Chapter 3

Geological faults in the CSZ and their correlation with earthquakes

Index

3.1 Introduction

The relation between the earthquakes of the Charlevoix Seismic Zone (CSZ) and the St. Lawrence rift faults is a major element in the seismic zoning of Eastern Canada. It is generally assumed that CSZ earthquakes cluster along or between the St. Lawrence paleo-rift faults (also called Iapetan faults; Anglin, 1984; Adams et al., 1995). This relationship is obscured by the incomplete knowledge of geological faults, especially under the St. Lawrence River. In this chapter, the geological faults of the CSZ are described, in some cases for the first time, using geophysical, geological and remote sensing information. This analysis relies on new geoscientific information (gravity, radar images, seismic profiles) and more powerful analysis techniques of the potential fields. Based on this information, the spatial correlations between geological faults and earthquakes are reexamined. The chapter is divided in three parts. First, the geoscientific data sets and the interpretation methods are presented. Second, regional faults are described with newly-acquired remote sensing imagery, potential fields, seismics and geological mapping. Finally, the relationships between the earthquakes and faults of the CSZ are discussed. Figure 3.1 displays the main place names used in text.

3.2 Data sets and methodology

Three major types of information are used to refine knowledge of the positions of geological faults: remote sensing imagery (airborne and satellite-borne radar); geophysical data sets (seismics, magnetics and gravity); and field mapping of the north shore (Roy, 1978; Rondot, 1979; 1989).

3.2.1 Remote sensing imagery

On land, remote sensing imagery, mainly radar, is central to analysis and interpretation. Passive remote sensing systems, such as SPOT or Landsat, record the energy reflected by the Earth's surface at frequencies roughly equivalent to those detected by our eyes. Landsat images of the region were examined but did not yield any structural information not already described by Rondot (1979). Radar sensors, on the other hand, are active systems that send a microwave pulse towards the Earth's surface and measure the amount of energy reflected (Figure 3.2 ). Unlike passive systems, radar is not affected by atmospheric conditions, such as darkness, clouds, rain, dust, or haze. For geological applications, the incidence angle of the beam enhances terrain topography, often related to structural lineaments.

Over the course of this study, radar coverage of the CSZ became available: Canadian airborne SAR (Synthetic Aperture Radar), the satellite-borne ERS-1 (European Remote Sensing Satellite), JERS-1 (Japanese Earth Resources Satellite) and Radarsat (Canada) (Figure 3.3 ), each one with its own look directions and incidence angles. Most images were corrected geometrically using ground reference points, an acceptable approximation for the low relief areas near the centre of the impact structure but approximate for the steep elevation gradients of the Laurentians. The Radarsat data were ortho-corrected by the Canada Centre for Remote Sensing, resulting in accurate lineament orientations (T. Toutin, pers. comm.). The various radar images partly overlap over the crater.

On the radar images, the most striking regional feature is the impact crater. Inside the crater, the topography is generally subdued, due to the intense fracturing that favoured erosion. In the Laurentians, north-south lineaments are sub-parallel to the main direction of the glacier flow (Roy et al., 1993; De Sève et al., 1994), suggesting preferential glacial erosion along that trend. Other smaller lineaments match the regional lithologies.

3.2.2 Geophysical data sets

Geophysical data sets include the magnetic and Bouguer gravity fields and a series of seismic profiles acquired in the early 1970's, under the leadership of the Société Québécoise d'Initiatives Pétrolières (SOQUIP).

3.2.2.1 Magnetics

Regional magnetic maps show the spatial distribution of geological units with different magnetic susceptibilities. Sedimentary rocks have the lowest average susceptibility and basic igneous rocks, the highest (Telford et al., 1976). In the CSZ, these differences in magnetic susceptibilities result in the north shore having a highly variable magnetic field, due to the outcropping Precambrian rocks, and the south shore having a very smooth field due to the deep Precambrian basement under the almost magnetically transparent Appalachian nappes (Figure 3.4 ). The magnetically contrasting areas correspond to variations in Precambrian lithologies and in Precambrian basement depth under the Appalachians. Steps in the Precambrian basement depth can correspond to normal faults. Our depth estimates are not dependent on the magnitude of the anomaly, but rather on the anomaly shape, especially the sharpness. An outcropping Precambrian unit will produce a sharp anomaly, whereas the same body will have a low frequency signature if buried under a few kilometres of Appalachian nappes.

With the Euler deconvolution technique, steep gradients in the potential fields can be interpreted as sources of various shapes and depth (Figure 3.5 ). Assuming a shape for the source of the magnetic anomaly, the Euler deconvolution provides the depth and the location of the source (Thomson, 1982; Reid et al., 1990). The technique is based on Euler's homogeneity equation

where (x0, y0, z0) is the source position, whose total field (T) is detected at (x, y, z), with a regional field value (B). The factor (N) is the order of homogeneity, also called the structural index. The structural index depends on the geometry of the causative body: for magnetics, a sub-vertical contact corresponds to N = 0; for gravity, N = 1. Other possible shapes for the magnetics are: sphere (N=3), vertical pipe (N=2), and dike (N=1). The method, explained in more detail in Reid et al. (1990), is as follows. First, the gradients T/x, T/y, T/z are calculated from the gridded data. Second, a square window within the grids of gradient values is determined (10 X 10 for example). Third, for a given structural index, all points in the window are used to solve Euler's equation (3.1) for a source position (x0, y0, z0) and a background value B using least-squares estimates. The solution is kept if the depth uncertainty is less than a set proportion of the estimated depth. Four, the previous steps are repeated for all possible window positions over the map area. Five, maps of the solutions are produced with a symbol showing the solution's depths.

