A classification of induced seismicity

In order to adopt the best safety procedures, man-made earthquakes should be differentiated as a function of their origin. At least four different types of settings can be recognized in which anthropogenic activities may generate seismicity: (I) fluid removal from a stratigraphic reservoir in the underground can trigger the compaction of the voids and the collapse of the overlying volume, i.e., graviquakes; the deeper the reservoir, the bigger the volume and the earthquake magnitude; (II) wastewater or gas reinjection provides the reduction of friction in volumes and along fault planes, allowing creep or sudden activation of tectonic discontinuities, i.e., reinjection quakes; (III) fluid injection at supra-lithostatic pressure generates hydrofracturing and micro-seismicity, i.e., hydrofracturing quakes; (IV) fluid extraction or fluid injection, filling or unfilling of artificial lakes modifies the lithostatic load, which is the maximum principal stress in extensional tectonic settings, the minimum principal stress in contractional tectonic settings, and the intermediate principal stress in strike-slip settings, i.e., load quakes; over given pressure values, the increase of the lithostatic load may favour the activation of normal faults, whereas its decrease may favour thrust faults. For example, the filling of an artificial lake may generate normal fault-related seismicity. Therefore, each setting has its peculiarities and the knowledge of the different mechanisms may contribute to the adoption of the appropriate precautions in the various industrial activities.     

水文地质结构与水库诱发地震

本文基于断层渗透结构和岩体渗透稳定性两个概念,提出了水文地质结构及其分类方案.通过理论分析和水库地震案例解析,阐述了水文地质结构对诱发地震的制约作用.提出了基于水文地质结构进行水库诱发地震危险性评价的思路.对水库诱发地震危险性评价具有一定借鉴意义.     

A review of theories of mechanisms of induced seismicity

Theories of the physical processes leading to the stimulation of seismic activity by underground explosions, fluid injection, and reservoir impoundment are summarized. In all cases, the materials must be pre-stressed to a substantial fraction of their breaking strength in order for seismicity to be induced. Stress concentrations due to the presence of old faults or to inhomogeneities in the material properties play an important role in localizing induced seismicity.

For the few cases for which data are available, the stimulation of earthquakes by fluid injection in bore holes is adequately explained by a Coulomb-Mohr failure criterion and the concept of effective pressure in a water-filled porous mechanism. Reservoir-related earthquakes are most likely due to the same mechanism, but, in view of the low injection pressures, additional physical or chemical effects of the water on the materials may play an important role. There may be a weakening of the materials in old fault zones by the introduction of water or static fatigue in silicate rocks due to stress corrosion.     

Mechanical and hydraulic properties of rocks related to induced seismicity

Witherspoon, P.A. and Gale, J.E., 1977. Mechanical and hydraulic properties of rocks related to induced seismicity. Eng. Geol., 11(1): 23–55.The mechanical and hydraulic properties of fractured rocks are considered with regard to the role they play in induced seismicity. In many cases, the mechanical properties of fractures determine the stability of a rock mass. The problems of sampling and testing these rock discontinuities and interpreting their non-linear behavior are reviewed. Stick slip has been proposed as the failure mechanism in earthquake events. Because of the complex interactions that are inherent in the mechanical behavior of fractured rocks, there seems to be no simple way to combine the deformation characteristics of several sets of fractures when there are significant perturbations of existing conditions. Thus, the more important fractures must be treated as individual components in the rock mass.In considering the hydraulic properties, it has been customary to treat a fracture as a parallel-plate conduit and a number of mathematical models of fracture systems have adopted this approach. Non-steady flow in fractured systems has usually been based on a two-porosity model, which assumes the primary (intergranular) porosity contributes only to storage and the secondary (fracture) porosity contributes only to the overall conductivity. Using such a model, it has been found that the time required to achieve quasi-steady state flow in a fractured reservoir is one or two orders of magnitude greater than it is in a homogeneous system. In essentially all of this work, the assumption has generally been made that the fractures are rigid.However, it is clear from a review of the mechanical and hydraulic properties that not only are fractures easily deformed but they constitute the main flow paths in many rock masses. This means that one must consider the interaction of mechanical and hydraulic effects. A considerable amount of laboratory and field data is now available that clearly demonstrates this stress-flow behavior. Two approaches have been used in attempting to numerically model such behavior: (1) continuum models, and (2) discrete models. The continuum approach only needs information as to average values of fracture spacing and material properties. But because of the inherent complexity of fractured rock masses and the corresponding decrease in symmetry, it is difficult to develop an equivalent continuum that will simulate the behavior of the entire system. The discrete approach, on the other hand, requires details of the fracture geometry and material properties of both fractures and rock matrix. The difficulty in obtaining such information has been considered a serious limitation of discrete models, but improved borehole techniques can enable one to obtain the necessary data, at least in shallow systems. The possibility of extending these methods to deeper fracture systems needs more investigation. Such data must be considered when deciding whether to use a continuum or discrete model to represent the interaction of rock and fluid forces in a fractured rock system, especially with regard to the problem of induced seismicity. When one is attempting to alter the pressure distribution in a fault zone by injection or withdrawal of fluids, the extent to which this can be achieved will be controlled in large measure by the behavior of the fractures that communicate with the borehole. Since this is essentially a point phenomenon, i.e., the changes will propagate from a relatively small region around the borehole, the use of a discrete model would appear to be preferable.     

