The spatial and temporal evolution of the Portland and Tualatin forearc basins, Oregon, USA

The Portland and Tualatin basins are part of the Salish-Puget-Willamette Lowland, a forearc trough along the Cascadia subduction zone that is formed by oblique convergence of the Juan de Fuca plate beneath North America (Fig. 1; Evarts et al., 2009). The inland Salish-Puget-Willamette Lowland trough is separated from offshore basins by the Coast Range, and in that respect, it is similar to the Nankai margin of southwest Japan, southern Alaska, and southern Chile (Bassett and Watts, 2015). Such inland seas and lowlands are population centers, and they are tectonically active, producing, for example, the upper-plate 1995 M 6.9 Kobe earthquake in southwest Japan and Cascadia's M 7+ Seattle earthquake ca. 900 A.D., and in the lower plate, Cascadia's 2001 M 6.8 Nisqually earthquake and the 2019 M 7 Anchorage earthquake (Wald, 1996; Kao et al., 2008; Nelson et al., 2014; West et al., 2020).

Rogers (2002) suggested that these margins with inland seas were the product of young, warm slab subduction, and more recently, Bodmer et al. (2019) have correlated Cascadia's Coast Range elevation with buoyant asthenosphere along strike. The Coast Range appears to be primarily a late Neogene feature, as flood basalts of the Columbia River Basalt Group (CRBG) sourced from the backarc flowed across the forearc into the ocean across a broad front from the central Oregon coast to the central Washington coast between 16 and 12 Ma (Snavely et al., 1973; Beeson et al., 1979). The uplift of the Coast Range may be superimposed on or partly coincident with a Neogene forearc transition from Paleogene transtension and mafic magmatism to deformation dominated by north-south transpression due to impingement of the northward- migrating forearc against the Canadian Coast Mountains (Snavely and Wagner, 1963; Snavely et al., 1993; Snavely and Wells, 1996; Wells et al., 1998; McCaffrey et al., 2013; McPhee et al., 2014; Wells et al., 2014).

The record of Oregon Coast Range development, forearc migration, and Cascade magmatism as recorded in the stratigraphy of the Portland and Tualatin basins is poorly understood. In this study, we provide constraints on the structural and tec-tonic evolution of the Cascadia forearc by tracking basin depocenters (areas of maximum sediment accumulation) in the Portland and Tualatin basins through space and time (e.g., Ingersoll, 1978; Zak and Freund, 1981; Heller et al., 1988; Flemings and Jordan, 1990). Previous geological and geophysical studies have been conducted in the Tualatin basin (Beeson et al., 1989a; Popowski, 1996; Wilson, 1997, 1998; McPhee et al., 2014) and in part of the Portland basin (Beeson et al., 1989a; Roe and Madin, 2013), though an integrated geologic model of both the Portland and Tualatin basins currently does not exist. We synthesize well log, outcrop, seismic, aeromagnetic, and gravity data to better understand stratigraphic architecture and the spatio-temporal evolution of accommodation space within the Portland and Tualatin forearc basins. We then use these observations to provide temporal constraints on the emergence of the Oregon Coast Range and initiation of transpressional deformation within the Portland Metropolitan area. Our basin model also provides information needed for ongoing seismic hazard and resource assessments in the Portland metropolitan and surrounding areas (e.g., Givler et al., 2009; Roe and Madin, 2013).

The Portland and Tualatin basins cover an area of ~3900 km² and are part of the 900-km-long Salish-Puget-Willamette Lowland, a forearc trough situated between the Coast Range and the Cascade volcanic arc (Fig. 1; Evarts et al., 2009; Bassett and Watts, 2015). The basins are elongated in a NW-SE orientation and are generally bound by northwest-striking dextral strike-slip faults (Beeson et al., 1989a; Blakely et al., 1995, 2000; Wong et al., 2001; Liberty et al., 2003; Evarts et al., 2009; Walsh et al., 2011; Wells et al., 2020a). A dextral sense of motion on these bounding faults is compatible with the modern stress field where the maximum horizontal compressive stress is oriented roughly north-south, essentially 45° oblique to the northwest striking faults (Werner, 1990; Yeats et al., 1991). In Paleogene time, the Tualatin and Portland basins were part of a large, continuous marine basin that stretched from the Portland basin west to the present-day ocean (Evarts et al., 2009).

