A Case Study from Northern Ontario, Canada

Bulletin of the Seismological Society of America, Vol. 98, No. 6, pp. 2828–2848, December 2008, doi: 10.1785/0120080134

Ⓔ

Intraplate Seismicity of a Recently Deglaciated Shield Terrane: A Case Study from Northern Ontario, Canada by Shutian Ma, David W. Eaton, and John Adams

Abstract

The economy of northern Ontario, Canada, is heavily dependent on mining, so accurate knowledge of seismicity is important for the safe design and operation of mines and other critical facilities, including a proposed underground repository for nuclear waste. In this study, we analyzed 537 cataloged earthquakes that occurred from 1980 to 2006. Seismicity is mainly concentrated in topographically elevated Archean terranes northwest of Lake Superior and in the James Bay and Kapuskasing regions. We analyzed waveforms to determine the focal depth for 331 recorded events, using the regional depth-phase modeling (RDPM) method coupled with surface-wave relative-amplitude analysis. The majority of events are shallow (< 6 km) and concentrated in areas of relatively high elevation (> 350 m), although in the eastern part this pattern breaks down and some deeper earthquakes (> 12 km) are observed. Based on a moving-window event-counting technique, we show that distinct spatial clusters of seismicity can be delineated that are statistically significant relative to background seismicity levels. A particularly active cluster is located within James Bay, where focal depths range from a few kilometers to more than 20 km. Another cluster near Kapuskasing contains deep-focus events and may occur along a hot spot track that runs through western Quebec. Near Dryden, a shallow (∼1 km) earthquake swarm concentrated in a 1 × 1 km region commenced in May 2002, faded, and then started up again in February 2003. Shallow mining-induced events are also common around Sudbury, a major world center for nickel mining. The overall pattern of seismicity appears to correlate with upper-mantle P-wave velocity anomalies, suggesting that lateral variations in mantle rheology may play a significant role in controlling intraplate seismicity of shield areas. It is also likely that crustal stresses caused by glacial isostatic adjustment are an important factor, although the correlation of seismicity with uplift rate is not as clear. Online Material: Focal depth solutions for earthquakes in northern Ontario, Canada.

Introduction Northern Ontario (Canada), a region extending northward from the Great Lakes to Hudson Bay, covers an area of ∼1:0 × 106 km2 in the continental interior of North America. The physiography of most of northern Ontario exemplifies a recently deglaciated shield terrane, with numerous lakes, extensive exposures of crystalline bedrock, and glacial till deposits of variable thickness. Although not as seismically active as other parts of eastern North America, there is, nevertheless, a significant level of intraplate seismicity (Fig. 1). Accurate knowledge of the distribution and rates of regional seismicity are important for the safe design and operation of numerous underground mines, dams, and other critical facilities, including a proposed underground repository for nuclear waste (Fenton et al., 2006).

Wetmiller and Cajka (1989) described the seismic activity of northern Ontario, mainly for the period from 1983 to 1987 subsequent to installation of the northern Ontario seismograph network for the Canadian Nuclear Fuel Waste Management Program. The majority of the earthquakes that they located in northern Ontario (1983–1988) fell within a broad zone in the eastern part of our study region (Fig. 1), suggesting an extension into northeastern Ontario of more active tectonics from the western Quebec seismic zone (WQSZ). Near-surface stress measurements (mostly in mines) and earthquake focal mechanisms show that much of the region is subject to high horizontal stress, oriented predominantly in the northeast–southwest azimuth (Wetmiller and Cajka, 1989; Adams and Bell, 1991; Adams, 1995).

2828

Intraplate Seismicity of a Recently Deglaciated Shield Terrane: A Case Study

2829

Figure 1. Seismicity of northern Ontario from 1980 to 2006, showing all cataloged earthquakes (537), of which 309 occurred from 1980 to 2003 (see the Data and Resources section). Green squares show active mines. Geological domains are denoted as follows: WB, Williston Basin; HBB, Hudson Bay Basin; THO, trans-Hudson orogen; Sup, Superior Province; MCR, midcontinent rift; SP, Southern Province; and GP, Grenville Province. A small section of the WQSZ extends into the study region.

Fenton et al. (2006) used a method of statistical spatiotemporal substitution to augment incomplete geologic data and a short seismicity record for a number of shield areas around the world. This approach allowed them to make a more reliable estimate for the rate of M ≥ 6 earthquakes in northern Ontario. Historical records for this region are relatively poor, extending for only the last 100 yr at most. Using their derived rate of seismicity, they were also able to make a statistical estimate of the potential for surface fault rupture through or near a site. These previous studies of seismicity parameters in northern Ontario have been hampered by relatively sparse seismic network coverage. Commencing in 2003, a major expansion of the seismograph networks in this region took place as part of the POLARIS and FedNor initiatives (Eaton, Adams, et al., 2005). This expansion, together with the advent of new techniques for focal-depth estimation from sparse regional data (e.g., Ma and Atkinson, 2006), has provided renewed motivation to investigate regional seismicity characteristics. Hence, this article aims to characterize spatial clustering patterns and focal-depth distribution of regional seismicity and, thus, to improve our understanding of the underlying causes of intraplate seismicity in shield areas that

have undergone deglaciation in geologically recent times. Here, we do not make extensive use of multievent hypocenter relocation methods such as HypoDD (Waldhauser and Ellsworth, 2000) to obtain a precise estimate of the fine-scale distribution of events primarily because of the large scale of our study, which focuses mainly on regional patterns of seismicity. In the section Tectonic Setting, we describe the tectonic setting of northern Ontario to provide a geologic context for interpreting patterns of seismicity. In the section Data and Methods, we apply a statistical approach to delineate spatial clusters of seismicity, including a number of small, ephemeral earthquake swarms. In addition, we describe two techniques to estimate earthquake focal depth based on analysis of regional depth-phase observations and the relative amplitude of high-frequency Rg waves. The section Characterization of Regional Seismicity contains a summary of regional seismicity patterns and a classification of seismicity into categories of sequences, clusters, and swarms. In the section Discussion, we discuss possible relationships between seismicity and stresses induced by glacial isostatic adjustment, as well as variations in upper-mantle velocity structure.

2830

S. Ma, D. W. Eaton, and J. Adams

Tectonic Setting The Precambrian shield of northern Ontario was assembled during the Proterozoic between ca. 2.4 billion and 1.0 billion years ago (Ga) by accretionary continental growth around an Archean cratonic nucleus. The bedrock of this region is dominated by the Archean Superior Province, which in this area is comprised of east–west trending subprovinces formed between 3.1 and 2.7 Ga, containing granitegreenstone belts separated by high-grade basement gneiss terranes (Thurston et al., 1992). Most underground mining activity in northern Ontario is concentrated in or near greenstone belts of the Superior Province. Crust of Archean age is bounded to the north by the ∼1:8 Ga trans-Hudson orogen (Hoffman, 1988), which outcrops locally but is largely obscured by younger sedimentary rocks in the Hudson Bay and James Bay lowlands. To the southeast, the Superior Province abuts younger Precambrian rocks of the Southern Province (∼2:2 Ga) and Penokean orogen (∼1:9–1:6 Ga; Ludden and Hynes, 2000). The extreme southeast of our study region is located within the younger Grenville orogen (Eaton, Dineva, and Mereu, 2005). The 1.1 Ga midcontinent rift, which underlies Lake Superior and much of Michigan, extends northward into the Lake Nipigon area immediately north of Lake Superior. A very thick (up to 30 km) sequence of flood basalts interlayered with associated igneous rocks and metasediments was emplaced into this failed rift structure (Hinze et al., 1997). The Laurentide ice sheet repeatedly blanketed northern Ontario during the past million years. The ice sheet began to recede near the end of the last phase of glaciation 21 ka, although final vestiges of ice persisted in this region until ∼10 ka. According to model ICE-5G (Pelter, 2004), at the last glacial maximum ice thickness varied from > 5 km in western areas to < 2 km in the southeast of our study region. Glacial isostatic adjustment (GIA) is an ongoing process that includes rebound initiated by removal of this surface ice load. The current GIA uplift (or subsidence) rate and induced crustal stresses vary across the study region, due to regional differences in deglaciation history, load magnitude, and the flexural strength of the lithosphere (e.g., Wu, 2001).

