Chapter 6
Summary and discussion
6.1 Summary of the findings of this thesis
The CSZ is the most active seismic zone of eastern Canada. The region contains geological faults of the St. Lawrence paleo-rift system and of a Devonian impact structure. Between October 1977 and December 1997, monitoring of the earthquakes by a local network has defined a 30 by 85 km active zone of elliptical shape with major axis along the St. Lawrence River. Hypocentral depths are distributed from near the surface to 29 km, with about two thirds between 5 and 16 km. A prominent weakly seismic volume exists along the North Shore of the St. Lawrence River. Most shallow (< 5 km depth) and deep (> 20 km) events occur in specific subareas of the CSZ. Similarly to the pattern observed for the 1924-1978 period, 6 out of the 7 magnitude > 4 events in the period 1977-1997 occurred in the SW and NE ends of the CSZ. The improvement in the recording capabilities of the local seismograph network has brought the number of located earthquakes from about 100/year in the late 70's to more than 200/year since 1995. The number of located earthquakes varies between 0 and 30 per month, with no obvious correlation with magnitude > 4 occurrences, at either the regional or local level. Using microearthquakes recorded both on regional and local stations, an empirical relation between the Nuttli magnitude (mN) and the Richter magnitude (ML, used for events only recorded locally) was defined: mN = 0.63 ML + 1.41 (Section 1.4.3). With a unified magnitude scale, the CSZ magnitude-recurrence curve was extended down to magnitude mN 1.5, which is the estimated magnitude for completeness of reporting for the 1977-1997 period. On a magnitude recurrence curve, the rate of magnitude > 4 earthquakes exceeds the level suggested by the lower magnitude events recorded between 1977 and 1997. This may suggest that the magnitude > 4 events have overestimated magnitudes, or that there is a real change in occurrence rates at higher magnitudes.
On the basis of the heat conduction equation and Grenvillian surface heat flow, the temperature at the cutoff depth of CSZ earthquakes (D99% = 25 km), has been estimated to be between 215 and 355oC (Chapter 2). These temperatures and the inferred mafic mid- and lower-crustal compositions imply a brittle-ductile transition deeper than 25 km. The transition could occur at 25 km only for surface heat flow higher than the average Grenville value (in the 50 mW m-2 range). Currently, two low quality measurements suggest such a high value. With the assumed absence of quartz at mid- and lower crustal depths, the earthquake cutoff depth may correspond to the passage from velocity strengthening to velocity weakening if the onset of flow of hydrated feldspars occurs at temperatures in the 300-350oC range. With an assumed maximum crustal stress difference of about 200 MPa, fault reactivation at mid-crustal depth can occur with a low friction coefficient and/or with a high pore fluid pressure, or with a change to a high-pressure type of fracturing (see also discussion in Section 6.2 in this chapter).
A series of normal faults associated with the St. Lawrence paleo-rift system breaks the CSZ crust. On land, most of these faults correspond to lineaments in the remote sensing imagery. Some of these faults, but not all, have geophysical expressions in the magnetic and gravity fields (see discussion in Chapter 3). Offshore and under the Appalachian sequence, some regional-scale normal faults are revealed in the Precambrian basement, using magnetic, gravimetric and seismic reflection data. The structure of the CSZ was studied along four profiles perpendicular to the St. Lawrence River, revealing regional as well as local characteristics. Long normal faults, such as the St. Lawrence faults, and the newly-revealed Charlevoix and South Shore faults, cut across the whole CSZ. The Charlevoix fault is a few kilometres offshore and parallels the north shore. It may represent the structure reactivated by some of the largest CSZ earthquakes, including the 1925 M 6.2 earthquake. The South Shore fault appears to bound the seismic activity to the SE. The attitude of the Precambrian-Appalachian interface changes along the strike of the St. Lawrence River. The NE part of the CSZ, where most magnitude > 4 events occur, is an area with linear normal faults defining a series of steps in the Precambrian basement. The impact crater, on the other hand, coincides with an interpreted graben under the river, possibly created by the impact readjustment. The SW part, which is weakly seismic, is cut by numerous normal faults parallel and perpendicular to the river's trend. It is suggested that the Charlevoix fault could be one major reactivated fault, while the St-Laurent and South Shore faults bound highly fractured blocks. Reprocessing of a seismic line revealed deep reflectors in the Precambrian basement.