Euler deconvolution is very sensitive to the gradients of the field and is very useful to identify contacts, or body edges. Reid et al. (1990) provide examples of its use to delimit faults and geological units. With its down-faulted Precambrian basement under the St. Lawrence River, the CSZ geology is ideally suited for this method. The possibilities offered by the Euler solutions were demonstrated with the gravity field (Keating, 1998). Using the CSZ magnetic map, Euler solutions with N=0 provides an estimate of the location and depth of a sub-vertical contact separating lithologies with contrasted magnetic susceptibilities. Hence, where the Precambrian outcrops, Euler solutions should be shallow. Under the St. Lawrence River, along with the Precambrian basement, Euler solutions progressively deepen. Consequently, the Euler solutions provide estimates of the position and depth of the Precambrian contacts.

The Euler solutions for the CSZ are shown in Figure 3.5 . As one can see, most of the shallow solutions are found on, or near, the north shore, where the Precambrian either outcrops, or is near the surface. The solutions can be used to map the depth variations of the Precambrian basement. From seismic reflection profiles, no Euler solutions exist where the Precambrian is deeper than about 6 km.

3.2.2.2 Gravity and rock densities

Bouguer anomaly maps represent the gravity field, which depends mainly on the crustal distribution of rock densities. Until 1994, the CSZ Bouguer anomaly map was based solely on 300 land measurements, leaving out the most seismogenic part of the CSZ, over the St. Lawrence River. Following a request from the Seismicity Section, the GSC gravity group completed the coverage in August 1994. In the CSZ, the St. Lawrence River is too shallow and narrow for a conventional ship gravity survey. Consequently, a LaCoste and Romberg dynamic gravity meter controlled by a portable computer was installed on a small boat (Figure 3.6 ; Lamontagne et al., 1995). Differential Global Positioning System (GPS) phase measurements were used to compensate uneven boat motions in the strong tidal currents. To minimize the corrections due to the rotation of the Earth, survey lines were mostly oriented north-south and east-west. Upstream from the CSZ, where navigation is constrained by water depth, the north and the south channels were surveyed. With this survey, the offshore coverage is now better than that on land (Figure 3.7A ).

To model gravity anomalies, rock densities were measured on 288 samples from the rock collections of the GSC and the Québec Department of Natural Resources, and some hand specimens. Densities were determined using standard techniques for non-porous rocks (weight in air and water). Our results are comparable to other studies (Table 3.1; Figure 3.7C ). The densities of the Grenville rocks show considerable scatter, due to the highly variable lithologies. Our mean density value for the Grenville, 2.71 g·cm-3, is somewhat lower than elsewhere in the Grenville (2.74 g·cm-3, Keary and Thomas, 1979; 2.76 g·cm-3, Loncarevic et al., 1990; 2.85 g·cm-3, Feininger, 1993). The lower density may be due to the intense fracturing induced by the impact. The densities of the Ordovician limestones and the Appalachian nappes are clustered around a mean 2.65 g·cm-3. The density for the hard, compacted and boulder rich Quaternary sequence was established at 2.3 g·cm-3 (B. Long, pers. comm.). For modeling purposes, density contrasts with the Grenville were -0.15 g·cm-3 for the Ordovician-Appalachian rocks, and -0.5 g·cm-3 for the Quaternary sequence.

3.2.2.3 Seismic profiles

During the early 70's, SOQUIP lead a series of offshore seismic reflection surveys to define the hydrocarbon potential of eastern Québec (Figure 3.8 ). SOQUIP kindly provided paper copies of the profiles, and later on, the digital data from one seismic line for reprocessing (line 13). The number of folds generally determine the quality of the seismic profiles (Table 3.2). In most profiles, seismic energy trapped in the shallow water of the St. Lawrence River produced multiple reflections that masked deeper reflectors. Most profiles show sub-horizontal reflectors, some in the Quaternary deposits, some in the underlying Precambrian basement and in the Appalachian nappes. The Precambrian-Appalachian interface is generally difficult to identify, probably due to the small acoustic impedance contrast between the indurated Ordovician carbonates and the underlying Precambrian. In the figures and discussions, a seismic P-wave velocity of 6.0 km/s is used to convert time to depth (from seismic refraction results; Lyons et al., 1980).

Table 3.1: Rock Densities

Rock units Number of samples Mean

g·cm-3

Standard Deviation g·cm-3 Minimum

g·cm-3

Maximum

g·cm-3

GRENVILLE

205 2.71 0.16 2.52 4.05
Granite 10 2.66 0.05 2.61 2.78
Anorthosite 6 2.74 0.05 2.67 2.85
Charnockite-mangerite 10 2.75 0.10 2.64 2.81
Gabbro 6 2.95 0.19 2.66 3.23
Gneiss 35 2.77 0.13 2.64 3.14
chap3.htmAPPALACHIANS 21 2.65 0.05 2.55 2.75
Ordo. LIMESTONES 31 2.67 0.04 2.54 2.73



Table 3.2 Seismic profiles of the CSZ



Line No. Company-Date No. of folds Qual Source Appr.

