New cosmogenic 3He chronologies and geologic mapping of faulted glacial drift provide new constraints for the slip rates of active faulting in the central Cascade arc, Oregon, USA. The White Branch and Dilman Meadows fault zones cut deposits created by three distinct glacial advances, which provide timing, kinematics, and rate constraints for fault motion. New cosmogenic 3He data from landforms comprising the youngest and most widespread deposits have ages between 19.4 +10.1/–6.2 ka and 21.3 ± 4.9 ka; therefore, they were deposited during the last glacial maximum (LGM). A second, older outwash surface reveals an age of 74.2 ± 3.8 ka, which suggests glaciation possibly associated with marine isotope stage (MIS) 5b. Dip-slip displacement across fault scarps expressed by lidar data reveal similar magnitudes of extensional deformation for LGM and older glacial deposits on the White Branch fault zone, which implies a lack of earthquake ruptures between the oldest and LGM advances. In contrast, scarp profiles along the Dilman Meadows fault zone reveal progressive cumulative slip for surfaces of increasing age. Taken together, our measurements provide the first constraints on the rate of extensional faulting derived from Quaternary geochronology along the White Branch and Dilman Meadows faults, which total 0.1–0.4 mm/yr since ca. 75 ka and 0.6 ± 0.04 mm/yr since the LGM, respectively. The White Branch fault zone accommodates predominately fault-normal extension, whereas right-oblique slip characterizes the Dilman Meadows fault zone. Active deformation across the central Cascade Range thus reflects the combined effects of ongoing crustal block rotation and arc magmatism.

Crustal faults in the U.S. Pacific Northwest accommodate both long- and short-term deformation reflecting interactions of crustal blocks delineated by both geology and geodesy (Wells et al., 1998; Fig. 1). Long-term deformation determined by paleomagnetic data reveals continuous clockwise rotation of the Oregon forearc with respect to stable North America about a pole near the Washington–Oregon–Idaho, USA, border since the Neogene (Simpson and Cox, 1977; Wells and Heller, 1988; McCaffrey et al., 2007). Similarly, short-term deformation revealed by GPS geodesy paints a picture of clockwise rotation, which indicates the longevity of this velocity field and deformation pattern (Fig. 1; Miller et al., 2001; Hammond and Thatcher, 2005; McCaffrey et al., 2013). Given the broad and diffuse nature of intraplate deformation in western North America (Fig. 1), however, the relative contribution of individual crustal faults in accommodating overall plate motion is not well understood.

Extensional faulting and associated arc volcanism in the Oregon Cascades arise from the combined effects of subduction and distributed shear across the Cascadia subduction zone (Fig. 1; Wells and McCaffrey, 2013). How extension and shearing of the crust accommodates the oblique motion remains largely unclear (Waldien et al., 2019). Low rates of fault slip (generally <0.5 mm/yr), dense tree cover, and ubiquitous young volcanic deposits from the Cascade volcanic arc mute the surface expression of active faults. Thus, fault slip rates, kinematics, and earthquake histories are only known for a small number of faults regionally (e.g., Pezzopane and Weldon, 1993; Brocher et al., 2017).

Advances in both the coverage of high-resolution airborne lidar and cosmogenic radio nuclide geochronology offer incremental progress in closing this knowledge gap. Recent work in the Oregon Cascades highlights the utility of airborne lidar data for identifying and characterizing Quaternary active faults (Speth et al., 2018; Vadman and Bemis, 2019; Waldien et al., 2019). Additionally, development of the cosmogenic 3He exposure systems allows for new chronometric constraints on formerly undated landforms and deposits sourced from the volcanic landscape (Kurz et al., 1990; Deligne et al., 2016; Speth et al., 2018; Bacon and Robinson, 2019). These advances create new opportunities to assess the timing and style of active faulting in the central Oregon Cascades and its role in accommodating North American plate crustal deformation.

Here, we provide new constraints on active faults spanning the crest of the Oregon Cascades volcanic arc. These structures include the White Branch fault zone (Conrey et al., 2002) on the west side of the arc crest and the Dilman Meadows fault zone, situated in the Shukash Basin, a sub-basin in the La Pine graben east of the crest (Fig. 2). New surficial lidar-based geologic mapping of fault scarps and offset Quaternary landforms and 3He exposure age dating provide quantitative bounds on slip rate and kinematics. These new data reveal that the two fault zones accommodate ~20–30% of the arc-parallel transtension implied by models of crustal block rotation.

Regional Tectonics

Oblique subduction of the Juan de Fuca plate beneath the North American plate and dextral shear across the Pacific–North American plate boundary drives crustal deformation in the Pacific Northwest (Pezzopane and Weldon, 1993; Fig. 1). In Oregon, the Cascade volcanic arc straddles the structural boundary where the rotating Cascadia forearc block separates from the backarc (Fig. 2; Wells et al., 1998). East–west extension within the Basin and Range, and northwest-directed transtensional dextral shear across the Walker Lane belt, reflect distributed oblique shear across the backarc from the California, USA, border to central Oregon (England and Wells, 1991; Scarberry et al., 2010; McCaffrey et al., 2013; Waldien et al., 2019).

The vertical axis rotational deformation field and relative motion of the forearc block changes orientation relative to the arc from central Oregon northward (Brocher et al., 2017) (Fig. 1). From ~42°N to 44.5°N, GPS and paleomagnetic data indicate that the forearc block moves roughly parallel to the arc relative to stable North America (Fig. 1). This observation indicates that crustal faulting should include a significant component of right lateral strikeslip (McCaffrey et al., 2007; Fig. 2). However, extensional faulting within the Cascades arc is well-established (Smith et al., 1987; Keach et al., 1989; Conrey et al., 2002; Sherrod et al., 2004; Schmidt and Grunder, 2009). Whether those faults include a component of lateral separation is unknown. The White Branch fault zone is one of the structures that accommodates deformation along the arc axis (Conrey et al., 2002).

Normal faults along the Cascades arc represent the western limit of the Basin and Range extensional province (Sherrod and Pickthorn, 1988). At the latitude of central Oregon, major Basin and Range structures to the east of the arc include the Sisters fault zone (Mark-Moser, 2018) and the La Pine graben (Lite and Gannett, 2002; Sherrod et al., 2004). In addition, aligned chains of north–south-trending volcanic cones and fissures mark the region immediately to the east of the arc axis (Pezzopane and Weldon, 1993; Personius, 2002; Weldon et al., 2003; Vadman and Bemis, 2019). The north–south orientations of the volcanic chains are similar to known Quaternary faults in the La Pine graben and surrounding areas. In detail, the La Pine Basin on the east and the Shukash Basin on the west comprise the two sub-basins of the La Pine graben (Lite and Gannett, 2002). The Dilman Meadows fault zone, the Wampus fault zone, and other sub-parallel, N–NNE-trending faults form the structural boundary between the La Pine and Shukash Basins (Fig. 2) (Lite and Gannett, 2002; Vadman and Bemis, 2019).

