A Broad, Distributed Active Fault Zone Lies beneath Salt Lake City, Utah

Although the Wasatch fault is currently known to have a high-seismic hazard from motion along range-bounding faults, new seismic data reveal faulted and folded 13,000 – 30,000-yr-old Lake Bonneville strata beneath Salt Lake City (SLC). Coupled with previous excavation trench, borehole, and other geologic and geophysical observations, we conclude that a zone of latest Pleistocene and/or Holocene faulting and folding kinematically links the East Bench and Warm Springs faults through a 3 km wide relay structure and transfer zone. We characterize faults beneath downtown SLC as active, and these faults may dis-place or deform the ground surface during an earthquake. Through offset but linked faults, our observations support throughgoing ruptures across faults of the Wasatch fault zone (WFZ) and an elevated risk of earthquake-induced building damage.


Introduction
The active, 350 km long Wasatch fault zone (WFZ) defines the eastern boundary of the Basin and Range Province in northern Utah (Fig. 1). The urbanized part of the WFZ is composed of five tectonic segments with a complex earthquake record (summarized by DuRoss et al., 2016). Evidence for at least 24 largemagnitude earthquakes since about 7 ka supports the possibility of a large magnitude (M ∼ 7) surface-rupturing earthquake along the Salt Lake City (SLC) segment. The downtown SLC area lies within a left step between the East Bench fault (EBF) and Warm Springs fault (WSF) of the SLC segment (Personius and Scott, 1992;Fig. 2). A major earthquake on these faults could produce significant ground amplification of low-velocity basin sediments in the SLC metropolitan area (e.g., Roten et al., 2011;Moschetti et al., 2017). Although rupture of the entire SLC segment has been considered, the behavior and interaction of individual faults is unknown. Connecting two faults of the SLC segment supports a multifault rupture model and the possibility of earthquake-induced surface rupture. Here, we present new seismic reflection data that image late Quaternary lacustrine strata, and new P-wave (V P ) and S-wave (V S ) tomographic images that characterize water saturation and sediment properties beneath downtown SLC. When we combine our analysis with existing data and models, we characterize the fault stepover region beneath SLC.

Setting
The Salt Lake basin is bounded by the Wasatch range to the east and Oquirrh Mountains to the west (Fig. 1). These ranges contain Precambrian bedrock, Paleozoic and Mesozoic sedimentary rocks, and Tertiary igneous and sedimentary rocks (Davis, 1983). The west-dipping WSF and Cottonwood fault appear as prominent topographic scarps separating the Salt Lake basin from the Wasatch range to the east (e.g., Bruhn et al., 1992). The EBF lies between these two faults and forms a prominent intrabasin scarp. These three faults compose the 40 km long SLC segment of the WFZ. Little is known about the subsurface fault geometry of any of these faults, but geologic mapping, structural models, gravity data, industry seismic data, and constraints from the 2020 Magna earthquake suggest a ∼70°fault near the surface and listric fault geometry at depth (e.g., Smith and Bruhn, 1984;Bruhn et al., 1992;Pang et al., 2020;Kleber et al., 2021). The ∼3 km left step between the active EBF and the WSF lies beneath the most densely urbanized parts of Utah, and the slip distribution and interactions between the fault strands are unclear (DuRoss et al., 2016). Lake Bonneville, the predecessor to Great Salt Lake, occupied the region from 30 to 13 ka, and consisted of transgressive, overflowing, and regressive phases (Oviatt, 2015). The deposits of Lake Bonneville are relevant to our seismic study, as they form much of the shallow stratigraphy for the Salt Lake basin consisting of alternating fine-and coarse-grained lacustrine and alluvial strata in the lower elevations, and coarse-grained, near shore deposits and alluvium near the Lake Bonneville margins (Personius and Scott, 1992;Fig. 2). Identifying deformed or vertically displaced Bonneville deposits would provide compelling evidence of active faulting and linkage between separate faults of the WFZ.
Based on circumstantial evidence, the WSF extends south of its topographic expression (DuRoss et al., 2014;Fig. 2). Construction excavations at the Salt Palace Convention Center (SP,200 south) and exploratory trenches at the Metropolitan Hall of Justice (HJ, 500 south) revealed complex faulting within latest Pleistocene post-Bonneville alluvial and Lake Bonneville deposits (Osmond et al., 1965;Simon and Shlemon, 1999). Although some have argued that the faults could have formed from liquefaction-induced lateral spreading, cone penetrometer test (CPT) soundings along 400 south provided evidence of a 3-11 m change in the elevation of Lake Bonneville sediments over about 1.2 km along 400 south, presumably related to folding or faulting (Leeflang, 2008). In addition to trench and CPT measurements, evidence to link the WSF and EBF has been hypothesized via the west-northwest-trending Virginia Street fault that lies northeast of downtown SLC (Van Horn and Crittenden, 1987) and through a normal fault relay structure (Machette et al., 1992;Fossen and Rotevatn, 2016). The latter explanation would produce strain accumulation in the fault overlap zone, resulting in subparallel fractures, folds, and subsidiary faults. These relay zones can provide a conduit or barrier for fluid flow, consistent with groundwater models (e.g., Thiros, 2003) and Interferometric Synthetic Aperture Radar (InSAR) results of Hu and Bürgmann (2020). They identify a ∼N20°W-trending, lowpermeability boundary that connects the WSF and EBF. These circumstantial observations and models are consistent with active faults that lie beneath downtown SLC. Here, we provide direct evidence for these faults through seismic imaging.