Data and Methods Figure 1 shows the distribution of cataloged seismicity (N 537) from 1980 to 2006. We used the online earthquake catalog created by the Geological Survey of Canada (GSC), see the Data and Resources section. We remark that from 1980 to 2003 only 309 events are contained in the catalog for our study region, whereas from 2004 to 2006 there are 228 events. As discussed in the following, this increase in number of observations reflects a reduction in magnitude detection threshold due to the recently improved station distribution. The seismicity is not uniformly distributed but instead appears to be concentrated in several discrete areas. Although the study area was not monitored evenly for the

smallest earthquakes throughout this 26 yr interval (Hayek et al., 2007), the pattern of seismicity described in the following persists even for a realistic magnitude detection threshold of mN 2.5 (Mw ∼2:1).1 In particular, seismicity appears to be more abundant in the James Bay, Kapuskasing, and Sudbury regions, as well as in higher-elevation areas of northwestern Ontario (the Severn Highlands). Before we classify and interpret this pattern seismicity, we will describe techniques used to characterize the seismicity distributions and to constrain focal depth. Spatial Clustering Analysis Using Poissonian Statistics Dineva et al. (2004) and Ma and Eaton (2007) have developed a simple statistical approach to characterize spatial clustering of seismicity. The method is implemented by subdividing the area of interest into square bins and counting the number of events that fall within each bin. For a random set of points in two dimensions, the resulting histogram is expected to approximate a Poisson distribution (Swan and Sandilands, 1995), n N 1 N=v e ; Pv n v v n!

(1)

where Pv n denotes the expected number of bins containing n events, v is the total number of bins in the study region, and N is the number of epicenters in the catalog. In contrast, a pseudoregular (ordered) distribution is expected to yield a Gaussian distribution, whereas event distributions that are clustered at the scale of the cell dimensions are expected to deviate from both distributions. For this statistical test to be valid, the dimensions of each bin must be larger than the uncertainty in epicentral location but significantly smaller than the total area of the study. We have tested bin sizes that satisfy these conditions in a range from 10 to 50 km, and although the details differ, all of these bin sizes support a similar interpretation of the results. Figure 2 shows an occurrence histogram obtained from the earthquake catalog using a 20 × 20 km bin size. The distribution does not resemble a Gaussian, and relative to a Poisson distribution (curved line on the figure), the histogram shows a major deficiency of bins containing one event, whereas bins with three or more events are more abundant than expected. We define this crossover point with respect to a Poisson distribution as the events-per-bin threshold value (2) that distinguishes significant spatial concentration seismicity from areas of random background seismicity. By using 2 to represent the lowest contour level in an event density map, concentrations of seismicity appear as closed regions (Fig. 3). In the Severn Highlands region in mN is also written as mbLg and is adapted from the Nuttli magnitude (Nuttli, 1973). For the moderate earthquakes in the study area, Mw ∼ mN 0:4 units. 1


2831

Figure 2. Seismicity occurrence histogram obtained using the 1980–2006 seismicity catalog and 20 × 20 km bins. Approximately 90% of the bins are empty, as expected because the number of events (N) is considerably smaller than the number of bins (ν). Relative to a random spatial distribution, the histogram is deficient in bins containing a single event, but this is balanced by a surplus of bins containing three or more events. The crossover at two events per bin represents a contour that can be used to delineate spatial clusters of seismicity from the catalog. the western part of the study area, a set of discrete concentrations is evident. As discussed in the following, these appear to be ephemeral features characterized by shallow seismicity localized in both space and time. To the east, other clusters of seismicity are evident in the James Bay, Kapuskasing, and Sudbury regions, and an isolated cluster of seismicity is apparent to the north in Hudson Bay. Between these regions of seismicity in the east and west, seismic activity is more scattered and diffuse. Focal-Depth Determination Focal depth is a critical earthquake source parameter for understanding regional seismicity and seismic hazards; it plays a particularly important role in relating seismicity to geologic structures, in delineating active faults, and in estimating seismogenic thickness. Even after recent seismic network expansion, it is not possible to estimate focal depth accurately using the traditional approach of simultaneously fitting focal depth and epicenter location, due to the sparse distribution of stations. However, certain depth phases sPg, sPmP, and sPn, with their reference phases Pg, PmP, and Pn, respectively, are often observable on regional records (see Ma and Atkinson, 2006, and references therein). These re-

gional depth phases and their reference phases can be used to estimate focal depth for small earthquakes. This method is based on the calculation of synthetics using the reflectivity method (Randall, 1994) with a default focal mechanism and a crustal model derived from the Canadian Shield region in western Quebec (Mereu et al., 1986). This velocity model is illustrated in Figure 4 and is similar to other crustal velocity models obtained in other parts of our study region (e.g., Kay et al., 1999). By comparing synthetics with observations for selected station distances over a reasonable range of focal depths, the depth that provides the best agreement (in relative arrival times) between the synthetics and the observations can be identified. The specific choice of depth and reference phase depends on epicentral distance, and the use of differential time between these two phases, rather than a single time measurement, greatly reduces the sensitivity of the result to the choice of reference model (Ma and Atkinson, 2006). An illustration of this technique is provided in Figure 4 (upper panel). This example shows that focal depth of a small earthquake can be reliably determined to within about 10% from a single record at a station as far as 250 km away, though confirmation from seismograms on other azimuths is preferred.

2832


Figure 3. Seismicity clusters (red areas) outlined by isoseismic contours of two events per 20 × 20 km bin. Additional contours at four and six events per bin are also indicated for reference. Triangles and diamonds show stations used to derive the catalog of seismicity (Hayek et al., 2007). Diamonds denote stations from the Canadian National Seismograph Network (CNSN), and triangles denote temporary FedNor and POLARIS stations. In northern Ontario the preceding method is effective for many small earthquakes that generate regional depth phases. In other cases, however, earthquakes do not produce a recognizable regional depth phase (likely because for very shallow earthquakes the various phases are not sufficiently different in travel time to be distinct; see Fig. 4, lower panel), but they do produce a strong, high-frequency Rg phase (Fig. 4, middle panel). Because the amplitude of the Rg phase relative to the body waves depends strongly on focal depth, it can be used to estimate focal depth quantitatively (Båth, 1975; Langston, 1987; Kafka, 1990). Figure 4 (lower panel) shows the attenuation of amplitude of synthetic Rg phase with depth. We see that below 4 km the amplitude becomes weak. Because the scaling of Rg in relation to other phases depends on focal mechanism as well as depth, it is not simple to quantify focal depth from a small number of waveform observations. In particular, the absence of Rg at a single station does not necessarily imply that the earthquake is deep. On the other hand, if a prominent Rg phase is observed for small events, it provides a strong indication that focal depth is less than ∼4 km. We combined the preceding two techniques to analyze waveform records in northern Ontario. Out of 537 earthquakes (1980–2006), we obtained focal depths for 331 events. Many events recorded from the east and north of

Kapuskasing lack focal-depth solutions because they have neither clear Rg phases nor regional depth-phase records. Further study is required to establish the underlying reasons for this, but we suspect that this may reflect a paucity of shallow-focus earthquakes and/or complexities in crustal structure or both. Figure 5 shows a histogram of the focal-depth distribution. The main peak at 3 km is characteristic of seismicity in the Severn Highlands of northwestern Ontario, as well as more scattered events throughout the region. A secondary peak at 13 km reflects depths for clusters of seismicity in the eastern parts of our study region. This secondary peak may be significant, because it is also a characteristic depth of moderate earthquakes in surrounding parts of eastern North America, such as the 1990 mN 5.0 Mont Laurier earthquake (Quebec), the 2000 mN 5.5 Kipawa earthquake (Ontario), and the 2002 mN 5.5 Plattsburgh earthquake (New York) (Ma and Atkinson, 2006). Figure 6 displays the distribution and depth of earthquakes for which focal-depth solutions were obtained. The overall distribution is similar to Figure 1, with the exception that onshore events in the north—especially 200 km northwest of Kapuskasing—are underrepresented. The map shows that earthquake concentrations in the eastern and western parts of our study region


2833

Figure 4. Methods used to estimate focal depth. (Upper panel) Regional depth-phase modeling for a small earthquake that generated the trace BUKO/05Jul10. Trace 020 was generated with depth 2.0 km; trace 025 was generated with depth 2.5 km; etc. The modeled focal depth for this small earthquake is 2.5 km. The inset shows the crustal velocity model used. (Middle panel) Examples of regional depth phases and strong Rg phases generated by three small shallow earthquakes in Sudbury, northern Ontario (swarm 2005 in Fig. 13, lower panel). Along trace BUKO/05Mar13, sPmP and PmP phases merge; the Rg phase is very strong. These phenomena show the earthquakes are extremely shallow. (Lower panel) The amplitude attenuation of synthetic Rg phase with focal depth at high frequencies (here the upper limit to generate the synthetics is 10 Hz). Trace 015 was generated with depth 1.5 km. Other numbers attached to the traces have similar meaning. The Rg phase is weak for focal depth below 4 km.

