Rowena Lohman, Cornell University
This project involved the use of satellite-based interferometric synthetic aperture radar (InSAR) observations of ground deformation and how they can help us understand the contribution of earthquakes vs. aseismic professes to the growth of individual folds in southeastern Iran – a region of active hydrocarbon extraction. Our focus area includes two M6 earthquakes with very well-characterized coseismic deformation fields that occurred within three years and 20 km of eachother, suggesting that some amount of triggering of the 2nd earthquake by the first may have occurred. We use Coulomb stress changes predicted by our models of these events to determine whether they are likely to have affected fold growth. Since the area is marked by numerous salt domes, we are also pursuing finite element modeling of the complicated regional elastic structure to determine how a more realistic earth model may affect our conclusions. The significance of this work will include characterization of how individual earthquakes bring regions of the fold closer to or further from failure along existing bedding planes, as well as the volume around the coseismic rupture that would be expected to experience widespread fracturing.
Overall, the most interesting finding to come out of our work is that we suggest that InSAR data can result in incorrect interpretations of an earthquake sequence if seismic data is not also taken into account. For two earthquakes with well-recorded aftershocks, it initially appeared that the ground deformation during the earthquake as constrained by InSAR was in the upper 10 km or sedimentary section, while the earthquakes were all in a distinctly separated region beneath, in the basement. This sort of separation of aftershocks and mainshock is unprecedented. However, under further analysis, we find that the noise in the InSAR data is sufficiently high that it could “hide” an earthquake occurring in the basement rocks, suggesting that the shallow signal observed in the InSAR imagery is aseismic. This suggests that an equal amount of deformation occurs , aseismically, above earthquakes occurring in the basement rocks, which goes a long way towards resolving the apparent discrepancy between GPS-observed convergence rates and seismicity rates in Iran. This sort of aseismic behavior is not unexpected, given the abundant evaporate layers within the sedimentary section.
Our progress on the tasks we set for ourselves for this project is as follows:
Task 1: Predict Coulomb stress changes associated with the two Qeshm Island earthquake using earthquake locations and slip distributions previously studied by the PI, under the assumption of an elastic half space: We have developed a workflow for ingesting both the best-fit fault plane geometries and slip distributions, as well as their errors, so that we can compare many thousands of combinations of candidate models during our evaluation of Coulomb stress triggering. As a result of this effort, we found that the ground deformation data is inconclusive in terms of the exact fault plane that ruptured and whether it brought the second fault plane closer to failure, but that triggering is more consistent with fault geometries constrained by the InSAR observations of ground deformation rather than the more remotely-determined teleseismic fault geometries.
Task 2: Generate model of folding associated with these earthquakes, based on maps of surface geology (which are publicly available) and published cross sections through other folds in the region: We have examined the depth range of seismicity predicted by our analysis of the ground deformation observations, and find that, based on nearby cross sections, it is likely that the observed deformation is limited the region that is actively folding rather than in the basement rocks below. This is in conflict with typical models of the area, which infer that most of the earthquakes occur at greater depths. We find that the location of inferred aseismic slip is consistent with folding models.
Task 3: Calculate shear stress changes along bedding planes in fold model, as well as regions where randomly oriented faults would be stressed to the point where they would be likely to slip. Identify the zones with the highest stress changes (positive or negative) and compare with the general geometrical characteristics of the Zagros folds: We find that the deep earthquakes are likely to increase stress on the areas that we believe then moved aseismically during a short period of time (several weeks) after the event.
Task 4: Set up finite element (using combination of Pylith and CUBIT). Beginning with a model containing no contrasts within the crust, benchmark stress change results with those generated in Task 1: We have performed the initial benchmarking exercise, comparing our results to those in a more simple elastic half space, for which there is an analytical solution.
Task 5: Using the actual distribution of salt domes in the Qeshm Island region, rerun the finite element stress change calculation with columns of salt that are more compliant than the surrounding rock. Explore how this addition changes the predicted stress field and it’s relationship to structures within the fold. We performed this test and found that most of the stress changes were immediately in the vicinity of the salt.
Task 6: Big picture- Revisit the problem of whether currently observed earthquakes could actually build the geological structures found in nature. Given the range of stress changes predicted by our modeling, would we expect deformation that agrees with the shape of folding updip and downdip of the earthquakes? Within how large of a volume would we expect stress changes that have the potential to fracture the rock?: The new observation of significant postseismic slip (separated spatially from the coseismic rupture, which differentiates this scenario from normal postseismic slip) suggests that models of how stress is distributed throughout the seismic cycle in Iran need to be revisited.