Glacial History

A composite ice cap and individual alpine glaciers occupied the central Oregon Cascade crest in the Late Quaternary (Scott, 1977). Glacial moraines, outwash, and other deposits record the extent of at least three major Quaternary advances away from the crest of the range (Scott, 1977). Local deposits include the Abbot Butte Formation, the Jack Creek Formation, and the Cabot Creek Formation (Scott, 1977). The Cabot Creek Formation records the last glacial maximum (LGM, ca. 18−22 ka) in central Oregon (Scott, 1977). 3He exposure dating of well-preserved moraines in the Klamath Basin to the south represent the onset of LGM deglaciation at 17.6 ± 2.1 ka during marine isotope stage (MIS) 2 (Fig. 1; Speth et al., 2018). Crandell and Miller (1974) correlate the Jack Creek Formation with the Hayden Creek advance at Mt. Rainier (40–80 ka), which are potentially correlative with glacial moraines of 97.6 ± 12.7 ka in age in the Klamath Basin (Speth et al., 2018). The Abbot Butte Formation is loosely estimated at 200–900 ka based on soil development, weathering rind thickness on clasts in deposits, and lack of moraine crest preservation (Scott and Gardener, 1992; Sherrod et al., 2004). If correct, the Abbot Butte Formation is older than the oldest moraines in the Klamath Basin, which are dated as 168.9 ± 22.3 ka (MIS 6; Speth et al., 2018).

To the east of the Cascade crest, two major outwash surfaces in the Shukash Basin grade westward to moraines formed by glaciers flowing eastward from the Cascade crest (Fig. 2; Scott and Gardener, 1992). Mazama tephra (7682–7584 cal yr B.P.) mantles both surfaces (Egan et al., 2015). Geophysical data suggest a third, older outwash deposit is interbedded within the lacustrine deposits in the subsurface (Miller and Markiewicz, 2000; Lyon, 2001). The deposit imaged in the seismic reflection survey was not sampled but is proposed to be from a glaciation older than Jack Creek based on surficial mapping by Scott (1977) (Lyon, 2001).

To characterize the style and rate of deformation across the central Cascade Range, we mapped fault traces, volcanic and surficial deposits, and landforms using Quick Terrain Modeler and Esri ArcGIS software and new lidar data (DOGAMI, 2015a, 2015b, The field mapping focused on two fault zones. Maps of cross-cutting relations between faults and glacial moraines and associated deposits reveal the character of the White Branch fault zone near, but to the west of, the Cascade crest. Faulted glacial drift, fluvial terraces, and Mazama tephra provide constraint on the timing and kinematics of motion on the Dilman Meadows fault zone in the Shukash Basin on the east flank of the Cascade arc.

Cascade Crest–White Branch Fault Zone

The White Branch fault zone is expressed over a 750 km2 area to the west of the Cascade crest (Fig. 3A, Files S1–S21). A complex and highly distributed zone of faulting ~30 km long and ~10 km wide characterizes the White Branch fault zone (Figs. 3A, 4, and 5). The faults are dominantly north–northwest-striking, steeply dipping to the east and west, and are highly segmented with typical lengths of <~2.5 km for individual fault strands. To utilize glacial landforms as strain markers for fault-related deformation, we present the surficial deposit stratigraphic framework followed by analysis of fault surface expression.

Stratigraphic Framework

Glacial drift and moraines developed in association with both alpine glaciers flanking volcanic edifices and on valley floors near larger trunk glaciers. Young moraines qualitatively exhibit higher boulder frequencies and sharper crests than older moraines (Scott, 1977; Speth et al., 2018). We utilize and build upon the stratigraphic framework of Scott (1977) east of Three Fingered Jack (Fig. 2) and other mapping in the region (Black et al., 1988; Walker and Duncan, 1989; Sherrod et al., 2004; Deligne et al., 2016).

We delineate three generations of glacial drift and landforms affected by the White Branch fault zone (Qg2, Qg5, and Qg6) (Fig. 3A). The oldest glacial drift (Qg6) is preserved only at the lowest elevations in the northwestern portion of the mapping area. This drift is distinguished from younger glacial drift by the northeast–southwest glacial fluting of the landscape, the lack of preserved moraine crests, lack of large boulders on the surface, and by a thicker soil development of up to ~3 m (File S3, see footnote 1). Unit Qg6 is inferred to correlate with the oldest Abbott Butte Formation of Scott (1977).

Well-formed, but muted and rounded, moraine crests characterize the intermediate glacial deposits (Qg5). Slopes of Qg5 moraines are gentle, and boulders are small and sparse. Road cuts into Qg5 reveal a well-developed A soil horizon with organic debris, which has a variable depth. Cross-cutting relationships indicate that the intermediate glacial deposit includes two discrete moraine deposits, Qg5a and Qg5b (Fig. 3A). Moraine crest sharpness and position in the landscape suggest that Qg5 correlates with the intermediate-age Jack Creek Formation of Scott (1977) and Sherrod et al. (2004).

The youngest glacial deposits in the White Branch fault zone (Qg2) consist of a suite of recessional terminal and lateral moraine sequences distinguished by sharp lateral moraine crests and hummocky topography of the terminal moraines (Figs. 3C3D). Boulders on the surface are slightly weathered and can exceed 2 m in diameter. The relative degree of soil formation varies across the unit but is dominated by an A soil horizon containing organic debris in the upper 30 cm of the ground surface. The Qg2 unit is the only glacial deposit observed on the White Branch Valley floor and is subdivided into Qg2a–e based on cross-cutting relationships of moraine crests (Figs. 3B3C). Similar to Qg5, the moraine crest sharpness and position in the landscape makes the Qg2 deposit resemble the youngest age glacial deposit of Scott (1977) and Sherrod et al. (2004).

Cascade East Flank–Dilman Meadows Fault Zone

The Dilman Meadows, Wampus, and other fault zones mark the boundary between the La Pine and Shukash Basins of the La Pine graben (Lyon, 2001; Lite and Gannett, 2002). The Dilman Meadows fault is a 12-km-long, down-to-the-east normal fault that is well exposed in a cut bank of the Deschutes River (Figs. 6B and 6D). Previous work on the Dilman Meadows fault suggests variable slip rates since ca. 140 ka ranging from 0.04 mm/yr to 0.15 mm/yr (Lyon, 2001). Vadman and Bemis (2019) identified several other normal fault strands that rupture the surface (<1.5 m) across the Shukash Basin and suggested that these fault splays appear to cut the Mazama tephra, which indicates rupture during the Holocene. For this study, we consider the faults recognized by Vadman and Bemis (2019) along with the Dilman Meadows fault as the Dilman Meadows fault zone.