2834


Figure 5. Histogram of earthquake focal depths (2 km bins) showing a bimodal distribution. The main peak at 3 km is representative of both shallow earthquake swarms in northwestern Ontario and random background seismicity throughout northern Ontario. The secondary peak at 13 km is representative of a subset of events within earthquake clusters in James Bay and Kapuskasing. are characterized by distinct focal-depth distributions. In the west, earthquakes have exclusively shallow foci (all < 12 km, more than 95% of them < 6 km), whereas in the east we find a range of focal depths extending to 20 km. Ⓔ A full list of these events and their estimated focal depth is provided in the electronic edition of BSSA.

event. Swarms are confined in time and space, with irregular variations in seismicity rate on a timescale of months or a few years. Examples of each of these distributions are discussed in the following.

Earthquake Sequences

Characterization of Regional Seismicity It is convenient to classify the observed seismicity concentrations into one of three categories, based on spatiotemporal characteristics (Mogi, 1963). The first category is an earthquake sequence, which represents a set of events with a recognizable mainshock–aftershock series, with or without foreshocks. The second category is an earthquake cluster, which we define as a zone of microseismic activity characterized by a seismicity rate that is significantly above the regional background rate on a timescale of decades or longer. The third category, earthquake swarms, represents a set of events that occur during a prolonged period of microseismic activity of small to moderate magnitude, with no principal

Earthquake sequences in this area are rarely observed due to the small magnitude of earthquakes that have occurred from 1980 to 2006. This, coupled with the relatively sparse distribution of stations, limits the size and duration of the aftershock record available for study. Aftershocks following the 1 January 2000 mN 5:2 event in Kipawa near the southeast corner of the study area were reported by Bent et al. (2002). This earthquake sequence was located near the epicenter of the 1935 Timiskaming M 6.2 earthquake. A total of 13 aftershocks were recorded within 1 day after the mainshock, ranging from mN 0:1 to mN 2.2. By waveform comparison at the station EEO (distance ∼29 km), it was found that the mainshock and the aftershocks were confined to a very small volume at a depth of ∼13 km. Compared to


2835

Figure 6. Epicenters of 331 earthquakes in northern Ontario (1980–2006) for which focal depths have been determined. Focal depth is indicated by color, with deepest earthquakes plotted last. Compare with the full distribution of earthquakes in Figure 16. On the Severn Highlands the focal depths are generally shallow, whereas in the Kapuskasing and James Bay clusters, focal depths range from a few kilometers to deeper than 20 km.

the mainshock, the aftershocks were small, with a maximum magnitude of mN 2.2 (i.e., 3 magnitude units smaller than the mainshock). The focal mechanism obtained from surfacewave modeling was a thrust type, with a northwest striking fault plane. A shallow (2 km) mN 4.1 event on 29 November 2006 at the Creighton Mine near Sudbury produced an aftershock sequence that persisted for several days. Two of these aftershocks (mN 3:2 and mN 2:5) were observed using the regional network, but numerous tiny aftershocks (mN < 0) were recorded due to the fortuitous close proximity of a deep underground seismometer to the earthquake hypocenter (Atkinson et al., 2008). By fitting the aftershock data using a generalized Omori’s law (Shcherbakov et al., 2004), Atkinson et al. (2008) found an unexpectedly rapid decline in the initial aftershock rate, punctuated by discrete subsequences that were triggered by the two larger aftershocks. This pattern is interpreted to reflect an epidemic-type aftershock sequences model for aftershock occurrence, which includes triggering events within a cascading sequence (Ogata, 1999).

Although not classified in GSC’s earthquake catalog as a mining event, the close proximity (< 1 km) of this event to an active underground mine suggests that it may have been induced by mining activity. Earthquake Swarms in the Severn Highlands In the western part of our study area the land surface lies 300–500 m above sea level, forming a broad upland shield region called the Severn Highlands (Barnett, 1992). The largest earthquake magnitude in this area between 1980 and 2006 was mN 3:9. As a result of improved seismic network coverage resulting from installation of the FedNor stations, the magnitude detection threshold has been reduced in the past few years to mN ∼2:0, from a historical value of mN ∼2:5. As noted previously, there appears to be a number of shallow-focus seismicity concentrations, although there is no clear correlation between the distribution of seismicity and regional faults, including the 1.1 Ga Nipigon Embayment rift structure (Fig. 7). Each of these seismicity concentrations was active over only a limited time period, giving

2836


Figure 7. Epicentral distribution of all available earthquakes (1980–2006) on the Severn Highlands of northwestern Ontario. Background image is shaded-relief total-field magnetic data (the source is the Geoscience Data Repository, Geological Survey of Canada), which shows the dominant east–west structural fabric (mainly in the upper few kilometers of the crust) associated with regional faults and subprovince boundaries of the western Superior Province (Thurston et al., 1992). The Nipigon Embayment is a rift structure that formed 1.1 Ga, synchronous with formation of the midcontinent rift (Thurston et al., 1992). There is no clear evidence for control of seismicity from structures that are imaged by the magnetic data. Distance from the Dryden Swarm to station SOLO is ∼38 km. them the defining characteristics of a swarm. Earthquake focal depths are generally < 6 km, with ∼5% in the range 6–12 km. Swarm 1 (Fig. 7) contained ∼10 small earthquakes that occurred between 2000 and 2001, primarily between June–July 2000. The largest magnitude in this swarm is mN 3:1. Swarm 2 contained ∼14 events between 1994 and 1995, primarily in November 1994. Here, the largest magnitude is also mN 3:1, but several other events in the swarm have similar magnitudes (mN 2:9, 2.7, and 2.5). Swarm 3 contained 12 events between 1999 and 2001 and has a largest magnitude of mN 3:9. The Dryden swarm, lying to the north of Dryden, occurred between 2002 and 2003 and produced 22 events in two subswarms, with a largest magnitude of mN 3:2 in each subswarm (Table 1; Fig. 8). The first subswarm of 8 events

began in May 2002 and persisted until July 2002. After a 7 month break, the second subswarm of 14 events began in February 2003 and continued until May 2003 (Fig. 8, upper panel). We have analyzed waveforms generated by most of these events. The waveforms are displayed in Figure 9, aligned on the Pg phase. In every case, time differences between Pg and the regional depth phase (sPg) are very small, indicating shallow focal depths, similar to swarms such as the Anjalakoski swarm in southeastern Finland (Uski et al., 2006). Shallow focal depths are also indicated by the relatively strong Rg phases produced by events from the swarm. Note that the Rg waveforms are more complex than the modeled Rg waveforms in Figure 4; we attribute this waveform complexity to velocity heterogeneity such as shallow layering, which is not incorporated into

Intraplate Seismicity of a Recently Deglaciated Shield Terrane: A Case Study Table 1 Events in the Dryden Swarm Date (yyyy/mm/dd)

Time (UTC)

Latitude (°)

Longitude (°)

Depth (km)

mN

2002/05/01 2002/05/01 2002/05/21 2002/06/13 2002/06/13 2002/07/04 2002/07/04 2002/07/04 2003/02/08 2003/02/08 2003/02/08 2003/02/08 2003/02/11 2003/02/13 2003/02/13 2003/02/13 2003/02/13 2003/02/18 2003/02/21 2003/03/10 2003/03/28 2003/05/16

01:42:16 01:46:32 06:04:08 13:51:30 13:52:25 01:27:49 01:27:59 22:20:03 17:10:17 18:05:31 18:07:25 18:16:12 12:32:08 00:30:55 01:54:58 20:02:23 23:42:28 14:23:59 21:52:07 16:25:58 09:35:52 16:41:17

49.97 49.98 50.10 50.07 50.07 50.18 50.12 49.98 50.01 50.02 50.00 50.04 50.21 50.15 50.06 50.02 50.05 49.97 50.00 50.09 50.18 49.93

92:67 92:77 92:72 92:68 92:68 92:64 92:73 92:68 92:76 92:73 92:74 92:75 92:69 92:70 92:72 92:72 92:64 92:66 92:66 92:75 92:68 92:65

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

2.4 2.2 3.2 1.7 1.5 2.0 1.9 2.5 3.0 2.7 2.4 2.3 3.0 2.5 2.6 2.9 2.1 1.7 2.1 3.2 2.4 1.8

The source for epicenter and magnitude data is the GSC online catalog (see the Data and Resources section); the focal depths were assigned based on surface-wave amplitudes.

our simple crustal model. The events in the first subswarm appear to be shallower than those in the second, while subtle differences between the differential times sPg–Pg in the second subswarm show that the corresponding focal depths are more varied.