Length (km)(1)

Remarks.
A SOQUIP - 1971 12 (SW)

24 (NE)

A Vibroseis 88 Line in two sections. One part inside the crater was never processed.
13 SOQUIP - 1971 24 B Vibroseis 13 Line was reprocessed.
16 SOQUIP - 1971 12 B Vibroseis 9
17 SOQUIP - 1971 12 B Vibroseis 8
18 SOQUIP - 1971 12 B Vibroseis 10
21 SOQUIP - 1971 12 B Vibroseis 9
22 SOQUIP - 1971 12 A Vibroseis 12 Clear and continuous PC-Appalachian boundary at 1.0 sec on most of the southern part of the profile.
23 SOQUIP - 1971 12 B Vibroseis 13 Structures in the basement. The north channel shows discontinuous but poorly defined reflectors from 0.20 to 0.40 sec.
27A CDRO- 1971. 1 C Air Gun 14
28 CDRO, 1971. 1 C Air Gun 8
30 CDRO, 1971. 1 B Air Gun 16
31 CDRO, 1971. 1 B Air Gun 16
35 CDRO, 1971. 1 B Air Gun 19
37 CDRO, 1971. 1 B Air Gun 19
42F CDRO, 1971. 1 C Air Gun 6
42N CDRO, 1971. 1 B Air Gun 11
130 SOQUIP-1972 6 A Air Gun 15
140-A SOQUIP - 1972 6 B Airgun 8

Note: CDRO: Central-Del Rio Oils Ltd

1. Length inside the map area.


This velocity, acceptable for the Precambrian, is too high for the Appalachian (5.6 km/s) and Quaternary sequence (2 km/s). Thus, near surface reflectors in the Quaternary and Appalachian sequences are shallower than suggested by the depth values on the right of the profiles. The profiles are discussed in more detail below.

The marine vibroseis seismic line 13, shot across the St-Lawrence River in 1970, was reprocessed using up-to-date processing techniques. The initial seismic section showed up to 4 seconds (TWT) of data contaminated with water-bottom related multiples (energy bouncing between the bottom of the river, and/or within the sedimentary layers beneath, and the top of the water column). The seismic profile was reprocessed to image upper-crust and mid-crustal reflectors, which can be seismogenic faults, and to refine the regional geological model.

To maximize the possibility of finding deep reflectors, an extended correlation of the raw seismic records and the marine vibroseis sweep was performed, yielding a 13 second long data set (TWT). In theory, this procedure could image reflectors down to a depth of 30 km. At first, the processing sequence involved the extended correlation and integration of the survey geometry. The traces were inspected one by one to discard noisy recordings and to pick the first breaks. From a spectral analysis, most of the signal was contained between 8 and 55 Hz.

Minimum phase spiking deconvolution, followed by predictive deconvolution based on first break energy, were performed on shot gathers to compress the signals, level the spectrum and remove water-bottom related multiples. Velocity analysis turned out to be the most important processing step. Semblance and constant velocity analyses were performed. Due to the presence of strong multiples, picks were made on constant velocity stacks. Best seismic velocities vary between 1500 m/s in the unconsolidated sediments to more than 5000 m/s in the Precambrian shield. The 24-fold data set was stacked and coherency filtering was applied. The resulting section is shown in Figure 3.13G and Figure 3.13H , and discussed in section 3.2.4.

3.2.3 Testing the methodology on the magnitude mN 5.1 Cap-Rouge earthquake

On November 6, 1997 a magnitude mN 5.1 earthquake occurred near Cap-Rouge, in the western suburbs of Quebec City (Nadeau et al., 1998). Since the depth to basement is well-assessed by high quality SOQUIP seismic profiles, this earthquake occurrence provided an opportunity to ground truth the methodology developed for Charlevoix. Geologically, the Quebec City region is very similar to the CSZ; it is an area where the Appalachian nappes were thrust over the rifted Precambrian.

Since the Appalachian nappes are nearly magnetically transparent, the magnetic field partly depends on the depth of the Precambrian Shield. Where the Precambrian Shield outcrops, such as to the North and West of Quebec City, the magnetic field is made up of very high frequency variations in its intensity. Where the Precambrian basement is deeper, such as under the south shore of the St. Lawrence River, the high frequency variations disappear, leaving a smooth field with low frequency variations. This change in wavelength of the magnetic anomalies is due to an abrupt change in the depth of the Precambrian rocks, most likely due to large throw on normal faults.

The correlation between the magnetic data and the normal faults inferred from the seismic profiles is good. These faults, which have a vertical throw of 2 km or more, are easily identified in the magnetic data. In general, they correspond to elongated magnetic anomalies whose response is slightly higher on the upthrow side than on the downthrow side. They trend at about N45oE.

The Cap Rouge epicentre is located on an elongated magnetic anomaly which is a normal fault of the Precambrian basement with vertical slip exceeding 2 km (Figure 3.9). If the faults dip steeply towards the southeast, as it is generally assumed for the St. Lawrence rift faults, the Neuville fault, located to the NW, may be the reactivated fault, an interpretation supported by the focal mechanism (Nadeau et al., 1998). The mapped Neuville fault is imaged in the magnetics and in the seismics, with a 2 km throw. The good correlation between the solutions of the Euler deconvolution and the seismics support the validity of our integrated approach to delineate normal faults in the CSZ, a region where seismic profiles are of lesser quality.