We mapped a 175 km2 area encompassing the Dilman Meadows fault zone to establish the fault slip sense and rate (Fig. 6A and Files S4–S5, see footnote 1). The Dilman Meadows fault zone is ~10 km wide and contains dominantly steeply dipping normal faults that dip both to the east and west (File S4). The Dilman Meadows fault zone is differentiated from the Wampus fault zone on the basis of strike. Dilman Meadows fault strands generally strike to the north, whereas Wampus fault zone faults strike NNE. The map presented in Figure 6A includes new mapping and incorporates the work of Vadman and Bemis (2019). Fault segments vary in length from 0.5 km to 3 km. The stratigraphic framework revealed by our mapping is presented first to provide context for the ensuing structural analysis.

Stratigraphic Framework

Our work refined bedrock mapping by MacLeod and Sherrod (1992) and surficial mapping of Quaternary deposits by Lyon (2001). The nearest moraines are located ~15 km west of the map area, which indicates that Quaternary sediment represents an outwash plain sourced from glacial catchments to the west (Lyon, 2001). Seismic reflection studies indicate that as much as 250 m of fill occupies the Shukash Basin. Deposits include glacial outwash (~50 m thick), lacustrine sediment, and tephra layers (~200 m) (Lyon, 2001; Fig. 6C). Lyon (2001) distinguished two glacial outwash units with laterally varying thicknesses.

An older unit (Qgo5) is >5 m thick and consists of fine gravels and interbedded sand with crossbedding; a well-developed soil profile is present in the upper 60 cm. Grain sizes range from 1 mm to 5 mm in sandy layers and 1 mm–250 mm in sandy gravel layers. Clasts include a mix of rock types but predominantly consist of basalt to basaltic andesite. Roots are present throughout the soil layer and are not observed at depth.

The younger outwash unit (Qgo2) is a 1.5–13.4-m-thick deposit that is inset within channels incised into the Qgo5 outwash surface (Lyon, 2001). The deposit consists of gravel and interbedded sand with roots observed to depths of 75 cm in some locations. The relative degree of soil development is less than in Qgo5 and observed only to depths of ~45 cm grading gradually to sands and gravels similar to those observed in Qgo5. The region is blanketed by <3-m-thick Mazama tephra (7682–7584 cal yr B.P.; Egan et al., 2015).

A flight of fluvial terraces cuts the basin fill (Fig. 6C). The surface of the oldest outwash deposit, Qgo5, is marked by numerous abandoned fluvial channels and is incised by terrace surface T4 (Fig. 6A). Cut-in-fill terraces T2 and T3 developed in the Qgo2 outwash fill (Figs. 7A7B). Late Holocene fluvial terraces (T0 and T1) comprise modern channel deposits (Qal) (Fig. 6D). Unit Qal is found on either side of modern channels and includes outwash gravels, reworked Mazama tephra, alluvium, and colluvium (Lyon, 2001). The T1 terrace formed in Qal and is a fill terrace. Terrace T0 is a cut-in-fill terrace that represents the active channel of the Deschutes River.

Scarp profiles extracted from the lidar data enable measurement of fault scarp height along the White Branch and Dilman Meadows fault zones. We use Quick Terrain Modeler v8.0.7.1 to extract 3-m-wide topographic swath profiles from lidar orthogonal to fault scarps to measure the vertical separation and calculate dip-slip displacement across faults. Piercing lines such as moraine crests and terrace tread-riser intersections constrained the direction and magnitude of the slip vector. We follow Thompson et al. (2002) by using a Monte Carlo simulation to incorporate uncertainties in age, fault dip, fault position, and geomorphic surface projection to in-the-fault slip calculations. Fault dip was determined by multiple three-point problems, and the dips in the models ranged from 60° to 90°. Fault-normal profile locations were selected where both the units and surface geometries could be reliably correlated across the fault trace. Fluvial terraces and associated risers offset by the Dilman Meadows fault constrain both the lateral and vertical components of offset. We use the LaDiCaoz MATLAB tool (Zielke and Arrowsmith, 2012; Haddon et al., 2016), which calculates both vertical and lateral offset using the shape of correlative landforms on each wall of a fault. The tool calculates the optimal offset and creates a back-slipped lidar hillshade that allows for visual assessment of offset measurement.

White Branch Fault Zone

A total of 140 scarp measurements reveal broad distributions of purely dip-slip displacements across all mapped units (Files S7–S9, see footnote 1). Fault scarp profiles from offset glacial drift in unit Qg2a–e, Qg5a–b, and Qg6 show slip values that range from 0.3–12.7 m, 0.3–7.6 m, and 0.4–6.6 m, respectively (Fig. 8A). Deposit age does not correlate with the amount of measured offset (Fig. 4). The largest offsets in the oldest unit (Qg6) are within the mean of the offsets observed in the youngest unit (Fig. 8B). This observation is supported by a Kruskal-Wallis ANOVA test χ2 value of 0.05.

Individual fault strands exhibit variable displacement measurements along strike (Fig. 9A). These data are plotted using a normal kernel density estimate to determine whether there is a typical value of displacement (Fig. 9). This analysis reveals one group of fault peaks that has a similar vertical separation as a function of distance where 0–4.5 m characterizes the vertical separation on 85% of all of the scarp measurements. In the second population, 50% of all measurements range from 1.9 m to 12.3 m. The overlap in individual measurements of the two populations leads to double counting in the percentages in some cases.

We explore whether the populations of scarps result from multiple generations of surface rupturing earthquakes by calculating the mean average vertical separation using a 1 km moving window with a 2σ error envelope (Fig. 9). Gaps (dashed lines) in the along-strike profile represent areas with no measurement. The differences in populations are most apparent along the 11–18 km distances from the fault end points (Fig. 9A). The two populations and along-strike variability in offset are also apparent in map view of separation (Fig. 8A). The north and south extent of the faulting reflects either the limits of the fault zone, burial by young lava flows, gaps in the lidar coverage, or slip transfer to another fault zone.