On self-similarity of premonitory patterns in the regions of natural and induced seismicity

Anticipating the scale invariance of rock fracturing processes, we applied Keilis-Borok’s algorithm M8, originally designed for identifying times of increased probability (TIPS) of occurrence of strong earthquakes (M < 8.0), retrospectively to Koyna earthquakes which occurred in the region after the impoundment of the Shivaji Sagar reservoir in 1962. The algorithm which enables diagnosis of TIPS from the 7th year onwards after the commencement of the earliest available data set showed that the 5.3 magnitude earthquake of 20 September 1980 indeed occurred within a time of increased probability. This result, apart from its potential application to recognizing future TIPS in the region, points to selfsimilarity between the premonitory patterns of natural and induced earthquakes and to scale-invariant nature of their processes. Further, a typical precursory rise in seismicity followed by a relative quiescence was also found to precede all the three larger earthquakes of the sequence.     

A database for worldwide seismicity quantification

The first step in a seismicity analysis usually consists of defining the seismogenic units, seismic zones or individual faults. The worldwide delimitation of these zones involves an enormous effort and is often rather subjective. Also, a complete recording of faults will not be available for a long time yet. The seismicity model presented in this paper therefore is not based on individually defined seismic zones but rather on the assumption that each point in a global 1/2° grid of coordinates represents a potential earthquake source. The corresponding seismogenic parameters are allocated to each of these points. The earthquake occurrence frequency, one of the most important parameters, is determined purely statistically by appropriately spreading out the positions of past occurrences. All the other significant seismicity characteristics, such as magnitude-frequency relations, maximum possible magnitude and attenuation laws including the dependence on focal depth are determined in a global 1/2° grid of co-ordinates. This method of interpreting seismicity data allows us to establish a transparent, sufficiently precise representation of seismic hazard which is ideally suited for computer-aided risk analyses.     

Synthesis of seismicity studies for western Alaska

An examination of the Alaskan earthquake catalogs revealed that from 1928 through 1965, twelve earthquakes of magnitudes (M) in the range 5.6 to 7.3 were located in and around Seward Peninsula region of western Alaska. Moreover, a number of earthquakes of M < 5.0 were found to distribute over the same area. Further investigation of the seismicity employing a local seismographic network in the above area showed a higher level of onshore and offshore seismic activity than had been previously recognized. A number of clusters of earthquakes have been identified. Some of them are located in the epicentral areas of past strong earthquakes (M > 5.5) and some are located in areas traversed by mapped faults. The nature of the seismicity as identified with the local network data is primarily crustal over the entire study area. Investigation of focal mechanisms of isolated strong earthquakes or clusters of small earthquakes show normal faulting as the dominant mode of strain energy release in the western part of Alaska. Moreover, in areas lying, approximately, south and north of Kotzebue Sound, the principal components of horizontal stresses tend to orient in the NW-SE and WNW-ESE directions, respectively.     

An investigation of seismicity for western Anatolia

In order to investigate the seismicity of western Anatolia limited with the coordinates of 36°–40° N, 26°–32° E, Gutenberg–Richter magnitude–frequency relation, seismic risk and recurrence period have been computed. The data belonging to both the historical period before 1900 (I₀ ≥ 5.0 corresponding to M_S ≥ 4.4) and the instrumental period until the end of 2006 (M_S ≥ 4.0) has been used in the analysis. The study area has been divided into 13 sub-regions due to certain seismotectonic characteristics, plate tectonic models and geology of the region. All the computations have been performed for these sub-regions, separately. According to the results, a and b values in the computed magnitude–frequency relations are in the intervals 3.19±0.17 – 5.15±0.52 and 0.42±0.05 – 0.66±0.07, respectively. The highest b values have been determined for sub-regions 3 and 12 (Demirci-Gediz and Gökova Gulf-Mu?la-Gölhisar). The lowest b values have also been determined for sub-regions 1 and 9 (Bal?kesir and Bodrum-?stanköy). Finally, seismic risk and recurrence period computations from a and b values have shown as expected that sub-regions 1 and 9 which have the lowest b values and the highest risks and the shortest-recurrence periods.     