Although the CSZ has been subject to at least five earthquakes of magnitude 6 or greater, no surface rupture was found under the St. Lawrence River (see Chapter 4). A systematic search for surface faulting was done on seismic reflection lines for oil exploration and for Quaternary mapping. Some 20 CSZ focal mechanisms for magnitude > 2 events show mostly reverse to reverse-oblique faulting in the CSZ, with no preferential orientation of the P-axis, but with a higher number of nodal planes in the NE quadrant. Local variations in the stress system appear to occur across the CSZ. Most earthquakes occur in clusters of two events or more, that may or may not have preferred orientation. One of these groups, located under the St. Lawrence, contains about one third of CSZ events, including the 1979 mN 5 event. It defines one of the best alignments of hypocentres, possibly related to the newly-defined Charlevoix fault. Comparison of these groups with focal mechanisms suggest multiple orientations for the reactivated faults. For the period November 1988 to August 1997, cross-correlation of seismic traces has revealed that less than 15% of CSZ events are multiplets. Most of these events occurred within a few days of each other. Comparing earthquakes groups and multiplets with focal mechanisms has revealed some fault planes. Reactivated structures can be of local or regional importance, with orientations not necessarily sub-parallel to the Iapetan rift faults. Since most earthquakes are small, the actual dimension of the fracture, or reactivated fault surface, is small (generally in the tens of meters in radius). This aspect, together with the varied orientations of the fault planes, suggest that most earthquakes occur in highly fractured volumes. Study of the various sub-zones of the CSZ also suggest that faults with varied orientations can be reactivated, including some related to the impact crater.
The microearthquake data of the local Charlevoix network was used in an inversion procedure to determine the local velocity structure (Chapter 5). Using some 171 events recorded between November 1988 and December 1997, the best 1-D crustal model for the CSZ is very close to the "standard" GSC model, i.e., 6.2 km/s and 3.57 km/s P- and S-wave velocities. Station corrections, defined for the local network, imply lower velocities inside the impact crater. The data were also used to obtain a pseudo 2-D velocity model, i.e. two velocity models corresponding to the north and the south shores respectively. The model resolves the Appalachian sequence of the south shore (6 km thick sequence with velocities in the 5.6 to 5.8 km/s) and the Precambrian basement (increase in velocity between 8 to 14 km depth from 6.2 to 6.6 km/s). The best 2-D velocity model reduces significantly the RMS of the solutions. With the improved velocity model, locations of most CSZ events move towards the SE by about 1.5 km. Using CSZ events recorded on the regional station DAQ, a 1-D velocity model was defined for the Laurentides Park region, to the West of the CSZ. The model shows a gradual P-velocity increase from the surface to 10 km from about 5.9 to 6.7 km/s. The CSZ velocity structure is different from that of the Laurentides Park. While the Vp/Vs ratio is 1.73 for the CSZ, the Vp/Vs is higher (1.81) in the Laurentides Park. Lithology and possibly the degree of fracturing appear to be the main factor controlling these variations.
6.2 Why do earthquakes occur in the CSZ?
Over the last 30 years, a number of studies have proposed factors that may lead to CSZ earthquakes. The proposed causative factors can be grouped in four classes. First, some studies refer to the inherent geological weakness of the CSZ, either primarily due to the rift faults (Adams and Basham, 1991) or to the combined meteor impact-rift faults (Roy and Du Berger, 1983; Anglin, 1984; Lamontagne, 1987). A second group refers to the combination of high stress difference and crustal weakness: either postglacial rebound-impact crater (Leblanc et al., 1973; Leblanc and Buchbinder, 1977; Anglin and Buchbinder, 1981); mafic intrusion-impact crater (Stevens, 1980); or stress concentration-impact crater (Lyons et al., 1980; Hasegawa and Wetmiller, 1980). A third group invokes mainly the effect of stress modifiers: postglacial rebound stress (that can favour thrust earthquakes in the CSZ: James and Bent, 1994; Wu and Hasegawa, 1996); crustal subsidence in the Laurentides Park (which is anomalous compared with the postglacial uplift witnessed in eastern Canada: Frost and Lilly, 1966(1)); or the effect of past major earthquakes (Bent, 1992; Ebel, 1998). Finally, some refer to possible anomalies in the rock properties such as high pore-fluid pressures at depth (Hasegawa, 1986; Sibson, 1989; Zoback, 1991; Lambert et al., in prep.) and/or low coefficient of friction (Lamontagne and Ranalli, 1996).
