The rupture process of fluid‐induced microseismic events is still poorly understood, mainly due to usually small magnitudes and sparse monitoring geometries. The high‐quality recordings of the earthquake sequence 2006–2007 at the enhanced geothermal system at Basel, Switzerland, constitute a rare exception, allowing a systematic directivity study of 195 events using the empirical Green's function method. We observe clear directivity signatures for about half the events which demonstrates that rupture directivity persists down to small magnitudes (ML∼1). The predominant rupture behavior is unilateral. We further find evidence that directivity is magnitude dependent and varies systematically with distance from the injection source. Whereas pore pressure seems to play the dominant role close to the injection source and no preferred rupture direction is observable, directivity aligns parallel to the event distribution with increasing distance ( ≳100 m) and is preferably oriented away from the injection point. The largest analyzed events (ML∼2) show a distinct behavior: They rupture toward the injection source, suggesting that they nucleate in the vicinity of the pressure front and propagate backward into the perturbed volume. This finding is of particular relevance for seismic hazard assessment of georeservoirs, since it implies that maximum event size is related to dimension of the fluid‐perturbed volume. Our study also resolves rupture complexities for a small group of events. This shows that small fault heterogeneities exist down to a scale of a few tens of meters. The observation of directivity and complexity in induced microseismic events suggest that future source studies account for these phenomena.

• A comprehensive stress drop distribution for the Iquique Earthquake rupture region is computed using a spectral ratio approach. • The stress drops estimates reveal no large scale pattern or major trend such as a depth dependency. 10 • We describe minor stress drop variations in greater detail and find a relatively strong 11 scaling with moment for the entire data set.

We perform a systematic directivity analysis of local seismic events west off the coast of northern Chile. An empirical Green's function technique is applied to a selection of events from a time period from 2008 to 2016 in the vicinity of the rupture area of the MW8.1 Iquique megathrust earthquake in 2014. We compute rupture directivity for 293 events of magnitudes between ML 2.6 and ML 5.3. We find a strong preference of rupture orientations subparallel to the convergence vector of the Nazca plate relative to the South‐American plate. The preferred rupture direction is downdip. We speculate that the reason for the dominating rupture direction could be a lateral limitation of available rupture directions by the repeating earthquake‐like nature of the observed events combined with a material contrast at the subduction interface which, according to the bimaterial effect, favors the downdip rupture direction.

We have developed a method for tracing the rupture propagation of microseismic events. We referred to it as microseismic rupture propagation imaging (MRPI), which is an adaptation of the back projection technique from global seismology. Hence, we shifted back recorded waveforms to a grid of possible source locations and obtained a coherent phase stack that migrated according to the migration of the rupture front. Using synthetic ruptures and the corresponding waveforms obtained by finite-difference modeling, we tested the viability of the approach for a reservoir model with the properties and geometry of the monitoring system of the Basel-1 geothermal reservoir. First, we found that an estimation of the rupture location, orientation, direction, and length was feasible in this environment. The method was then applied to the four largest events ([Formula: see text]) recorded at the Basel-1 reservoir. We found that the obtained rupture lengths and orientations were reasonably consistent with independent estimates from seismic moments, stress drops, and fault-plane solutions. MRPI allowed us to solve the ambiguity between the actual fault plane and the auxiliary plane. The derived fault planes and rupture directions for the three best-determined events indicated that the failure process was directed preferentially from the periphery toward the injection source. This agreed with the observation that hypocenters of large-magnitude-induced events tend to occur on the edges of the stimulated volume. The results also corroborated the recently proposed idea that induced events were more probable to occur on preexisting faults if the potential rupture surface lay within the stimulated volume.

No abstract