Sage Journals: Discover world-class research

Abstract

Canada’s 2020 national seismic hazard model (CSHM 2020) provides hazard estimates for interface Cascadia Subduction Zone (CSZ) earthquakes in southwestern Canada using four ground motion models (GMMs) with equal weights. Two out of the four GMMs were derived using data primarily from subduction earthquakes in Japan so their use in CSHM 2020 includes a “Japan-to-Cascadia” factor to account for local site conditions. Despite this regional factor, the GMMs do not explicitly consider the amplification effects from the Georgia sedimentary basin below Metro Vancouver. This study benchmarks ground motion shaking from a suite of 30 physics-based simulations of M9 CSZ earthquakes, which explicitly account for basin effects for periods exceeding 1 s, to corresponding estimates from interface GMMs in CSHM 2020. Using this comparison, this study proposes site-specific and period-dependent basin amplification factors for a range of sites located within the Georgia sedimentary basin. The average basin amplification factors at the deepest basin location reach values as high as 2.24 and 6.29 at a 2 s period with respect to basin-edge and outside-basin reference locations, respectively. We evaluate the proposed factors by comparing them against long-period amplifications observed in empirical strong motion recordings from the 2001 M6.8 Nisqually earthquake. A framework to incorporate the proposed basin amplification factors in the calculation of CSHM 2020 uniform hazard spectra (UHS) is also proposed. The modified UHS ordinates at the deepest basin location reach values that are 58% to 271% higher, at a 2 s period, than non-basin UHS estimates when amplification factors are calculated with respect to basin-edge and outside-basin reference locations, respectively.

Keywords

Cascadia Subduction Zone simulated ground motions deep sedimentary basins Georgia sedimentary basin basin amplification factors

Introduction

Ground motion shaking intensity can increase significantly due to the amplification effects of deep sedimentary basins. These amplification effects are most notable at longer periods, which can have detrimental impacts on medium-to-long period buildings and infrastructure. For instance, observations from the 1985 Mexico City earthquake showed there was a strong correlation between the heavily damaged areas and the basin depth at those sites (Kobayashi et al., 1986). Historically, basin amplification effects have not been explicitly accounted for in national seismic hazard models, and thereby not accounted for in building codes. However, in recent years, the amplification effects of deep sedimentary basins on long-period earthquake ground motions have been studied using recorded (Graves et al., 1998; Rekoske et al., 2021) and simulated motions (Choi et al., 2005; Day et al., 2008; Kakoty et al., 2021; Marafi et al., 2019; Molnar et al., 2014a, 2014b; Olsen, 2000) across various geographical regions. The United States Geological Survey (USGS) has included basin effects (for periods exceeding 1 s) in the 2018 version of the US National Seismic Hazard Model (Petersen et al., 2020) and basin effects are now included in the building code, that is, ASCE 7-22 (American Society of Civil Engineers (ASCE), 2022). While progress has been made in the United States, Canada’s national seismic hazard model and the National Building Code of Canada (NBCC) do not explicitly account for these effects.

Graves et al. (1998) evaluated the effect of the basin-edge structure in ground motion amplification in the Santa Monica area using observations from the M6.7 1994 Northridge earthquake. Basin amplification factors (BAFs) were calculated by normalizing simulated spectral response values by the response calculated at a rock site adjacent to the basin. The study found large amplifications at the shallow basin-edge structure (1 km deep) due to basin-edge effects (i.e. constructive interference of direct waves with the basin-edge generated surface waves). A recent study by Rekoske et al. (2021) quantified basin amplification in the Puget lowland and the Seattle basin using a large set of recorded crustal and inslab strong-motion data. Unlike the study by Graves et al. (1998) that calculated basin amplifications using spectral response values, this study used Fourier amplitude spectrum (FAS) of ground acceleration to estimate amplification. Reference ground motions were taken as predicted values by the Bayless and Abrahamson (2018) ground motion model (GMM), which was developed for crustal earthquakes in California. Their results suggest maximum average BAFs in the Seattle basin of approximately four for both crustal and inslab earthquakes.

The advent of robust physics-based simulations of earthquake ground motions have facilitated studies of the expected level of basin amplification as a function of basin properties (Choi et al., 2005; Day et al., 2008; Molnar et al., 2014a, 2014b; Olsen, 2000), as well as studies of the resulting impact of basin amplification on the seismic performance of buildings (Bijelić et al., 2019; Kakoty et al., 2021; Kourehpaz et al., 2021; Marafi et al., 2017, 2019; Monfared et al., 2021). Olsen (2000) quantified basin amplification in the Los Angeles basin as a ratio of peak velocity obtained from three-dimensional (3D) simulations (considering basin effects) and predictions using a regional one-dimensional (1D) model (neglecting basin effects) for nine shallow crustal scenario earthquakes. This study estimated average amplification factors up to a factor of four, with considerable variation across individual scenarios (by a factor of up to 2.5). Olsen (2000) also found that average basin amplification correlated well with basin depth and that the amplification generally increased with increasing basin depth expressed in terms of Z_2.5, that is, the depth to soils with a shear wave velocity of 2.5 km/s.

Choi et al. (2005) investigated the relationship between basin amplifications and basin depth, as well as the location of the seismic source relative to the basin extents. Amplification factors for 5% damped response spectral acceleration (SA) were derived from strong-motion recordings (from basin and non-basin sites in southern California and the San Francisco Bay Area). The basin depth parameter considered was Z_1.5, that is, the depth to soils with a shear wave velocity of 1.5 km/s. The basin edge was defined by a Z_1.5 boundary of 500 m. BAFs were calculated as the ratio of SA with simulations that included the 3D basin geometry, and simulations that utilized a 1D layered-rock model. The reference 1D model used in the simulations had a surface shear velocity of about 3000 m/s (i.e. seismological rock). This is significantly higher than typical engineering rock, which has a value around 600–700 m/s.

Day et al. (2008) developed a model for basin effects on long-period response spectra for the Los Angeles basin by isolating the effects of basin depth and period. The authors used 3D numerical simulations of long-period (2–10 s) ground motions for a suite of 60 scenario earthquakes ranging from M6.3 to M7.1. Amplification factors were calculated with reference to a horizontally stratified 1D model. The stratified reference model corresponds to an artificially high velocity, that is, a hard-rock site constructed by laterally extending a vertical profile of the Southern California Earthquake Center (SCEC) community velocity model (CVM; Magistrale et al., 2000) in the San Gabriel Mountains. The source-averaged effect of basin depth on SA was characterized using Z_1.5 as the predictor variable. This study also found good correlation between basin amplification effects and basin depth, with maximum amplification factor of eight at deepest basin sites.

In the US Pacific Northwest, Marafi et al. (2019) computed BAFs in the Seattle basin using 30 physics-based simulations of M9 CSZ earthquakes by Frankel et al. (2018). These simulations use the Pacific Northwest CVM (Stephenson et al., 2017) to explicitly account for basin effects. SA values from the M9 simulations were normalized by the SA predictions from the BC Hydro GMM (Abrahamson et al., 2016), which does not explicitly account for the effects of basins on SA. The computed BAFs were dependent on a reference Z_2.5 value or Z_2.5,o. For a Z_2.5,o value of 1 km, average BAFs reached a maximum value of 6.5 at a period of 3 s. For a Z_2.5,o of 3 km, the BAFs reached a maximum value of 2.4 at 4 s, around 62% lower than the basin amplification computed for a Z_2.5,o value of 1 km. This work highlighted the impact of the choice of reference site on the resulting BAFs. Using an updated version of the Pacific Northwest CVM (Stephenson et al., 2017) that integrates local high-resolution model of the Georgia sedimentary basin (Molnar et al., 2014a), Roten et al. (2020) generated a suite of M9 CSZ ground motions to study peak ground velocities, and frequency-dependence of near-surface plasticity and rupture directivity effects. The authors found that plasticity of near-surface sediments have minimal effects on shaking at frequencies below 1.25 Hz, and rupture directivity can have significant impacts on shaking in the basin at lower frequencies $(\leq 0.4 Hz)$ .