The Mohr diagram can be used as a graphic illustration of the factors leading to shear failure (Figure 6.1). Sliding occurs when the Mohr circle is tangent to the Coulomb envelope, i.e. when the shear stress exceeds the strength along the plane of failure. This instability can occur by changing the ambient stress field and/or the stability conditions of the pre-existing fracture. First, the Mohr circle (i.e. the stress conditions) can be brought closer to the failure envelope by modifying either its radius (i.e. the stress difference) or its position. The stress difference can be increased with a larger principal compressive stress (Sigma1), or inversely, by a decrease in the least compressive stress (Sigma3). The Mohr circle can approach failure (i.e. a shift to the left) with higher pore-fluid pressures, a process responsible for reservoir-induced seismicity. Secondly, fault reactivation can occur if failure conditions are modified. The failure envelope can be brought closer to the Mohr circle, by lowering the cohesion (S) of the pre-existing fractures, or by lowering the coefficient of friction (µ). In the following sections, I discuss separately the factors affecting the stress system, and those affecting the failure strength of rocks in the CSZ. A summary is given in Figure 6.2
6.2.1 The stress system in the CSZ
In Eastern North America, the inferred maximum horizontal compressive stress axis correlates well with plate tectonics interpretations (Zoback and Zoback, 1991). Over most of the so-called mid-plate stress province, plate driving and resisting forces are primarily responsible for the predominantly ENE orientation of the maximum compressive stress axis. The reorientation in the CSZ may be due to glacioisostatic effects (Wu, 1998). The tectonic stresses can be amplified in the brittle upper crust leading to absolute levels in the 100 MPa range (Hasegawa et al., 1985). Thus, over most of eastern North America, the fundamental situation could be described by the Mohr diagram (Figure 6.1). The situation is stable with stress differences arising from the plate-wide forces which are generally too small to cause reactivation. Unstable situations, such as the one existing in the CSZ, can come about by adding local sources of stress to the ambient stress field (such as flexural stresses and lateral density contrasts/buoyancy forces) and/or by concentrating existing stresses.
Flexural stresses are caused by changes in the loads on or within the lithosphere (erosion, sedimentation, postglacial readjustment). Of these three, only postglacial rebound is significant in continental Eastern Canada. In areas formerly covered by continental glaciers, glacial rebound stresses can perturb the ambient stress field for thousands of years after glaciers have retreated. Interestingly, the change in the faulting style from reverse faulting in Eastern Canada to strike-slip faulting in the eastern United States occurs approximately at the limit of the former glacial maximum (Zoback, 1992; James and Bent, 1994). Consequently, it is evident that postglacial rebound stresses are perturbing the eastern Canadian stress field. The level of induced stresses, however, is low: the deviatoric stress field is increased only by about 5 to 10 MPa (Wu and Hasegawa, 1996). In the CSZ, the radial component of the postglacial stress field is compressive, oriented NW-SE and larger than the transverse component (James and Bent, 1994). Due to their orientation and small magnitude, postglacial stresses may contribute to thrust faulting in the St. Lawrence Valley (Hasegawa et al., 1985; Wu and Hasegawa, 1996; James and Bent, 1994). In other words, they can favour reverse faulting earthquakes, without constituting the primary cause.