Dilman Meadows Fault Zone

Scarps in the Dilman Meadows fault zone are blanketed by the Mazama ash. Whereas Vadman and Bemis (2019) report offset of the Mazama ash across the Twin Lakes maar fault 8 km to the west, it is unclear whether other faults in the Dilman Meadows fault zone cut the tephra. Thus, uncertainty for each measurement is introduced because it is unclear whether the Mazama ash drapes preexisting scarps or if the relief reflects post-Mazama surface faulting. Diffusion of the Mazama tephra across the scarps potentially reduces the scarp height and the associated fault slip. We conservatively infer that the scarp heights in glacial units below the tephra have displacements similar to those of scarp height at the surface. Displacement measurements from 120 scarp profiles (Files S10–S12, see footnote 1) reveal the dip-slip component of motion. Slip values range from 0.3 m to 5.3 m. Displacement patterns on individual faults in the Dilman Meadows fault zone are comparable to those of the White Branch fault zone in that the faults are short, segmented, and variable in height along their length. In contrast to the White Branch fault zone, offset does vary as a function of age. Offset inferred from scarps across Qgo2 is less than from scarps in Qgo5 (Fig. 10). Offset is largest near the center of the fault traces in general, and displacement monotonically tapers toward the fault ends (File S13).

Oblique slip characterizes displacement on the Dilman Meadows fault. Determination of the lateral component of slip is complicated by the vertical component of separation because of uncertainties in how to project piercing lines into the fault plane. Apparent lateral separation of terrace risers on the Dilman Meadows fault suggests an overall ratio of lateral to vertical offset as large as 3:1 (Figs. 7C7F and File S6, see footnote 1). The Monte Carlo and LaDiCaoz approaches were used to measure the vertical and lateral components of slip, respectively. Lateral offset of the T1 terrace riser calculated by the LaDiCaoz tool is 1.5 +0.5/–1.5 m, and the vertical offset is 1.5 ± 0.1 m. The vertical offset measured across the T1 terrace using the Monte Carlo approach was 1.7 ± 0.1 m, which is comparable to the vertical offset measured using LaDiCaoz. Lateral and vertical offsets of the T3 terrace are 6.5 +2/–0 m and 1.6 +0.2/–0.4 m, respectively, and the vertical offset measured using the Monte Carlo approach was 1.6 ± 0.03 m. Lateral and vertical offsets of the T4 terrace are 11.1 +3.4/–0.2 m and 3.4 +0.4/–0.1 m, respectively, and the vertical offset measured using the Monte Carlo approach is 3.4 ± 0.1 m.

Establishing the timing and rate of slip on the two fault systems requires knowledge of surficial deposit and landform age. We applied 3He cosmogenic dating to establish chronologic constraints on sediment and landform ages. The surface boulder method was used for moraines offset by the White Branch fault zone at the Cascade crest sites. In the Shukash Basin on the east flank of the Cascades, the depth profile method was appropriate because the dating objectives were outwash surfaces.

The boulder method uses the following criteria in sample selection: (1) the size of the boulder is >1 m in diameter; (2) the location is within 2 m of the crest of the moraine; (3) the degree of exhumation is low, as evidenced by a similar degree of weathering on all sides; (4) the degree of weathering is low, as evidenced by the presence of intact weathering rinds that show the original outer edge of the boulder; and (5) ideally a minimum of four boulders is sampled for a calculated age (Putkonen and Swanson, 2003). In the White Branch fault zone, boulders on the older moraines (Qg5 and Qg6) did not meet these criteria, and we were only able to date three moraines from the youngest Qg2 moraine complex.

A single depositional unit underlying an outwash surface is sampled at fixed intervals in the depth method (Anderson et al., 1996). We collected at least six samples at 15 cm intervals beginning at the top of the outwash surface, except in the case of sample DM_DP_01, where root stirring had potentially occurred, and the loose nature of the Mazama tephra made it impossible to collect the outwash deposit without also mixing in the Mazama tephra. At each sampling horizon, we used a trowel to expose fresh, undisturbed material, being careful not to incorporate material from above. We collected samples from both of the larger outwash surfaces (Qgo2 and Qgo5) in the Shukash Basin.

Boulder Exposure Dating in the White Branch Fault Zone

Assessing the ages of surfaces and landforms using exposure dating methods relies on nuclide production in minerals as a result of cosmic ray bombardment (Dunai et al., 2007). Given the dominance of basalt and basaltic andesite in the Oregon Cascades, we targeted 3He, which is produced in pyroxene and olivine phenocrysts in intermediate volcanic rocks (Gosse and Phillips, 2001). Recent studies (Speth et al., 2018) highlight the utility of this technique in central Oregon. Inherited concentrations of 3He in a moraine boulder include: (1) prior exposure to cosmic rays and (2) non-cosmogenic pathways, and for glacially transported boulders, we expect that they will have no prior exposure to cosmic rays, so we make a correction only for noncosmogenic 3He. Inherited concentrations lead to biasing toward older model ages. Erosion, including erosion of the boulder surface and erosion of the landform, leading to exhumation or diffusion, lowers the observed concentration of 3He and therefore biases to younger ages.

We collected and analyzed 22 boulder samples from Qg2, and one shielded sample, to estimate the inherited noncosmogenic 3He concentration in the boulders (Fig. 3). There were few boulders >1 m in the area, which resulted in minimal sampling opportunities along the moraine crests and left us to sample several boulders that were <1 m in diameter. We used a hammer and chisel to remove ~2.5 kg of the outer few centimeters of intact weathering rind from the tops of the boulders (Files S14–S16, see footnote 1). On average, snow covers this region for four months of the year, but because of the dense forest cover we are unable to estimate the thickness of the annual winter snow cover and thereby the effect of snow shielding. We do not apply a snow correction because of a lack of constraints, and as a result, our ages may somewhat underestimate the actual ages (Licciardi et al., 2004).

We separated a mixture of pyroxene and olivine grains from crushed rock by sieving to 125–425 µm and applying magnetic and heavy liquid separations. We then powdered the resulting separates in ethanol to minimize trapped gases in fluid inclusions and then measured total 3He concentrations in the resulting powders by noble gas mass spectrometry at Berkeley Geochronology Center. Analytical methods are described in detail in Speth et al. (2018).

We calculate exposure ages using version 3 of the online exposure age calculator described by Balco et al. (2008), which was subsequently updated ( The shielded sample had a 3He concentration of 4.8 × 105 atoms/g 3He, which is an order of magnitude less than total 3He in the boulder samples, so corrections for noncosmogenic 3He are quite small; subtracting the shielded concentrations from the total 3He concentrations in each boulder had little effect (Amidon and Farley, 2011). Due to variability in texture and vesicularity of the source rock, actual erosion rates are most likely variable depending on where in the lava flow each boulder originated. Boulders that originated in a more vesicular section of the flow might have higher erosion rates than a boulder originating from a less vesicular section of the flow. In the absence of independently known specific erosion rate values for each boulder, we applied three different erosion rate scenarios to the age calculations: (1) no boulder erosion, (2) an erosion rate of 1 mm/kyr, and (3) a maximum plausible erosion rate of 3 mm/kyr (Speth et al., 2018). We report exposure ages calculated using the 1 mm/kyr scenario based on weathering rind thickness observed in the field (Table 1).