Seismic Data
We acquired approximately 35 km of seismic data along heavily trafficked SLC streets (Fig. 2). The data were collected at a nominal 2 m shot spacing using a 200 kg accelerated weight drop source and two-component (vertical and inline), 1.25 m spaced sensors with 4.5 Hz baseplate-coupled geophones embedded within a fire hose. The source produced broadband impulsive signals between 3 and 300 Hz. We pulled our seismic streamer at a distance of 5 or 10 m behind the seismic source. Thus, we collected more than 15,000 48 channel shot gathers with a 60 m receiver aperture. The length of the streamer was restricted to less than the length of a city block, and this restricted length limited our imaging capabilities to late Quaternary and younger strata. Although we emphasize results along select profiles here, Liberty et al. (2018) showed results from all profiles acquired as part of this study. We extracted vertical-component first-motion signals to obtain V P measurements to 20-30 m depth using the approach of St. Clair (2015), two-component fundamental Rayleighwave dispersion curves to estimate V S to 30 m depth using the approach of Gribler et al. (2020), and vertical-component reflection signals to map structure and stratigraphy from about 20 m depth to 300 m depth (Fig. S1). Reflection processing for land streamer data include a mix of land and marine processing approaches. Depths from travel-time images were estimated using a smoothed stacking velocity and refraction model. We utilized a differential Global Navigation Satellite System to obtain decimeter-scale position measurements during data collection. We then used 0.5 m light detection and ranging-derived (lidar) elevations for subsequent data processing. Additional details regarding our seismic approach are summarized in the supplemental material.