3.2.4 Profiles across the CSZ

To define the geological structures along the St. Lawrence River, four profiles were defined in the CSZ, each one with measured and computed Bouguer gravity anomalies, crustal density model, Euler solutions, structural interpretation and one or more nearby seismic profiles (Figure 3.10 to Figure 3.14). Gravity modelling has shown that the Quaternary deposits controlled most of the high frequency variations, masking the signal from the Appalachians. The thickness of the Quaternary was resolved using the seismic reflection profiles. The longer period variations of the gravity field are related, for the most part, to the depth of the Precambrian-Appalachian interface. Along the north shore of the St. Lawrence River, there appears to be a change in the density of the Precambrian basement, possibly related to the St-Laurent fault. For this reason, the depth of the Precambrian basement there was based on the Euler solutions. The structural interpretations include information from gravity, magnetics, and seismics and display major faults and structures, with approximate fault dips (sub-vertical) and basement depth.

The four models are presented in Figure 3.11A - 3.14. The Ile-aux-Coudres profile presents many complications, modeling Ile-aux-Coudres being one of them. Near the north shore, there is a sharp gradient in the gravity field, which is interpreted to be due to the combined Quaternary sequence, Ordovician-Appalachian sequence, and possible lower density of the Precambrian basement. Constraints for Ile-aux-Coudres are the depth-to-basement clearly seen on the land profile (Figure 3.11F), and the interpreted step-like structure southeast of the island (Figure 3.11G / H ). The island appears to rest on an important normal fault, shown in the Euler solutions. South of the island, the Shield appears to deepen (gravity low with no gradient in the magnetics), while near the south shore, it becomes shallower (Euler solutions and gravity). These suggest the presence of an offshore graben. Under the south shore, there is a large step in the basement, interpreted from the gravity low and the lack of magnetic gradient. The modelled depth of 8 km is similar to that suggested by seismic refraction and reflection profiles (Lyons et al., 1980).

The crater profile presents characteristics similar to the Ile-aux-Coudres profile, i.e., a central portion with no magnetic gradient, and a possible shallower basement towards the south shore. The gravity data suggest variations in the depth of the Precambrian basement under the St. Lawrence River.

The Line 13 profile appears to match a relatively smooth Appalachian-Precambrian interface for the central portion (Figure 3.13A-D). This interpretation is supported by the reprocessed seismic line (Figure 3.13G and 3.13H). There is a suggestion of a step fault, a few kilometres offshore from the north shore, that brings the Precambrian to a depth of about 2 km. Again, under the south shore, a large normal fault is interpreted from the gravity and the magnetics.

The Ile-aux-Lièvres profile is simpler than the previous profiles (Figure 3.14A-D). A series of three steps in the Precambrian basement fits the gravity field and the magnetic gradient.

3.3 Geological faults of the CSZ

Based on the various geoscientific data sets, the main structural aspects of the CSZ can now be outlined. Numerous lineaments exist in the radar images and the geophysical data sets. On land, most faults are described in geological reports (Roy, 1978; Rondot, 1979; 1989), while offshore, most structures are interpreted for the first time. In this section, regional faults are discussed first, followed by the local structures, grouped by sub-areas with similar characteristics: northeast, central and southwest (Figure 3.15). Figure 3.16 shows the adopted fault model for the Precambrian structure of the CSZ.

3.3.1 Regional Faults

The CSZ is cut by the St. Lawrence paleo-rift system, a series of regional normal faults that run parallel to the St. Lawrence River (Kumarapeli, 1978; Rondot, 1979; also called Iapetan normal faults: Wheeler, 1995). Most faults that outcrop on the north shore of the St. Lawrence River, such as the St-Laurent and the Gouffre NW faults are described in geological reports (Rondot, 1979; 1989). Others lie beneath the St. Lawrence River where their existence is revealed by geophysics.

3.3.1.1 The St-Laurent fault

The St-Laurent fault is the most prominent CSZ structure in radar imagery and in the potential field maps. With a length exceeding 150 km and a vertical throw of 2 km (Rondot, 1970), it is a major fault of the St. Lawrence paleo-rift system. Created as an Iapetan fault, the St-Laurent fault was reactivated during the Taconian orogeny (Rondot, 1979). The St-Laurent fault is probably made up of multiple fault segments, as suggested by its jagged course between the Ile d'Orléans and the Saguenay River. The gravity and magnetics suggest three main subdivisions of the St-Laurent fault: a southwest and northeast areas separated by a middle portion centred on the impact structure.

In the central uplift of the impact structure, the St-Laurent fault is oriented N40oE and dips steeply (70o) towards the SE (Rondot, 1979). Here, the St-Laurent fault separates mangerite/charnockite assemblages to the NW from granodioritic to granitic migmatite assemblages to the SE. The lithological change is reflected in the magnetic field, where the generally high values to the NW abruptly change to low values SE of the fault. The Bouguer gravity field gradually changes from high values to the NW (-30 mgal or more) to low values to the SE (-40 to -60 mgals). On land, the gravimetric low reflects solely the lower density rocks of the Precambrian basement, while offshore, it combines with the low densities of the platform sediments, the Appalachian sequences and the unconsolidated Quaternary sediments. The density change across the fault makes gravity modelling more difficult, as described above for the Ile-aux-Coudres profile.

To the SW of the impact, the St. Laurent fault is parallel to the north shore (oriented N40oE). It may line up with a series of Euler solutions in the magnetics under Ile d'Orléans. There, seismic reflection profiles indicate that, over a 10 km distance, a series of normal faults oriented north-south brings the depth to Precambrian basement from about 2.1 km to about 4.7 km (SOQUIP, pers. comm.).