The age results of the boulder samples are displayed using normalized kernel density plots (Fig. 11). The plots use Gaussian distributions to show individual ages and uncertainties for each sample based on the 1 mm/kyr erosion rate scenario, which are calculated using Gaussian distribution methods described in Zechar and Frankel (2009) with 2σ uncertainties. The ages that do not contribute to the highest peak in the cumulative probability function are excluded from the preferred calculated ages reported (Speth et al., 2018). A reduced χ2 statistic computed shows that uncertainty within the measurements is likely due to scatter in the data and not geomorphic processes that would tend to artificially lower the apparent age of the unit (Table 2).

We sampled 10 boulders from a terminal moraine in unit Qg2b (Figs. 3C3D and Files S15–S16, see footnote 1). The ages of the boulders range from 19.4 ± 2.6 ka to 22.7 ± 2.9 ka, not including outliers. The age reported for the Qg2b moraine excludes four outlier ages from samples WB18-05 (35.1 ± 3.1 ka), WB18-06 (52.1 ± 6.6 ka), WB18-09 (6.5 ± 1.3 ka), and WB18-16 (14.3 ± 2.1 ka), and we chose to exclude these ages as they do not contribute to the largest peak in the normal kernel density estimates. We report an age of 20.3 +6.2/–5.3 ka for the Qg2b moraine (Fig. 3C). We collected seven samples from the Qg2c moraine that do not show any cluster of ages but a relatively even distribution that ranges from 16.1 ± 2.3 ka to 25.9 ± 3.4 ka; and for this reason, we included all seven ages in the reported age of 19.4 +10.1/–6.2 ka (Fig. 3C). From the lateral moraine in Qg2d, we excluded two of the five ages from samples we collected, WB18-14 (36.5 ± 3.3 ka) and WB18-15 (41.6 ± 3.7 ka); the similarity of the ages excluded could reflect a geologic process that resulted in older ages or scatter in the data. In either case, the ages did not contribute to the largest peak in the normal kernel density estimates and were excluded. The ages used for the reported age of the Qg2d moraine range from 20.7 ± 2.0 ka to 22.1 ± 2.1 ka, and the age we report for this moraine is 21.3 ± 4.9 ka (Fig. 3D). Based on the stratigraphy, the Qg2 advances observed must be different; however, they all have indistinguishable LGM ages.

Depth Profiles in the Shukash Basin

3He cosmogenic nuclide dating methods can also be applied to estimate the residence time of an outwash deposit at the surface. In contrast to glacially transported boulders, inherited 3He in outwash is likely to contain both noncosmogenic 3He and also cosmogenic 3He produced earlier in the exposure history of the sediment. However, for the depth profile method, it is not necessary to know the source of the inherited 3He. Given the assumption that the deposit is well mixed so that inherited 3He is constant with depth, it can be estimated from the measurements. The nuclide concentration-depth relationship is given by:


where N(z) is the concentration (atoms/g) of 3He at a depth z (cm), No is the 3He concentration at the surface produced after emplacement of the deposit, ρ is the density (g/cm3) of the outwash material, Λ is the mass attenuation length for production by high-energy spallation (150 g/cm2), and Ninh is the inherited 3He concentration (Anderson et al., 1996). Fitting Equation 1 to the data yields estimates for No, Ninh, and ρ, and the resulting value for No can then be used to compute an exposure age for the outwash surface.

An additional complication at our sites is that the outwash in which the depth profiles were measured is overlain by Mazama tephra. Thus, the cosmic-ray shielding depth of all samples changed abruptly at the time of tephra deposition. We account for this by observing that:


where P (atoms/g/yr) is the production rate of 3He (obtained from Balco et al., 2008, as discussed in the Methods), E is an erosion rate (cm/yr), which we assume is the same before and after tephra deposition; zt and ρt are the thickness and density of the tephra; and tpre and tpost are the exposure times of the surface (yr) before and after tephra deposition (Lal, 1988). We measured zt and ρt (File S17, see footnote 1), tpost is known from the age of the tephra (7682–7584 cal yr B.P.; Egan et al., 2015), and we assume a value for E. This leaves tpre as the only unknown, and the exposure age of the outwash is tpre + tpost. As we do not have an independent estimate for E, we compute the exposure age using several values (see Figs. 1213 and discussion below).

In the Dilman Meadows fault zone we sampled two glacial outwash surfaces: (1) Qgo5, the highest terrace surface in the mapping area, and (2) Qgo2, a separate terrace deposit inset into Qgo5 (Fig. 6A). As with the moraine boulders from White Branch fault zone, we targeted pyroxene and olivine minerals in the clasts because of the abundance of basalt and basaltic andesite. We dug two ~2.5-m-deep pits into the outwash surfaces and collected ~2.5 kg of material at each 15 cm interval. The samples were prepared using the same methods as those for the boulder samples at Middlebury College, Middlebury, Vermont, USA, and at Western Washington University, Bellingham, Washington, USA. We analyzed the samples at the Berkeley Geochronology Center, Berkeley, California, using the same methods as those for the boulder samples.

After determining the best fit values of No for each surface, we interpreted the exposure age of the surface using Equation 2. To do this, we need an estimate of the erosion rates for each surface. Without a way to directly measure the erosion rates in the Shukash Basin, we look for evidence of what the erosion rate could be. Both terrace surfaces are flat; both have soil development beneath the ash, as seen in the depth profile pits (Figs. 12A and 13A), which implies that the erosion rate is likely very low. In the case of the older surface (Qgo5), there appears to be a silt cap where an inflation soil indicates that rather than erosion of the surface there is actually material being added to the surface.

The depth profile pit for the younger glacial outwash surface (Qgo2) was dug into the upthrown block east of a fault scarp in the center of unit Qgo2 (Fig. 6A). This location was chosen based on permits for digging. From the Qgo2 pit, we collected six samples at 15 cm intervals beginning at the top of the outwash unit; the tephra at this location had more structure, and we were able to collect samples without mixing materials (Fig. 13A). Small roots occur throughout the deposit but do not appear to have significantly mixed materials (Fig. 13B). There were few signs of surface weathering, and the grains contained higher percentages of pyroxene than olivine. There was 0.45 Matoms/g of inherited 3He in the sample when the best fit equation was calculated for the unit, which was similar to the amount calculated in the shielded sample from the White Branch fault zone. There was an exponential decrease in the concentration of 3He, which indicates that the outwash was deposited rapidly (Fig. 13C).