Two crustal earthquakes of >M 5 have been recorded in or near the Portland and Tualatin basins in the past 60 years. A M 5.2 earthquake occurred in the Portland basin in 1962 (Yelin and Patton, 1991), and the 25 March 1993 M 5.7 Scotts Mills earthquake that occurred ~58 km south of Portland indicates that damaging earthquakes are possible (Thomas et al., 1996; Givler et al., 2009). The Gales Creek fault west of the study area shows evidence for Quaternary activity (Redwine et al., 2017; Horst et al., 2019, 2020; Wells et al., 2020b), as clockwise rotation and northward migration of the Oregon Coast Range results in dextral shear on faults in the study area and across the forearc (Fig. 2) (McCaffrey et al., 2007; Wells et al., 2020b).

The stratigraphy of the Portland and Tualatin basins records a history of volcanism and sedimentation in both fluvial and marine depositional environments (Fig. 3). Oceanic basalts and basaltic sedimentary rocks of the Siletz River Volcanics, commonly referred to as Siletzia, comprise the Eocene basement underlying Cenozoic basin fill of the Portland and Tualatin basins (Snavely et al., 1968; Duncan, 1982; Yeats et al., 1996; Wells et al., 2014). Accretion of the Siletzia terrane to North America (NAM) at the latitude of Oregon occurred between 51 and 49 Ma (Wells et al., 2014). West of the study area along the east flank of the Coast Range anticline, basement rocks exposed are lower Eocene submarine basalt of the Siletz River Volcanics, associated diabase sills, and subaerial basalt of the mid-Eocene Tillamook Volcanics (Wells et al., 1984 [their fig. 1], 2014, 2020a; Trehu et al., 1994; Blakely et al., 2000). The southern part of the Portland Hills anticline separating the Portland and Tualatin basins is underlain by the Eocene basalt of Waverly Heights, best exposed adjacent to the Willamette River near the Waverly Heights area (Fig. 1; Beeson et al., 1989b; Blakely et al., 2004). Waverly Heights basalts are similar to the Tillamook Volcanics and are evidence of Eocene Siletzia basement below (McPhee et al., 2014). Cascade arc volcanism was near its present location by ca. 45–40 Ma, after the accretion of Siletzia and westward migration of the subduction zone (Snavely and Wells, 1996; Schmandt and Humphreys, 2011; Wells et al., 2014). Trehu et al. (1994) suggested that the western Cascades erupted over a well-developed sedimentary basin and that the eastern boundary of Siletzia exists beneath the western Cascades. Paleoaltimetry results from central Oregon support interpretations that the Oregon Cascades were high by Oligocene time (Bershaw et al., 2019). Following the accretion of Siletzia to Oregon, at least 2 km of marine and marginal-marine sedimentary rocks were deposited in both the Portland and Tualatin basins from Eocene to Oligocene time (Popowski, 1996; McPhee et al., 2014; Wells et al., 2020a). In the Portland basin, deposition of these sediments was coeval with eruptions from an incipient western Cascade volcanic arc and their eastern extent delineates the Paleogene coast (Evarts et al., 2009, 2010). In mid-Miocene time, the Columbia River Basalt Group (CRBG) erupted 210,000 km³ of flood basalt from a series of dike swarms near the present-day junction between Oregon, Washington, and Idaho (Reidel et al., 2013). At 15.5–16.5 Ma, flows of the Grande Ronde Basalt passed through the Cascade Range via the ancestral Columbia River into the nascent Portland and Tualatin basins (Beeson et al., 1985, 1989a; Beeson and Tolan, 1990). CRBG unconformably overlies volcanic rocks of the older, eroded Paleogene and Neogene western Cascades and uplifted marine sedimentary rocks as flows inundated the Portland and Tualatin basins (Beeson et al., 1989a; Liberty, 2002; Wells et al., 2020a).

