The MacKenzie Mountains, Northwest Territories, earthquake sequence of 1953 to 1957 Open Access

The Fold and Thrust Belt of the northern Cordillera is one of the most seismically active regions in Canada. Understanding the seismic hazard in this area is critical for decision-making and design considerations for infrastructure design and development in the north (Hyndman et al. 2005). However, most of the largest earthquakes (that dominate the hazard in this region) occurred before modern seismic instrumentation was in place. Between January 1953 and December 1957, a series of M5.8–6.4 earthquakes struck the northern Mackenzie Mountains in the Northwest Territories. In this paper, we utilize regional and teleseismic waveforms from historical seismograms to determine the location, depth, magnitude, source mechanism, and aftershock characteristics of these earthquakes to better understand the seismicity and seismic hazards of this area.

The Fold and Thrust Belt of the Canadian Cordillera represents the eastern limit of Cordilleran deformation. The Mackenzie Fold and Thrust Belt resulted from Late Cretaceous to early Tertiary east–west compression (Cook et al. 1991). This curved foldbelt (Fig. 1) extends for approximately 950 km (Norris 1985) from the British Columbia–Northwest Territories border in the south, to the northern Yukon Territory. The northern part of the Mackenzie Fold and Thrust Belt in the vicinity of the 1953–1957 earthquake sequence is about 125 km wide. The structures are east–west striking and dominated by three broad anticlines that are cut by steeply dipping thrust faults. Offsets within the Proterozoic and Paleozoic strata provide evidence for dextral strike-slip movements during the formation of the Fold and Thrust Belt (Cook et al. 1991). The structural style here, with steeper thrust faults that project into the Precambrian rocks below, perhaps even to 20 km depth (Cook 2000), suggests a lower level detachment, in contrast to regions farther south in the Cordillera (south of ∼56°N, Cook et al. 1991). In the southern part of the Mackenzie Mountains, near the 1985 Nahanni earthquake sequence (Fig. 1), structures strike north–south. Some recent geophysical studies in this region are presented in Schutt et al. (2023), Audet et al. (2020), and Estève et al. (2021).

The Northern Fold and Thrust Belt of the Cordillera is one of the most seismically active regions in Canada. Small (M < 3) earthquakes occur on near-daily basis, and some of Canada’s largest historic earthquakes (M6–6.9) have occurred here (Fig. 1). These large earthquakes (more than 600 km from the active plate boundary along the west coast) represent a very different pattern from the observed seismicity in the Fold and Thrust Belt of the Southern Canadian Cordillera (see Cassidy et al. 2001). This raises the question “could large thrust earthquakes also occur in the southern Cordillera” and hence the importance of studying these large historic earthquakes in the northern Cordillera.

The largest earthquakes in the Northern Fold and Thrust Belt are the 1940 M6.2 and M6.5 earthquakes that occurred in the Richardson Mountains (see Cassidy and Bent 1993), the M5.8–6.4 northern Mackenzie Mountains earthquake sequence of 1953–1957 (documented here), and the 1985–1988 Nahanni earthquake sequence of the Southern Mackenzie Mountains (Choy and Boatwright 1988; Wetmiller et al. 1988; Horner et al. 1990; Ghahramani and Evans 2018). The 1940 Richardson Mountains earthquakes are strike-slip events that occurred only 6 days apart. They were associated with the north–south striking Richardson Fault System (Cassidy and Bent 1993) and had no recorded aftershocks. The 1985–1988 earthquake sequence, dominated by the M6.6 and M6.9 earthquakes of October and December, 1985, occurred in the southern Mackenzie Mountains (Fig. 1). These earthquakes, at 4–10 km depth, represent the largest events ever recorded in the area and were “surprise” events in that no earthquakes of that magnitude had occurred, or were expected in that region. These earthquakes represented thrust faulting along a north-striking, shallow west-dipping fault (Choy and Boatwright 1988; Wetmiller et al. 1988). The Nahanni earthquake sequence is similar to the 1953–1957 sequence, in that hundreds of aftershocks were recorded, including large (M5.8) aftershocks. It is noteworthy these large earthquakes in the Mackenzie Fold and Thrust Belt have occurred 50–150 km to the west of the deformation front.