To the NE of the impact, the St-Laurent fault is not as prominent in the geophysics. The St-Laurent fault never comes ashore there and its position is assumed to follow the north shore from the Gros Cap-à-l'Aigle to Tadoussac (Rondot, 1979). In contrast to the situation within the impact crater, the magnetics do not show any change across the fault. This can be explained by a smaller throw than to the SW or by a much thinner sedimentary cover. The fault may die out NE of the impact.

3.3.1.2 The Gouffre NW fault

The Gouffre NW fault (sometimes called Gouffre) is named after a small creek that flows into the Gouffre River near Baie-St-Paul. To the SW of the crater, this fault is a prominent lineament on the radar images and in the magnetics (Figures 3.17). Within the impact crater and to the northeast, the fault position is not obvious. Geologically, its position inside the impact is based on stratigraphic evidence that suggests that the fault existed at the time of deposition (Rondot, 1968). To the NE, its extension is based on lineaments with similar trends. Compared with the St-Laurent fault, the Gouffre NW fault appears to have a small throw with little or no geophysical signature inside the CSZ. Its position is approximate within the impact and uncertain to the NE.

3.3.1.3 The Charlevoix fault (new feature)

A few kilometres offshore from Ile d'Orléans to Tadoussac, a normal fault parallels the St-Laurent fault. The fault separates a 2-8 km wide Precambrian basement plateau from a deeper basement to the southeast. In the magnetics, the plateau corresponds to high frequency anomalies, while the deeper basement correspond to low frequency anomalies. The fault's gravity signature is less conspicuous due to the proximity of the St-Laurent fault and to the sediment-filled valleys. In the southwestern portion of the fault, the SE shallow-dipping Precambrian-Appalachian plateau is imaged at 3.0 km depth (Figure 3.18). Under Ile-aux-Coudres, the fault may also correspond to the truncated Precambrian reflector of Figure 3.11E. The position of the Charlevoix fault coincides with the overlying Logan's thrust, suggesting that the footwall may have acted as a buttress, or that the fault was reactivated by the weight of the advancing nappes.

3.3.1.4 The South Shore fault (new feature)

Along the south shore of the St. Lawrence River, the gravity and the magnetic fields suggest a large normal fault in the Precambrian basement. On all profiles, gravity modelling suggests that the Precambrian depth drops from about 3-4 km to 7-8 km. The latter depth value corresponds to the seismic refraction results (Lyons et al., 1980). In the southern part of the fault, slightly outside the CSZ, the Bouguer anomaly low implies an even deeper basement than to the NE . The gravity suggests that this fault may terminate near Ile-aux-Grues.

3.3.2 Northeast zone

In the NE section of the CSZ, the NS trend of lineaments seen on land continues under the River. While the St-Laurent fault trends mostly NE, in this northern portion of the "Jacques Cartier block", a large proportion of lineaments trend NNE (Roy et al., 1993). Under the St. Lawrence River, a similar trend is found in the Euler solutions (Figure 3.5C). Most of these lineaments appear to relate to Precambrian lithologies.

South of Ile-aux-Lièvres and oriented parallel to the St. Lawrence River axis, a normal fault is suggested by the gravity and by the Euler solutions (Ile-aux-Lièvres fault). Gravity modelling (Figure 3.14A-D) suggests a 2 to 3km throw for this fault, i.e. comparable to the St-Laurent fault. This is supported by the magnetics and the Euler solutions that get progressively deeper southeast of this fault. The fault runs from the outer rim of the impact crater, near La Malbaie, and extends beyond the NE boundary of the CSZ.

The Palissades fault (PAL), the southernmost fault of the Saguenay graben, is supposed to correspond to the NE extension of the seismic zone (Anglin, 1984). This structure of regional extent can be seen on ERS-1 imagery to the NW, outside the seismic zone. The Palissades fault my continue offshore, as suggested by the Euler solutions and the gravity.

3.3.3 The impact structure faults

The 55 km diameter Charlevoix astrobleme of Devonian age ( 350 Ma; Rondot, 1979) is a "complex" impact structure characterized by a central uplift surrounded by an annular trough and a peripheral down-faulted structurally complex rim (Grieve, 1993). Across the structure, the nature and the extent of the block movements vary from upward movement in the central uplift to subsidence in the rim. The faults directly created by the impact are found near the central uplift, where the shock has been most intense, and where the vertical movement can reach 9 km (Robertson, 1975). This area is made up of a series of blocks separated by normal faults (Roy, 1978). In the rim, structures are a series of locally sub-parallel normal faults of about 11 km depth extent (Rondot, 1994; similar to the Siljan impact structure of Sweden; Juhlin and Pedersen, 1987). Whereas most rim faults are conspicuous in the radar images, inside the Charlevoix impact structure, lineaments are less clear (Figure 3.3D). This is possibly due to the erosion of highly fractured rocks. Beneath the zone affected by the impact, faults without surface expressions probably exist.

Beneath the St. Lawrence River, geophysical data suggest structures possibly related to the impact. The impact structure deformation is illustrated by the SW part of seismic Line A (Figure 3.19). Moving towards the impact structure, a series of sub-horizontal reflectors is present until the 60 km radius line is reached. There, a small basin, possibly controlled by shallow faults, is found near the theoretical position of the peripheral graben.