Late Miocene volcaniclastic rocks of the Rhododendron Formation overlie the CRBG in the southeast portion of the Portland basin on the west flank of the Cascade Range (Trimble, 1963). The lower Sandy River Mudstone was deposited in the basin during the last half of the Miocene (Evarts et al., 2009). This unit is interpreted as lacustrine, consisting of silt and very fine sand (Trimble, 1963). At the end of the Miocene, the Columbia River began to deposit coarse sandstone and conglomerates of the lower Troutdale Formation. Clast composition suggests an extrabasinal source in pre-Cenozoic rocks in eastern Washington and Idaho (Evarts et al., 2009). The upper Sandy River Mudstone was also deposited during this time, suggesting the ancestral Columbia River was a meandering system with low-energy floodplains (Tolan and Beeson, 1984; Evarts et al., 2009). Low-potassium tholeiite (LKT) flows erupted at 3.5 Ma in the Cascade Range to the east, generating hyaloclastite as the LKT flowed into the ancestral Columbia River. This resulted in deposition of a hyaloclastic sandstone member of the Troutdale Formation, deposited as a volcaniclastic alluvial fan in the eastern portion of the Portland basin (Evarts et al., 2009). Unconformably overlying the CRBG in the Tualatin basin are ~450 m of late Miocene and younger fluvial siltstone, sandstone, mudstone, and the Hillsboro formation of Wilson (1998), deposited under low-energy conditions (Yeats et al., 1996; Wilson, 1997, 1998; McPhee et al., 2014). On the basis of gross lithology and stratigraphic position, Madin (1990) considered these sediments equivalent to the Sandy River Mudstone. However, Wilson (2000) suggests that the Portland and Tualatin basins have been relatively isolated from each other since late Miocene time based on differences in elemental geochemistry plots of silt and clay samples from Neogene nonmarine clastic deposits in both basins. The present-day topography of the Portland and Tualatin basins is influenced by underlying structure and geologic events of the past ~2.5 m.y. Faults trend NW-SE throughout the study area, influencing stream erosion and the distribution of high topography. The Boring Volcanic Field, consisting of cinder cones and associated lava flows, small shields, and lava cones, erupted west of the Cascade arc axis during the latest Pliocene (Treasher, 1942; Conrey et al., 1996; Evarts et al., 2009, 2010). More recently, late Pleistocene glacial outburst floods (Missoula floods) inundated and scoured the study area, depositing mud, sand, and gravel on the valley floors (Waitt, 1985).

The modern boundaries of the Portland and Tualatin basins were established by mid-Miocene time based on distribution maps of the CRBG and inversion of gravity data (Beeson et al., 1989a; Evarts et al., 2009; McPhee et al., 2014). Cross sections based on a gravity survey through a light-rail tunnel (Blakely et al., 2004) in the Portland Hills show step-like anomalies that are consistent with steeply dipping reverse faults bounding the Portland Hills anticline, a structure consisting of CRBG that separates the Portland and Tualatin basins (Fig. 1). The Sylvan-Oatfield and Portland Hills faults comprise part of the larger, northwest-oriented Portland Hills–Clackamas River structural zone described by Beeson et al. (1985, 1989a) and Blakely et al. (1995); this zone has experienced folding and faulting since mid-Miocene time (Beeson et al., 1985; Blakely et al., 2004). South of the Beaverton fault, Cooper Mountain and related sub-surface structures are interpreted as E-W–oriented anticlinal folds in CRBG forming the hanging wall of the Beaverton fault (Fig. 5; Wells et al., 2020b). These folds reflect the northward motion of western Oregon in Neogene time.