South
The 1.3 km long west to east 500 south profile typifies our seismic observations. Personius and Scott (1992) mapped late Pleistocene Lake Bonneville and younger Holocene deposits along the profile, with lateral spread deposits identified between about 200 east and 400 east (Fig. 2). Osmond et al. (1965) identified liquefaction, lateral spread deposits, and west-trending and north-trending high-angle faults, with up to 2.5 m of post-Bonneville sediment displacement from excavations at the HJ building site. However, they provided no direct evidence that the deformation was rooted in active faults, instead relating softsediment deformation possibly to liquefaction-induced lateral spreading.
V P ranges from about 600 m=s at the surface, gradually increasing with depth (Fig. 3). The 1500 m=s contour, our water table proxy, ranges from 8 to 15 m depth and is consistent with static water well levels. This contour parallels the ground surface along much of the profile, but it deepens and undulates along the western portion of the profile (locations f 1 and f 2). V S measures mostly less than 240 m=s in the upper 10 m, except for between position 700 and 1000 m, where we map >240 m=s soil at a few meters depth. We note a V S decrease with depth between positions 700 and 900 m, consistent with sediment repacking from lateral spreading (Personius and Scott, 1992;Obermeier, 1996). Near the western portion of the profile, Leeflang (2008) interpreted two CPT soundings (CP-9 and CP-10) that show large lateral changes in physical properties over a distance of about 100 m. Leeflang (2008) concluded that latest Pleistocene sediments of relatively different ages and origins have been juxtaposed between the two adjacent CPT soundings. The shallow high over low-tip resistance measurements of CP-9 directly relates to faster (more stiff) V S values overlying slower (less stiff) V S soils. The combined CPT, V S , and excavation observations suggest that earthquake-induced deformation followed late-stage Lake Bonneville deposition. Combining regionally low V S values with shallow water observations, liquefaction potential remains high during future earthquakes.
The 500 south reflection image shows west-dipping and vertically offset reflectors from the near surface to about 300 m depth (Fig. 3). The Wasatch Front Community Velocity Model of Roten et al. (2011) places the Tertiary rock surface (R1) at about 130-230 m below the surface, suggesting our reflection data have captured late Quaternary structure and stratigraphy. Centered near 250 m distance, we identify the crest of a ∼250 m wide fold, that lies slightly east of a similarly wide fold noted with a CPT transect located about 200 m to the north (Leeflang, 2008), and west of a similar fold identified on 700 south and 800 south seismic profiles (described subsequently). This fold is bounded by faults that offset and tilt late Quaternary strata. Although it is difficult to track reflectors across some faults, we generally observe increasing displacement with depth across both east-and west-dipping high angle faults. For example, the down-to-the-east fault below the HJ excavation site (position 950 m) shows 5-10 m reflector offsets at 100-150 m depth that is consistent with an active growth fault. We conclude that ground displacements observed at the HJ site are likely related to motion along the underlying 60°-85°eastdipping faults that offset upper Lake Bonneville and older deposits. Reflector offsets across this profile support a broad deformation zone that lies between the WSF and EBF. The high V S zone beneath HJ is consistent with a 300 m wide zone of reworked sediments related to earthquake-induced liquefaction.