Lateral mass anomalies within or just beneath the lithosphere are additional sources of stress. In the CSZ, upper crustal density variations exist between the Canadian Shield and the Appalachians (evident in Bouguer anomaly maps, e.g. Figure 3.7B). Two approaches have been used to model their impact on the local stresses. In the first one, an elastic model suggested near vertical maximum compressive stress corresponding to the negative gravity anomaly (which in the CSZ case, is approximately the St. Lawrence River; Goodacre and Hasegawa, 1980(2)). In the second approach, an elastic multi-layer FE model was used to compute stresses induced by lateral density variations (Assameur and Mareschal, 1995). At 10 km depth, average stress difference can reach about 27 MPa in regions with the highest density contrasts (east of the Grenville Front, Thetford Mines and the lower St. Lawrence). Of these three areas, only the lower St. Lawrence is seismically active. In the CSZ, induced stress differences are about 10 MPa at 10 km depth. Assuming a 20 MPa horizontal compression oriented N60oE, the ambient stress difference increases to 24 MPa when the orientation of the local stress field is considered (Assameur and Mareschal, 1995). Other sources, such as crustal thinning and lithosphere thickening, should not apply to the relatively constant crustal thickness in the CSZ (as inferred from seismic refraction; Lyons et al., 1980). The fact that regions with the highest induced stress differences are mostly aseismic suggests that mass anomalies cannot be the main factor of the CSZ seismicity.
At least two factors can increase the crustal stress difference: inclusions of materials with elastic parameters different from the surroundings, and fault intersections. Stresses can concentrate around a weak inclusion in a rigid and brittle material. Weak hydrated materials such as serpentinized gabbros at mid-crustal depth are particularly good stress concentrators due to the high rigidity contrast they provide with the surrounding rocks (Campbell, 1978). Appropriate orientation of the long axis of the intrusive with respect to the stress field and large size of the intrusive also enhance the effect. In Eastern Canada, it has been noticed that large intrusions such as the Morin anorthosite in the Western Quebec Seismic Zone and the Sept-Iles layered mafic intrusion in the Lower St. Lawrence Seismic Zone, are apparently aseismic areas in epicentral maps (Lamontagne et al., 1989; Lamontagne et al., 1994). In the CSZ, the massive St-Urbain anorthosite body does not seem to act as a stress concentrator; it is not seismic and it is not surrounded by earthquake activity. However, it is possible that a highly fractured volume (i.e. the volume beneath and around the impact crater) may have average elastic parameters lower than the surrounding, less fractured, rock. The analysis by Campbell (1978) shows that, if the body is elliptical in plan view with long axis perpendicular to the direction of maximum compressive stress, the largest stress differences occur just outside the ellipse along the prolongation of its long axis. This situation shows some similarity with the one in the CSZ, where the largest earthquakes occur at the two extremities of the area. The possible effect of contrast in elastic moduli cannot therefore be excluded.
Upper crustal stress concentration of 10 to 100% can arise due to the strength decrease of a localized area of the lower crust subject to ductile creep (Long and Zelt, 1991). This idea is somewhat different than the model of Hasegawa et al. (1985) which showed upper crustal stress amplification for plate-wide motions. In Chapter 2, the strength of a ductile layer was shown to vary with the lithology, the fluid content and the heat flow. In the CSZ, while lateral heat flow variations are unlikely, lateral changes of ductility in the lower crust cannot be excluded.
Stresses can also concentrate around intersecting faults (Talwani, 1988). In a rock subject to stress difference, stresses build up in fracture tips. Stress concentration is highest at the locked portions of intersecting faults, reaching eight times the ambient stress level (Talwani, 1988). The intersecting fault model could explain the high earthquake activity in the highly fractured zone within the impact structure. The intersection of these fractures could lead to enhanced stress levels, favouring ruptures, whereas areas with healed fractures would not give rise to the stress build-up necessary to cause an earthquake.
6.2.2 Physical conditions of the faults
On the whole, local sources of stress are relatively modest contributors (Sigma1 - Sigma3 < or = to 10-30 MPa) to the CSZ stress field. Since these contributions are about the same everywhere in Eastern Canada, the CSZ faults must be inherently weak compared to aseismic faults elsewhere. This could be accomplished by either a decrease in friction, an increase in pore-fluid pressure, or a combination of both.