The Ile-aux-Coudres corresponds to an area of elevated basement, an interpretation supported by seismic reflection profiles and by higher magnetic and gravity values (Figure 3.11A-D). East of Île-aux-Coudres in the centre of the crater, the smooth magnetic field and the low gravity values suggest a deep Precambrian basement, which is interpreted as a graben structure on Figure 3.11A-D and Figure 3.12A-D.

3.3.4 The SW section

To the southwest of Ile-aux-Coudres, the gravity and the magnetics suggest a deep basin, probably controlled by a series of normal faults. Under Ile-aux-Grues, where a gravimetric low is found, a SOQUIP seismic line suggests a Paleozoic-Precambrian interface at about 5.5 km depth. This basin is controlled by at least two normal faults, one oriented nearly E-W inferred from the seismics and one parallel to the north shore, inferred from the Euler solutions.

3.4 Faults and seismicity of the CSZ

This section examines the relationships between the geological faults described above and the CSZ earthquakes. Earthquake magnitudes and fault dimensions for intraplate earthquakes may be related through proposed scaling laws (Table 3.3). During the period 1977-1997, only eight earthquakes exceeded magnitude mN 4.0. Thus, most CSZ earthquakes reactivated relatively small fault surfaces, that may not have a surface expression. As described below, regional faults bound highly seismically active blocks. Geographically, the level of activity varies: whereas the highly fractured central zone has frequent, but small magnitude, earthquakes, the SW and NE areas, with their long normal faults, produce infrequent, but larger, events.

Table 3.3 Fault dimensions for stable continental earthquakes of various moment magnitudes (M; after Johnston,1993)

M Length (L) Width (W) slip (d)
0.0 5.8 m 5.8 m 1 mm
1.0 18 m 18 m 3 mm
2.0 58 m 58 m 9 mm
3.0 183 m 183 m 2.9 cm
4.0 577 m 577 m 9.1 cm
5.0 1.8 km 1.8 km 29 cm
6.0 8.0 km 4.1 km 91 cm
6.5 15 km 7.0 km 1.6 m
7.0 30 km 11 km 2.9 m

3.4.1 Faults of regional extent

The St-Laurent fault (SL) appears to bound the earthquake distribution, especially in the Les Eboulements-Cap-à-l'Aigle section. There, most shallow (z < 10 km) earthquakes locate to the NW of a south-east boundary, which is interpreted to be the sub-vertical SL (Figure 3.20). Below 13 km depth, most earthquakes occur SE of the fault in a central corridor bound by two weakly seismic areas. Near Ile-aux-Coudres, earthquakes are generally shallower than 10 km depth northwest of the fault, whereas to the SE, deeper earthquakes are found (down to 15 km depth). To the NE of Cap-à-l'Aigle, the St-Laurent fault does not appear to affect the earthquake distribution. There, another rift fault, the Rang-Ste-Mathilde fault, separates shallow activity to the west from deeper activity to the east.

To the NE, the Charlevoix Fault (CH) may be currently reactivated by the earthquake activity. The NE area is one of the two zones where magnitude 4.0 earthquakes in the period 1924-1978 occurred, including the M 6.2 1925 Charlevoix earthquake (Stevens, 1980). Under the river, most earthquakes occur within a river-parallel planar volume which includes most mN 4.0 CSZ events, including the magnitude 5.0 earthquake of 1979, and the recent October 28, 1997 mN 4.7. A cross-section through the 1978-1997 hypocentres shows a remarkable alignment with a dip of about 50 (Figure 3.21), in agreement with the focal mechanisms and depths (10 km) of the 1925 M 6.2 and the 1979 mN 5.0 earthquakes (Bent, 1992; Lamontagne, 1997). Inside the impact crater, the Charlevoix Fault may be a boundary to the earthquake activity rather than the active structure. Under Ile-aux-Coudres for example, events northwest of the fault are shallower than to the southeast.

To the SE, and possibly to the NE, regional faults act as boundaries. To the SE, the South Shore fault (SS) delimits the seismicity, without being active itself. To the NW, the Gouffre NW (GNW) fault has been proposed to limit the earthquake activity (Anglin, 1984). This correlation may only be apparent: the position of the Gouffre NW fault is approximate and hypocentres appear independent of a clear and linear NW boundary. Remote sensing images and aeromagnetics also suggest a NE lineament that lines up with a portion of the St-Laurent fault (named Crater Lineament, CRL; Figure 3.17).

3.4.2 The NE portion of the CSZ

In the NE, most earthquakes appear to correlate with faults. As described above, offshore hypocentres line up with the Charlevoix fault. In addition, this area also has multiple lineaments oriented NNE. It is possible that some faults with this orientation are reactivated (such as the 1989 doublet of Wetmiller and Adams, 1990). Anglin (1984) noticed that the "B" fault (Figure 1.4) matches numerous surface-projected hypocentres. This fault is one of many faults, parallel to the St. Lawrence River, seen in ERS-1 imagery. One of these lineaments could be the Rang-Ste-Mathilde fault (RSM) of Rondot (1989). Offshore, the Ile-aux-Lièvres fault (L) is not seismically active as it lies to the southeast of the activity previously linked with the Charlevoix fault. The NE extremity of the CSZ contains mainly deep events (> 20 km). The earthquake activity appears to terminate at the offshore extension of the Palissades fault (PAL).