This study models three key stratigraphic horizons in the Portland and Tualatin basins (top CRBG, base CRBG, and Eocene basement) using well-log, outcrop, seismic, aeromagnetic, and gravity data. Model outputs for this study cover ~3885 km² of the Portland and Tualatin basins. A geologic workflow similar to that in Burns et al. (2011) is employed, in which all available data are compiled as inputs for trend surface (horizon) generation. Kingdom Suite, a geological interpretation software package, was used to generate a series of structure and isochore maps, which were converted to metric in Petrosys. Kingdom's Flex Gridding algorithm uses a system of differential equations whose solution yields a grid of points that must pass through (or very close) to the data in XYZ space, resulting in low residuals. Bicubic interpolation, which uses slope information, was employed to produce smoother output grids. Data used to model stratigraphic surfaces are summarized in Figure 4 and included in Tables S1 and S2¹.

A total of 2336 wells were used to model the top of CRBG (Table S2). Many of the wells contained in these data sets do not penetrate the top of CRBG; however, the deepest of these wells provides minimum elevation estimates of CRBG. The best well-log control for the top of CRBG exists at locations along the margin of the basins, where top CRBG elevations range from 0 to 250 m (Fig. 1). In addition to well data, depth-converted seismic profiles in both the Portland and Tualatin basins were used to estimate top CRBG elevations (Popowski, 1996; Wilson, 1997; Liberty, 2002; Ma et al., 2012). Outcrop data surrounding the basins were integrated using a new, regional geologic map of the study area (Wells et al., 2020a) superimposed onto a 10-m-resolution digital elevation model (DEM). Short-wavelength aeromagnetic anomalies caused by surface and near-surface presence of CRBG and geologic field relations helped qualitatively delineate an interpreted eastern boundary of CRBG in the Portland basin (Fig. 4; Blakely et al., 1995, 2000).

A total of 52 wells and outcrop exposures from geologic mapping were used to model the base CRBG in this study (Table S2). While only four wells penetrate the entire CRBG section, seven wells reach into the Wapshilla Ridge unit of the CRBG and are interpreted to be close to base CRBG based on stratigraphy (Beeson et al., 1989a). Seismic and well-log data in the Portland basin were used to estimate the thickness of CRBG along a 2D seismic line shot along the Columbia River (Liberty, 2002). The base of CRBG is exposed along the margins of the Tualatin basin and around the Dutch Canyon anticline (Fig. 6). As with the top CRBG horizon, surface geology was used to guide interpolation. Previously modeled gravity data were also used to interpret base CRBG elevations, where thinner flows overlie gravity (basement) highs, and thicker flows overlie gravity lows (McPhee et al., 2014). Exposures of pre-CRBG sedimentary rock and basement (i.e., Waverly Heights, southwest flank of the Portland Hills, and Dutch Canyon) provide additional constraints on the thickness and areal extent of this unit in the study area (Fig. 4).

Our Eocene basement surface is derived from the gravity inversion of McPhee et al. (2014). They calculated the depth to the top of the Eocene volcanic basement from the density contrast between the sedimentary basin fill and the oceanic basalt of the Siletz terrane, which is exposed in the Coast Range and in a few deep hydrocarbon exploration wells. The exposure of Eocene basement at the surface (i.e., Waverly Heights basalt) provides important constraints for both the gravity inversion and geometry of overlying units (Fig. 4; Beeson et al., 1989b). The gravity inversion map is likely inaccurate in the southeast portion of the Portland basin, where modeled basement crosses the over-lying top and base CRBG horizons. Sedimentary and volcaniclastic rocks on the eastern margin are largely derived from the Cascade Mountains and are denser than stratigraphically equivalent sedimentary units to the west. McPhee et al. (2014) suggest that Paleogene Cascade volcanic arc rocks underlie parts of the Portland basin.

Major faults in the study area were modeled as sub-vertical planes, based on their linear traces as mapped in Wells et al. (2020a), geophysical data, and evidence for strike-slip motion (Fig. 4). Most of the displacement on the northwest-striking and steeply dipping Sylvan-Oatfield, Portland Hills, and East Bank faults is likely dextral (Blakely et al., 1995, 2000, 2004). Fault surfaces were modeled down to Eocene basement, where offsets in the gravity inversion grid of McPhee et al. (2014) can be correlated to mapped faults at the surface.