Earthquake Clusters in Eastern Regions The geology of the eastern part of our study region is dominated by the Abitibi subprovince, the world’s largest greenstone belt, and an area of intense mining activity. This area lies northwest of the WQSZ and south of Hudson Bay. The seismicity in this region contains a number of distinct clusters, with hypocenters extending to greater depth than those in the Severn Highlands. James Bay Cluster. The James Bay cluster is located in the northeast corner of our study region. There was little information about this cluster prior to augmentation of the regional seismic network beginning in 2002. Indeed, the documentation of this poorly known seismic zone in central Canada provides a good illustration of the benefits of such network densification for detailed hazard studies. Figure 10 (upper panel) shows the epicentral distribution of all detected earthquakes during the interval 1980–2003. Only 55 earthquakes, with a largest magnitude of mN 3:6, were detected in the region over this 23 yr period when seismograph network coverage was relatively sparse. During the interval from

2837

2004 to 2006 (Fig. 10, lower panel), after expansion of the seismograph network in northern Ontario, 118 earthquakes were detected with a largest magnitude of mN 2:9. This increase in rate of detection, from ∼2 events per year to ∼40 events per year, was enabled by the reduction in magnitude detection threshold (to mN ∼1:5, from a historical value of mN ∼2:5), allowing better recognition of this cluster of seismicity. The post-2003 distribution is more tightly clustered around 53° N, 80.7° W than the previous distribution. Furthermore, the post-2003 activity now underlies Akimiski Island (i.e., it extends about 30 km to the southwest of the older locations; see the dashed reference lines in Fig. 10). We have not yet evaluated the reliability of the older epicenters, so they may represent a biased estimate for the extent of the cluster As evident in Figure 6, the focal depths for the James Bay clusters are typical of the other seismicity clusters in the eastern part of the study region, varying from < 4 to ∼20 km depth. The cause of seismicity in this area is uncertain, but the presence of midcrustal earthquakes suggests that deep-rooted (up to 20 km) geological structures may be present. As discussed in the following, like WQSZ (Ma and Eaton, 2007) this cluster is also located along the track of a hot spot. Kapuskasing Cluster. The Kapuskasing cluster is located northwest of, and approximately along strike with, the WQSZ. It is separated from the more seismically active region of western Quebec by a relatively aseismic gap of ∼100 km. Moderate earthquakes have occurred previously in the area of the Kapuskasing cluster, including an ML 5:0 earthquake on 1 December 1928 (Smith, 1966; Basham, et al., 1979; Hayek et al., 2007) and events of mN 4:1 on 13 April 1980 and mN 4:2 on 7 December 2006. Improved seismic network coverage has lowered the magnitude detection threshold in this region to mN ∼1:5, from a historical value of mN ∼2:5. We have analyzed waveforms and obtained focal-depth solutions wherever possible for the cataloged events in this cluster (Table 2). Figure 11 shows the epicentral distribution of the 62 earthquakes detected from 1980 to 2006, including 37 for which focal depths were obtained. The hypocenters occur within a depth range from a few kilometers to more than 21 km, similar to that in the WQSZ (Ma and Eaton, 2007). The mN 4.2 event on 7 December 2006 had relatively good station coverage. We estimated a focal mechanism for this, the largest recent event, using the interactive waveform fitting method described by Ma and Eaton (2007). This method solves for double-couple parameters and stationdependent time shifts using body waves, because this event did not produce strong surface waves. The waveform fitting process is carried out using synthetic seismograms computed using the wavenumber integration method of Herrmann (2002). Figure 12 shows a lower-hemispheric projection of the focal mechanism, which is of thrust type. It is similar

2838


Figure 8.

(Upper panel) Epicentral distribution of earthquakes in the Dryden swarm and vicinity. The focal mechanism for the 1984 event is from Wetmiller and Cajka (1989). (Lower panel) History of swarm events. The timelines show the occurrence of the events in the swarm; the height of the line shows the magnitude. There are two subswarms: around May 2002 and around February 2003. Table 1 lists the events in the swarm; the largest event is mN 3.2. The apparent elongation of the swarm and the scatter in the subswarms in the north–south direction is probably an artifact of the station distribution.

to a composite P-nodal focal mechanism of 15 events (1975– 1984) obtained in the same region by Wetmiller and Cajka (1989), thus, supporting a compressive stress regime with a northeast principal stress axis. Also of significance is the northwest strike of the nodal planes, which is parallel to both the elongation of the Kapuskasing cluster and to the track of the hot spot trace (see the following). Sudbury Cluster. Sudbury, located in the southeastern part of our study region, has been a major nickel mining district

for more than a century. The current magnitude detection threshold in this region is mN ∼1:0. Figure 13 (upper panel) shows the distribution of the 916 mining-related events for the period from 1985–2006, as taken from the GSC catalog. The largest such event occurred on 29 November 2006, with magnitude mN 4.1, previously discussed in terms of its aftershocks. This event took place near the Creighton Mine just outside of Lively, Ontario, and was felt within a 50 km radius. A focal depth of 2.5 km was obtained using the regional depth-phase modeling (RDPM) method and was confirmed


Figure 9.

2839

Recorded waveforms at station SOLO for 18 of the 22 events from the Dryden swarm, aligned on reference phase Pg. Top panel shows P wave train; lower panel shows a longer time window to include the Rg phase. The first five events (the first subswarm) show a strong waveform similarity. The second subswarm events have more variation and appear to be deeper on average than the first subswarm events (based on the time difference between Pg and sPg).

2840


Table 2 Earthquakes in the Kapuskasing Cluster with Focal Depths Estimated Using the RDPM Technique

Figure 10.

Epicentral distribution of earthquakes in the James Bay region from 1980–2003 (upper panel) and from 2004–2006 (lower panel). Because of expansion of the seismograph network, the location rate increased from ∼2 to ∼40 per year. Note stronger clustering and the change in mapped distribution.

by inspection of underground mine workings (Atkinson et al., 2008). A number of active mines in this district extend as deep as 2.5 km. Because of the intensity of mining activity as well as the unusual mining depth, information about background seismicity is critical for mining safety, including rates, focal depths, and possible tectonic controls. After removal of mine blasts and demonstrably mining-induced seismic events, some 30 events classified by the GSC as earthquakes remain in the 1980–2006 catalog; some of these events near Sudbury may still be mining related. The events yield a distinct seismicity cluster (Fig. 13, lower panel), with a largest mag-

Date (yyyy/mm/dd)

Time (UTC)

Latitude (°)

Longitude (°)

Depth (km)

mN

1980/04/13 1987/12/20 1992/08/11 1994/12/25 1995/12/06 1996/04/12 1996/04/20 1996/08/16 1996/08/22 1999/02/01 2000/07/24 2000/11/05 2001/01/09 2002/10/13 2002/10/16 2004/05/17 2004/05/26 2004/10/11 2004/12/07 2005/01/27 2005/04/17 2005/04/20 2005/06/01 2005/06/11 2005/07/05 2005/10/25 2005/12/19 2006/01/03 2006/02/22 2006/03/04 2006/05/20 2006/06/05 2006/09/09 2006/12/01 2006/12/07 2006/12/07 2006/12/14

22:40:23 07:52:48 19:50:44 21:31:22 22:52:49 02:40:00 01:43:03 04:56:46 20:19:06 22:22:05 08:14:16 16:14:03 16:23:58 03:07:37 19:10:07 19:25:19 04:21:22 08:33:50 16:07:33 10:31:08 18:48:28 07:03:30 00:56:54 07:40:23 02:47:21 00:01:44 04:47:21 11:05:09 07:46:42 02:13:10 17:29:40 17:44:23 11:39:39 22:55:15 04:44:59 04:59:09 09:58:25

49.59 48.66 48.72 49.75 49.55 48.99 49.18 49.24 49.25 49.26 50.24 49.67 50.00 49.91 49.88 48.63 49.90 49.62 49.25 49.66 49.82 49.48 48.80 48.78 49.56 49.65 49.60 49.33 49.56 49.51 49.22 49.82 48.95 48.55 49.52 49.51 50.11

81:76 80:34 80:79 81:71 81:83 81:45 80:97 82:95 82:97 80:94 82:37 81:47 82:34 81:60 82:00 80:79 82:00 81:47 81:61 81:46 81:45 81:58 80:30 81:40 81:67 81:79 81:50 81:10 81:56 81:55 81:47 82:00 82:91 82:32 81:53 81:56 82:45

