Seismic hazard due to small shallow induced earthquakes

We present the basic methodology for a Probabilistic Seismic Hazard Analysis (PSHA) applied to small and shallow earthquakes, i.e. 1.5 < ML < 3.9 and depth < 4 km, in The Netherlands. Such small shallow earthquakes occur as induced events due to gas exploitation, mainly in the north of The Netherlands. Few studies have applied a PSHA for small and shallow events and we indicate some specific complications. Most important is that relatively high Peak Ground Accelerations (PGA) are predicted and observed. PGA above 0.2g is not unusual, but damage is mostly restricted to cracks in masonry. The Peak Ground Velocity (PGV) is a more appropriate hazard parameter. Observed response spectra show a fairly stable peak around 10 Hz and are, consequently, used in the specified velocity hazard parameter. We predict that, for response spectra with 50% damping, peak values up to 20 and 30 mm/sec may be exceeded with an annual probability of 0.1 and 0.01 respectively above the Groningen field. In some small areas (about 3-4 km2 ) above the Roswinkel and Bergermeer field values around 35 and 60 mm/sec may be exceeded with an annual probability of 0.1 and 0.01 respectively. For response spectra with 5% damping, the corresponding peak values may reach 50 and 80 mm/sec, and 85 and 140 mm/sec. These values, which do not include possible site-specific effects, can be related to the existing vibration guidelines in use in The Netherlands.

For the three regions, where regular induced seismicity is observed, we have computed tentative hazard maps for return periods of 10 and 100 years, i.e. around Roswinkel, Bergermeer and Groningen. These hazard maps do not include the influence of specific site responses. Also a more general model has been computed under the assumption that seismicity is an inevitable consequence of gas exploitation, but that additional models, explaining the physical relation between gas extraction and seismicity, are lacking. In all our hazard models we find significant uncertainties. Some of these uncertainties may be constrained with additional information, like for example: refining local attenuation relations, relating seismicity to specific faults, including exploitation-seismicity relations, a priory focal mechanism information, etc. Other uncertainties, mostly related to the attenuation will remain. The inherent nonstationary character of induced seismicity can yet not been modeled.

Continuous monitoring of seismic activity shows that the seismic hazard due to induced seismicity remained stable during the last five years. The introduction of a systematic hazard estimation methodology may be appropriate for fields with clearly defined seismicity, but in other cases it requires additional information in order to be valuable for pragmatic purposes. However, the systematic quantitative methodology of a PSHA is an excellent tool to clarify crucial knowledge gaps and helps, consequently, to define additional research priorities. Hereby the continuous and improved monitoring provides, with the geophysical data from the exploited reservoirs, the essential database. Improved hazard estimates may be obtained by investigating statistical correlation of mechanical parameters with seismicity, attenuation and site response models, possible favored earthquake mechanism and, for the long-term, microearthquakes (M < -3) and their relation to small earthquakes (1.5 < M < 3.9).