Development and application of statistical and seismological tools for the discrimination of natural and fluid injection-induced earthquakes in the vicinity of oil and gas reservoirs

Project leaders

Mariarosaria Gallipoli, Leo Eisner

Agreement

REPUBBLICA CECA - CAS (ex AVCR) - Czech Academy of Sciences

Call

CNR-CAS (ex AVCR) 2016-2018

Department

Earth system science and environmental technologies

Thematic area

Earth system science and environmental technologies

Status of the project

New

Research proposal

Seismicity induced by fluid extraction or injection has been becoming a debated scientific topic worldwide (Ellsworth, 2013; Clarke, 2014), due to the seismic hazard that such activities are able to cause, concerning not only the occurrence of microseismic events (Stabile et al., 2014) but large events felt on surface (Keranen et al., 2013).
Nevertheless, it was shown that seismicity occurring in the vicinity of oil and gas reservoirs not necessarily has to be induced (or triggered) by anthropic activities, since seismicity naturally occurs on pre-existing faults, which may form boundaries of reservoirs (Davis and Frohlich, 1993; Janska and Eisner, 2013). Understanding of processes which lead to induced seismicity requires a good differentiation between natural and induced seismicity (e.g. Opsal and Eisner, 2014, Telesca et al., 2015, Dahm et al., 2015) in order to identify processes which lead to felt seismicity. Such understanding will serve not only the energy industry in the oil and gas sector but also more environmentally friendly geothermal sector which is also limited by induced seismicity (e.g. Haring 2008). Furthermore, it would help energy industry towards a safer exploitation of resources as well as would enable governments to create balanced policies that meet the national energy requirements without induced earthquakes affecting population.
Therefore, it is crucial to develop robust methods that allow discriminating natural earthquakes from those induced by anthropic activities of fluid extraction/injection.
Although recently relevant publications have been growing, still only a few studies have been focused on the way to differentiate natural from induced earthquakes. Based on McGarr (1971) relationship derived from mining, Shapiro et al. (2010) have introduced the so-called seismogenic index that expresses the potential of a reservoir to induce seismicity per injected unit fluid volume. The analysis carried out by McGarr (2014) and Hallo et al. (2014) of numerous case histories of earthquake sequences induced by fluid injection reveals that the total volume of injected fluid plays a key role in determination of maximum magnitude resulting from given injection. Such studies can be useful to evaluate the probability of inducing given-magnitude earthquakes and the maximum expected magnitude of induced earthquakes. Using simple relations between peak ground acceleration, magnitude, and stress drop inferred from standard scaling relations and vibration theory, Hough (2014) has shown that stress-drop values for injection-induced events are lower by a factor of 10 than stress drops of regional tectonic events. Even if such information could be used to differentiate between natural and induced events, it is known that the variation in stress drop for natural earthquakes is huge and it is hard to determine for induced events unless a large number of stations nearby the study area is available.
Recently, probabilistic discrimination criteria have been suggested in order to develop clear, quantitative, and testable discrimination rules such as the discrimination of induced seismicity by full moment tensor inversion and decomposition (Cesca et al., 2013a), the use of cross-correlation (Opsal and Eisner, 2014; Telesca et al., 2015), and the development of methods based on physical-statistical seismicity models (Passarelli et al., 2013; Dahm et al., 2015). Furthermore, while quantitative models exist for estimation of the production-related stress perturbation, two main challenges remain: i) underground properties over large scales (e.g. permeability, storativity, etc.) and ii) understanding if a certain perturbation is significant in comparison to the pre-existing tectonic stress field or not (Suckale, 2010; Dahm et al. 2015). Therefore, the main scientific challenge that this proposal addresses is investigation of criteria separating natural and induced events using advanced statistical and seismological methods.
Both the Czech and Italian groups have relevant experience and competence in seismological and statistical research, concerning, in particular, seismic data analysis, seismic measurements, seismological network design and installation, development and application of advanced statistical methodologies for space-time seismicity analysis. Furthermore, both teams have complete datasets of fluid injection/extraction-induced seismicity in several areas of the world that will be shared and jointly analysed in order to benchmark main scientific results, focused on the discrimination between natural and induced seismicity. For instance, the Italian group will contribute a seismic dataset from the High Agri Valley in Southern Italy, where the exploitation of the biggest onshore oil field in West Europe takes place, managed by the Eni Company, with an average production of about 3.6 × 109 kg of oil per year and 9.6 × 108 m3 of natural gas per year. Prague group will contribute with a dataset of the microseismicity monitored during the hydraulic fracturing of a Woodford shale gas reservoir in Oklahoma, USA. This dataset is using ten lines of a star-like surface array of approximately 1600 geophones and accelerometers.
The availability of these seismic datasets along with the newly developed seismological and statistical experience of the involved research teams enable to benchmark and generalize developed methodologies and provide new insights on the complex relationship between anthropic activity and triggering of earthquakes, hopefully providing information, methodologies and data useful for a more balanced seismic hazard assessments in areas close to energy underground extraction.

Research goals

The goals of this project are improvement of understanding of the phenomena of induced seismicity mainly through advanced data processing and comparative analysis of mutual relationship between human activity and seismicity. The goals will be achieved through advanced analysis of the existing seismic data, then the processed results will be tested and benchmarked on two available datasets that the parties agreed to share to achieve following:
1. Improve locations and detections of seismic events: we shall use both datasets to detect weaker seismic events to achieve greater completeness of the earthquake catalogues.
2. The complete catalogue will be analysed with the Epidemic Type Aftershock Sequence (ETAS) model with the aim to also determine the stress relaxation of seismicity. We shall compare and contrast datasets of natural seismicity and attempt to identify distinct differences between natural and induced seismicity.
3. We carry out source mechanism analysis of suitable characteristic events to determine stress orientation and interaction on the faults. This information will be useful to propose a geomechanical explanation of the observed seismicity analogously to the method developed by Dahm et al. (2015).
4. We will investigate and develop further seismological and statistical methods useful to establish criteria for differentiation between natural and induced seismicity.

Last update: 14/05/2024