6.2.2.1 Coefficient of friction
From a fault mechanics point of view, a low friction coefficient decreases the fault strength, making it susceptible to reactivation at plausible stress difference values (< 200 MPa). As shown in Figure 2.9B, without near-lithostatic pore-fluid pressures, a fault with a coefficient of friction µ = 0.75 is too strong to be reactivated in the current stress field. A low coefficient of friction (for instance, µ = 0.3) makes the faults weaker, i.e. more susceptible to reactivation, and allows the reactivation of steeply-dipping faults ( 70, as in the CSZ). Although the coefficient of friction is assumed to be decreasing with increasing depth as illustrated by the shallower slope of the Mohr envelope with increasing pressure (Figure 6.1), values of µ of less than 0.3 cannot be reached without a fault gouge consisting of hydrated clay minerals (Lockner, 1995).
Fault gouges are documented along regional faults of the CSZ. On the St-Laurent paleo-rift fault, for example, a fault gouge exists and was probably created in the normal faulting regime that prevailed in mid- to upper-Ordovician times (Rondot, 1979). The fault is a highly fractured zone, tens of meters wide, and includes breccias and diabase veins (Brassard, 1990). Down to at least 400 m (maximum depth of the well), the fractures are interconnected and linked to the surface, implying relative permeability. Alteration (silicification, chloritization and carbonatization) is present. In the Quebec City region, St. Lawrence rift faults show hydraulic breccias created during the opening of the Iapetus Ocean when relatively impermeable rocks of the Precambrian basement were shattered by the fluid pressure exceeding the rock strength (Lachapelle, 1993).
Although fault breccias are found along the St. Lawrence paleo-rift faults, the coefficient of friction on these faults is probably not sufficiently small as to bring the crustal strength in the 200 MPa range. A coefficient of friction of 0.3 corresponds to clays and other sheet silicates inter-layered with water, minerals that dehydrate and lose their lubricating properties at mid-crustal depth (Lockner, 1995; Scholz, 1990). Although the weakness of CSZ faults cannot be entirely due to the breccias, CSZ earthquake epicentres correspond to the locations of these breccias on the St. Lawrence rift system (Lachapelle, 1993). This may be a partial explanation of the confinement of CSZ earthquake clusters along the St. Lawrence River, and not further inland in the Precambrian Shield. The fault gouge, mapped on numerous rift faults, may sufficiently lower the coefficient of friction to favour reactivation. It could, for example, lower it from high values found with rock on rock friction (µ = 0.85) to lower values associated with smoother surfaces (µ = 0.5-0.6).
6.2.2.2 Pore-fluid pressure
As discussed in Chapter 2, high pore-fluid pressures (near lithostatic level) can bring the crustal strength to within plausible levels ( 200 MPa), even with an average value for the coefficient of friction (µ = 0.75; Figure 2.9B). Worldwide, high pore-fluid pressures appear to explain the relative crustal weakness of seismically active zones, the San Andreas fault being the most prominent example (Rice, 1992).
In the CSZ, evidences for crustal fluids exist(3). In the late 70's, fluid flow along fault zones was suggested as an explanation for the temporal disappearance of a micro-gravity anomaly and a change in P-wave velocities (Buchbinder et al., 1988). The anisotropy of travel-time changes in the CSZ observed between 1979 and 1980 was explained by the closing of saturated cracks (Kirsch et al., 1987). Finally, near La Malbaie, magneto-telluric anomalies deeper than 1400 m can be explained by water or solutions in a zone of high porosity (Chouteau, 1985).
In intraplate environments, such as the CSZ, the sources of these near-lithostatic pore-fluid pressures at mid-crustal depths are subject to debate. In the upper crust, the recognized sources of high pore-fluid pressures are compaction of saturated sediments overlain by low-permeability rocks, and dehydration reactions in metamorphism (Scholz, 1990). These two mechanisms can hardly apply to mid-crustal depth in the CSZ: no sediments are present there and the absence of local tectonic activity does not support active metamorphism. According to some petrological arguments, lower crustal rocks are essentially dry due to hydration reactions that occur at temperatures above 250C (Frost and Bucher, 1994). This opinion, however, is not shared by all (Hyndman and Shearer, 1989). We can speculate that fluids exist along fault zones, from evidence such as along the San Andreas fault (Sibson et al., 1988). Within the fault zone, high pore-fluid pressures exist, as opposed to the surrounding dry rocks. The source for these fluids could be the ductile roots of faults (Rice, 1992) or the lower crust (Sibson, 1992). For the CSZ mid-crust, two possibilities exist: fluids may be trapped by ancient tectonic processes (but the trapping of these fluids for millions of years remain enigmatic); or fluids may originate from below, possibly within the ductile layers of the lower crust (Rice, 1992) or the mantle (Sibson, 1989). Evidence for the deep origins of these fluids includes the distribution of CO2-rich springs in present-day seismic belts (Irwin and Barnes, 1980) and the anomalously high proportion of gases associated with deep origin (such as helium). These anomalies are interpreted as chemicals from the mantle that percolate to the surface through the region's rift faults (from a yearly report of activities by the GHK Group, 1988).
