strong ground motions from a large number of real earthquakes. The events exhibit various magnitudes and degrees of complexity.
We explore various stress drop measures estimated following seismological practice and true stress drop evaluated directly from the dynamic models. While the estimated stress drops agree with empirical findings, the true stress drop exhibits larger values and lower variability.
Our study suggests that the discrepancy between the various stress drops is related to the too simplistic assumptions on the source model used in empirical studies. The individual rupture process of each event is more complex and, therefore, not entirely captured by a single stress drop parameter.