3.4.3 The impact structure

The relationship between CSZ earthquakes and the impact structure remains unclear. The majority of CSZ earthquakes locate within the volume of the impact structure defined by the 60 km diameter of the outer rim and the 10 km depth extent of the faults. Two focal mechanisms correspond to the trend of the rim faults (Lamontagne and Ranalli, 1997). This suggests that some impact-related faults are currently reactivated. One of the faults of the astrobleme ("A" fault; Figure 1.4) was thought to cause an offset to the seismicity (Anglin, 1984). Within the impact structure, clouds of earthquakes without any particular alignment are seen (except near St-Joseph-de-la-Rive, where micro-earthquakes define an alignment, which does not correlate with radar and geophysics lineaments). Under the river as well, the highest concentration of activity of the CSZ corresponds to the highly fractured graben, possibly created by the impact. It is possible that these highly fractured areas give rise to numerous low magnitude events with highly variable focal mechanisms. The central uplift of the astrobleme, where fracturing and faulting have been the most intense, however, is not more particularly active.

3.4.4 The SW section

Earthquakes occur very infrequently in this area and their correlation with faults is more difficult to establish. Since the area is outside the seismograph network, hypocentral depths are not as precise as in the rest of the CSZ. The area is important, however, since it is one of the two areas where magnitude 4.0 events of the period 1924-1978 have occurred (Stevens, 1980). Iapetan faults are known to exist there, with some possibly at high angle to the river axis. With conditions similar to the rest of the CSZ, the lower rate of earthquake occurrences is enigmatic. The normal fault near Ile-aux-Grues, or possibly the impact crater rim, may act as a boundary to the higher level of activity to the NE.

3.5 Discussion

This analysis has brought to light the main structural patterns of the CSZ. In the NE area, the Precambrian basement is cut by a series of long normal faults parallel and at high angles to the river axis. In the centre, the impact crater controls the faults, at least down to about 10 km depth. Finally, in the SW zone, the Precambrian basement, deep under the Appalachians, is cut by normal faults parallel and at high angles to the river axis. The CSZ is a transition zone between the SW part, with its steeply-dipping faults with large normal throw, and a NE part, with its en echelon normal faults with progressive deepening of the basement under the Appalachians.

In Anglin (1984), CSZ earthquakes were correlated with five geological faults (Figure 1.4): St-Laurent; Gouffre NW; Palissades and faults "A" and "B". These correlations were established on the basis of projections of the hypocentre distribution to the surface along a trend (N52E) and a dip (70SE) that provided the tightest cluster of hypocentres. The main assumption was that most earthquakes occur on planar structures that cross the entire seismic zone. After some 20 years of micro-earthquake recording, it is now apparent that earthquakes do not concentrate only on planar structures. Locally, earthquakes define depth-limited volumes of enhanced activity. In this context, most large faults seem to bound these active volumes rather than being active themselves.

Three Iapetan faults bound seismically active sub-zones: the St-Laurent, South Shore and Charlevoix faults. These faults separate blocks possibly with different fracturing levels. The foot-wall of the St-Laurent fault, for example, has most of the micro-seismic activity, whereas the hanging-wall has weaker and deeper activity. Conceptually, the foot-wall may be highly fractured, and/or subject to high pore-fluid pressures. Iapetan faults do not seem particularly active (for micro-earthquake activity): hypocentres do not concentrate along the St-Laurent and the South Shore faults. The best correlation between an Iapetan fault and earthquakes is found in the NE part of the zone. There, hypocentres define a narrow steeply-dipping volume, possibly the Charlevoix fault. Inside this volume, earthquakes may or may not occur directly on the main fault, as shown by the variability in the focal mechanisms.

Contrasting with its conspicuous surface expression, the SE-dipping St-Laurent fault itself is not particularly active. However, it bounds the earthquake activity over part of the CSZ. In the central part, most of the deep (> 20 km) events occur to the east of the St. Laurent fault. Our analysis suggests that the St-Laurent fault acts more like a boundary to the activity than as an active structure. Similarly, the South Shore fault bounds the seismicity to the SE. To the NW, a fault-controlled boundary is not as clear.

The relation between the Charlevoix impact structure and the current earthquake activity has been subject to much debate. On the one hand, those who disclaim any relationship observe that earthquakes do not distribute over the whole impact structure, but concentrate in zones parallel to the St. Lawrence paleo-rift faults (Adams and Basham, 1989). In addition, most impact structures are currently aseismic (Salomon and Duxbury, 1987). According to others, however, the structure might contribute to the inherent weakness of the region (Leblanc and Buchbinder, 1977; Lamontagne, 1987; Kumarapeli, 1987). "Impacting has undoubtedly complicated the pattern of potential failure in the area and consequently may have created an environment for stress concentration in an otherwise throughgoing Iapetan fault system" (Kumarapeli, 1987). At present, it is unclear why earthquakes do not occur over the whole impact structure. Instead, they appear to concentrate in two bands elongated along the St. Lawrence. Some earthquakes, however, appear to occur on faults, possibly created as Iapetan faults, but clearly reactivated by the impact (Lamontagne and Ranalli, 1997). Interestingly, most of the large events tend to concentrate at both ends of the CSZ, outside the impact structure. Within the impact structure, the highly fractured Grenville releases strain energy in small earthquakes only (mbLg 3.5), with highly variable fault plane orientations.


> > > To Chapter 4


Figure Captions

Figure 3.1 Location map of the CSZ surroundings and geographic entities mentioned in the text.

Figure 3.2 Principle of the side-looking radar system (Radarsat in this case). Each beam is defined by the area it covers and the level of detail (resolution) available.