Once present in the middle crust, fluid pressures must build up to reach near-lithostatic values to cause fault reactivation. This can be achieved in permeable volumes capped by impermeable material. The motion on the fault breaks the seal and allows fluids to migrate upwards, in a fault-valve fashion (Sibson, 1990). One can speculate that the paleo-rift faults of the St. Lawrence rift system represent the conduits to these fluids. As shown by the mapped hydraulic breccias, the Precambrian Shield (outside the meteor impact) is relatively impermeable and restricts fluid flow to these fault zones. In addition to increasing the pore pressure, the fluids can weaken fault zones by chemically corroding gouge material, lowering the coefficient of friction.
6.3 Discussion
The deviatoric stress field in the CSZ does not appear anomalously high compared to that of eastern North America. As elsewhere, the CSZ is subject to the mid-plate ambient stress field that provides a large proportion of the deviatoric stress. The CSZ stress difference may be marginally higher than elsewhere due to postglacial rebound, a factor contributing to reverse faulting (James and Bent, 1994). Mass anomalies and stress concentration factors (inclusions with different moduli and fault intersections) can modestly contribute to the stress field. Since most of these conditions exist throughout the Grenville Province, their impact in the CSZ can only be related to the inherent weakness of the mid- to upper crust.
Since the CSZ stress field is comparable to surrounding areas, the Charlevoix earthquake activity is likely due to a local crustal weakness, caused by a low coefficient of friction and/or high pore-fluid pressures. A friction coefficient in the 0.2 to 0.3 range, necessary to bring the fault strength to plausible levels, is improbable due to the dehydration of gouge clay minerals at mid-crustal depths. The presence of a fault gouge may, however, bring the coefficient of friction towards the low end of the measured values (µ 0.5-0.6). High pore-fluid pressures appears as a likely explanation to many characteristics of the CSZ, such as the low crustal strength (Chapter 2), the sub-zones of enhanced activity (Chapter 4), the tendency of aftershocks to be shallow and outside the immediate rupture of the main shocks (Section 4.6.2.1), the variations in the stress field and in the reactivated fault orientations at the sub-zone level (Section 4.6), and the mismatch of the orientations of reactivated faults with respect to the mid-plate stress field (Zoback, 1992).
In order to take into account the seismic characteristics of the CSZ, any seismotectonic model must be centred on crustal weakness, created in all likelihood by high pore-fluid pressures. Such a model is proposed in Figure 6.3. First, fluids are assumed to exist along some of the major rift faults. The exact source of these fluids is uncertain but they could ascend from the mantle or possibly from the ductile lower crust (Chapter 2). One may speculate that the approximate linear zone defined by the 20 km depth earthquakes (Figure 1.21E) represents the mid-crustal conduit where fluids ascend to the surface. These fluids migrate to mid-crustal levels, where locally-sealed zones cause local over-pressures (Sibson et al., 1988). The over-pressures, mainly found near the major faults, can reach near-lithostatic levels, inducing fault weakening and favouring reactivation. When an earthquake is triggered, the motion on the fault breaks the seal, allowing fluids to migrate upwards, in a fault-valve fashion (Sibson, 1990). If the shock is sufficiently large, aftershocks can be induced where the migrating fluids create instability conditions. Where the surrounding rocks are impermeable, due to fault healing for example, the fluids concentrate, and remain, along the rift faults. This situation appears to be found outside the impact crater, where hypocentres define a clear alignment, and where most magnitude > 4 earthquakes occur. Where highly fractured zones exist, fluids diffuse in all directions, locally increasing the pore-fluid pressure and eventually creating numerous but small earthquakes. This situation is found within the highly fractured zone inside the crater where hypocentre groups are diffuse, and active faults extremely variable in orientations (as shown by focal mechanisms, earthquake groups and multiplet analysis). Near the central peak of the impact crater, the dense network of intersecting faults can also amplify the local stress level favouring earthquake occurrences.