In the Georgia sedimentary basin, studies by Molnar et al. (2014a, 2014b) estimated average peak ground velocity (PGV) BAFs of 4.1 and 3.1 for shallow blind-thrust North America plate and deep Juan de Fuca plate scenario earthquakes, respectively. Kakoty et al. (2021) quantified BAFs for a series of sites in the Georgia sedimentary basin by comparing SA estimates from the simulated M9 CSZ ground motions by Frankel et al. (2018) with corresponding estimates from the BC Hydro GMM (Abrahamson et al., 2016). This article extends the work by Kakoty et al. (2021) by incorporating SA estimates from all interface earthquake GMMs in Canada’s 2020 national seismic hazard model (CSHM 2020; Kolaj et al., 2020b). This article also explores the effects of reference sites used for estimating BAFs and presents the findings with respect to two sets of reference locations clustered according to their location relative to the basin. A framework to incorporate the period-dependent and site-specific BAFs in uniform hazard spectra (UHS) is also proposed, such that these effects can be applied to SA values currently used in design practice (National Research Council Canada (NRC), 2020). Furthermore, the computed BAFs are compared against observed long-period amplification (LPA) from the M6.8 2001 Nisqually earthquake to understand the expected amplifications from the simulations and limited empirical data. SA amplification from empirical observations, that is, the inslab 2001 Nisqually earthquake recordings, are referred to as LPA to differentiate them from the simulated BAF from the M9 ground motions that represent subduction interface earthquakes. Although both parameters represent amplification in SA at basin sites, LPA is from one scenario only (2001 Nisqually earthquake) and can have contributions from basin structure, azimuthal direction, rupture parameters, and so on. By contrast, these effects are averaged in the calculation of BAF by utilizing 30 different simulated scenarios.

Canada’s sixth generation seismic hazard model

As an essential part of the national seismic risk reduction strategy, Natural Resources Canada (NRCan) has been periodically updating Canada’s national seismic hazard model since its first edition was introduced in 1953 (Kolaj et al., 2020a). The first generation of the seismic hazard model was deterministic in nature, that is, it was created with qualitative assessments of previous earthquakes and the potential spatial extent of the damage. In 1970, the first fully probabilistic assessment of seismic hazard was incorporated into the national seismic hazard model with a single intensity measure of peak ground acceleration (PGA) and a probability of exceedance of 40% in 50 years (return period of ∼100 years). In the 1985 version of the seismic hazard model, the probability of exceedance was lowered to 10% in 50 years (return period of ∼475 years) and one additional intensity measure, PGV, was added. The following three editions (released in 2005, 2015, and 2020) of the seismic hazard model introduced additional intensity measures, including SAs at different periods. The probability of exceedance of the seismic hazard was further reduced to a 2% probability of exceedance in 50 years (return period of ∼2475 years). The capabilities of the seismic hazard model have increased multi-fold with the introduction of epistemic uncertainty, increased number of maps, hazard estimates at multiple probabilities of exceedance, and so on. Seismic hazard estimates from the national seismic hazard model are adopted by the NBCC for use in building design.

The 2005 version of the seismic hazard model included, for the first time, the Cascadia Subduction Zone (CSZ; Kolaj et al., 2020a), a 1000 km long megathrust fault spanning from Northern Vancouver Island, BC to Cape Mendocino, CA, with an estimated average recurrence of ∼500 years (Atwater et al., 1995). However, it was not until the 2015 version that the hazard estimates from the CSZ were incorporated within a probabilistic framework. The 2020 national seismic hazard model (Kolaj et al., 2020b), also referred to as the sixth generation seismic hazard model, provides hazard estimates from interface CSZ earthquakes in southwestern Canada using four GMMs with equal weights: Abrahamson et al. (2016), Zhao et al. (2006), Ghofrani and Atkinson (2014), and Atkinson and Macias (2009). This is in contrast to the previous edition of the seismic hazard model where the Atkinson and Macias (2009) GMM was given a weight of 50%, and the other models were assigned unequal weights on the basis of their perceived relevance to the CSZ. Because the studies by Zhao et al. (2006) and Ghofrani and Atkinson (2014) were derived using data primarily from subduction earthquakes in Japan, their use in Canada’s national seismic hazard model includes a “Japan-to-Cascadia” factor (Atkinson and Adams, 2013) to account for the average site and path differences between Japan and Cascadia. These factors were obtained by taking the average of two sets of factors available in the literature by (1) Atkinson and Boore (2003), who calculated the factors by considering region-based residuals to global GMMs for subduction events, and (2) Atkinson and Casey (2003), who examined these factors in detail by comparing recorded motions for an event in Japan to those for a comparable event in western North America. Despite this regional factor, the GMMs do not explicitly consider basin amplification effects from the Georgia sedimentary basin below Metro Vancouver. Table 1 lists the four GMMs considered in Canada’s sixth generation seismic hazard model to characterize ground motion shaking from CSZ earthquakes, including features of each of the GMMs.

Table 1.

List of ground motion models (GMMs) used in CSHM 2020 to characterize the hazard from subduction interface earthquakes

GMM	Weight	V_s30 (m/s)^a	T_max (s)^b	Japan-to-Cascadia Factor^c
Abrahamson et al. (2016)	0.25	180–1500	10	No
Zhao et al. (2006)	0.25	180–1500	5	Yes
Ghofrani and Atkinson (2014)	0.25	180–1500	9	Yes
Atkinson and Macias (2009)	0.25	360–1500	5	No

V_s30 (time averaged shear wave velocity at 30 m depth) values are interpreted from site class descriptors from NEHRP Site Classification (Building Seismic Safety Council (BSSC), 1994). The range in the table indicates the V_s30 at the sites from which data were considered for a particular GMM.

T_max is the maximum period for which coefficients to calculate SA are defined in the GMM.

Japan-to-Cascadia factors range from 0.44 (at T = 0.04 s and T = 0.1 s) to 1.51 (at T = 2 s) as per recommendations from Atkinson and Adams (2013).

As seen in Table 1, only two of the GMMs estimate SAs at periods greater than 5 s, and only one estimates shaking at a period of 10 s. Estimates of SA at long periods are of special significance because the CSZ dominates the seismic hazard at long periods and the Metro Vancouver region in southwest British Columbia has a high concentration of tall buildings and other long period infrastructure (Kakoty et al., 2021; Monfared et al., 2021). SA estimates between T_max and 10 s are obtained by extrapolation of the ratio of SA at T_max to be consistent with the study by Abrahamson et al. (2016), which is defined up to 10 s (i.e. SA estimates beyond T_max follow the same slope as that of the study by Abrahamson et al. 2016).

Simulated M9 CSZ earthquake ground motions

The paucity of quantitative observations of CSZ interface earthquakes led a collaborative group of researchers from the USGS and the University of Washington (UW) to develop a set of 30 simulated M9 CSZ earthquake ground motions at closely spaced grid points across the Pacific northwest region as part of the USGS-UW M9 project (Frankel et al., 2018). The simulated ground motions were derived from 3D finite-difference simulations for periods exceeding 1 s that explicitly considered basin effects, and stochastic synthesis for periods lesser than 1 s. The methodology to produce the simulated M9 ground motions was validated using recorded ground motions from the Maule and Tohoku earthquakes (Frankel et al., 2018). However, the authors acknowledged the limitation of the method to capture basin edge–generated surface waves, especially at 1–2 s. A logic tree of rupture parameters (i.e. down-dip rupture extent, hypocenter location, slip distributions, and subevent locations) was adopted in the development of the set of simulations to systematically account for the variation in the expected level of ground motion shaking. These synthetic ground motions were developed using the Pacific Northwest CVM developed by the USGS (Stephenson et al., 2017), which explicitly considers deep sedimentary basins in the region (e.g. Seattle basin, Georgia sedimentary basin, and so on.). The velocity model was utilized in the 3D finite difference simulations, thus the simulated ground motions account for basin amplification effects at periods greater than 1 s. While basin amplification is not explicitly considered at periods less than 1 s, empirical ground motion recordings at basin sites examined by Wirth et al. (2019) suggest that basin amplification is prominent only at longer periods and not significant in the 0–1 s period range.