Three isochore maps were generated for time intervals that span the basin history: (1) post-CRBG sedimentary and volcaniclastic rocks (0–15 Ma), (2) Miocene Columbia River Basalt Group (CRBG, 15.5–16.5 Ma), and (3) early Miocene to late Eocene sedimentary and volcaniclastic rocks (ca. 17–35 Ma) (Fig. 3). Each isochore map is computed as the difference between a geologic unit's top surface and its bottom. As in Burns et al. (2011), interpolation guides were introduced into the modeling process to create surfaces that are consistent with geologic conceptual models and inference from geological and geophysical data (Roe and Madin, 2013).

In addition to well data and geologic mapping, the CRBG isochore map is based partly on Tualatin basin multichannel seismic-reflection data that were collected in 1984–1985 as part of a search for hydrocarbons (red lines in Fig. 4; Popowski, 1996; Wilson, 1997; Oregon Department of Geology and Mineral Industries, 2012). We converted seismic thicknesses from time to meters for this study using a CRBG velocity of ~5300 m/s (Spitzer et al., 2008). This time-to-depth conversion produces thickness estimates that are consistent with seismic data in the Tualatin basin correlated to wells (WASH 55816, Cooper Mountain) that penetrate ~305–323 m of CRBG (Table S2 [^{footnote 1}]).

The top CRBG horizon is relatively well constrained due to the large amount of subsurface data available (Fig. 4). This surface reveals two distinct relative lows delineating the Portland and Tualatin basins separated by a large northwest-trending anti-cline associated with the Portland Hills (Fig. 5). The top CRBG horizon reaches a greater depth in the Portland basin with a −500 m elevation compared to that of the Tualatin basin, the depth of which is −390 m.

The top CRBG horizon is exposed at the surface along the perimeter of the study area (Fig. 1; Wells et al., 2020a). The horizon reaches a maximum elevation of roughly 640 m at Dutch Canyon, northwest of the Portland Hills (Fig. 5). Structure map rugosity reflects the resolution of the 10 m regional DEM in areas where CRBG is exposed at the surface, particularly along the Portland Hills, Dutch Canyon, and the margins of the Tualatin basin. Mapped faults, which generally strike NW-SE, coincide with steep gradients on the structure map (Wells et al., 2020a). Elevation decreases markedly northeast of the East Bank fault, with a gently sloping surface between it and the Portland Hills fault. The edge of the top CRBG surface terminates near the dextral Lacamas Lake and reverse Prune Hill and Blue Lake faults on the eastern side of the Portland basin. In the Tualatin basin, the top CRBG surface is elevated at Cooper Mountain, in the hanging wall of the Beaverton thrust (reverse) fault reaching a maximum elevation of ~213 m (Fig. 5). This is nearly ~600 m higher than the same surface in its footwall to the north. South of the Beaverton fault, the CRBG forms a complex fold belt in the hanging wall (Wells et al., 2020a, 2020b; diagonal-hatch pattern in Fig. 5). The dextral Canby-Molalla fault (Blakely et al., 2000) links up with the Beaverton fault to the southeast, near the eastern edge of this complex domain, following the interpretation of Wells et al. (2020a).

There are a few locations in the study area where CRBG is missing, and pre-CRBG sedimentary rock is exposed at the surface (cross-hatch pattern in Fig. 5). Dutch Canyon, the core of an eroded anti-cline, exposes older Paleogene and early Miocene sediments at the surface. These same sedimentary rocks are exposed in a small area on the southwest flank of the Portland Hills along the Sylvan-Oatfield fault. CRBG is also missing where it laps onto Eocene basalt exposed south of the Portland Hills near Waverly Heights (Fig. 5).

The base CRBG horizon is exposed at the surface around the western margin of the Tualatin basin, in Dutch Canyon, and the southwest flank of the Portland Hills (Wells et al., 2020a). Exposed pre-CRBG sedimentary rock is denoted with a cross-hatch pattern on Figure 6. In the centers of the basins, the base CRBG horizon (Fig. 6) is poorly constrained because fewer wells penetrate the horizon in the subsurface, and base CRBG is not as easily resolved in seismic images (Wilson, 1997; Liberty, 2002). Despite these uncertainties, our results show that the base CRBG horizon reaches lower elevations in the Portland basin at −820 m relative to the Tualatin basin, which is −730 m, similar to the pattern observed from our results of the top CRBG. The surface reaches a maximum elevation of roughly 570 m at Dutch Canyon (Fig. 6).