Outside the CSZ, the lack of seismicity could be due to a combination of factors such as the absence of fluids or of a conduit if fractures are sealed or without fault breccias. The absence of breccias can possibly explain the boundaries of the CSZ. To the NW of the CSZ, the stable craton was never reactivated after the Grenvillian orogeny, whereas the CSZ crust was reactivated in the Paleozoic. According to Wheeler (1996), the SE boundary of the seismicity corresponds to the transition from relatively intact Precambrian rocks to the NW to a zone with thinner Precambrian crust to the SE where faults have been healed. Outside the CSZ, but still along Iapetan rift faults, a much slower rate of pore-fluid pressure build-up (and possibly a lower level of fracturing) could explain the rarity of earthquake occurrences. Using the CSZ model, fluid flow from the lower crust can explain the occurrence of two deep earthquakes: the 1988 Saguenay earthquake (29 km depth) and the 1997 Cap-Rouge earthquake (22 km depth).
The absence of any apparent CSZ surface rupture suggests that the current rates of seismic strain release are geologically recent. Had these rates existed over millions of years, evidence of this activity would have been mapped in the field and seen in the seismic reflection profiles. The earthquake activity likely started some thousands of years ago, possibly right after the last glaciation, when conditions were most favourable for reverse faulting in the CSZ (Wu and Hasegawa, 1996). In all likelihood, the current conditions that create CSZ earthquakes may persist for thousands of years in the future. The current stress field depends mainly on factors that change on geological time scales, such as plate motions, and to a lesser extent, postglacial rebound and other stress contributors that will continue to play a role for thousands of years to come.
Figure CaptionsFigure 6.1 Mohr diagram representing relationship between stresses and failure parameters. The parametres shown are: shear stress (Tal); stress (Sigma); maximum (Sigma1) and minimum (Sigma3) compressive stress and their modified values (Sigma1' and Sigma3' respectively); cohesion (S0); internal angle of friction (Phi). Reactivation occurs when the Mohr circle is tangent to the Coulomb criterion line. The Mohr envelope represents the observed change in friction coefficient with increased pressure.
Figure 6.2 Factors that can lead to earthquakes in the CSZ with the section numbers where they are discussed.
Figure 6.3 Schematic model of the CSZ earthquakes shown on a cross-section perpendicular to the St. Lawrence River axis. Major faults of the area are presented as permeable (solid black lines) or impermeable (i.e. healed; dotted lines). Fluids can move along the permeable faults, represented as zones of enhanced fluid pressures (in blue). The origin of these fluids could be the ductile layers in the lower crust or the mantle. The fluids ascend towards the surface, triggering earthquakes where temporary seals halt their progression, causing local pressure build-up. Near the surface, within the impact crater, the fluids migrate into highly fractured areas, causing repetitive, but small magnitude, earthquakes. See text for details.
1. However, a recent re-evaluation of this subsidence has shown that it is within the error margins of measurements, and consequently of doubtful significance (T. Lambert, pers. comm.).
2. In a thrust environment, additional vertical stress translates into more stable faults. Inmost papers, paired gravity low and high are usually referred to as a perturbing addition to the stress field, without mentioning the additional stability obtained in a reverse faulting environment. Consequently, the Canadian Shield-Appalachians boundary should not enhance earthquake probability in the CSZ.
3. Fluctuations of the water table creating hydraulic pulses were proposed as a means totrigger earthquakes in a prestressed environment at the limit of failure (Tsoflias et al., 1995). Since water levels were measured near Montreal, one may wonder if these measurements apply to the CSZ, some 300 km downstream, where the St. Lawrence River is 20 km wide. The diffusion of the water pressure fluctuations to mid-crustal depth implies an unlikely interconnection of the pores from the surface down to 20-25 km. In any case, the pore-fluid pressures could only be hydrostatic in such a model.