Georgia sedimentary basin

The Georgia sedimentary basin in southwest British Columbia is one of a series of basins extending from California to southern Alaska along the Pacific margin of North America (England and Bustin, 1998) and is relatively shallow and wide (130 by 70 km wide by 5 km in depth) compared to basins further south in Seattle, WA (75 by 30 km wide by 8 km in depth; Frankel et al., 2007) and Los Angeles, CA (50 by 30 km wide by 5 km in depth; Magistrale et al., 1996). Properties of the Late Cretaceous and Tertiary sedimentary rocks within the Georgia basin and its basement are known from seismic surveys (White and Clowes, 1984), particularly seismic tomography results of the 1998 Seismic Hazards Investigations in Puget Sound experiment (Brocher et al., 2001; Dash et al., 2007; Ramachandran et al., 2004a). The presence of the Georgia basin is expected to cause amplification of ground motions in Metro Vancouver, based on amplification observed in areas with sediment-filled basins (Campillo et al., 1989; Frankel et al., 2009; Olsen, 2000). The amplifications are likely to occur in the medium-to-long period range and can have adverse consequences on the seismic performance of structures, particularly with periods in the 1–5 s range (Kakoty et al., 2021). In previous studies, average Georgia BAFs were estimated to be 4.1 and 3.1 for shallow blind-thrust North America plate and deep Juan de Fuca plate scenario earthquakes, respectively (Molnar et al., 2014a, 2014b). These amplification factors determined by Molnar et al. (2014a, 2014b) are the ratios of the synthetic PGV ground motions for the same earthquake scenario using a velocity model that includes the Georgia basin to that of a non-basin velocity model (depth to primary wave velocity (V_p) of 5.5 km/s).

The USGS-UW M9 project uses a CVM of the Pacific Northwest (Stephenson et al., 2017), originally developed by Stephenson (2007), spanning from northern California to southern British Columbia with dimensions of 337.5 km north-south by 200 km east-west by 60 km up–down and a spatial uniform-grid resolution of 500 m. The physical model is represented by six geologic units (continental basin sediments, crust and mantle, as well as oceanic sediments, crust, and mantle) characterized by unique primary wave velocity (V_p), shear wave velocity (V_s), and density. The oceanic crust is set to a thickness of 5 km. The 3D sedimentary basin structure in the Georgia basin region is primarily constrained by the tomographic V_p model (1 km grid; Ramachandran et al., 2004b, 2006). The V_p/V_s ratio for Quaternary basin sediments varies from 2.5 at the surface to 2.2 at 1 km depth. Tertiary sediments are set to a V_p/V_s ratio of 2, and their base is taken as the 4.5 km/s isovelocity contour (Ramachandran et al., 2006). Densities are derived from the V_p model using the Nafe–Drake relation (Ludwig et al., 1970). Surface topography is not included. The minimum V_s is set to 600 m/s for computational feasibility. The surface of the 3D basin model, therefore, represents over-consolidated Pleistocene and older glaciated sediments or stiff soil sites, that is, post-glacial sediments are not included in the model, the thickest of which occur in the Fraser River delta (up to 300 m) in southwestern Greater Vancouver. For 3D simulations of M9 ground motion, Frankel et al. (2018) applied a V_s gradient in the top 200 m of the velocity model by Stephenson et al. (2017) for hard rock areas outside of the Puget Lowland. Therefore, an effective V_s30 of 960 m/s is obtained for hard rock sites with Z_2.5 < 200 m. Wirth et al. (2021) and Hu et al. (2022) showed that calibrated geotechnical layers enhance the accuracy of expected ground motions. Wirth et al. (2021) developed ShakeMaps for M9 CSZ earthquakes, applying site-specific amplification factors to the simulated motions from the study by Frankel et al. (2018). The authors observed higher estimates of shaking with consideration of the geotechnical layer, particularly at deep sedimentary basin sites. Hu et al. (2022) observed significant improvement in bias reduction of Fourier amplitude spectra in the Los Angeles area by modifying the SCEC CVM (Small et al., 2017) to include the geotechnical layer. Appropriate geotechnical layer considerations in the Georgia sedimentary basin model are needed especially in the deeper basin areas as most sites in the deep Georgia basin correspond to a measured in situ Site Class D (V_s30 = 180–360 m/s). Data from an ongoing microzonation project (Assaf et al., 2022) in the region is expected to fill this gap. In the absence of such data, current estimates from simulated M9 ground motions may under-predict shaking at these sites and implications on the resulting BAFs need to be studied.

Figure 1 illustrates the model by Stephenson et al. (2017) in terms of depth to a V_s of 2.5 km/s (Z_2.5,) and shows the sites selected for this study as listed in Table 2. Molnar et al. (2014a, 2014b) updated the Stephenson (2007) model for the Georgia sedimentary basin in southern British Columbia. However, their updates were not incorporated into the model used by Stephenson et al. (2017) to generate the 30 M9 CSZ synthetic ground motions by Frankel et al. (2018).

Figure 1.

Spatial variation of Z_2.5 from the Pacific Northwest community velocity model overlaid with selected sites of this study (listed in Table 2).

Table 2.

Summary of site characteristics at locations within the Georgia sedimentary basin, basin edge, and outside basin.

Zone	Site name or code	Site ID	Latitude (°N)	Longitude (°E)	Model V_s30^a (m/s)	In situ V_s30^b (m/s)	Z_1.0^a (m)	Z_2.5^a (km)
Basin sites	Ladner	LAD	49.09	–123.08	609	189	168	3.76
	English Bluff Terminal	EBT	49.01	–123.09	609	377	167	3.73
	Richmond	RIC	49.17	–123.13	609	192	168	3.22
	Delta	DEL	49.09	–123.03	609	170	163	3.27
	Blodel Conservatory	BLO	49.24	–123.11	609	514	146	2.21
	New Westminster	NEW	49.21	–122.91	609	219	150	2.23
	Surrey	SUR	49.19	–122.85	609	Class C	150	2.23
	Ingledow Substation	ING	49.16	–122.87	609	801	153	2.26
	Burnaby	BUR	49.25	–122.98	609	504	145	1.74
	Langley Municipal Hall	LAN	49.10	–122.62	609	185	144	1.77
Basin-edge sites	University of British Columbia	UBC	49.26	–123.25	609	451	123	1.14
	Vancouver	VAN	49.28	–123.12	609	442	133	1.22
	Cathedral Square Substation	CSQ	49.28	–123.11	609	576	133	1.22
	Murrin Substation	MUR	49.28	–123.10	609	199	135	1.70
	North Vancouver	NOR	49.32	–123.07	609	585	130	1.19
	Barnard Substation	BND	49.26	–122.90	609	441	143	1.73
	Meridian Substation	MDN	49.31	–122.81	609	417	139	1.26
Outside-basin sites	West Vancouver	WES	49.33	–123.16	609	371	118	0.66
	Cleverland Dam	CLE	49.36	–123.11	609	298	121	0.68
	Y00SHB	SHB	49.60	–123.88	609	Class A	82	0.39
	Y0NLLB	NLLB	49.23	–123.99	960	Class B	90	0.43
	Y0SYMB	—	48.56	–123.80	960	Class A	0	0
	Y00VGZ	VGZ	48.41	–123.32	960	Class A	0	0
	Y00ERW	ERW	48.45	–122.63	960	Class B	0	0
	Y0MBKE	BKE	48.92	–122.14	960	—	0	0
	Y0GOBB	GOBB	48.95	–123.51	609	Class B	130	1.23
	Y00SNB	SNB	48.77	–123.17	609	Class B	121	1.11
	Y0LUMI	—	48.72	–122.71	609	Class B	140	1.22
	Y07043	—	48.72	–122.50	960	—	0	0.74
	Y0HNBB	HNB	49.29	–122.57	609	Class A	135	1.25
	Pacific Geoscience Center	PGC	48.65	–123.44	960	Class A	0	0

Determined from the model by Stephenson et al. (2017) as updated by Frankel et al. (2018).

Determined from in situ V_s measurements when available or inferred site class (Assaf et al., 2022).