Since 1985, there have been more than 224 M > 4 earthquakes in the MacKenzie Mountains region, including 26 of M > 5. All have shallow focal depths (<10 km) and the largest event (excluding Nahanni sequence aftershocks) was M5.7 on 5 March 2006. Focal mechanisms for the largest events, derived from regional centroid tensor (RCMT) inversion (Kao et al. 2012; Kao (personal communication, 2024)), shown in Fig. 2, are largely consistent with thrust faulting on shallow-dipping faults with a strike direction ∼parallel to the deformation front. Seismicity of the region (1985 to present) is shown in Fig. 1.

Data were obtained from the Canadian Seismogram Archives, the National Earthquake Information Center, Natural Resources Canada (NRCAN), and by making requests to stations known to be operating relatively high-gain instruments in the early to mid 1950s. A total of 26 seismograms, both long-period and short-period (in many cases three-component data), were obtained from 12 stations for the 1953 earthquake, 60 seismograms from 24 stations for the 1955 earthquake, 34 seismograms from 12 stations for the 1956 earthquake, and 39 seismograms from 10 stations for the 1957 earthquake. It should be noted that, for all earthquakes, the closest seismic station (Resolute (RES), NT) was at a distance of ∼1700 km. A list of stations used in this study (and instrument types) is given in the Supplement (Table S1). Arrival time information and in some cases first arrival polarities and earthquake magnitudes were obtained from the Canadian Earthquake Epicentre File (CEEF), bulletins compiled by the International Seismological Summary (ISS), the Bureau Central Institute de Seismologie (BCIS), the United States Coast and Geodetic Survey (USCGS), the Seismological Service of Canada (Seismological Service of Canada 1956a, 1956b, 1957,1959), and from individual seismograph stations (e.g., California Institute of Technology Seismological Bulletins).

Felt information for these earthquakes was collected from the Seismological Series of the Dominion Observatory (Meidler 1961; Smith 1961), from the US Geological Survey (Brockman et al. 1988) and the US Coast and Geodetic Survey, and by doing a search of newspapers from communities in the Yukon and Northwest Territories that are archived at the National Library of Canada. A number of articles were found and are summarized here. The felt area cannot be accurately defined due to the sparse population in the epicentral area. However, relative comparisons can be made between the felt intensities and areas for the earthquakes of the 1953–1957 sequence (Fig. 3). In addition, this felt information can be compared with intensities for other large historic events in the northern Cordillera, such as the 1940 Richardson Mountains M6.5 earthquakes (about 200 km to the north-northeast) and the 1985 M6.9 Nahanni earthquake (about 500 km to the southeast).

This earthquake occurred at 2:53 p.m. local time. The Whitehorse Star newspaper reported that it was felt in the communities of Elsa, Keno, and Mayo in the Yukon Territories (Fig. 3), about 230 km to the southwest of the epicentre. It was reported as a “severe tremor” that lasted for about 4 min causing telephone poles to sway, and cracking some windows in these communities. These reports indicate a Modified Mercalli Intensity (MMI) of IV–V. It was also felt (MMI IV) by several people in Northway, AK (∼520 km to the southwest) where light fixtures swayed gently, and a duration of “only a second” was estimated.

This earthquake occurred at 8:42 p.m. local time. Newspaper reports indicate that this earthquake was felt (but no details available) in the communities of Mayo and Keno, Yukon Territory (220 km southwest of the epicentre), Fort Good Hope, Northwest Territories (230 km to the northeast of the epicentre), and Aklavik, N.W.T. (340 km to the north-northwest of the epicentre). It was also felt (MMI IV) by several in Northway, AK (∼520 km to the southwest) where light fixtures swayed gently, and a duration of 2 s was estimated. This indicates a minimum felt area of ∼850 000 km². The earthquake was not felt at Dawson, Whitehorse, or Atlin, B.C. (distances of 350–640 km). There were no reports of any aftershocks being felt at these communities.

This earthquake occurred at 07:41 a.m. local time. The Whitehorse Star newspaper reported that a number of residents of Mayo were awakened by a slight tremor. This suggests an MMI of IV.