South/800 South
The 5 km long 700 south and 4.5 km long 800 south profiles cross the southern projection of the WSF through downtown SLC between 200 West and State Street, and crosses the EBF https://www.seismosoc.org/publications/the-seismic-record/ • DOI: 10.1785/0320210009 The Seismic Record near position 3800 m (Fig. 2). Although there is no topographic or geologic evidence for faulting or liquefaction linked to the WSF along either profile, there is a prominent topographic scarp related to the EBF (Fig. 4). Along each profile, Personius and Scott (1992) mapped an eastward progression of Holocene to latest Pleistocene lacustrine, marsh, and alluvial deposits (Fig. 2). West of position 3500 m on both the profiles, the surface topography slopes about 1.5°-2°to the west. Near the EBF, the surface slope increases to about 10°, then tapers off farther east. Personius and Scott (1992) (Fig. 4).
To the west of position 1500 m along 700 south and west of position 2000 m along 800 south, we measure V S < 180 m=s in the upper 20 m (Fig. 4), consistent with Holocene lacustrine and marsh deposits (Personius and Scott, 1992). Here, borehole V S measurements are consistent with our surface measurements (McDonald and Ashland, 2008;Gribler et al., 2020). Above about 1295 m elevation, east of position 2000 m on both profiles, we observe an increase in V P and V S . We relate the abrupt lateral change in V P to a shallower water table or a transition from confined to unconfined groundwater systems. Thiros (2003) noted this as a transition from a recharge to a discharge groundwater system, and InSAR data are consistent with a ∼N20°W low-permeability, fault-derived groundwater barrier (Hu and Bürgmann, 2020).
Above an elevation of about 1310 m, or east of position 3500 m and a few hundred meters west of the EBF, we observe slower V P and faster V S in the upper 10-20 m depth (Fig. 4). We attribute the slower V P to an increase in depth to watersaturated sediments and to the presence of colluvium at the base of the EBF. The faster V S or stiffer soils are consistent with older late Pleistocene lake strata at higher basin elevations  (Personius and Scott, 1992). Immediately to the west of the EBF, we observe slower V P and V S , consistent with the presence of alluvium, colluvium, and lacustrine deposits. Low V P and high V S within the narrow faulted region suggests that dry stiff soil occupies the fault zone. In the footwall of the EBF at position 4000 m, we observe V P > 1500 m=s at about 3 m depth. This V P increase is consistent with groundwater springs (Fig. 2), and that strata in the footwall part of the EBF provide a barrier to lateral groundwater flow. Between position 4500 and 4600 m, a sharp V S increase, coupled with a change in surface slope, is consistent with the Mount Olivet fault (McKean, 2020). We identify no V S inversion along either profile that would support the presence of lateral spread deposits. We map faults from our reflection profile that extend to 20 m depth, or shallower, along the length of each profile (Fig. 4). We observe kilometer-wavelength folded strata in the upper 300 m depth, consistent with late Quaternary syn-or post-deposition deformation (Fig. 4). The crest of one anticline is located between position 2500 and 3000 m, where an artesian spring is present (Fig. 2). We link this anticline to the one observed along the 500 south seismic profile and the 400 south CPT transect to identify a N20°W structural trend (also constrained from 200 south to 400 east seismic profiles; Liberty et al., 2018). Reflectors close to the EBF appear folded and have a 40 m offset along the shallowest reflector (about 20 m depth) over a distance of about 400 m. The presence of tilted strata within the EBF zone is consistent with the monoclinal warping documented in the nearby Dresden Place trenches (Machette et al., 1992; D on Fig. 2). A synclinal inflection in reflectors at position 4500 m (best seen on 800 south) is consistent with either a second strand of the EBF or related to the subparallel Mount Olivet fault (McKean, 2020). We note that fault locations and characteristics differ between the 700 south and 800 south profiles, consistent with soft sediment, noncataclastic deformation (e.g., Fossen and Rotevatn, 2016). These faults may localize and merge at depth.
Offset reflectors to within 20 m of the surface along both 700 south and 800 south profiles are consistent with a distributed zone of folding and faulting to the west of the EBF (Fig. 4). Reflector discontinuities and changing reflector dips are clear and consistent with late Pleistocene faulting beneath SLC. This observation is consistent with post-Bonneville faulting that extends to at least 1 km to the south of the current WSF mapped limit ( Fig. 2; DuRoss et al., 2014). Unlike our 500 south V S result, we find no seismic evidence consistent with lateral spreading along either profile. This observation may support a tapering of slip along the southern limits of the WSF and transfer to the EBF beneath this part of downtown SLC.