We dug the depth profile pit for the older glacial outwash surface (Qgo5) in the side of an abandoned quarry (Fig. 6A). From the Qgo5 pit, we collected seven samples at 15 cm intervals beginning 30 cm below the top of the outwash unit (Fig. 12A). The loose structure of the Mazama tephra above the outwash unit made sampling the surface of the outwash impossible without mixing large amounts of tephra in with the outwash gravels, and several large roots in the upper 30 cm of the outwash surface made sampling at those depths less than ideal due to the potential for root mixing of the deposit affecting the concentration of 3He measured (Fig. 12B). The surface of the outwash has significant soil development to depths of 60 cm, and weathering of individual grains is apparent, which implies that the surface has not been stripped. The grains collected had high percentages of olivine relative to pyroxene. Similarly, in the Qgo5 profile there was an exponential decrease of the 3He concentration in the Qgo2 profile, which shows that the outwash was also deposited rapidly (Figs. 12C and 13C).

To understand the importance of erosion rates for these depth profiles, we report ages using three separate erosion rate scenarios. In the first scenario we assume no erosion has occurred since deposition of the gravels. In the second we apply a high erosion rate of 10 mm/kyr, and in the third we assume a moderate amount of erosion has occurred and apply a rate of 5 mm/kyr since deposition (Fig. 12D and Fig. 13D). The age results for the younger outwash terrace, assuming no erosion, is 20.3 ka, and the age of the older is 64.9 ka. The age results of the younger outwash terrace assuming a 10 mm/kyr rate (~20 cm of total erosion) is 21.3 ka, and the age of the older outwash terrace assuming the 10 mm/kyr rate (~90 cm of total erosion) is 88.8 ka. The age results of the younger outwash terrace assuming a 5 mm/kyr rate (~10 cm of total erosion) is 20.8 ± 3.2 ka, and the age of the older outwash terrace assuming the 5 mm/kyr rate (~30 cm of total erosion) is 74.2 ± 3.8 ka. The level of pedogenic development in the older outwash terrace pit is inconsistent with a 10 mm/kyr erosion rate, and it is unlikely that no erosion has occurred in the last ca. 65 ka; we use the intermediate erosion rate of 5 mm/kyr and ages calculated using the 5 mm/kyr rate for the slip calculation in the following sections.

Fault Slip Rates

We calculate slip rates from piercing lines such as moraine crests and terrace risers offset by faulting. We then compare these to the horizontal GPS velocity field to determine the contribution of long-term surface faulting to accommodation of clockwise rotation across the forearc–backarc transition of the upper plate along the Cascadia plate margin (McCaffrey et al., 2013; Brocher et al., 2017; Mark-Moser, 2018). The ages of glacial deposits from the LGM, MIS 5, and slip measurements calculated from the Monte Carlo simulation are used to calculate slip rates for both the White Branch and the Dilman Meadows fault zones.

White Branch Fault Zone

Offset measurements in the White Branch fault zone exhibit no correlation between age and the amount of offset measured (Fig. 8). Moreover, offset landforms reveal no evidence of lateral separation. Dip-slip measurements in the White Branch fault zone range from 0.3 ± 0.1–12.7 ± 0.6 m, with variability exhibited by three independently dated geomorphically distinct Qg2 moraines. The Qg2 offset measurements show the same magnitude of offset as is found in the Qg5 moraines and in Qg6 deposits. The fact that topographic relief is similar across scarps in old moraines and young moraines suggests that deformation occurred entirely post LGM.

To further understand the post-LGM deformation, we sum the offset across the fault zone in 1 km windows along a north–south transect (Fig. 14A). Along this transect, we measure dip slip (resolved to a dipping plane), extension, vertical separation (measured from projected surface), and the absolute values of vertical separation (Thompson et al., 2002) to compare how the magnitude of offset varies from north to south. The values of each component of offset increase toward the center of fault strands and indicate an overall net displacement down to the west. We use the window with the largest summed slip (41.2 ± 1.8 m) for a maximum summed dip-slip rate of 2.1 ± 0.1 mm/yr and a maximum summed extension rate of 1.7 ± 0.1 mm/yr since the LGM. If we consider deformation of the Qg5 moraine complex for the White Branch Valley using our age from the Shukash Basin (74.2 ± 3.8 ka), the summed slip rate is reduced to 0.6 ± 0.1 mm/yr; although again, we consider the overlap for individual scarp offset measurements in Qg2 vs. Qg5 as strong evidence that all scarps in the White Branch fault zone result from post-LGM earthquake surface ruptures. Furthermore, similar scarp heights are measured across multiple generations of glacial deposits Qg2b and Qg5b (Figs. 3A and 9) on the westernmost fault trace north of the White Branch Valley over a 2.5 km distance along strike. There is a high (99%) likelihood that where similar scarp heights are measured between two points ~2.5 km apart, the scarp being measured was generated during the same rupture (Biasi and Weldon, 2006).

We compare this summed rate to a transect along a single Qg5 lateral moraine crest that is cut by multiple fault traces (Fig. 14B). This moraine was deposited by a valley glacier on the south side of the White Branch Valley and is the only single geomorphic feature that spans the width of the fault zone (Figs. 5B and 14B). The White Branch fault zone has a dominant net displacement down to the west, despite the Qg5 moraine having a net displacement of 1.2 m down to the east; this is likely because the faults cutting the moraine primarily accommodate extension (Fig. 14B). A larger summed dip-slip offset (23.5 ± 1.2 m) across this transect than the corresponding summed dip-slip offset from the fault zone window (18.5 ± 0.8 m) implies that the slip and corresponding rate revealed by considering only the Qg5 moraine and post-LGM onset of faulting are nominally larger. The values of summed dip-slip offset between the transect of a moraine with an inferred MIS 5 age and the summed fault zone window at that location are similar, which provides more evidence that observed deformation in the White Branch fault zone is entirely post LGM.

Similar styles of distributed faulting in Searles Valley and Death Valley, California, and along the Sierra El Mayor, Baja California, typically involve graben structures formed by normal faulting that sole into a master fault at depth (Axen et al., 1999; Hayman et al., 2003; Numelin et al., 2007). The graben-style structures observed in the White Branch fault zone could be explained by a master fault at depth similar to that of the West Klamath Lake fault zone in southern Oregon, where along-strike variability on sub-parallel, surface-rupturing normal faults is likely controlled by a single fault (Speth et al., 2018). The change in net displacement of the White Branch fault zone along strike from down to the west to down to the east implies the complexity observed at the surface continues with depth. The proximity of the White Branch fault zone to the Three Sisters volcanic complex could also affect the rupture patterns similar to the pattern of the 2019 Ridgecrest earthquake, where the ruptures terminated close to the Coso volcanic field due to high temperatures and greater pore pressure, which weakened the crust and inhibited slip propagation (Chen et al., 2020).