This earthquake occurred at 2:07 p.m. local time and was felt mildly (MMI of III–IV) in Mayo, YT. To compare with other significant earthquakes in the northern Cordillera, the 1940 Richardson Mountains earthquakes (M6.2, May 29 and M6.5, June 5) were felt to distances of 70 and 360 km, respectively (Cassidy and Bent 1993). The M6.5 June, 1940 earthquake had an approximate felt area of 410 000 km². The M6.9 December 1985 earthquake was felt to distances of more than 1500 km (Wetmiller et al. 1988), with an elongation of felt intensity contours in a northwest–southeast direction (∼parallel to the cordillera). The approximate felt area for these earthquakes was ∼1.5 million km² (October 1985) and >1.5 million km² (December 1985) (Wetmiller et al. 1988).

Previous magnitude estimates for these earthquakes are M_S 6.3 (International Seismological Commission (ISC)) and M_L 6.5 (CEEF) for the 1953 event, M_S 6.3 (ISC) and M_L 6.6 (CEEF) for the 1955 event, M_L 4.7 (CEEF) for the largest aftershock (1 March 1955 at 14:02 UTC) of the 1955 event, M_L 6.5 (CEEF) for the 1956 earthquake, and M_L 5.8 (CEEF) for the 1957 event.

Magnitudes determined in this study are Mw (moment magnitude of Hanks and Kanamori 1979) using seismic moment obtained by modelling broadband body waves (see “Source parameters” section for more detail). For completeness, we also include magnitude estimates (albeit poorly constrained) based on an estimate of the total felt area (see Toppozada 1975), a magnitude estimate based on the number of ISS stations reporting the event (see Johnston 1996) and, where available, recent International Seismological Centre–Global Earthquake Model (ISC–GEM) catalogue magnitude re-evaluations (Di Giacomo et al. 2015, 2018; Storchak et al. 2015). ISC–GEM catalogue magnitude “Quality factors” are as defined by Di Giacomo et al. (2018) and International Seismological Centre (2024).

The original CEEF magnitude of M_L 6.5 is believed adopted from BCIS and USCGS catalogues. ISS provides no magnitude estimate for this event, but BCIS indicates M6 (Berkeley, Uppsula, Strasbourg); M6.25 (Kiruna); M6.33 (Praha); M6.38 (Rome); and M6.5 (Pasadena). The more recent ISC–GEM catalogue (Di Giacomo et al. 2015) estimates a magnitude of Mw 6.31 ± 0.2 (with a quality factor B—as described in table 1 of Di Giacomo et al. 2018).

Our estimated M_L, based on an approximate felt area of ∼170 000 km², is 6.1. A magnitude estimate based on the number of stations reporting to the ISS (214 stations) is 6.0. Our estimated Mw, based on seismic moment, Mo, estimated from body wave modelling, is 6.3. Our preferred magnitude for this event is Mw 6.3.

The original CEEF magnitude of M_L 6.5–6.75 is believed adopted from BCIS and USCGS catalogues. ISS provides no magnitude estimate for this event, but BCIS indicates M6 (Skalnate, Pleso); M6.75 (Kiruna); M6–6.25 (Praha); M6.4 (Rome); M6.5–6.75 (Pasadena); and M6.5 (Uppsala). The more recent ISC–GEM catalogue (Di Giacomo et al. 2015) estimates a magnitude of Mw 6.35 ± 0.39 (with a quality factor C—lower quality, as described in table 1 of Di Giacomo et al. 2018) and also lists a M6.8 from station PAS.

Our estimate of M_L, based on an approximate felt area of ∼360 000 km², is 6.6, and a magnitude estimate based on the number of stations reporting to the ISS (234 stations) is 6.3. Our Mw, based on Mo estimated from body wave modelling, is 6.4. Our preferred magnitude for this event is Mw 6.4.

The CEEF magnitude for this aftershock is M_L 4.7. There are no details provided in the CEEF catalogue, but this magnitude is likely based on the RES recording only, as was the case for other events in this sequence. The ISC–GEM catalogue (Di Giacomo et al. 2015) has a location for this event, but no magnitude estimate.

The ISS, BCIS, and USCGS do not provide a magnitude estimate for this event.