Similar to the top CRBG horizon, faults are generally coincident with steep elevation gradients on the base CRBG structure map (Fig. 6). The Sylvan-Oatfield and Portland Hills faults bound a northwest-trending structure of higher elevation coincident with the Portland Hills. The inferred northwest extension of the Portland Hills fault matches a steep gradient that offsets the Portland basin down to the southeast. Similar to the top CRBG surface, elevation decreases markedly northeast of the East Bank fault. A gently sloping surface lies between the East Bank and the Portland Hills fault. At the location of Cooper Mountain in the Tualatin basin, this surface reaches an elevation of about −100 m in the hanging wall of the Beaverton thrust fault and decreases to −488 m in its footwall to the north.

The gravity-derived depth to basement grid of McPhee et al. (2014) reveals a deep depression underneath the Tualatin basin, which gradually increases in elevation toward the western Cascades to the east (Fig. 7). The surface is over twice as deep in the Tualatin basin at −5.7 km compared to the Portland basin, the depth of which is −2.1 km. In the southeastern portion of the Portland basin, western Cascade arc rock is more dense than that assumed for basin fill in the gravity inversion (McPhee et al., 2014). This causes an erroneous shallowing of basement in that area and is denoted by a diagonal-hatch pattern on Figure 7. The surface increases to a local high of roughly −0.9 km at the Dutch Canyon anticline.

In many cases, modeled faults are also coincident with gradients on the Eocene basement structure map, suggesting they deform basement. The Sylvan-Oatfield and Portland Hills faults follow two northwest-trending basement highs at the boundary between the Portland and Tualatin basins. Basement elevation ranges from roughly −1.7 km to −2.1 km in a low that coincides with the Portland Hills anticline. In the Tualatin basin, the basement surface lies at −2.3 km elevation in the hanging wall of the Beaverton fault, decreasing from roughly −3.4 km to −4.5 km in its footwall to the north (Fig. 7).

The post-CRBG (0–15 Ma) isochore map reveals two distinct northwest-trending depocenters coincident with the Portland and Tualatin basins (Fig. 8). Slightly more sediments were deposited in the Portland basin at roughly 500 m thick compared to the Tualatin basin, which reaches 445 m thick over this time period. Sedimentary rocks thin onto the Portland Hills where CRBG is exposed (Fig. 1). Isolated “bull's-eyes” of 430–460-m-thick basin fill in the southern portion of the Portland basin incorporate volcanic cones of the post-CRBG Boring Volcanic Field. Relatively thick sedimentary deposits in the southernmost portion of the Portland basin continue out of the study area, into the Northern Willamette Valley.

Modeled faults in the study area mark abrupt changes in 0–15 Ma sediment thickness (Fig. 8). The inferred northern continuation of the Portland Hills fault reveals a steep thickness gradient along its footwall with thicknesses up to ~30 m in its hanging wall (to the southwest). Similarly, the East Bank fault bounds a steep thickness gradient to its northeast, shown by a transition from warm colors (thick) to cool colors (thin) to the southwest (Fig. 8). A significant change in thickness is also observed across the Beaverton fault in the Tualatin basin.

The CRBG (15.5–16.5 Ma) isochore map reveals multiple depocenters across the mid-Miocene Portland and Tualatin basins, with basalt flows generally thinning west of the NW-striking Portland Hills fault zone (Beeson et al., 1989a; Fig. 9). On average, thickness of CRBG is comparable between the centers of the basins, with ~275–330 m in the Portland basin and ~275–365 m in the Tualatin basin. CRBG remains relatively thick south toward the Northern Willamette Valley where the Trans-arc Lowland and Sherwood trough of Beeson et al. (1989a) extend southwest toward a gap in the Coast Range. The CRBG thins prominently over the structural highs of Waverly Heights and Dutch Canyon (Fig. 9).