Benchmarking simulated M9 spectra against GMMs in CSHM 2020

Selection of reference sites within and outside the basin

The choice of a reference site is pivotal to estimating basin amplification. A reference condition can have a significant impact on the shape (both the peak amplitude and peak frequency/period) of the calculated basin amplification function. When non-basin synthetic ground motions are not available, the selected reference condition is commonly taken as a site (or sites) outside the basin and near the interface of Z_2.5≈ 1 km ( Kakoty et al., 2021; Marafi et al., 2019; Wirth et al., 2018). This selection is based on the concept that the basin effect is not present or minimal when Z_2.5 < 1 km. For example, using Z_2.5≈ 3 km as a reference condition around the Seattle basin, Wirth et al. (2018) showed that the mean BAF from the M9 project’s simulations was ∼1.5 at 1 s, increasing to ∼2 at periods greater than 2–3 s (see their Figure 6). When the reference condition is changed to Z_2.5≈ 1 km, BAFs determined by Frankel et al. (2018) ranged from about 1.5 to 5 at periods of 1–10 s, with peak basin amplification of 5 at about 3 s (see their Figure 25).

For this study, we define three zones in the model by Stephenson et al. (2017; refer back to Figure 1): outside of basin with Z_2.5 < 1 km, basin edge with 1 km ≤ Z_2.5 ≤ 2 km and (deep) basin with Z_2.5 > 2 km. We therefore introduce two options as the reference condition to determine basin amplification using the synthetic broadband ground motions from the set of 30 CSZ simulations of the M9 project (Frankel et al., 2018): (1) relative to the basin-edge sites and (2) relative to the outside-basin sites. Furthermore, to minimize the risk of error arising from localized effects at any particular station, we select a number of reference sites at various azimuthal distributions within each of the two reference condition groups.

The selected non-basin rock reference condition where Z_2.5 is < 1 km include several areas (as seen in Figure 1) on Vancouver Island including Greater Victoria and westward to Port Renfrew, as well as much of the area west of Duncan/Nanaimo. In the Lower Mainland, non-basin rock conditions occur along Vancouver’s North Shore (West and North Vancouver) and near the eastern model limit straddling the United States–Canada international border. As the first layer of the Stephenson et al. (2017) model is 500 m, the shear wave velocity in this layer serves to characterize the V_s30 at each site of interest according to the NBCC seismic site class (as listed in Table 2). The range of V_s30 for non-basin reference sites is 609–960 m/s or seismic site class C (V_s30 = 361–760 m/s) to B (V_s30 = 761–1500 m/s).

Basin-edge reference sites are locations with Z_2.5 between 1 and 2 km. Both basin-edge and basin sites have the same V_s30 of 609 m/s in the velocity model by Stephenson et al. (2017). However, in reality, basin-edge sites correspond to consistently higher in situ V_s30 (> 400 m/s) relative to basin sites (as listed in Table 2). Since the M9 simulations correspond to spectral estimates for V_s30 values from the study by Stephenson et al. (2017), to maintain consistency, the same V_s30 assumptions are used for GMMs from the CSHM 2020 in the BAF calculations. Figure 2 compares V_s profiles for the three different site categories (i.e. basin sites, as well as basin-edge and outside-basin reference sites). There are consistent within- and between-group trends in the V_s depth profile. For instance, overall stiffness increases in the upper 10 km from basin sites to basin-edge to outside-basin reference sites, and differentiation in the V_s profile for basin and basin-edge sites occurs below 500 m.

Figure 2.

V _s profiles at selected locations as per Stephenson et al. (2017) grouped by the three selected zones up to a depth of (a) 60 km and (b) 6 km. The vertical dashed black line in Figure 2b indicates a V_s of 2500 m/s and the corresponding depth for each site yields Z_2.5.

Comparison of SA estimates inside and outside the basin

At all sites of interest, SAs of the simulated M9 ground motions are benchmarked against the average of the SA estimates from the four interface GMMs used in CSHM 2020 (refer to Table 1), assuming consistent magnitude (M9), distance, site class (consistent with Model V_s30 as listed in Table 2), and other relevant parameters. Figure 3 illustrates the average of the geomean spectra of the suite of 30 simulated M9 CSZ ground motions against corresponding CSHM 2020 interface GMMs estimates at three distinct locations representative of (1) outside-basin sites, (2) basin-edge sites, and (3) basin sites, as previously introduced in Table 2. Since the GMMs do not explicitly account for basin amplification, while the simulated M9 motions do, the comparison provides insights into the expected amplification of ground motion shaking throughout the basin.

Figure 3.

Average geomean response spectra of the 30 simulated M9 CSZ earthquake ground motions and average geomean response spectra from the four interface GMMs as per CSHM 2020 for the same scenarios at (a) outside-basin (PGC), (b) basin-edge (MUR), and (c) basin (DEL) sites with site ID listed in Table 2.

As shown in Figure 3a, SA estimates are consistent between the simulated M9 ground motions and the interface GMMs for periods above 1 s for outside-basin sites, for example, Pacific Geoscience Center (PGC). When comparing the spectral amplitudes in Figure 3a to c, it is apparent that the simulated M9 ground motion spectra increases considerably for periods above 1 s due to the presence of the Georgia basin. The comparison across zones illustrates that the amplification in the SA is strongly correlated with basin depth, that is, Z_2.5, as identified in previous studies (Kakoty et al., 2021; Marafi et al., 2019). For instance, the ratio between the SAs of the M9 simulations and the CSHM 2020 interface GMMs estimates at a 2 s period are 1.06, 2.41, and 5.13 for the non-basin (PGC), shallow basin-edge (Murrin Substation (MUR)), and basin (Delta (DEL)) sites, respectively. These observations serve to highlight how the CSHM 2020 GMMs underestimates shaking at basin locations, particularly in the of 1–5 s period range.

Basin amplification factors

Formulation

In this study, an extension of the site-specific and period-dependent BAF formulation employed by Kakoty et al. (2021) is used to quantify amplification in SA at basin sites. Equation 1 illustrates the BAF calculation, which uses the simulated set of M9 CSZ earthquake ground motions (denoted as M9 in Equation 1) and corresponding CSHM 2020 interface GMM estimates (denoted as GMM_j in Equation 1). The amplification factors are calculated for sites within the basin with respect to a reference location (denoted as ref_k in Equation 1), within the basin-edge or outside-basin zones. The numerator in Equation 1 represents the ratio of SA estimates from the simulated M9 CSZ ground motions to the corresponding estimates from each of the four interface GMMs used in CSHM 2020 (denoted by $S A_{GM M_{j}}$ ) at the site of interest i and at period T. The denominator represents the same ratio at the reference location. BAFs are calculated for each of the 30 (denoted by n) simulated ground motions and the geometric mean is computed for each basin site with respect to one reference location. This calculation is repeated for each GMM and each reference location within the basin-edge and outside-basin zones, and the average BAF across reference locations (denoted by n_s in Equation 1) is computed as:

B A F_{i}^{T} = \frac{1}{n_{s}} \sum_{r e f_{k} = 1}^{n_{s}} \frac{1}{G M M_{n}} \sum_{G M M_{j} = 1}^{G M M_{n}} \prod_{x = 1}^{n} {((\frac{S A_{M 9}^{i, x} (T)}{S A_{G M M_{j}}^{i, x} (T)}) / (\frac{S A_{M 9}^{r e f_{k}, x} (T)}{S A_{G M M_{j}}^{r e f_{k}, x} (T)}))}^{\frac{1}{n}}

(1)

Figure 4 highlights variability in the resulting BAF calculated at a basin site due to differences in GMM estimates, as seen in Figure 4a, and due to differences in reference sites within a zone, as seen in Figure 4b. Figure 4a illustrates calculated BAF for one basin site (DEL) with one reference site (PGC) to highlight the difference in BAF estimated using the four GMMs as listed in Table 1. Whereas Figure 4b illustrates the calculated BAF for one basin site (DEL) using one GMM (Abrahamson et al., 2016) to highlight the variation that occurs due to the selection of different reference sites.

Figure 4.

Variability in average (μ) BAF with one standard deviation (σ) for a site in Delta due to (a) variation in GMM estimates for a given reference site (i.e. PGC) and (b) 13 reference sites within the outside-basin zone as listed in Table 2 for a given GMM (Abrahamson et al., 2016).