A magnitude estimate based on the number of stations reporting to the ISS (87) is 5.2. There are no reports of this aftershock being felt, suggesting that the magnitude of this aftershock is less than ∼M_L 5.8 (the magnitude of the 9 December 1957 felt event). Our estimated Mw, based on Mo estimated from body wave modelling, is 5.9.

The CEEF magnitude of M_L 6.5 for this earthquake was obtained from unpublished index cards by W.E.T. Smith (a federal government seismologist in the 1960s)—known as “Smith’s cards”—and appears to be entirely based on a single measurement at RES—a distance of ∼1700 km and recorded on an SPZ Sprengnether seismograph.

This event is not listed in the ISS, and the BCIS and USCGS do not provide a magnitude estimate. There is no ISC–GEM magnitude estimate for this event.

Our M_L estimate, based on an approximate felt area of ∼170 000 km², is 6.1, and Mw, based on Mo estimated from body wave modelling, is 5.8. Our preferred magnitude for this earthquake is Mw 5.8.

The original CEEF magnitude of M_L 6.7 from information on Smith’s cards (Earthquakes of the Canadian Arctic) is based on a single measurement at RES—at a distance of 1700 km and recorded on an SPZ Sprengnether seismograph. In the 1988 CEEF, this magnitude was listed as M_L 5.7, and it is not clear what prompted the change. We note, however, that there was a change in the instrumentation/vault that occurred just before this earthquake, on 7 November 1957 (Seismological Service of Canada 1959).

The ISS and USCGS provide no magnitude estimate for this event. The BCIS provides estimates of 5.75 at Matsushiro, Japan, 5.3 at URSS, and “5.1 (?)” at Quetta, Pakistan. The ISC–GEM catalogue (Di Giacomo et al. 2015) provides a location, but no magnitude estimate for this event.

Our M_L estimate, based on an approximate felt area of ∼230 000 km², is 6.1, and a magnitude estimate based on the number of stations reporting to the ISS (106 stations) is 5.3. Our Mw estimate (and preferred magnitude), based on Mo estimated from body wave modelling, is 5.8.

The locations of these earthquakes, like other historic events, were based on global seismic data and early processing by the ISS, BCIS, or the USCGS in the pre-computer era. Solutions provided in the CEEF are largely based on these early teleseismic solutions (e.g., Seismological Service of Canada 1956a, 1956b, 1957,1959), sometimes updated with additional readings (for arrival times and magnitude estimates) from Smith’s unpublished index cards (Smith was a Canadian federal government seismologist in the 1960s). Table 1 and Fig. 4 provide a summary of all known solutions for these earthquakes. Note that many of these solutions are not independent—for example, for some of these events the BCIS solutions are taken from the USCGS catalogue, and some of the CEEF catalogue solutions are based on the ISS/BCIS/USCGS solutions. The ISC–GEM (Di Giacomo et al. 2015, 2018) solutions are modern estimates based on a reevaluation of global seismic data. In our study we relocate these historic earthquakes in both a relative and absolute sense using some modern “calibration events” to better constrain the teleseismic solutions.

Given the lack of seismic stations in the epicentral region in the 1950s (the closest Canadian seismic station was at RES, NT, ∼1700 km away), estimating the locations of these earthquakes requires primarily teleseismic data. To estimate earthquake locations, we have used the program “EPDET”, an iterative least-squares program (Weichert and Newton 1970) with both P- and S-wave travel times. The Dziewonski and Andreson (1981) “PREM” earth model (which, compared to other earth models such as the Jeffreys-Bullen (1958) model, provided the lowest average residuals) was used for the locations. Initially, stations having travel-time residuals greater than 60 s were rejected. Then, stations were successively culled at residual thresholds of 10.0, 6.0, and 4.0 s. To improve the teleseismic locations, we have utilized four well-recorded modern events (with the largest and best-recorded being the 5 February 2001 M5.3 earthquake at 64.33 -130.91 located ∼120 km to the southeast) and computed “path corrections” that were applied to the historic earthquakes (teleseismic waves having the same travel path). As an example, modern events have much more precise locations based on local and regional data. Using the same earth model and teleseismic location method that we use for the historic earthquakes, and using a similar data distribution, we find that the modern events are consistently mislocated to the north by ∼5–25 km.