6.0 4.5 6.0 12.0 5.0 5.0 3.0 4.5 4.5 12.0 17.0 10.0 9.0 8.0 12.0 8.0 12.0 14.0 14.0 12.0 10.0 13.0 8.0 14.0 16.0 11.0 8.0 22.0 12.0 16.0 15.0 7.0 3.0 3.0 16.0 16.0 8.0

4.1 2.7 2.5 2.8 3.5 2.6 3.3 3.7 2.8 3.4 2.2 2.7 2.5 2.5 2.8 2.2 1.9 2.3 1.7 2.0 2.1 1.4 2.0 2.0 1.7 1.9 1.3 3.7 1.8 3.4 2.3 2.5 1.8 1.8 4.2 2.7 1.8

The source for epicenter and magnitude data is the GSC online catalog (see the Data and Resources section).

nitude of mN 3:6. Within this cluster some sets of events have swarmlike characteristics, including 10 events (largest magnitude mN 2:9) at the western end of Lake Nipissing in 1986 and 5 events (largest magnitude mN 3:6) near Sudbury in 2005. Like the small earthquake swarms observed in northwestern Ontario (but unlike James Bay and Kapuskasing), waveform analysis indicates shallow focal depths (< 3:5 km) for the Sudbury cluster.

Discussion Pseudorepeating Earthquakes At least three of the events from the Kapuskasing cluster have waveforms that are very similar in appearance (Fig. 14). The inferred high degree of similarity is confirmed by nor-


2841

(Fig. 14). The relative time difference between depth and reference phases is consistent with progressive rupture migration down the fault, on a length scale of the order of hundreds of meters. To evaluate this model in more detail, we relocated the three small events using the HypoDD algorithm (Waldhauser and Ellsworth, 2000), using cross correlation to obtain the time differences. The initial focal depth 16 km from the RDPM method was used. The resulting hypocenters (Table 3) are extremely close each other (Fig. 15), supporting (but not proving) the presumption that each initiated on the edge of the previous rupture. Note that the depth solutions from HypoDD were inconsistent with the depthphase changes in Figure 14, implying trade-off errors within HypoDD. By carefully analyzing regional waveform data from the 1990 Mont Laurier earthquake, Haddon and Adams (1997) inferred that the event was comprised of several ellipseshaped subevents that sequentially ruptured adjacent parts of the fault plane. Using this as an analogy, if we suppose that the mN 2.6 event initiated at the edge of the mN 4.2 rupture, then we can try to determine if the inferred rupture dimensions are realistic. To make this assessment, we consider a circular source region of radius r, for which the stress drop is given by (Brune, 1970) Δσ 7M0 =16r3 :

Figure 11. (Upper panel) Epicentral distribution of all cataloged earthquakes (62) in the Kapuskasing region for the period 1980–2006, as well as the approximate epicenter of the 1928 moderate earthquake. (Lower panel) Epicentral distribution of earthquakes that have focal-depth solutions (37). The color scale on the right-hand side shows the focal-depth range. malized cross-correlation amplitudes, which are > 82% for filtered traces (1.9–19 Hz) within a time window from 0.2 sec before to 30 sec after the P arrival. Given the sensitivity of observed seismic waveforms to the source–receiver path and focal mechanism, this degree of waveform similarity indicates that the events originated from very close locations and have essentially the same source mechanism. Accordingly, we initially considered these events as probable repeating earthquakes, because repeating earthquake phenomena have been documented in the nearby WQSZ (Ma and Eaton, 2007). Elsewhere, repeating earthquakes have been interpreted to reflect creep loading of stuck fault patches (Schaff and Richards, 2004). When arranged in order of occurrence, however, we observe that the focal depths (∼16 km) of these earthquakes successively increase

Taking M0 ∼6:3 × 1014 N m (Mw ∼3:8) and r as the HypoDD slant separation to the mN 2.6 aftershock (165 m, from Table 3) yields a Δσ ∼ 60 MPa for a circular rupture. For dip slip on a rectangular fault, the stress drop is given by (Stein and Wysession, 2003) Δσ 8M0 =3πw2 L: For a square fault with the same area as the preceding circular fault, this yields a stress drop of ∼21 MPa. The latter value is within the stress-drop range reported for shield regions (e.g., Singh et al., 1999). Allowing for an aspect ratio greater than unity (L > w) and/or realistic uncertainty in the relative location of the epicenters (∼100 m) from HypoDD, these values of stress drop are not unreasonable for a shield environment. We remark that the purpose of this calculation is not to compute stress drop per se, but rather to demonstrate that the length scale inferred from apparent changes in hypocenter is broadly consistent with simple estimates of rupture scale. This result is also consistent with the notion that an event of this type ruptures a fault patch that is on the order of hundreds of meters in scale length and adjacent to the previous rupture. We refer to this type of event as pseudorepeating; that is, the location and mechanism are about the same but new areas of the fault plane are ruptured each time. We find it gratifying that new analysis methods allow such precision

2842


Figure 12. Waveform comparison and inferred focal mechanism for the 7 December 2006 mN 4.2 earthquake in the Kapuskasing seismic zone. For each pair of waveforms, the upper trace is the recorded waveform, and the lower trace is the synthetic seismogram generated with the focal mechanism solution on the upper left-hand side. Both the observed and the synthetic waveforms were filtered with passband 0:4–4 Hz. At the left-hand side of each pair, the symbols and numbers from top to bottom indicate station name, vertical component (Z_P is the P phase; Z_D is the depth phase), distance in kilometers, station azimuth in degrees, and the ratio between observed maximum amplitude and that of the synthetic. The stations appear on the lower-hemispherical projection of the focal mechanism at the upper right-hand side. Records at stations KAPO and MALO on both sides of the event played a critical role for controlling the solution. and conclusions, even for regions with very sparse seismograph coverage. Seismotectonic Subdivisions of Northern Ontario There are several striking differences between the western and eastern parts of our study area. For example, focal

depths in the western region are overwhelmingly shallow (< 6 km), whereas seismicity in the eastern region spans a depth range up to 20 km. Also, more of the seismicity in the western part is swarmlike in character. From the tectonic history of the region and distribution of tectonic domains, it is difficult to explain these differences based on surface geol-


2843

Figure 13. (Upper panel) Epicentral distribution of all available mine-related events (916) for 1985–2006 in the Sudbury region. The largest event is mN 4.1, and occurred on 29 November 2006. The small triangles show the locations of seismographs, not all of which operated at the same time. (Lower panel) Epicentral distribution of all cataloged earthquakes (30) in the Sudbury region for 1985– 2006. The largest magnitude is mN 3.6. There are two possible swarms, in 1986 and 2005. ogy alone, as both regions occur within essentially similar parts of the Canadian Shield. Here, we consider two possible explanations for these differences: variations in glacial ice load (and, thus, postglacial rebound response) and narrow transitions in properties of the underlying mantle that might influence the elastic behavior of the lithosphere. Present-day seismicity in eastern North America is probably strongly influenced by crustal stress perturbations associated with GIA (e.g., Adams, 1989; Zoback, 1992; James and Bent, 1994; Wu and Hasegawa, 1996a,b; Wu, 1998; Mazotti et al., 2005). It has been suggested that both tectonic stresses and GIA are necessary to explain the nature and distribution of seismicity in formerly glaciated terranes of eastern Canada (Stewart et al., 2000). Figure 16 (left-hand panel) shows a map of present-day uplift rates derived from Global

Positioning System (GPS) data (Sella et al., 2007) and tidegauge data combined with satellite altimetry (Lee et al., 2008). The uplift rates are greatest (> 13 mm=yr) in the Hudson Bay region and generally decrease with distance from Hudson Bay (Sella et al., 2007). The hinge line separating uplift from subsidence occurs south of the Great Lakes. Although seismicity in our study region appears to be mainly concentrated in areas with uplift rates between 0 and 10 mm=yr (Fig. 16, left-hand panel), there is no obvious correlation between the broad east–west differences in seismicity characteristics that we observe here and this regional pattern of uplift due to GIA. We remark, however, that stresses due to GIA are probably not large enough to rupture intact rock but may be sufficient to reactivate preexisting faults or cause rupture in already weakened zones (Wu and

2844


Figure 14.