The measured slip rates for the White Branch fault zone are comparable to others throughout central Oregon. Across central Oregon, several fault zones were measured that show average slip rates of 0.3–1 mm/yr (Pezzopane and Weldon, 1993; Mark-Moser, 2018; Speth et al., 2018). The White Branch fault zone is similar to the West Klamath Lake fault zone in that individual faults are small, but cumulatively in the fault zone, the slip rate is higher and varies from 0.03 mm/yr to 2.6 mm/yr. The average slip rate across the White Branch fault zone is 0.6 ± 0.04 mm/yr, which agrees well with other slip rates reported throughout the region. The White Branch fault zone is a short fault zone with relatively large displacement. A lack of lidar to the northwest limited the opportunity to explore whether and how the fault zone continues into the Western Cascades.

Dilman Meadows Fault Zone

Offset varies as a function of surface age across the Dilman Meadows fault zone (Fig. 10). Therefore, we were able to consider slip rate across multiple time intervals (Fig. 15). Evidence of faulting on multiple time scales includes post-Mazama tephra (ca. 7.6 ka) earthquakes in the Dilman Meadows fault zone (Lyon, 2001; Vadman and Bemis, 2019) and greater offsets in the pre-Mazama deposits Qgo5 than in Qgo2.

We combined terrace offsets, our new outwash exposure ages, and the age of Mazama tephra deposition to calculate slip rates (Fig. 7). In the Qal unit we recognize the lowest elevation terrace (To) as the active channel deposit, which is not offset by the Dilman Meadows fault. Terrace tread T1, a higher and older terrace formed within the Qal unit, is offset by the Dilman Meadows fault. The terrace records a 1.7 ± 0.1 m vertical and 2.0 ± 0.5 m lateral offset. On the south side of the Deschutes River, terrace tread T2 is cut into the Qgo2 outwash, but it does not extend to the fault. Terrace tread T3, the abandonment surface of the Qgo2 outwash deposit, has a vertical offset north of the Deschutes River of 2.0 ± 0.0 m and 1.6 ± 0.0 m south of the river (File S6, see footnote 1). On the southern exposure, we measured 6.5 ± 2.0 m of lateral offset. Terrace T4, the Qgo5 abandonment surface south of the Deschutes River, is the highest terrace tread. Offsets are 3.1 ± 0.1 m vertically and 11.1 ± 3.4 m laterally. We interpret the similarities in offset values between T1 and T3 to mean that the terrace surfaces have experienced the same number of earthquakes since deposition of Mazama tephra and use the 20.8 ± 3.2 ka age of Qgo2 to calculate oblique slip rates of 0.1–0.3 mm/yr. We infer from the larger offset measured across Qgo5 that terrace T4 has experienced more earthquakes than the younger, lower terraces. Using the 74.2 ± 3.8 ka age of the Qgo5 deposit yields an oblique slip rate of ~0.2 mm/yr for this terrace. A trench north of the Deschutes River revealed evidence of at least one pre-Mazama earthquake (Fig. 7B) (Lyon, 2001). Previous studies of the Twin Lake maar fault just west of the map area have shown evidence of post-Mazama tephra rupture (Vadman and Bemis, 2019).

Offsets measured in the glacial units of Qgo5 are slightly higher than in Qgo2. Distributed faulting among the numerous short, low-relief scarps likely formed from single events and may explain the small difference (Fig. 10B). In addition, the Mazama tephra mutes, but does not completely obscure, the surfaces and scarps, which complicates the use of scarp height to infer the vertical component of separation. One post-Mazama unit, Qal, is offset, which allows for use of the Mazama tephra as a maximum age of the unit to calculate a minimum rate since 7.6 ka (Table 3; Lyon, 2001; Vadman and Bemis, 2019).

Using offsets measured in the trench north of the Deschutes River across the Dilman Meadows fault and general ages of glaciation, Lyon (2001) reports intervals of slip rates for the Dilman Meadows fault. The slip rate reported by Lyon (2001) for the Dilman Meadows fault, which is between 18 ka and 7.6 ka, is 0.3 mm/yr, which is within the 0.1–0.3 mm/yr range of oblique slip rates that we report. The Dilman Meadows fault could account for nearly half of the total slip rate across the entire Dilman Meadows fault zone.

Regional Tectonic Deformation

A continuous clockwise vertical axis rotation deformation field in the Pacific Northwest is evidenced by paleomagnetic rotations and geodesy (Wells et al., 1998; McCaffrey et al., 2007). Fault-bounded crustal domains are defined using seismicity patterns, GPS data, and known faults (McCaffrey et al., 2007; Savage and Wells, 2015). East of the Three Sisters volcanic complex, McCaffrey et al. (2007) defined a block boundary between the southern Oregon Coast Range domain (Western Cascades) and the eastern Oregon domain (Cascade backarc; Fig. 2). The slip vector calculated at this block boundary is ~1.1 mm/yr to the southeast, nearly parallel to the White Branch fault zone. Dextral slip on faults is predicted to accommodate this motion at the boundary, but the components of this vector can also be broken down into velocity parallel (vp = 1.0 mm/yr) and velocity normal (vn = 0.4 mm/yr) motion (Fig. 16; Mark-Moser, 2018). The average summed horizontal extension rate across the White Branch fault zone calculated from the 1 km increment divisions of the fault zone is 0.4 ± 0.04 mm/yr and indicates that the White Branch fault zone is capable of accommodating the vn component of the slip vector calculated for the block boundary. However, the 1.04 mm/yr of fault-parallel motion must be accommodated by other faults. Along the arc axis, dextral shear is likely being accommodated through the White Branch fault zone extensional array, which could explain the absence of lateral motion on the individual faults (Crider, 2001; Wesnousky et al., 2012). The Dilman Meadows fault zone accounts for some of the deficit. Importantly, the Dilman Meadows fault has a 3:1 strike- to dip-slip ratio. The post-76 ka oblique slip rates of 0.1–0.3 mm/yr indicate that oblique slip is distributed across the backarc in response to the vertical axis deformation field. Thus, both the White Branch and Dilman Meadows fault zones are likely seismogenic and potentially accommodate ~20–30% of Oregon forearc rotation in the High Cascades.

Glacial Chronology of the Central Oregon Cascades

The 3He cosmogenic surface exposure dates on moraines and outwash provide new absolute age constraints on the glacial chronology of the central Oregon Cascades. The map observations and age data from the crest and east flank of the Cascades of this study compare well with the three-part glacial stratigraphy of Scott (1977). Our data suggest that the main glacial advances are related to advances during MIS 2, 5b, and 6.

The LGM (MIS 2) is well represented west of the Cascade crest by moraines that are ca. 20 ka, and east of the crest by an outwash deposit (Qgo2) that is 20.8 ± 3.2 ka (Mix, 2001). Our ages compare well with the ages of moraines and outwash in the Klamath Basin in southern Oregon and moraines in the Wallowa Mountains in eastern Oregon (Speth et al., 2018; Laabs et al., 2020; Licciardi et al., 2004).