We use the fixed location of the modern event to calculate “path corrections” for each of the stations (209) that recorded this event. Then, for each of these teleseismic stations that recorded the 1953–1957 earthquakes, the path corrections were applied, and the historic events were relocated. This resulted in a reduced data set (14–27 stations for each solution) but, we believe, a more accurate location estimate. These locations are provided in Table 2 and are shown in Fig. 4.

For historic earthquake relocations, it is important to remember that arrival time picks are made from old, low-resolution paper records. The largest factor influencing the accuracy of the locations is the uncertainty in the arrival time picks, and these can easily be as large as ±4 s. Timing uncertainty at individual seismic stations is another important factor. The lack of local or regional stations and lack of details in earth models further adds to the location uncertainties. While the locations presented here are considered more accurate than those in the historic catalogues, they are still subject to uncertainties on the order of 5–10 km (or more). As local network densification increases, better earth models and more well-recorded earthquakes become available, the locations of these historic earthquakes could improve over time, with more detailed studies.

As a quick check on the earthquake locations (in a relative sense), we do a simple comparison of seismic waveforms at common stations. For example, a comparison of waveforms for the two largest earthquakes (1953 and 1955 mainshocks) shows that all waveforms are very similar, including short-period vertical components from a selection of seismic stations around the world. Seismic recordings (Fig. 5) including both the P- and S-waves for the 1953 and 1955 mainshocks recorded at station WES (north–south component) show both waveform similarity (for P-, S-, and intermediate phases) and there is a 1 s (or less) S–P time at this station, requiring the two events to be within ∼10–20 km of one another.

Focal mechanism solutions based on P- and SH-wave first polarity data (from the ISS, USCGS, and Seismological Series of the Dominion Observatory (Canada) bulletins and as read directly from seismograms, where available) were determined for the two largest earthquakes in the sequence (1953 and the 1955 mainshock) using the program FOCMEC and grid search method of Snoke et al. (1984). For both events, a thrust-faulting solution was obtained, consistent with those determined using body-wave modelling techniques.

Trial-and-error forward modelling of teleseismic P- and SH-waveforms (see Supplement for stations used) was employed to constrain the source depth, focal mechanism, source time function, and seismic moment of the mainshock and the largest aftershocks. The method of Langston and Helmberger (1975) was used with a simple layer over a half-space earth model to generate synthetic seismograms. For body wave attenuation, we used a Futterman (1962) operator, t^*, of 1.0 s for P-waves and 4.0 for SH-waves. We concentrated on modelling the displacement body waveforms to determine the source parameters of this event. We also considered some higher-frequency waveforms to more accurately constrain the focal depth, and to examine source complexity. More information on the stations/instruments used for the analysis is provided in the Supplement (Table S1).

A total of 22 teleseismic waveforms having a good azimuthal distribution were modelled. This includes eight long-period P-waveforms, five short-period P-waveforms, and nine SH-waveforms. The results of the body-wave modelling are shown in Fig. 6. The waveforms were best fit using two subevents, each having a triangular source-time function with a rise time of 1 s. The second subevent (Fig. 6) is only slightly larger than the first, and 3 s later than the first. This double source is required to match the double pulses observed in the short-period waveforms (e.g., see TUC, CSC, WES, etc.). Both subevents have the same thrust mechanism (strike of 116°, dip angle of 62°, and a slip of 100°) and a focal depth of 9 km. The total long-period seismic moment was 3.1 × 10¹⁸ N m.

A total of 24 teleseismic P- and SH-waveforms covering a wide azimuthal and distance range were modelled for this earthquake. Select observed and synthetic waveforms are given in Fig. 7. These waveforms were well-fit with a single source duration of seconds. A thrust focal mechanism (strike of 116°, dip angle of 52°, and slip of 90°) and a focal depth of 9 km were obtained. The seismic moment estimated from long-period waveforms is 3.9 × 10¹⁸ N m.