Results

Figure 5a and b illustrate the resulting variability in BAFs as a function of the choice of reference condition, that is, basin-edge or outside-basin zones. Each geometric mean BAF is computed using 30 simulated ground motions at the site of interest with respect to a reference site located in the basin-edge and outside-basin zones. The average of the geometric mean BAFs captures the azimuthal variability for each of the two reference conditions where the n value in Figure 5 indicates the number of reference sites within each zone. The maximum average BAF among the sites of interest occurs at the Ladner (LAD) basin site and are 8.09 and 2.74 at a 1.61 s period, with respect to outside-basin and basin-edge reference sites, respectively. The difference in the average BAF between the two reference conditions is consistently 2.6–3.1 times higher for the outside-basin when compared to the basin-edge reference condition. The variability in the BAF estimates is generally lower when using basin-edge reference sites than outside-basin reference sites, as observed by comparing Figure 5a and b. The lower variability in BAF for basin-edge reference sites is attributed to lower variability in the SA estimates from the GMMs resulting from the proximity of the site of interest and reference locations as inferred from Figure 1.

Figure 5.

Average (μ) and one standard deviation (σ) basin amplification factors for the 30 M9 CSZ scenarios (gray lines) at a basin site in delta (DEL) with respect to reference sites located (a) in the basin edge (n = 7) and (b) outside of the basin (n = 13). Average basin amplification factor for all sites of interest with respect to reference sites located (c) in the basin edge (n = 7) and (d) outside of the basin (n = 13). All reference sites are as listed in Table 2.

Generally, the period-dependent BAFs in Figure 5c and d result in two types of shape: broad and narrow peak amplification. The narrow peak amplification shapes (e.g. DEL, EBT, LAD sites) are associated with sites with Z_2.5 greater than 3 km, whereas the broad peak amplification shapes (e.g. BLO, ING, BUR) are associated with sites with Z_2.5 between 2 and 3 km. The outside-basin reference sites are much more scattered geographically compared to the basin-edge reference sites, which results in higher variation in SA estimates due to the difference in local site conditions as well as a higher degree of variation in the source-to-site distances than basin-edge sites. These effects combine in the BAF calculations resulting in greater BAF variation for outside-basin reference sites than basin-edge reference sites. Average BAF at the selected basin sites generally increase with increasing Z_2.5, especially in the period range of 1–5 s. These findings are consistent with those from Olsen (2000). Figure 6 shows the distribution of BAFs at 3 s period for each basin site in decreasing order of Z_2.5, with respect to outside-basin and basin-edge reference sites.

Figure 6.

Distribution of basin amplification factors (BAFs) for all basin sites in decreasing order of Z_2.5 (lighter shading corresponds to shallower Z_2.5) at a period of 3 s with respect to reference sites located within the (a) basin-edge (n = 7) and (b) outside of the basin (n = 13). Note different BAF axis scaling of (a) and (b).

Development of modified uniform hazard spectra

Currently enforced seismic design provisions of the 2020 NBCC (NRC, 2020) utilize a 5% damped uniform hazard spectrum (UHS) with a 2% probability of exceedance in 50 years to define the design level earthquake. By definition, the UHS envelopes SA estimates at all periods with a specified hazard level (e.g. 2475 year return period) as obtained by probabilistic seismic hazard analysis, considering all potential earthquake sources, magnitudes and distance combinations. Currently, the UHS defined in the 2020 National Building Code does not explicitly account for basin amplification due to the Georgia sedimentary basin because the underlying GMMs within the 2020 National Seismic Hazard Model do not explicitly include these effects. As a result, in southwestern British Columbia, buildings within the Georgia Sedimentary Basin, particularly long-period structures, are currently being designed for a hazard level that is lower than anticipated when basin amplification effects are considered. To address this issue, a modified UHS is proposed to incorporate the BAFs previously computed to explicitly incorporate basin amplification effects in the calculation of the UHS used in design. Because the BAFs are derived from physics-based simulations from the M9 CSZ earthquakes, the amplification is only applied to the interface–source contribution to the hazard. Nevertheless, at long periods, where basin effects are most pronounced, the interface earthquake contribution dominates the hazard. For example, the interface source contributes 76% and 87% of the total hazard at a 4 and 5 s period, respectively, for a return period of 2475 years (Kakoty et al., 2021).

The methodology involves three steps: (1) disaggregating the UHS at the site of interest per seismic source contribution, that is, crustal, interface, and inslab earthquakes; (2) applying the calculated BAFs to the interface contribution to the hazard at each period; and (3) re-aggregating the hazard estimates to obtain the modified UHS. In doing so, the modified UHS explicitly accounts for the amplification effects of the Georgia basin in the interface contribution to the hazard. Because BAFs are computed using basin-edge and outside-basin reference conditions, the resulting UHS is also computed for these two reference conditions. An example is provided to showcase the methodology as applied to the UHS defined in the 2020 edition of the NBCC (NRC, 2020). However, because the seismic hazard disaggregation data from CSHM 2020 is not yet publicly available, we assume seismic hazard disaggregation per source will not vary significantly between the 2015 and 2020 CSHM.

To illustrate a practical implementation of the methodology, the modified UHS, considering the amplification effects of the Georgia basin on CSZ megathrust earthquakes, is calculated for the BLO site (as listed in Table 2). Figure 7 illustrates in a solid blue line the UHS for the BLO site of interest as derived from CSHM 2020. Figure 7 also illustrates the seismic source disaggregation of the UHS, highlighting how interface earthquakes dominate the hazard at periods greater than 1 s, with negligible contributions from crustal and inslab seismic sources at longer periods. The interface BAF calculated at this site, applied to the interface contribution of the hazard, is illustrated by the dashed brown line. This modified interface hazard contribution is then re-aggregated with the crustal and inslab hazard estimates to obtain the modified UHS as shown by the dotted blue line. Figure 7 illustrates the results of the calculation using basin-edge (Figure 7a) and outside-basin (Figure 7b) reference locations. Since this study only explores basin amplification for interface sources, the resulting factors are only applied to the interface contribution to the hazard. However, basin amplification for inslab earthquakes would raise the modified UHS because inslab earthquakes represent a considerable contribution to the hazard, particularly at the 1–2 s period range.

Figure 7.

Modified uniform hazard spectrum or UHS (dashed blue line) at BLO explicitly considering interface–source basin amplification calculated with (a) basin-edge and (b) outside-basin reference locations compared to the CSHM 2020 UHS (solid blue line).

As observed in Figure 7, the modified UHS results in considerably higher SA at longer periods than currently estimated by CSHM 2020. For the BLO site considered in Figure 7, the average increase in SA estimates in the 1–5 s period range is 29% and 141% when BAFs with respect to basin-edge sites and outside-basin sites are adopted, respectively. The SA amplifications follow the trends previously presented in Figure 5, resulting in the greatest amplifications in the 1–2 s period range. For instance, at a 2 s period, the modified UHS at the BLO site of interest is 0.32 and 0.69 g with respect to basin-edge and outside-basin reference sites, respectively, as compared to 0.26 g for the non-basin UHS, that is, there is 23% and 165% increase in SA, respectively, when basin effects are considered. Similarly, at a 5 s period, the modified UHS results in a 43% and 184% increase for basin-edge and outside-basin reference sites, respectively. Results in Figure 7 adopt BAFs for periods exceeding 1 s since basin amplification is explicitly considered only for that period range. Modified UHS calculations are carried out at other basin sites in an identical way, and these results are summarized in the Electronic Appendix.

Long-period amplification from recorded strong-motion data

Ground motion recordings of the 2001 M6.8 Nisqually earthquake

The 2001 M6.8 Nisqually earthquake is one of the largest magnitude inslab events recorded by strong-motion instrumentation in southwestern British Columbia. Other recorded large inslab earthquakes occurred beneath the Puget Sound in 1949 (M7.1) and 1965 (M6.5). Earthquake events beneath the Georgia Strait have generally been of smaller magnitude (M ≤ 5.5). The M6.8 Nisqually earthquake produced long-period shaking levels ≤ 5 cm/s (corresponding to Modified Mercalli Intensity IV) in the Seattle basin and resulted in US$2 billion worth of damage in Washington State (Meszaros and Fiegener, 2002). This earthquake triggered strong motion instruments at 32 sites in southwestern British Columbia (at epicentral distances from 150 to 300 km), including 20 in the Lower Mainland and 12 in Vancouver Island, thus providing the largest single strong motion data set collected to date in this region. Peak horizontal ground accelerations at these sites ranged from 0.3%g to 3.5%g (2.9–34.2 cm/s²) and peak horizontal ground velocities ranged from 0.17 to 1.5 cm/s (Cassidy et al., 2003). The Nisqually earthquake generated not only the highest quantity, but also the highest quality, empirical ground motion data set to date for southwest British Columbia with waveforms of 15–90 s in duration and sufficient signal-to-noise ratio. More importantly, sufficient long-period content (> 1 s) is present in the waveforms, which permits benchmarking the previously calculated BAFs for the Georgia sedimentary basin.