The mechanism of induced seismicity at mosul reservoir based on the first motion analysis

In 1986 shortly after the impounding of Mosul reservoir, shallow earthquakes began occurring in the immediate reservoir vicinity, with magnitudes up to ML 3.0, at rates of up to 3 events per week. These events were almost certainly reservoir-induced and coincided with steadily increasing water levels. Cluster of epicenters was observed in the area located within a complex fault zone called the Sinjar-Dohouk-Kuchuk fault system. The presence of such fault system considers a potential source of earthquakes. A composite fault plane solution, based on first p-wave motion analyses, indicates that the mechanisms of seismicity were right-lateral strike-slip faulting along N44°E plane dipping 58° NW, in conformity with the local tectonics.     

Control of hazard due to seismicity induced by a hot fractured rock geothermal project

In 2003 hydraulic stimulations were carried out in a geothermal field in eastern El Salvador, Central America, as part of a project to explore the feasibility of commercial hot fractured rock energy generation. A key requisite for this environmentally friendly energy source is that the fracturing of the hot rocks at depths of 1–2 km must not produce levels of ground shaking at the surface that would present a serious disturbance or threat to the local population. Thresholds of tolerable ground motion were inferred from guidelines and regulations on tolerable levels of vibration and from correlations between instrumental strong-motion parameters and intensity, considering the vulnerability of the exposed housing stock. The thresholds were defined in terms of peak ground velocity (PGV) and incorporated into a “traffic light” system that also took account of the frequency of occurrence of the induced earthquakes. The system was implemented through a dedicated seismograph array and locally derived predictive equations for PGV. The “traffic light” was used as a decision-making tool regarding the duration and intensity of pumping levels during the hydraulic stimulations. The system was supplemented by a small number of accelerographs and re-calibrated using records obtained during the rock fracturing.     

The mechanism of induced seismicity at mosul reservoir based on the first motion analysis

In 1986 shortly after the impounding of Mosul reservoir, shallow earthquakes began occurring in the immediate reservoir vicinity, with magnitudes up to ML 3.0, at rates of up to 3 events per week. These events were almost certainly reservoir-induced and coincided with steadily increasing water levels. Cluster of epicenters was observed in the area located within a complex fault zone called the Sinjar-Dohouk-Kuchuk fault system. The presence of such fault system considers a potential source of earthquakes. A composite fault plane solution, based on first p-wave motion analyses, indicates that the mechanisms of seismicity were right-lateral strike-slip faulting along N44°E plane dipping 58° NW, in conformity with the local tectonics.     

The present status of reservoir induced seismicity investigations with special emphasis on Koyna earthquakes

The status of Reservoir Induced Seismicity (RIS) has been reviewed periodically (Rothé, 1968, 1973; Gupta and Rastogi, 1976; Simpson, 1976; Packer et al., 1979). In the present paper, the significant work carried out during the last three years on RIS is reviewed.An earthquake of magnitude occurred on November 14, 1981 in the vicinity of Aswan Lake, Egypt, 17 years after the filling started in 1964. This event occurred 4 days after the seasonal maximum in the reservoir water level and was followed by a long sequence of aftershocks. Another event of magnitude occurred in the vicinity of Aswan Lake on August 20, 1982. Results of preliminary investigations indicate that this seismic activity is reservoir induced. Recent analyses of induced seismic events at Nurek Reservoir U.S.S.R., show that the second stage of filling during August to December 1976, increasing the maximum depth from 120 m to 200 m, was accompanied by an intense burst of shallow seismic activity. An outward migration from the centre of the reservoir, possibly associated with diffusion of pore pressure, is revealed by the temporal distribution of earthquake foci. A variety of investigations including the in situ measurement of tectonic stress, pore pressure, permeability, distribution of faults, etc., in addition to monitoring seismicity, have been undertaken in the vicinity of the Monticello Reservoir, South Carolina. The largest reservoir induced earthquake is predicted not to exceed magnitude 5.The Koyna Reservoir, India, continues to be the most outstanding example of RIS. Three earthquakes of magnitude 5 occurred in September 1980. Earthquakes of magnitude 4 occur frequently in the vicinity of Koyna, the latest being on February 5, 1983. Events that occurred during the period 1967–1973 have been relocated using better procedures and are found to be much shallower and the epicentres less diffused. Location of 12 earthquakes of M_s 4.0, their foreshocks and aftershocks, that occurred during 1973–1976, composite focal mechanism solutions and related studies are consistent with the delineation of a N-S trending fault through the reservoir area. In a couple of interesting studies it has been demonstrated that earthquakes of magnitude 5.0 in the Koyna region are usually preceded by several magnitude 4 earthquakes in the preceding fortnight. Also, a rate of loading of Koyna reservoir of at least 40 ft/week appears to be a necessary, although not sufficient, condition for the occurrence of magnitude 5 earthquakes. Smooth filling/emptying appears to be the key to reduce the hazard of RIS.A map and a table of the reported cases of reservoir induced changes in seismicity through 1982 have been compiled.     