A total of eight teleseismic P-waveforms were modelled (three long-period, five short-period) for the largest aftershock of the 1 March 1955 earthquake. Although the waveforms are very similar to the mainshock, there are a few differences. For example, the P-wave polarity reversal and very emergent arrivals at VIC and HBC (Fig. 8) suggest a slight change in the mechanism to strike 116°, dip 57°, slip 90°. There are also slight differences in the TUC LPZ waveform—suggesting a slightly greater depth of 11 km for this aftershock, compared to 9 km for the mainshock, and a source time function that is nearly 4 s in duration. The seismic moment estimated for this, the largest aftershock, is 9.0 × 10¹⁷ N m.

A total of 14 teleseismic P- and S-waveforms (6 long-period and 8 short-period) were modelled for this earthquake. The results (shown in Fig. 9) require a thrust mechanism with a strike of 116°, dip angle of 62°, and a slip of 90°. The waveforms were best fit using a focal depth of 11 km, slightly greater than that of the other events modelled, and a source time function of 2 s duration. The total seismic moment, based on long-period body waves, is 5.5 × 10¹⁷ N m.

A total of 12 teleseismic P- and S-waveforms (6 long-period and 6 short-period) were modelled for this earthquake. The results (shown in Fig. 10) also require a thrust mechanism with a strike of 116°, dip angle of 52°, and a slip of 110°. The waveforms were best fit using a focal depth of 7 km, and a source time function of 2 s duration. The total seismic moment, based on long-period body waves, is 5.1 × 10¹⁷ N m.

It is very likely that all of these moderate to large earthquakes (1953 through 1957) are linked in some way, similar to the 1985–1988 Nahanni earthquake sequence. Given the very sparse seismograph coverage at the time of the 1953–1957 earthquakes, only relatively large (e.g., ∼M4.5+) events would have been detected at regional or teleseismic distances. Here, we provide the basic information (as retrieved from the NRCAN earthquake catalog, the BCIS, USCGS, and ISS catalogs) on aftershocks for each of the earthquakes described in this article. For the 1 March 1955 mainshock, the first author reviewed the original paper seismograms from RES (a station that was consistently used for evaluating earthquake parameters from this region) for a period of 1 year following the mainshock to identify possible aftershocks. More information on aftershocks and sources of information is provided in the Supplement.

There are relatively few aftershocks found for this event. There were no ISS entries that could have been possible aftershocks, but the NRCAN catalogue (also USCGS and BCIS) list a total of five likely or potential aftershocks (13 January, two events on 12 February, and 10 January 1954—see Supplementary Files. No magnitude estimates are available for these events.

There are many confirmed aftershocks for this earthquake, including one (at 14:02) as large as Mw5.9 (and described in previous sections). The largest, and best recorded aftershocks (from USCGS, BCIS, and Meidler (1961) include events at 08:48 and 14:02 on 1 March, and events on 22 March, 29 March, 8 May, 26 August, and 22 October 1955 (see Supplement).

Aftershocks were also evaluated by looking at the RES original seismograms (the closest Canadian station) for 1 year after the event (see Supplement). RES is ∼1700 km from the mainshock location, and with an S–P time of ∼160 s. Considering an S–P time of ∼150–170 s, we find a total of 26 likely aftershocks (over the course of a year) recorded at RES, and an additional 18 events with emergent P arrivals and so a poorly constrained S–P time. In all cases, these would be significant aftershocks to be recorded at this distance of 1700 km. There are no felt reports for these events, due to the lack of communities in the region.

There are a few significant events (Smith 1961) in the epicentral region of the 7 January 1956 earthquake that may be aftershocks (or perhaps separate earthquakes, due to the time delay?). These events (see Supplement) occurred on 1 August 1956 (M5.8 OTT) and 30 January 1957 (M5.8 OTT). Neither of these events was listed in the ISS, but both are included in the USCGS catalogue.

There are very few aftershocks noted for this event—the most obvious being a significant (originally listed as M5.7 based on RES amplitudes) event on 10 December at 08:15 UT with the annotation “similarity of records at RES shows this to be an aftershock of Dec. 9, 1957 22:07”. This was subsequently downgraded to M_L4.8 in the1988 CEEF. Smith (1961) also notes two events in October 1958 in the epicentral region of the 1957 earthquake. Those occurred on 11 October (00:41 UTC) 1958 and listed as M5.7 (OTT) and 26 October (15:24 UTC) 1958 and listed as M5.6 (OTT). Both of these events are also listed in the USCGS.