Strong-motion recordings of the Nisqually earthquake in southwest British Columbia were documented by Cassidy et al. (2003). Molnar et al. (2014a) evaluated Nisqually earthquake recordings in the Georgia basin region of both British Columbia and Washington State. Nisqually earthquake ground motions in the Georgia basin region are significantly lower than in the Seattle basin region (Molnar et al., 2014a), as the deep earthquake is more than 150 km away from Metro Vancouver. The Nisqually recordings contain significantly more long-period (1–5 s) energy compared to other available strong motion data sets with sufficient signal-to-noise ratio for this region, for example, the 1996 Duvall, 1997 Georgia Strait, and 2015 Victoria earthquakes (Cassidy and Rogers, 1999; Jackson et al., 2017). Other recorded earthquakes with long-period energy generally do not have a sufficient signal-to-noise ratio for further consideration here (Assaf et al., 2022).

The Nisqually earthquake was recorded in the Georgia basin region by continuously recording Canadian National Seismograph Network (CNSN) and Pacific Northwest Seismic Network (PNSN) seismograph stations, as well as trigger-based strong-motion instruments. Strong-motion stations on rock and stiff ground (equivalent to non-basin and shallow basin-edge sites of the velocity model by Stephenson et al., 2017) either did not trigger and hence no recording is available, or triggered but the recording is short and captures the largest amplitude shear and/or surface wave content. The longest duration Nisqually earthquake waveforms were recorded on soft Holocene delta sediments (minimum V_s of 70 m/s) of the Fraser River delta, which are sediments not present in the velocity model by Stephenson et al. (2017; minimum V_s of 600 m/s), that is, stations grouped as basin sites have lower in situ V_s30 than the model V_s30 as previously discussed (refer back to Table 2). As a result, these recordings may reflect amplification from multiple sources, including those associated with the shallow Holocene Fraser River delta sediments as well as those associated with the deeper Georgia sedimentary rock basin. As it is not possible to divorce these two site effects, we refer to the amplification observed in these empirical recordings as LPA.

Long-period amplification factor

To estimate the LPA observed in strong-motion recordings of the 2001 Nisqually earthquake, the recorded SA estimates are benchmarked against SA estimates from subduction inslab GMMs in CSHM 2020. The subduction inslab hazard is estimated in CSHM 2020 as an equally weighted average of four GMMs: (1) Abrahamson et al. (2016), (2) Atkinson and Boore (2003), (3) García et al. (2005), and (4) Zhao et al. (2006). Consistent with the interface–source BAFs previously estimated, we maintain the “basin,”“basin-edge,” and “outside basin” identifiers for the selected seismic stations. Figure 8 shows the spatial distribution of the seismic stations from which Nisqually earthquake time histories are used to estimate the LPA. To maintain consistency with the empirical recordings when estimating SAs at the seismic stations from the GMMs, in situ V_S30 at the seismic stations are used, reported as in situ V_s30 in Table 2.

Figure 8.

Strong motion stations from which 2001 Nisqually earthquake recordings are used to estimate long-period amplification: Red (EBT, LAD, RIC, BLO, ING), black (CSQ, MDN, BND), and green (PGC) colors indicate sites in basin, basin-edge, and outside-basin zones, respectively, with site ID listed in Table 2.

Figure 9 illustrates the LPA estimates at two sites, LAD and EBT, previously listed in Table 2 and shown in Figure 8. Namely, Figure 9a and c highlight the LPA observed during the 2001 Nisqually earthquake ground motion recordings at the LAD and EBT sites in Vancouver relative to three basin-edge reference sites: BND, CSQ, and MDN. The observed long-period basin-edge amplification at LAD and EBT is overlaid with the interface–source basin-edge BAF estimates previously calculated (as seen in Figure 5a) enveloped by one standard deviation. LPAs observed from the Nisqually motions are consistent with simulated interface–source BAF computed at the same site up to a period of around 2 s. However, BAFs seem to underestimate amplification for periods greater than 2 s. Similarly, Figure 9b overlays the empirical LPA factors and simulated BAF estimates using an outside-basin reference location, namely, the PGC site. These results suggest that simulated BAF predicts higher amplification at shorter periods (≤ 2 s) than the observed LPA, but it is consistent at higher periods (> 2 s). Overall, when basin-edge reference sites are used, empirical LPA factors are greater than the simulated BAFs derived from the physics-based ground motion simulations. There is greater agreement between empirical Nisqually LPAs and the simulated basin amplification estimates with respect to outside-basin reference sites.

Figure 9.

Long period amplification for sites of interest: LAD with reference sites located at (a) basin-edge (BND, CSQ, and MDN) and (b) outside-basin (PGC); and EBT with reference sites located at (c) basin-edge (BND, CSQ, and MDN) and (d) outside-basin (PGC). The shaded area represents calculated average BAF ± one standard deviation with reference sites located in basin-edge and outside-basin, respectively, (as seen in Figure 5a and b).

The SA BAF computed using physics-based ground motion simulation are greatest in the 1–5 s period range (as seen in Figure 5), empirical LPA of SAs determined from the inslab Nisqually earthquake ground motion recordings are observed at longer periods and over a wider range, that is, 2–8 s. In addition, these are most predominant at periods of 2, 3, and 5 s. The most notable observations are (1) agreement in amplification factors (long-period vs basin) observed when outside of basin reference sites are used, as well as the (2) high amplification factors, which range from 5 to 10 over a 2–8 s period range.

It is important to note that different mechanisms of basin amplification would be expected for the Nisqually inslab earthquake and M9 CSZ earthquakes due to differences in rupture depth and azimuth (Choi et al., 2005). For the Nisqually earthquake, seismic waves enter the basin from beneath, and critical body wave reflections would not be expected to occur. This wave propagation problem is similar to classical 1D wave propagation, which has long been recognized as producing depth-dependent ground motion amplification (Chang and Bray, 1997; Hashash and Park, 2001; Salvati et al., 2001). For M9 CSZ scenario earthquakes, seismic body waves would enter the basin from the west, in the direction of basin thickening, and can become trapped (i.e. occurrence of critical body wave reflections and surface wave propagation across the basin). This could result in stronger depth dependence in expected basin response for the Nisqually earthquake than the M9 CSZ earthquake.

Conclusion

This study leverages broad-band synthetic ground motions for a suite of 30 M9 CSZ earthquake scenarios, which explicitly consider the effects of the Georgia sedimentary basin, to develop site-specific and period-dependent BAFs. SA estimates from this simulated set of ground motions are benchmarked against corresponding estimates using the four interface GMMs used in CSHM 2020. Based on their Z_2.5, a proxy commonly used to characterize the extents of deep basins, a range of locations within the Georgia basin (Z_2.5 > 2 km), as well as sites in the basin-edge (Z_2.5 = 1–2 km) and outside-basin (Z_2.5 < 1 km), are selected. A framework is proposed to incorporate these site-specific and period-dependent BAFs in the computation of the UHS currently used for the design of buildings in Canada. The computed BAFs are also benchmarked against LPA factors obtained from empirically recorded ground motions from the M6.8 2001 Nisqually earthquake.

SA BAFs generally correlate well with the Z_2.5 of the site of interest. Maximum BAFs occur at the deepest basin site Ladner (Z_2.5 = 3.76 km), with values of 8.09 and 2.74 at a 1.61 s period when reference sites located outside the basin and in the basin-edge, respectively, are used in the calculations. The average BAFs for the same site at a period of 2 s are 6.29 and 2.24 with reference sites outside the basin and in the basin-edge regions. Average BAFs at the same period for a site with Z_2.5 of 2.21 km are 3.53 and 1.21 with reference sites outside the basin and in the basin-edge regions. The modified UHS, which incorporate BAFs for the interface contribution of the hazard, result in 271% and 58% higher spectral ordinates at 2 s period, at the deepest basin site (Ladner), with respect to outside-basin and the basin-edge reference sites, respectively, when compared to the non-basin UHS as derived from CSHM 2020.