Three pseudorepeating earthquakes from the Kapuskasing cluster. Seismograms are aligned on the Pg phase and were recorded at station KILO, at an epicentral distance of ∼175 km. The bottom trace shows an aftershock of the event on 6 December 2007. Note that the regional depth phase sPg (and sPmP) arrived progressively later for each event, showing that focal depths became progressively deeper with time. The focal-depth difference between the 4 March 2006 event and the aftershock is about 300 m.

Hasegawa, 1996a,b). Thus, we cannot rule out a GIA contribution to the seismicity, despite the lack of strong regional correlation between seismicity concentrations and uplift rates. Frederiksen et al. (2007) have conducted a tomographic study of the mantle beneath northern Ontario, using the same set of stations used in this study as well as previous portable deployments. One of the key results from their study was the delineation of a profound, west–east mantle subdivision of the Superior Province that is not apparent in the surface geology. This subdivision is evident in Figure 16 (right-hand panel), which shows a map of P-wave velocity perturbation at 200 km depth. We remark that this pattern of mantle seismic velocities in the tomographic model of Frederiksen et al. (2007) persists to shallower mantle depths (see their 4-depth fig. 9), but a reduction in crossing rays above ∼200 km depth leads to distracting gaps in the image. Figure 16 (right-hand panel) shows that the western Superior region is underlain by a thick, high-velocity upper-mantle root that is typical of many Archean cratons (Eaton et al., 2008). Relatively subtle topographic features may reflect this lithospheric root. For example, as we have noted previously, most of the earth-

quakes in the Severn Highlands are confined to topographically elevated areas, whereas adjacent, lower-lying regions have fewer earthquakes and are underlain by a relatively lowvelocity mantle. These empirical seismicity-topography-root relationships are illustrated by the 350 m elevation contour, which approximately delineates higher-elevation areas, superimposed on the 200 km tomographic depth slice (Fig. 16, right-hand panel). Thybo et al. (2000) pointed out that stresses in the mantle associated with lateral density changes tend to be concentrated at the edges of thick (> 200 km) lithospheric roots and that in North America these root-edge regions are also characterized by enhanced levels of intraplate seismicity, presumably reflecting brittle deformation in the crust resulting from elevated stress. Here, we speculate that the seismicity of the Severn Highlands may be a local manifestation of elevated stresses along the edge of a thick lithospheric root, in accordance with the model proposed by Thybo et al. (2000). In contrast to the western Superior region, the eastern part of our study region overlies a low-velocity region in the upper mantle. Eaton and Frederiksen (2007) proposed that this feature is a residual anomaly caused by Mesozoic

Table 3 Comparison between Catalog Locations and Those Obtained by the HypoDD Technique (Waldhauser and Ellsworth, 2000) for Three Small Earthquakes in the Kapuskasing Region Number

Date (yyyy/mm/dd)

1 2 3 1 2 3

Time (UTC)

Latitude (°)

Logitude (°)

2006/03/04 2006/12/07 2006/12/07

GSC catalog 02:13:10.55 49.5107 04:44:59.24 49.5126 04:59:09.27 49.5192

2006/03/04 2006/12/07 2006/12/07

02:13:10.54 04:44:59.24 04:59:09.28

HypoDD 49.513976 49.514215 49.514300

Depth (km)

mN

81:5382 81:5332 81:5475

16.000 16.000 16.000

3.4 4.2 2.6

81:539641 81:539452 81:539818

16.091 16.029 15.867

3.4 4.2 2.6

The initial focal depth of 16 km was obtained by the RDPM technique.


Figure 15.

Comparison between the GSC catalog locations (circles) and those obtained by the HypoDD technique (diamonds) for three small earthquakes in the Kapuskasing region.

passage of North America over the Great Meteor hot spot, resulting from a combination of plume-induced compositional and residual thermal effects. Furthermore, various authors have postulated that the seismicity of the WQSZ is related to this hot spot (see Ma and Eaton, 2007, and references therein), either as a result of thermal rejuvenation of zones of weakness in the crust or by emplacement of igneous bodies into the middle crust that are characterized by a present-day rheology that is stronger than the host material.

2845

Previous studies of hot spot tracks in North America (Morgan, 1983; Heaman et al., 2004) have identified several tracks that pass through our study region (Fig. 16). In the context of the present study, there are several interesting features. First, hot spot tracks pass through both the Kapuskasing (Great Meteor track) and James Bay (Labrador track) seismicity clusters. The Labrador hot spot was under the James Bay cluster ca. 60 Ma, and the Great Meteor hot spot was under the Kapuskasing cluster ca. 130 Ma (Heaman et al., 2004). Both of these clusters have characteristics, including focal depths > 20 km that resemble the WQSZ where a hot spot model has been proposed to explain the seismicity (Ma and Eaton, 2007). Secondly, no known hot spot track transects the region of the western Superior Province, where high upper-mantle velocities provide evidence for a cold lithospheric root. This association is consistent with the notion that intact lithospheric roots tend to be best preserved in areas that have not interacted with a mantle plume (Morgan, 1983). Seismic Hazard Implications Improved understanding of seismicity such as we report improves estimation of seismic hazard in several ways. Earthquake Distribution in Depth. In low-seismicity regions the focal depth of the earthquakes has only a small

Figure 16. Comparison between seismicity, vertical uplift rate, and upper-mantle P-wave velocity perturbation in northern Ontario. (Left-hand panel) Map of smoothed vertical uplift rate from GPS (Sella et al., 2007) and tide-gauge–satellite-altimetry (Lee et al., 2008) data. The highest rates of uplift are in Hudson Bay. (Right-hand panel) A comparison between seismicity P-wave velocity perturbation obtained from travel-time tomography (Frederiksen et al., 2007). The 350 m elevation contour approximately outlines the area of higher elevation on the Severn Highlands. Small earthquake swarms in northwestern Ontario coincide roughly with an area of high upper-mantle P-wave velocity. Persistent clusters in James Bay and Kapuskasing fall along hot spot tracks (Great Meteor hot spot [GMH] and Labrador hot spot [LH], after Heaman et al. [2004] and Morgan [1983]). A low-velocity anomaly is offset slightly from the GMH track, possibly due to mantle creep driven by North American plate motion (Eaton and Frederiksen, 2007).

2846 impact on the seismic hazard, because earthquakes very close to the site contribute only a small part of the total hazard. For example, changing the best estimate of the focal depth from 15 km (a typical value for active eastern Canadian zones such as the WQSZ) to 3.5 km (the expected value we find for northwestern Ontario) increases the hazard by only 3% at short periods at the probabilities used for the National Building Code. The data to date do not eliminate the possibility of deep earthquakes under the Severn Highlands (consider the Uluru earthquake in the similar Australian craton at 31 km [Michael-Leiba et al., 1994]), but they do suggest that if a large earthquake were to occur it might be initiated by a small earthquake in the top 6 km. Such a scenario has a prototype in the Meckering, Western Australia, earthquake of 1968, which had a magnitude of 6.8 produced a 37 km long, 2 m high surface scarp and is thought to have ruptured downwards (Vogtfjord and Langston, 1987). Overall, the effect of depth on ground shaking hazard is probably less significant than its effect on fault rupture hazard at extremely low probabilities. Earthquake Distribution in Time. Current seismic hazard analyses treat the earthquake catalog as representative of a Poissonian process, a simplification that has worked reasonably well in low- to moderate-seismicity regions where most dependent events are aftershocks and the magnitude difference between mainshock and largest aftershock is large. The approximation will not work as well when much of the activity is swarmlike. The high-activity rate of swarms can lead (counterintuitively) to a lower hazard estimate through an unusually large b-value. Application of a declustering algorithm is probably warranted for the Canadian Shield. Earthquake Distribution in Space. Previous estimates of seismic hazard for the Canadian Shield (Adams and Halchuk, 2003; Fenton et al., 2006; Atkinson and Martens, 2007) have been challenged by the low rates of activity, and they used either worldwide analogous regions, the Canadian Shield as a whole, or a combination to determine the local seismic hazard. All studies recognized certain active clusters such as the James Bay and Kapuskasing clusters, but of necessity the models produce seismic hazard maps that are uniform over large areas—effectively nearly all of the other spatial variations discussed in the present paper are smoothed out. A preliminary analysis of the new observations and understanding (such as Figs. 3 and 16) would argue that (1) earthquake activity along the hot spot traces (with its greater focal depth) is distinct from other activity and (2) if associations between contemporary seismicity, topographic elevation, and lithospheric structure can be established, it may be possible to identify regions of the Canadian Shield that will be less active than the average. Speculatively, the area around Lake Nipigon may be one such region.