Discussion
When we combine our interpretations with past observations, we show that the WSF and EBF are linked. The best evidence for active faulting within this left-step region is at the HJ trench site. Here, both west-and north-trending faults offset the youngest Lake Bonneville strata, and these shallow faults lie above a steep, east-dipping growth fault. This presumed antithetic fault, coupled with the other synthetic and antithetic subsidiary faults, defines a diffuse deformation zone where fault dips and displacements change across distances of a few hundred meters.
Our interpretation of linked, subparallel faults that are capable of a throughgoing rupture is not unique. Wesnousky (2008) showed many examples in which an earthquake has triggered a rupture on an adjacent fault within a fault system of similar dimensions. Many of these examples show tectonic displacements within the stepover region. One difference here is that there is no topographic fault expression within downtown SLC, suggesting that either the identified faults have not produced measurable displacements during the past few earthquake cycles (DuRoss et al., 2016) or that the topography was reset during early settlement. Given the diffuse nature of faulting, we may expect only small tectonic displacements for any one fault during a large earthquake.
The prominent fold and related faults that we identify on all three seismic profiles lies within the hanging wall of the EBF and is consistent with a rollover anticline. These structures are common with listric normal faults with active deposition (Xiao and Suppe, 1992;Pang et al., 2020). To explain the broad zone of distributed faulting, we support the relay structural model of Fossen and Rotevatn (2016). They noted that when two subparallel fault segments interact, a complex deformation zone of subsidiary faults and folds can form a topographic relay ramp, tapering slip on each fault as they overlap (Fig. 5). High-Coulomb shear stress in the relay zone would favor fault linkage near the center of the overlapping fault system or beneath downtown SLC. The presence of north-and northwest-trending subsidiary faults observed with seismic and trench data are consistent with this structural model. Surface topographic data and depth estimates to the base of Tertiary strata (Roten et al., 2011) are consistent with a southwest-dipping unbreached ramp between the WSF and EBF (Fig. 5). The Virginia Street fault (Van Horn and Crittenden, 1987)  We use the Parihaka fault in the Taranaki basin, New Zealand, as an analog to the EBF-WSF left step (Fig. 5). Through 3D seismic data, Giba et al. (2012) showed two offset faults, mapped about 2 km apart at shallow depths. These two faults show increasing linkage with depth, with a prominent fold between the linked faults. Soliva et al. (2006) characterized the fault-zone-width (lateral distance between faults) to fault-overlap-length ratios for a range of similar fault systems. Their empirical database for a 3 km left step between the EBF and WSF would place a fault overlap length between 3 and 10 km. Regardless of how the fault tips are measured or the fault zone details, this model is consistent with the WSF extending south to at least 800 south, and through the SLC downtown corridor.
The complex distribution of faults identified at both trench sites and with our seismic data suggest a detailed stratigraphic and structural analysis is needed to fully assess each fault's slip history. Although this analysis is beyond the scope of this article and the presented dataset, the proximity of active faults, shallow water, and low V S (loose soils) suggest that M < 5 earthquakes may induce liquefaction beneath downtown SLC (Ambraseys and Menu, 1988), and larger earthquakes (M > 6:5) may induced surface ruptures that could impact buildings not designed to handle such deformation.

Conclusions
Seismic imaging through SLC provides a catalog of faults, soil properties, and fluid distributions within a left step of two faults associated with the WFZ. We show a strong relationship of V S to mapped geology and of V P to groundwater elevations. We place our seismic results in the context of paleoseismic trench investigations, geologic maps, and other geophysical data to highlight structural controls needed to assess seismic hazard for the SLC area. We identify a deformation zone and relay ramp that connects two active faults of the WFZ. These observations and models suggest that surface deformation within downtown SLC may accompany a large earthquake. In addition, we identify a broad zone of lateral spread deposits that lie above the transfer zone. The proximity of paleoliquefaction to the underlying faults suggests that M < 5 earthquakes may induce lateral spreading beneath downtown SLC. This fault connectivity model may also relate to other WFZ segments and could help refine the surfacerupture hazard and fault connectivity along other normal fault systems.

Data and Resources
Seismic data are available at the Incorporated Research Institutions for Seismology (IRIS) as an assembled dataset (http://ds.iris.edu/ds/nodes/dmc/forms/assembled-id/) and data products are archived at the Utah Geological Survey in the GeoData Archive (https://geodata.geology.utah.gov). The Wasatch Front Community Velocity Model (CVM) is available at https://geology.utah.gov/hazards/assistance/consultants/cvmgeophysical/. Light detection and ranging (lidar) elevation data are available at https://gis.utah.gov/data/elevation-and-terrain/ 2013-2014-lidar/. Earthquake epicenters are from the National Earthquake Information Center (NEIC) available at https:// earthquake.usgs.gov. Water well data are available at http:// www.waterrights.utah.gov/wellinfo/wellsearch.asp. We used the Pan American Center for Earth and Environmental Sciences (PACES, http://gis.utep.edu/subpages/GMData.html) database, in which gravity measurements were obtained using the 1927 North American Datum (NAD 27) and were terrain-corrected using a U.S. Geological Survey-derived (USGS) digital elevation model (DEM). All websites were last accessed in February 2021. The supplemental material for this article includes our seismic approach and discussion of model uncertainties.

Declaration of Competing Interests
The authors declare no competing interests.