LPA factors calculated for recorded ground motions from the M6.8 2001 Nisqually earthquake are significant at longer periods and over a wider range (∼2–8 s) when compared to BAFs calculated for the same basin sites, which are greatest in the 1–5 s period range. LPAs computed with recorded ground motions provide greater agreement with BAFs calculated with respect to outside-basin reference sites, especially for periods exceeding 2 s.

Future work should extend the simulated M9 CSZ earthquake ground motions to account for in situ V_s30 across southwestern British Columbia to allow for a more comprehensive quantification of basin amplification. In addition, the consequence of using a modified UHS for design should be explored to understand the impact of considering versus neglecting basin effects on the resulting seismic performance of buildings and other infrastructure. The results of such studies would support engaging stakeholders and communicating the importance of explicitly considering basin effects in future editions of Canada’s national seismic hazard models and the associated building codes.

Footnotes

Acknowledgements

The authors are very grateful to the M9 team at the University of Washington for sharing the results of their M9 simulations in southwest British Columbia, particularly Nasser A. Marafi, Marc O. Eberhard, Jeffrey W. Berman, Arthur D. Frankel, and Erin A. Wirth. The authors are also grateful for the feedback provided by three independent reviewers and the associated editor, which have greatly improved the quality of the article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by Canada’s Natural Sciences and Engineering Research Council under Discovery Grant No. RGPIN-2019-04599, as well as Canada’s New Frontiers in Research Fund—Exploration under Grant No. NFRFE-2018-01060. The authors would also like to thank the University of British Columbia’s Cascadia Engagement Fund for supporting this collaboration with the University of Washington.

Data availability statement

The data compiled as a companion to this article can be found by locating the article in the publications list found on the following web page () and accessing the supplied “Electronic Supplement” link.

ORCID iDs

Preetish Kakoty

Carlos Molina Hutt

Hadi Ghofrani

References

Abrahamson

Gregor

Addo

(2016) BC hydro ground motion prediction equations for subduction earthquakes. Earthquake Spectra 32: 23–44.

American Society of Civil Engineers (ASCE) (2022) Minimum Design Loads and Associated Criteria for Buildings and Other Structures. Reston, VA: ASCE.

Assaf

Molnar

El Naggar

Sirohey

(2022) Seismic site characterization in Fraser River Delta in Metropolitan Vancouver. Soil Dynamics and Earthquake Engineering 161: 107384.

Atkinson

Adams

(2013) Ground motion prediction equations for application to the 2015 Canadian national seismic hazard maps. Canadian Journal of Civil Engineering 40: 988–998.

Atkinson

Boore

(2003) Empirical ground-motion relations for subduction-zone earthquakes and their application to Cascadia and other regions. Bulletin of the Seismological Society of America 93: 1703–1729.

Atkinson

Casey

(2003) A comparison of ground motions from the 2001 M _6.8 in-slab earthquakes in Cascadia and Japan. Bulletin of the Seismological Society of America 93: 1823–1831.

Atkinson

Macias

(2009) Predicted ground motions for great interface earthquakes in the Cascadia subduction zone. Bulletin of the Seismological Society of America 99: 1552–1578.

Atwater

Nelson

Clague

Carver

Yamaguchi

Bobrowsky

Bourgeois

Darienzo

Grant

Hemphill-Haley

Kelsey

Jacoby

Nishenko

Palmer

Peterson

Reinhart

(1995) Summary of coastal geologic evidence for past great earthquakes at the Cascadia subduction zone. Earthquake Spectra 11: 1–18.

Bayless

Abrahamson

(2018) An Empirical Model for Fourier Amplitude Spectra Using the NGA-West2 Database. Berkeley, CA: Pacific Earthquake Engineering Research Center (PEER), University of California, Berkeley.

10.

Bijelić

Lin

Deierlein

(2019) Quantification of the influence of deep basin effects on structural collapse using SCEC cybershake earthquake ground motion simulations. Earthquake Spectra 35: 1845–1864.

11.

Brocher

Parsons

Blakely

Christensen

Fisher

Wells

ten Brink

Pratt

Crosson

Creager

Symons

Preston

Van Wagoner

Miller

Snelson

Trehu

Langenheim

Spence

Ramachandran

Hyndman

Mosher

Zelt

Weaver

(2001) Upper crustal structure in Puget Lowland, Washington: Results from the 1998 Seismic Hazards Investigation in Puget Sound. Journal of Geophysical Research B: Solid Earth 106: 13541–13564.

12.

Building Seismic Safety Council (BSSC) (1994) NEHRP Recommended Provisions for Seismic Regulations for New Buildings. Washington, DC: BSSC.

13.

Campillo

Gariel

Aki

Sanchez-Sesma

(1989) Destructive strong ground motion in Mexico city: Source, path, and site effects during great 1985 Michoacan earthquake. Bulletin of the Seismological Society of America 79: 1718–1735.

14.

Cassidy

Rogers

(1999) Seismic site response in the greater Vancouver, British Columbia, area: Spectral ratios from moderate earthquakes. Canadian Geotechnical Journal 36: 195–209.

15.

Cassidy

Molnar

Rogers

Mulder

Little

(2003) Canadian Strong Ground Motion Recordings of the 28 February, 2001 M=6.8 Nisqually (Seattle-Olympia) Washington, Earthquake. Vancouver, BC, Canada: Geological Survey of Canada.

16.

Chang

Bray

(1997) Implications of recent strong motion data for seismic building code design at deep, stiff soil sites. In: Proceedings of the CUREe Northridge earthquake research conference, Los Angeles, CA, August 20–22.

17.

Choi

Stewart

Graves

(2005) Empirical model for basin effects accounts for basin depth and source location. Bulletin of the Seismological Society of America 95: 1412–1427.

18.

Dash

Spence

Riedel

Hyndman

Brocher

(2007) Upper-crustal structure beneath the strait of Georgia, Southwest British Columbia. Geophysical Journal International 170: 800–812.

19.

Day

Graves

Bielak

Dreger

Larsen

Olsen

Pitarka

Ramirez-Guzman

(2008) Model for basin effects on long-period response spectra in Southern California. Earthquake Spectra 24: 257–277.

20.

England

Bustin

(1998) Architecture of the Georgia Basin southwestern British Columbia. Bulletin of Canadian Petroleum Geology 46: 288–320.

21.

Frankel

Stephenson

Carver

(2009) Sedimentary basin effects in Seattle, Washington: Ground-motion observations and 3D simulations. Bulletin of the Seismological Society of America 99: 1579–1611.

22.

Frankel

Stephenson

Carver

Williams

Odum

Rhea

(2007) Seismic hazard maps for Seattle, Washington, incorporating 3D sedimentary basin effects, nonlinear site response, and rupture directivity. USGS open file report 1-82, 27 July. Reston, VA: United States Geological Survey (USGS).

23.

Frankel

Wirth

Marafi

Vidale

Stephenson

(2018) Broadband synthetic seismograms for magnitude 9 earthquakes on the Cascadia Megathrust based on 3D simulations and stochastic synthetics, Part 1: Methodology and overall results. Bulletin of the Seismological Society of America 108: 2347–2369.

24.

García

Singh

Herraíz

Ordaz

Pacheco

(2005) Inslab earthquakes of Central Mexico: Peak ground-motion parameters and response spectra. Bulletin of the Seismological Society of America 95: 2272–2282.

25.

Ghofrani

Atkinson

(2014) Ground-motion prediction equations for interface earthquakes of M7 to M9 based on empirical data from Japan. Bulletin of Earthquake Engineering 12: 549–571.

26.

Graves

Pitarka

Somerville

(1998) Ground-motion amplification in the Santa Monica area: Effects of shallow basin-edge structure. Bulletin of the Seismological Society of America 88: 1224–1242.

27.

Hashash

YMA

Park

(2001) Non-linear one-dimensional seismic ground motion propagation in the Mississippi embayment. Engineering Geology 62: 185–206.

28.

Olsen

Day

(2022) Calibration of the near-surface seismic structure in the SCEC community velocity model version 4. Geophysical Journal International 230: 2183–2198.

29.

Jackson

Molnar

Ghofrani

Atkinson

Cassidy

Assatourians

(2017) Ground motions of the December 2015 M 4.7 Vancouver Island earthquake: Attenuation and site response. Bulletin of the Seismological Society of America 107: 2903–2916.