Evaluation of the risk of induced seismicity at the itzantun hydroelectric site, Chiapas, Mexico

Consolidation theory and concepts of rock failure can be used to evaluate the probable risk of induced seismicity as a result of filling of reservoirs. This evaluation indicates the safest way to fill a reservoir, and depends only on the geometry of the load, the rate of filling and the geological structures in the area. The stability function is actually a measure of the risk of having failure, with time, for a particular loading history in respect to a plane of weakness.

The stability function is applied to the area of the Itzantun reservoir, which will be in southern Mexico. Drawdowns can increase the risk of triggering earthquakes in this area, which is prone to thrust faulting. It is possible to estimate the stresses after a period during which the water level is maintained and a decrease in stresses with the depth of the observation point.

The estimates of the probable induced seismicity are limited as the residual stress in the area prior to the impounding is unknown. With a measure of the residual tectonic stress it will be possible to determine an optimal filling rate to reduce the probability of induced seismicity.     

Earthquake focal mechanisms of the induced seismicity in 2006 and 2007 below Basel (Switzerland)

To stimulate the reservoir for a “hot dry rock" geothermal project, that was initiated by a private/public consortium in the city of Basel, approximately 11500 m³ of water were injected between December 2nd and 8th, 2006, at high pressures into a 5 km deep well. More than 10500 seismic events were recorded during the injection phase, and minor sporadic seismic activity was still occurring more than two years later. The present article documents the focal mechanisms of the 28 strongest events, with M_L between 1.7 and 3.4, that have been obtained by the Swiss Seismological Service (SED) during and after the stimulation. The analysis is based on data that was recorded by a six-station borehole network, operated by the project developers, as well as by several permanent and temporary surface networks. The hypocenters of the events are located inside the stimulated rock volume at depths between 4 and 5 km within the crystalline basement. Of the 28 faultplane solutions two are normal faulting mechanisms and one is a strike-slip mechanism with a strong normal component. All others are typical strike-slip mechanisms with mostly NS and EW striking nodal planes. As a consequence, the T-axes are all nearly horizontal and oriented in a NE or SW direction (mean azimuth 46 ± 11 degrees) and the P-axes of the strike-slip events point in a NW or SE direction (mean azimuth 138 ± 13 degrees). Overall, the observed focal mechanisms agree with what would be expected from both the stress observations within the well and the stress field derived from the previously known natural seismicity.     

Induced seismicity in Canada

The annual and cumulative catalogues of Canadian earthquakes prepared by the Department of Energy, Mines and Resources, are critically examined to determine if they contain examples of induced seismic activity caused by fluid injection in oil fields, by the impounding of water in large reservoirs or by mining.

It is concluded that there is just one example each of induced seismic activity caused by fluid injection and by reservoir impoundment, but that there are many examples of induced seismic activity associated with mining. These include both mine bumps and rockbursts.     

Time-dependent seismicity in China

Precise zonation of the territory of China has been performed based on the active known faults, type of faulting and seismicity level. One hundred and forty seven seismogenic regions were defined, forming 10 larger seismic areas, and the seismotectonic characteristics in each one of them were investigated in detail. After checking for data accuracy and completeness of the shallow earthquakes (h≤60 km), the regional time and magnitude predictable model was applied and the model parameters were estimated. Based on the model applicability in the studied area, probabilities for the occurrence of strong (M≥6.0) earthquakes during the next 10 years were calculated for each seismogenic region. Statistical tests have been used proving the superiority of the model in comparison with the time independent one, as well as in comparison with the actual earthquake occurrence.