Conclusions Based on analysis of waveform and catalog data for the period 1980–2006, moderate intraplate seismicity of northern Ontario is broadly divisible into four region-specific categories: (1) small swarms of shallow-focus (< 6 km) microseismic events concentrated mainly in Archean terranes of the Severn Highlands in northwestern Ontario; (2) persistent earthquake clusters located in the James Bay and Kapuskasing regions, with focal depths in the range of 5–20 km; (3) shallow seismicity in the Sudbury area that may be causally linked to deep mining activity; and (4) sparse and apparently random background seismicity distributed over most of northern Ontario. Regardless of region and focal depth, available earthquake mechanisms (Fig. 12 and solutions obtained by Wetmiller and Cajka [1989]) are dominantly of thrust type with roughly northeast–southwest P axes, consistent with a regional stress regime characterized by highhorizontal stress. Careful waveform analysis of near-identical waveforms from earthquakes in the Kapuskasing cluster suggests smallscale migration of hypocenters for successive events. We refer to such sets of events as pseudorepeating earthquakes and propose that this phenomenon constitutes a type of progressive fault rupture, slower than but otherwise similar to complex-rupture models for moderate earthquakes such as that proposed by Haddon and Adams (1997) for the 1990 Mont Laurier earthquake. Small earthquake swarms in the Severn Highlands have remained active for periods from several months to a few years and have produced earthquakes up to mN 3.9. These small swarms are temporally separated by hiatuses of several years and are spatially separated by 50–200 km, with no clear trend in loci of swarm activity. Although precise hypocentral location techniques such as HypoDD have not yet been widely used (due to the sparse network distribution and generally weak signals), waveform similarities observed at proximal stations suggest that hypocenters within each swarm are tightly clustered. There is no obvious correlation between this region of swarm activity and ice thickness trends from model ICE-5G, suggesting that postglacial rebound is not a primary control on regional seismicity patterns. However, in addition to a positive elevation difference of ∼350 m, the Severn Highlands are underlain by a positive P-velocity anomaly in the upper mantle associated with a particularly thick, cold lithospheric root. By contrast, the sparse seismicity near Lake Nipigon–Geraldton is an area of lower elevation and a negative P-velocity anomaly. This contrast suggests that there may be a causal link between lithospheric roots and intraplate seismicity, as proposed by Thybo et al. (2000). Like the WQSZ, seismicity clusters at Kapuskasing and James Bay seem to occur along ancient hot spot tracks. This association may reflect either thermal rejuvenation of the crust by passage of the hot spot and/or rheological contrast between younger igneous bodies and the host medium. The

Intraplate Seismicity of a Recently Deglaciated Shield Terrane: A Case Study persistence and unusual depth extent of these clusters, together with historical and paleoseismic events within similar clusters in the WQSZ, support the interpretations that larger (albeit infrequent) earthquakes may be more likely near Kapuskasing and James Bay than elsewhere in northern Ontario. This article also represents a step toward improved seismic hazard estimates for the Canadian Shield, especially through the potential identification of regions that will be more active or less active than the average.

Data and Resources The seismograms and earthquake catalogs used in this article were retrieved from the Geological Survey of Canada (GSC) official Web site (http://earthquakescanada.nrcan.gc .ca, last accessed December 2006). Figures 1, 3, 6, 7, 8 (upper panel), 10, 11, 13, 15, and 16 were generated with Generic Mapping Tools (GMT) software. Figures 4, 9, and 14 were generated with Geotool software. Patrick Wu graciously provided the GIA uplift data used in Figure 16.

Acknowledgments Funding for POLARIS and FedNor seismic stations was provided by Natural Resources Canada (NRCan), FedNor, the Canada Foundation for Innovation, the Ontario Research and Development Challenge Fund, and Ontario Power Generation. We wish to express our sincere thanks to Isa Asudeh, Bernie Dunn, Kadircan Aktas, and other POLARIS and GSC staff who operate the seismograph networks and to GSC staff who maintain and develop the earthquake catalog and provide waveform digital records service. Funding for this research was provided by NRCan and the Natural Sciences and Engineering Research Council of Canada.

References Adams, J. (1989). Postglacial faulting in Eastern Canada; nature, origin and seismic hazard implications, Tectonophysics 163, 323–331. Adams, J. (1995). The Canadian crustal stress database—a compilation to 1994, Geological Survey of Canada Open-File Rept. 3122, 226 pp. Adams, J., and J. S. Bell (1991). Crustal stresses in Canada, in Neotectonics of North America, Decade of North American Geology, D. B. Slemmons, E. R. Engdahl, M. Zoback and D. Blackwell (Editors), Vol. DMV-1, Geological Society of America, Boulder, Colorado, 367–386. Adams, J., and S. Halchuk (2003). Fourth generation seismic hazard maps of Canada: values for over 650 Canadian localities for the 2005 National Building Code of Canada, Geological Survey of Canada Open-File Rept. 4459, 155 pp. Atkinson, G., and S. Martens (2007). Seismic hazard estimates for sites in the stable Canadian craton, Can. J. Civ. Eng. 34, 1299–1311. Atkinson, G., S. Kaka, D. Eaton, A. Bent, V. Peci, and S. Halchuk (2008). A very close look at a moderate earthquake near Sudbury, Ontario, Seism. Res. Lett. 79, 119–131. Barnett, P. J. (1992). Quaternary geology of Ontario, in Geology of Ontario, P. C. Thurston, H. R. Williams, R. H. Sutcliff and G. M. Stott (Editors), Ontario Ministry of Natural Resources, Peterborough, Ontario. Basham, P. W., D. H. Weichert, and M. J. Berry (1979). Regional assessment of seismic risk in eastern Canada, Bull. Seismol. Soc. Am. 69, 1567–1602.

2847

Båth, M. (1975). Short-period Rayleigh waves from near-surface events, Phys. Earth Planet. Inter. 10, 369–376. Bent, A. L., M. Lamontagne, J. Adams, C. R. D. Woodgold, S. Halchuk, J. Drysdale, R. J. Wetmiller, S. Ma, and J.-B. Dastous (2002). The Kipawa, Quebec “Millennium” earthquake, Seism. Res. Lett. 73, 285–297. Brune, J. (1970). Tectonic stress and seismic shear waves from earthquakes, J. Geophys. Res. 75, 4997–5009. Dineva, S., D. W. Eaton, and R. Mereu (2004). Seismicity of the southern Great Lakes: revised earthquake hypocenters and possible tectonic controls, Bull. Seismol. Soc. Am. 94, 1902–1918. Eaton, D. W., and A. Fredericksen (2007). Seismic evidence for convectiondriven motion of the North American plate, Nature 446, 428–431. Eaton, D. W., J. Adams, I. Asudeh, G. M. Atkinson, M. G. Bostock, J. F. Cassidy, I. J. Ferguson, C. Samson, D. B. Snyder, K. F. Tiampo, and M. J. Unsworth (2005). Investigating Canada’s lithosphere and earthquake hazards with portable arrays, EOS 86, no. 17, 169–176. Eaton, D. W., F. Darbyshire, R. L. Evans, H. Grütter, A. G. Jones, and X. Yuan (2008). The elusive lithosphere–asthenosphere boundary (LAB) beneath cratons, Lithos, doi 10.1016/j.lithos.2008.05.009 (in press). Eaton, D. W., S. Dineva, and R. Mereu (2005). Crustal thickness and Vp=Vs variations in the central Grenville orogen (Ontario, Canada) from analysis of teleseismic receiver functions, Tectonophysics 420, 223–238. Fenton, C. H., J. Adams, and S. Halchuk (2006). Seismic hazards assessment for radioactive waste disposal sites in regions of low seismic activity, Geotech. Geol. Eng. 24, 579–592. Frederiksen, W., S.-K. Miong, F. A. Darbyshire, D. W. Eaton, S. Rondenay, and S. Sol (2007). Lithospheric variations across the Superior Province, Ontario, Canada: evidence from tomography and shear-wave splitting, J. Geophys. Res. 112, B07318, doi 10.1029/2006JB004861. Haddon, R. A. W., and J. Adams (1997). Anatomy of a small intraplate earthquake: a dissection of its rupture characteristics using regional data, Geophys. J. Int. 129, 235–251. Hayek, S., J. A. Drysdale, V. Peci, S. Halchuk, J. Adams, and P. Street (2007). Seismic activity in northern Ontario portion of the Canadian Shield: annual progress report for the period January 01–December 31, 2006, Report TR-2007-02, Nuclear Waste Management Organization, Toronto, 21 pp., plus appendices. Heaman, L. M., B. A. Kjarsgaard, and R. A. Creaser (2004). The temporal evolution of North American kimberlites, Lithos 76, 377–397. Herrmann, R. B. (2002), Computer Programs in Seismology, Department of Earth and Atmospheric Sciences, Saint Louis University, St. Louis, Missouri. Hinze, W. J., D. J. Allen, L. W. Braile, and J. Mariano (1997). The midcontinent rift system: a major Proterozoic continental rift, R. W. Ojakangas, A. B. Dickas and J. C. Green (Editors), in Middle Proterozoic to Cambrian Rifting, Central North America, Geol. Soc. Am. Spec. Pap. 312, 7–36. Hoffman, P. F. (1988). United Plates of America, the birth of a craton: early Proterozoic assembly and growth of Laurentia, Ann. Rev. Earth Planet. Sci. 16, 543–603. James, T. S., and A. L. Bent (1994). A comparison of eastern North American seismic strain-rates to glacial rebound strain-rates, Geophys. Res. Lett. 21, 2127–2130. Kafka, A. L. (1990). Rg as a depth discriminant for earthquakes and explosions: a case study in New England, Bull. Seismol. Soc. Am. 80, 373–394. Kay, I., G. Musacchio, D. White, I. Asudeh, B. Roberts, D. Forsyth, Z. Hajnal, B. Koperwhats, and D. Farrell (1999). Imaging the Moho and Vp=Vs ratio in the western Superior Archean craton with wide-angle reflections, Geophys. Res. Lett. 26, 2585–2588. Langston, C. A. (1987). Depth of faulting during the 1968 Meckering, Australia, earthquake sequence determined from waveform analysis of local seismograms, J. Geophys. Res. 92, 11,561–11,574.