30.

Meszaros

Fiegener

(2002) Effects of the 2001 Nisqually Earthquake on Small Businesses in Washington State. Seattle, WA: Economic Development Administration, U.S. Department of Commerce, Seattle Regional Office.

31.

Kakoty

Dyaga

Molina Hutt

(2021) Impacts of simulated M9 Cascadia subduction zone earthquakes considering amplifications due to the Georgia sedimentary basin on reinforced concrete shear wall buildings. Earthquake Engineering & Structural Dynamics 50: 237–256.

32.

Kobayashi

Seo

Midorikawa

(1986) Estimated Strong Ground Motions in the Mexico City Earthquake: The Mexico Earthquakes 1985. Factors Involved and Lessons Learned. American Society of Civil Engineers, New York.

33.

Kolaj

Adams

Halchuk

(2020a) The 6th generation seismic hazard model of Canada. In: Proceedings of the 17th world conference on earthquake engineering, Sendai, Japan, 27 September–2 October, pp. 1–12.

34.

Kolaj

Halchuk

Adams

Allen

(2020b) Sixth generation seismic hazard model of Canada: Input files to produce values proposed for the 2020 National Building Code of Canada. Geological Survey of Canada, open file report 8630-15. Ottawa, ON, Canada: Geological Survey of Canada.

35.

Kourehpaz

Molina Hutt

Marafi

Berman

Eberhard

(2021) Estimating economic losses of midrise reinforced concrete shear wall buildings in sedimentary basins by combining empirical and simulated seismic hazard characterizations. Earthquake Engineering & Structural Dynamics 50: 26–42.

36.

Ludwig

Nafe

Drake

(1970) Seismic refraction. In: Maxwell

(ed.) The Sea, vol. 4. New York: Wiley-Intersciences, pp. 53–84.

37.

Magistrale

Day

Clayton

Graves

(2000) The SCEC southern California reference three-dimensional seismic velocity model version 2. Bulletin of the Seismological Society of America 90: S65–S76.

38.

Magistrale

McLaughlin

Day

(1996) A geology-based 3D velocity model of the Los Angeles basin sediments. Bulletin of the Seismological Society of America 86: 1161–1166.

39.

Marafi

Eberhard

Berman

Wirth

Frankel

(2017) Effects of deep basins on structural collapse during large subduction earthquakes. Earthquake Spectra 33: 963–997.

40.

Marafi

Eberhard

Berman

Wirth

Frankel

(2019) Impacts of simulated M9 Cascadia subduction zone motions on idealized systems. Earthquake Spectra 35: 1261–1287.

41.

Molnar

Cassidy

Olsen

Dosso

(2014a) Earthquake ground motion and 3D Georgia basin amplification in southwest British Columbia: Deep Juan de Fuca plate scenario earthquakes. Bulletin of the Seismological Society of America 104: 301–320.

42.

Molnar

Cassidy

Olsen

Dosso

(2014b) Earthquake ground motion and 3D Georgia basin amplification in southwest British Columbia: Shallow blind-thrust scenario earthquakes. Bulletin of the Seismological Society of America 104: 321–335.

43.

Monfared

Molina Hutt

Kakoty

Kourehpaz

Centeno

(2021) Effects of the Georgia sedimentary basin on the response of modern tall RC shear-wall buildings to M9 Cascadia subduction zone earthquakes. Journal of Structural Engineering: ASCE 147: 05021003.

44.

National Research Council Canada (NRC) (2020) National Building Code of Canada. Ottawa, ON, Canada: NRC.

45.

Olsen

(2000) Site amplification in the Los Angeles basin from three-dimensional modeling of ground motion. Bulletin of the Seismological Society of America 90: S77–S94.

46.

Petersen

Shumway

Powers

Mueller

Moschetti

Frankel

Rezaeian

McNamara

Luco

Boyd

Rukstales

Jaiswal

Thompson

Hoover

Clayton

Field

Zeng

(2020) The 2018 update of the US National Seismic Hazard Model: Overview of model and implications. Earthquake Spectra 36: 5–41.

47.

Ramachandran

Dosso

Zelt

Spence

Hyndman

Brocher

(2004a) Upper crustal structure of southwestern British Columbia from the 1998 Seismic Hazards Investigation in Puget Sound. Journal of Geophysical Research B: Solid Earth 109: B09303.

48.

Ramachandran

Hyndman

Brocher

(2004b) Structure of the Northern Cascadia subduction zone: A 3-D tomographic P-wave velocity model. In: Proceedings of the AGU fall meeting abstracts, San Francisco, CA; 13–17 December, 2004.

49.

Ramachandran

Hyndman

Brocher

(2006) Regional P wave velocity structure of the Northern Cascadia subduction zone. Journal of Geophysical Research B: Solid Earth 111: B12301.

50.

Rekoske

Moschetti

Thompson

(2021) Basin and site effects in the US Pacific Northwest estimated from small-magnitude earthquakes. Bulletin of the Seismological Society of America 112: 438–456.

51.

Roten

Olsen

Takedatsu

(2020) Numerical simulation of M9 megathrust earthquakes in the Cascadia subduction zone. Pure and Applied Geophysics 177: 2125–2141.

52.

Salvati

Lok

Pestana

(2001) Seismic response of deep stiff granular soil deposits. In: Proceedings of the international conferences on recent advances in geotechnical earthquake engineering and soil dynamics, San Diego, CA; 26–31 March, 2001.

53.

Small

Gill

Maechling

Taborda

Callaghan

Jordan

Olsen

Ely

Goulet

(2017) The SCEC unified community velocity model software framework. Seismological Research Letters 88: 1539–1552.

54.

Stephenson

(2007) Velocity and density models incorporating the Cascadia subduction zone for 3D earthquake ground motion simulations. US Geological Survey, open-file report 2007-1348, 31 October, 24 pp. Reston, VA: United States Geological Survey (USGS).

55.

Stephenson

Reitman

Angster

(2017) P- and S-wave velocity models incorporating the Cascadia subduction zone for 3D earthquake ground motion simulations, version 1.6—Update for open-file report 2007-1348. USGS open file report 2017-1152, 20 December, 17 pp. Reston, VA: United States Geological Survey (USGS).

56.

White

Clowes

(1984) Seismic investigation of the Coast Plutonic Complex—Insular Belt boundary beneath the Strait of Georgia. Canadian Journal of Earth Sciences 21: 1033–1049.

57.

Wirth

Chang

Frankel

(2018) 2018 report on incorporating sedimentary basin response into the design of tall buildings in Seattle, Washington. US Geological Survey, open-file report 2018-1149, 25 September. Reston, VA: United States Geological Survey (USGS).

58.

Wirth

Vidale

Frankel

Pratt

Marafi

Thompson

Stephenson

(2019) Source-dependent amplification of earthquake ground motions in deep sedimentary basins. Geophysical Research Letters; 46(12), 6443–6450.

59.

Wirth

Grant

Marafi

Frankel

(2021) Ensemble ShakeMaps for magnitude 9 earthquakes on the Cascadia subduction zone. Seismological Research Letters 92: 199–211.

60.

Zhao

Zhang

Asano

Ohno

Oouchi

Takahashi

Ogawa

Irikura

Thio

Somerville

Fukushima

(2006). Attenuation relations of strong ground motion in Japan using site classification based on predominant period. Bulletin of the Seismological Society of America 96: 898–913.

Spectral acceleration basin amplification factors for interface Cascadia Subduction Zone earthquakes in Canada’s 2020 national seismic hazard model

Abstract

Keywords

Introduction

Canada’s sixth generation seismic hazard model

Simulated M9 CSZ earthquake ground motions

Georgia sedimentary basin

Benchmarking simulated M9 spectra against GMMs in CSHM 2020

Selection of reference sites within and outside the basin

Comparison of SA estimates inside and outside the basin

Basin amplification factors

Formulation

Results

Development of modified uniform hazard spectra

Long-period amplification from recorded strong-motion data

Ground motion recordings of the 2001 M6.8 Nisqually earthquake

Long-period amplification factor

Conclusion

Footnotes

Acknowledgements

Declaration of conflicting interests

Funding

Data availability statement

ORCID iDs

References