The detailed analysis of this earthquake sequence (M_W 5.8–6.4) sheds new light on the details of each of these significant earthquakes, as well as the overall seismic hazards of the northern Canadian Cordillera. The 1953 and 1955 earthquakes, which occurred in a sparsely populated region, were felt (up to intensity V) to distances of 230–280 km. The magnitudes (Mw) of these events have been determined at 6.3 and 6.4, respectively. Epicentres have been re-determined using teleseismic path corrections obtained from well-recorded modern earthquakes (M4.5–5) in this region, and by waveform comparison techniques. The relocated epicentres are distributed in an east-southeast west-northwest direction over a distance of about 45 km. The first earthquakes in the sequence are located to the east-southeast and later events migrate to the west-northwest. The 1955 mainshock and largest aftershock are separated by about 20 km, with the aftershock being located to the north-northeast of the mainshock. The 1957 (Mw5.8) earthquake is located a similar distance and direction from the 1956 (Mw 6.1) earthquake, suggesting that it may be an aftershock of that event. Focal mechanisms and depths were determined using body wave modelling. All depths ranged between 9 and 15 km (in a region where the Moho depth is 36 km or more, as mapped by Audet et al. 2020), and focal mechanisms show thrust faulting along either a plane dipping at 30–40 degrees to the north-northeast, or a plane dipping at 50–60 degrees to the south-southwest. The orientation of the west-northwest east-southeast striking nodal planes is consistent with the distribution of epicentres and the regional geology. The north-northeast south-southwest striking, shallow dipping P axes are consistent with the regional stress field. Each of these earthquakes was followed by aftershocks (in some cases as large as Mw 5.9) as would be expected for shallow, thrust events. It is likely that these earthquakes are caused by stresses transmitted from the plate boundary more than 700 km to the west, activating favourably oriented thrust faults in the northeastern Canadian Cordillera. The 1953–1957 earthquake sequence in the northern MacKenzie Mountains is very similar to the 1985–1988 sequence in the Nahanni region, to the southeast. That sequence also consisted of large earthquakes (M5–6.9) continuing over a 3-year period with a rich aftershock sequence. Those were also shallow, thrust earthquakes in the Fold and Thrust Belt that were likely triggered by the transmission of crustal stress from the active plate boundary, hundreds of kilometres to the west, as suggested in Mazzotti and Hyndman (2002),Hyndman et al. (2005), and Hyndman (2023). It is noted that both the 1950s earthquake sequence presented here, and the 1985 Nahanni sequence, occurred in regions of relatively high crustal velocity (Estève et al. 2021; Schutt et al. 2023). Future studies could incorporate more detailed recent studies of crustal structure, heat flow, crustal stress, potentially active faults, and seismicity patterns to better understand the differences (and/or similarities) in seismic hazard between the more active Fold and Thrust Belt of the northern Canadian Cordillera and (what appears to be) lower seismic hazard in the southern and central portions of the Fold and Thrust Belt of the Canadian Cordillera (Cassidy et al. 2001).

The first author initiated this research as a Visiting Fellow at the Geophysics Division, Geological Survey of Canada. We thank Honn Kao for providing us with the most recent version of his Regional Centroid Moment Tensor catalog, and NRCan colleagues Camille Brillon for providing Fig. 4 and Robert Kung for providing Fig. 1. We thank Garry Rogers, Brindley Smith, Dianne Doser, and one anonymous reviewer for their very thoughtful and helpful reviews that have improved this manuscript. We are forever grateful to Dr. Roy Hyndman for his decades of encouragement, enthusiasm, and support for studying earthquakes, old and new, in this region.

Data generated or analyzed during this study are available from the corresponding author upon reasonable request.

Conceptualization: JFC

Formal analysis: JFC, ALB

Investigation: JFC

Methodology: JFC, ALB

Project administration: JFC

Writing – original draft: JFC, ALB

Writing – review & editing: JFC, ALB

This work was entirely funded by Natural Resources Canada, Geological Survey of Canada.

Supplementary data are available with the article at https://doi.org/10.1139/cjes-2024-0132.