The penultimate glaciation (MIS 5b) is represented by a 74.2 ± 3.8 ka-age outwash deposit (Qgo5) in the Shukash Basin. This age overlaps with moraine ages in the Klamath Basin and near Leavenworth, Washington (Speth et al., 2018; Porter and Swanson, 2008). Unable to date pre-LGM moraines on the west side of the Cascade crest, we relied on similar preservation and appearance of moraines along the Cascade crest. We use the existence of MIS 5b moraines in the Klamath Basin and MIS 5b outwash from east of the Cascade crest as evidence that the Qg5 drift west of the Cascade crest is likely from a MIS 5b glacial advance.

Deposits from glaciations older than MIS 5b are either not preserved or buried in the Shukash Basin, despite the fact that MIS 6-aged moraines are found in the Klamath Basin (Hawkins et al., 1989). MIS 6 is dated at 144 ± 14 ka in the Sierra Nevada (Rood et al., 2011), and in the Klamath Basin a moraine boulder age of 168.9 ± 22.3 ka was tentatively linked to MIS 6 (Speth et al., 2018). It is possible that the oldest Qg6 drift west of the Cascade crest represents a MIS 6 glaciation (138 ± 13 ka; Laabs et al., 2020), but we do not have exposure age data to confirm this inference. Morphological evidence, however, suggests that the Qg6 deposit is much older than the Qg5 and Qg2 deposits. Moraine crests are not preserved on the Qg6 surface, and the direction of ice flow was different than during younger glaciations, as evidenced by north–south glacial lineations preserved on the surface of Qg6 that are unlike the east–west glacial lineations observed in the younger drift (File S1, see footnote 1).

Fault-Volcanism Connections

It is likely that the same stresses explain arc-parallel faulting and arc-parallel cinder cone alignment (Muffler et al., 2011; Deligne et al., 2016). The trends of the cinder cones in the central Oregon volcanic field, ~8 km north of the White Branch fault zone, vary between N18°W and N11°W and N5°E and N7°E, and the average strike of the faults in the White Branch fault zone is N25°E. The strike of the northernmost faults in the White Branch fault zone is more northerly than that of faults farther south, which could be the result of the emplacement of feeder dikes ~200–300 m below the surface changing the stress orientations (Deligne et al., 2016). The faults in the White Branch fault zone are likely not due to intrusive episodes like those recorded around South Sister volcano that produce little to no seismicity, last days to years, and are separated by decades to centuries of quiescent periods (Dzurisin et al., 2006). Therefore, the faults in the White Branch fault zone are likely tectonic and accommodate some fraction of Oregon forearc rotation in the High Cascades.

The Dilman Meadows fault zone is located south of the Mt. Bachelor volcanic chain, and trenching evidence suggests colluviation followed by episodic motion on the Dilman Meadows fault, which indicates that some faults are seismogenic (Fig. 17A; Lyon, 2001). Graben-style faulting is apparent at the site, where sets of faults dip east and west across the older outwash surface (Qgo5) (Fig. 17B). A fault in the eastern part of the outwash units displays scissor motion (Fig. 17B), where across Qgo5 the fault dips to the west and as it continues into Qgo2 the dip direction changes to the east. A fault with its southern terminus in Qgo5 is segmented to the north, where north–south-oriented fissures strike parallel to the fault (Fig. 17C) and project toward a set of north–south-aligned cinder cones. The Twin Lakes maar fault is marked by a series of craters on both the north and south ends and shows evidence of post-Mazama surface rupture (Vadman and Bemis, 2019). These faults likely serve as conduits for ascending magma or could be the result of changing stresses as magma ascends, resulting in faulting.

New mapping, cosmogenic nuclide dating, and scarp profiling in the central Oregon Cascades provide new constraints on faulting, glaciation, and the distribution of strain related to the vertical axis rotational deformation field in the North American plate in the Pacific Northwest. From this work, we conclude the following:

  1. Surficial geologic mapping based on bare-earth airborne lidar indicates that LGM glacial deposits and two older glacial deposits are preserved on both sides of the Cascade crest. The glacial sequence proposed by Scott (1977) is represented by three generations of moraines at the Cascade crest and by outwash deposits in the Shukash Basin.

  2. Our cosmogenic ages suggest that glacial landforms and deposits correlate across the Cascade crest. The dating indicates glacial advances during MIS 2 (ca. 20 ka), MIS 5 (ca. 75 ka), and, although not directly dated, possibly MIS 6, which is correlative to the 144 ± 14 ka Tahoe glaciation in the Sierra Nevada.

  3. Offset measurements on the White Branch faults suggest that deformation postdates the LGM (20 ka), despite multiple scarps crossing landforms that are at least MIS 5b in age. This pattern suggests either seismogenic quiescence or magmatic upwelling quiescence prior to the LGM. Faulting in the White Branch Valley area is bracketed in age by the deposition of LGM drift and the most recent volcanism at 1.6 ± 0.1 ka. The average strike of the White Branch fault zone is similar to the alignment of cinder cones, likely because they share the same parent stresses.

  4. Faults cutting the surfaces in the White Branch fault zone accommodate primarily extension, with individual throws and slip rates being small, but when summed, they indicate regionally significant active deformation in this region. Slip rates summed in 1 km increments range from 0.03 mm/yr to 2.1 mm/yr, with an average of 0.6 ± 0.04 mm/yr. Extension rates summed in 1 km increments range from 0.01 mm/yr to 1.7 mm/yr and average 0.4 ± 0.04 mm/yr.

  5. The Dilman Meadows fault zone is characterized by right-oblique transtension. The Dilman Meadows fault has a 3:1 lateral to vertical slip ratio and a post-75 ka slip rate of 0.2 mm/yr. Other structures in the fault zone preserve evidence of dip-slip displacement. Previous trenching evidence suggests that the faults are seismogenic. In some locations, faults share along-strike continuity with fissures and are likely related to magma intrusion.

This research was supported by the Western Washington University Graduate School and the Western Washington Geology Department in Bellingham, Washington. All lidar digital elevation models were obtained from the Oregon Department of Geology and Mineral Industries. We are grateful to the students from Middlebury College for their assistance with sample preparation. Cosmogenic dating was performed at the Berkeley Geochronology Center. Special thanks go to Reyne K. Lesnau for assistance in the field. Reviews from Ashley Streig, Tom Brocher, and two anonymous reviewers greatly improved this manuscript.

1Supplemental Material. Includes map plates, cosmogenic radionuclide sample-site information, and scarp profile data tables. Please visit to access the supplemental material, and contact with any questions.
Science Editor: Andrea Hampel
Associate Editor: Jason W. Ricketts
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.