2848 Lee, H., C. K. Shum, C.-Y. Kuo, Y. Yi, and B. Alexander (2008). Application of TOPEX altimetry for solid Earth deformation studies, Terr. Atmos Ocean. Sci. 19, 37–46. Ludden, J., and A. Hynes (2000). The Lithoprobe Abitibi-Grenville transect: two billion years of crust formation and recycling in the Precambrian Shield of Canada, Can. J. Earth Sci. 37, 459–476. Ma, S., and G. M. Atkinson (2006). Focal depth distribution for earthquakes with mN ≥ 2:8 in western Quebec, southern Ontario and northern New York, Bull. Seismol. Soc. Am. 96, 609–623. Ma, S., and D. Eaton (2007). Western Quebec Seismic Zone (Canada): clustered, mid-crustal seismicity along a Mesozoic hotspot track, J. Geophys. Res. 112, B06305, doi 10.1029/2006JB004827. Mazotti, S., T. S. James, J. Henton, and J. Adams (2005). GPS crustal strain, postglacial rebound, and seismic hazard in eastern North America; the Saint Lawrence Valley example, J. Geophys. Res. 110, B11301, doi 10.1029/2004JB003590. Mereu, R., D. Wang, O. Kuhn, D. Forsyth, A. Green, P. Morel, G. Buchbinder, D. Crossley, E. Schwarz, R. DuBerger, C. Brooks, and R. Clowes (1986). The 1982 COCRUST seismic experiment across the Ottawa-Bonnechere graben and Grenville Front in Ontario and Quebec, Geophys. J. R. Astron. Soc. 84, 491–514. Michael-Leiba, M., D. Love, K. McCue, and G. Gibson (1994). The Uluru (Ayres Rock), Australia earthquake of 28 May 1989, Bull. Seismol. Soc. Am. 84, 209–214. Mogi, K. (1963). Some discussions on aftershocks, foreshocks and earthquake swarms—the fracture of a semi-infinite body caused by an inner stress origin and its relation to the earthquake phenomena, Bull. Earthq. Res. Inst. Univ. Tokyo 41, 615–658. Morgan, W. J. (1983). Hotspot tracks and the early rifting of the Atlantic, Tectonophysics 94, 123–139. Nuttli, O. W. (1973). Seismic wave attenuation and magnitude relations for eastern North America, J. Geophys. Res. 78, 876–885. Ogata, Y. (1999). Seismicity analysis through point-process modeling: a review, Pure Appl. Geophys. 155, 471–507. Pelter, W. R. (2004). Global glacial isostasy and the surface of the ice-age Earth: the ICE-5G (VM2) model and GRACE, Ann. Rev. Earth Planet. Sci. 32, 111–149. Randall, G. (1994). Efficient calculation of complete differential seismograms for laterally homogeneous earth models, Geophys. J. Int. 118, 245–254. Schaff, D., and P. Richards (2004). Repeating seismic events in China, Science 303, 1176–1178. Sella, G. F., S. Stein, T. H. Dixon, M. Craymer, T. S. James, S. Mazzotti, and R. K. Dokka (2007). Observation of glacial isostatic adjustment in “stable” North America with GPS, Geophys. Res. Lett. 34, L02306, doi 10.1029/2006GL027081. Shcherbakov, R., D. L. Turcotte, and J. B. Rundle (2004). A generalized Omori’s law for earthquake aftershock decay, Geophys. Res. Lett. 31, L11613, doi 10.1029/2004GL019808. Singh, S. K., M. Ordaz, S. R. Dattatrayam, and K. H. Gupta (1999). A spectral analysis of the 21 May 1997, Jabalpur, India, earthquake (Mw 5:8) and estimation of ground motion from future earthquakes in the Indian Shield region, Bull. Seismol. Soc. Am. 89, 1620–1630. Smith, W. E. T. (1966). Earthquakes of eastern Canada and adjacent areas, 1928–1959, Publ. Dom. Obs. Ott., 32, no. 3, 87–121. Stein, S., and M. Wysession (2003). An Introdution to Seismology, Earthquakes and Earth Structure, Blackwell, Oxford.

S. Ma, D. W. Eaton, and J. Adams Stewart, I. S., J. Sauber, and J. Rose (2000). Glacio-seismotectonics: ice sheets, crustal deformation and seismicity, Quat. Sci. Rev. 19, 1367–1389. Swan, A. R. H., and M. Sandilands (1995). Introduction to Geological Data Analysis, Oxford, Cambridge, Massachusetts, 446 pp. Thurston, P. C., H. R. Williams, R. H. Sutcliffe and G. M. Stott (Editors) (1992). Geology of Ontario, Ontario Geological Survey Special Volume 4, Ontario Ministry of Natural Resources, Peterborough, Ontario, 1525 pp. Thybo, H., S. Zhou, and E. Perchuc (2000). Intraplate earthquakes and a seismically defined lateral transition in the upper mantle, Geophys. Res. Lett. 27, 3953–3956. Uski, M., T. Tiira, A. Korja, and S. Elo (2006). The 2003 earthquake swarm in Anjalakoski, south-eastern Finland, Tectonophysics 422, 55–69. Vogtfjord, K. S., and C. A. Langston (1987). The Meckering earthquake of 14 October 1968: a possible downward propagating rupture, Bull. Seismol. Soc. Am. 77, 1558–1578. Waldhauser, F., and W. L. Ellsworth (2000). A double-difference earthquake location algorithm: method and application to the northern Hayward fault, California, Bull. Seismol. Soc. Am. 90, 1353–1368. Wetmiller, R. J., and M. G. Cajka (1989). Tectonic implications of seismic activity recorded by the northern Ontario seismograph network, Can. J. Earth Sci. 26, 376–386. Wu, P. (1998). Will earthquake activity in Eastern Canada increase in the next few thousand years?, Can. J. Earth Sci. 35, 562–568. Wu, P. (2001). Postglacial induced surface motion and gravity in Laurentia for uniform mantle with power-law rheology and ambient tectonic stress, Earth Planet. Sci. Lett. 148, 427–435. Wu, P., and H. S. Hasegawa (1996a). Induced stresses and fault potential in Eastern Canada due to a disc load: a preliminary analysis, Geophys. J. Int. 125, 415–430. Wu, P., and H. S. Hasegawa (1996b). Induced stresses and fault potential in Eastern Canada due to a realistic load: a preliminary analysis, Geophys. J. Int. 127, 215–229. Zoback, M. L. (1992). Stress field constraints on intraplate seismicity in eastern North America, J. Geophys. Res. 97, 11,761–11,782.

Department of Earth Sciences University of Western Ontario London, Ontario, Canada N6A 5B7 [email protected] (S.M.)

Department of Geoscience University of Calgary Calgary, Alberta, Canada T2N 1N4 (D.W.E.)

Geological Survey of Canada Ottawa, Ontario, Canada K1A 0Y3 (J.A.)

Manuscript received 13 December 2007