Figure 3.3 Remote sensing radar images of the CSZ. On these various radar images, regional and local lineaments can be recognized: (A) Radarsat image of the St. Lawrence River between Quebec City and Tadoussac that shows the numerous lineaments in the Precambrian Shield, north of the river (courtesy of the Ministère des ressources naturelles du Québec). (B) JERS-1 image of the area to the west of the CSZ; (C) ERS-1 image of the CSZ; (D) Airborne SAR with West-facing direction of the beam; (E) Airborne SAR with South-facing direction of the beam; (F) Enlarged view of E that shows the high resolution of the airborne-radar system. (G) Chromo-stereoscopic image that integrates a Radarsat image (texture) and topographic information (colours). Courtesy of Thierry Toutin, Canada Centre for Remote Sensing.

Figure 3.4 CSZ total magnetic field (colours) with shorelines (black).

Figure 3.5 Solutions of the Euler deconvolution (N=0) of the magnetic field. The circle sizes correspond to the relative depth of the solutions (shallow solutions are small). (A) All solutions; (B) Solutions between 0 and 500 m depth; (C) 0 to 1 km depth; (D) 1 to 2 km depth; (E) 2 to 3 km depth; (F) 3 to 4 km depth.

Figure 3.6 Photos taken during the August 1994 gravity survey: (A) The LaCoste and Romberg dynamic gravity meter, with the portable computer and the GPS receiver; (B) The launch used on the St. Lawrence River.

Figure 3.7 Gravity and density data. (A) The gravity stations in the CSZ (crosses) superimposed on the Bouguer anomaly map; (B) The Bouguer anomaly map produced with standard GSC contouring schemes; (C) Distribution of density values for the three main lithological groups of the CSZ.

Figure 3.8 Location map of the SOQUIP seismic reflection lines in the CSZ.

Figure 3.9 Euler solutions of the magnetic field in the Quebec City region (see description of solutions in Figure 3.5). The epicentre of the November 1997 mN 5.1 earthquake (star) lies on a NE-SW lineament (A). The Neuville fault (N) a paleo-rift fault dipping towards the SE may be the structure reactivated during the earthquake. Other magnetic lineaments (dashed blue lines) correspond to faults in the Precambrian basement evidenced in the SOQUIP seismic profiles.

Figure 3.10 Location of the four geophysical profiles discussed in text (red dotted lines) and presented in Figures 3.11 to 3.14. The yellow lines are the SOQUIP seismic profiles are presented along with the nearest profile.

Figure 3.11 Geophysical profile near Ile-aux-Coudres: (A) Bouguer anomaly (circles) and modelled field (red line); (B) Gravity model and density contrasts from the Precambrian; (C) Euler solutions within 5 km of the profile, plus locations of the seismic lines; (D) Structural model based on geophysical information; (E) Structural model with hypocentres located within 5 km on both sides of the profile; (F) Seismic profile on Ile-aux-Coudres with the red arrow indicating the surface of the Precambrian (SOQUIP, pers. comm.); (G) Seismic profile 140 on the St. Lawrence River; (H) Interpretation of the seismic profile.

Figure 3.12 Crater profile (A-E), with legend similar to 3.11A-E; (F) Seismic line 30 over the St. Lawrence River (top) and interpretation (below).The Quaternary sequence is evident, while deeper reflectors are not continuous.

Figure 3.13 Line 13 profile (A-E) with legend similar to 3.11A-E; (F) Original SOQUIP profile with arrows showing interpreted reflectors in the Precambrian shield; (G) Reprocessed Line 13; (H) Interpretation of G. The Appalachian sequence (yellow), possibly cut across by two faults (red) appears to have a sub-horizontal interface with the Precambrian. A number of reflectors can be seen in the Precambrian. The attitude of these reflectors changes with depth. A normal fault may exist to the NE of the line (right side).

Figure 3.14 Ile-aux-Lièvres profile (A-E), with legend similar to 3.11A-E; (F) Seismic line 35 (top) with interpretation (below).

Figure 3.15 The three main sub-regions discussed in text (NE, Impact crater (Central) and SW), and the main geological features of the CSZ.

Figure 3.16 Interpreted structures of the CSZ. Acronyms used: PAL: Palissades fault; RSM: Rang Sainte-Mathilde fault; SL: Saint-Laurent fault; CH: Charlevoix fault; L: Lièvres fault; SS: South Shore fault; G: postulated graben under the St. Lawrence; CRL: crater fault; GNW; Gouffre NW fault.

Figure 3.17 Major faults of the CSZ, mapped (red), interpreted (dashed blue), and lineaments (dashed yellow) in the Radarsat chromo-stereoscopic image (courtesy of T. Toutin, Canada Centre for remote sensing).

Figure 3.18 SOQUIP seismic line 22. The position of the Precambrian-Appalachian interface (red arrow) is at a depth of about 3 km (SOQUIP, pers. comm.). The syncline to the left of the profile corresponds approximately with the border of the impact structure.

Figure 3.19 SOQUIP seismic line A, south of Ile-aux-Coudres.

Figure 3.20 Geological faults and CSZ hypocentres for the period October 1977 to December 1997.

Figure 3.21 Cross-section across the 1979 M5 earthquake zone. Earthquakes are within 10 km of a line across the trend of the St. Lawrence River, and centered on the M5 event of 1979. Pure reverse faulting on the best hypocentre alignment is shown.