Abstrakt
The hypocentres of natural earthquake swarms and injection-induced seismicity usually show systematic migration, which is considered to be a manifestation of their triggering mechanism. In many of these cases, the overall growth of the earthquake distribution is accompanied by short episodes of rapid migration, the origin of which is still not sufficiently clarified.
We review the possible triggering mechanisms of these migrating episodes and propose a graphical method for distinguishing internal and external triggering forces. We also analyse the theoretical relationship between the evolution of the cumulative seismic moment and the rupture area and propose two models, the crack model and the rupture front model, which can explain the spreading of hypocentres.
We developed an automatic algorithm for detecting fast migration episodes in seismicity data and applied it to relocated catalogues of natural earthquake swarms in California, West Bohemia, and Iceland, and to injection-induced seismicity. Fast migration episodes is shown to be relatively frequent during earthquake swarms (8-20 per cent of all events) compared to fluid-induced seismicity (less than 5 per cent of the events).
Although the migration episodes were detected independently of time, they grew monotonically with time and square-root dependence of radius on time was found suitable for majority of sequences. The migration velocity of the episodes of the order of 1 m s(-1) was found and it anticorrelated with their duration, which results in a similar final size of the clusters scattering around 1-2 km.
Comparison of seismic moment growth and activated fault area with the predictions of the proposed models shows that both the rupture front model and the crack model are able to explain the observed migration and that the front model is more consistent with the data. Relatively low estimated stress drops in the range of 100 Pa to 1 MPa suggest that aseismic processes are also responsible for cluster growth.
Our results show that the fast migrating episodes can be driven by stress transfer between adjacent events with the support of aseismic slip or fluid flow due to dynamic pore creation.