An elongate region of local thickening coincident with the Portland Hills is bound by the Sylvan-Oatfield and Portland Hills faults. Between the faults, the basalt ranges from ~215–300 m thick. Along the trace of the faults, the basalt thins to ~90–180 m before thickening again toward the central Portland and Tualatin basins (to the east and west, respectively) (Fig. 9). The East Bank fault juxtaposes an area of thick basalt to the northeast, against basalt that is ~60–120 m thinner to the southwest. Relatively thick CRBG exists south of the Beaverton fault near Cooper Mountain and in the Sherwood trough. However, uncertainty in this area is high because of the poor constraints on the base of the CRBG.

The pre-CRBG early Miocene to late Eocene (ca. 17–35 Ma) isochore map reveals one distinct depocenter coincident with the western edge of the Tualatin basin (Fig. 10). Early Miocene to Eocene sedimentary rocks thin gradually from the Tualatin basin east toward the Portland basin and western Cascades. Maximum sediment thickness reaches ~5.2 km in the Tualatin basin and ~1.5 km in the Portland basin. We observe a minimum thickness of ~1 km in the core of the Dutch Canyon anticline. Relative uncertainty of sedimentary rock thickness in the southern portion of the Portland basin is denoted by a diagonal hatch pattern.

Similar to the CRBG isochore, it appears there is an area of local thickening that coincides with the Portland Hills bound by the Sylvan-Oatfield and Portland Hills faults. Here, sediment thickness ranges from ~1.5–1.8 km and decreases to a thickness of ~0.9–1.2 km on its flanks. In the southwest portion of the map, the isopach thickness decreases significantly across the Beaverton fault, suggesting it was active during this time. Significant changes in thickness are not observed across the Canby-Molalla fault.

Our results show that Portland and Tualatin basin depocenters have shifted in both location and shape over time. We interpret spatio-temporal changes in basin thickness as related to changes in stress within the Cascadia forearc and emergence of the Oregon Coast Range.

A cross section through the Portland and Tualatin basins (Fig. 11) based on well-log, seismic, and gravity data suggests substantial vertical displacement of basement in pre-CRBG time along northwest-striking faults now interpreted as Quaternary active strike-slip faults (Beeson et al., 1985, 1989a; Beeson and Tolan, 1990; Yelin and Patton, 1991; Blakely et al., 2000, 2004; Wells et al., 2020a). McPhee et al. (2014) interpreted the 5-km-deep Paleogene Tualatin basin as a releasing-bend step-over between the Gales Creek and Portland Hills faults, creating accommodation space prior to CRBG emplacement. However, there is sedimentary evidence that the Tualatin basin was on the continental shelf during this time and likely, an inboard extension of the Astoria-Nehalem basins (Niem and Niem, 1985; Niem et al., 1992b). During Paleogene time, both the Tualatin and Astoria-Nehalem basins are dominated by marine shelf and slope sedimentary strata (Niem et al., 1992b; Wilson, 1997). The thickness of the late Eocene to Oligocene Pittsburg Bluff formation does not change significantly across the Coast Range, where these rocks are documented in hydrocarbon exploration wells and exposed in out-crop (Niem et al., 1992b, 1992a), demonstrating that this unit was likely deposited at a relatively uniform thickness rather than thinning onto a preexisting high. Thinning of the ca. 17–35 Ma isochore map to the west onto the present-day Coast Range (Fig. 11) is primarily due to erosion, as ~4 km of Paleogene and early Neogene marine strata are mapped in the foothills of the Coast Range, dipping gently east into the Tualatin basin (Seismic line WV-1, Oregon Department of Geology and Mineral Industries, 2012; Wells et al., 2020a, 2020b).

We estimate an average sediment accumulation rate in the Tualatin basin of ~286 m/Ma from Paleo-gene to early Miocene time based on a thickness of ~5300 m and age range of ca. 17–35 Ma. This is similar to sediment accumulation rates in the Astoria basin during Oligocene time (~275 m/Ma), estimated from well-log data (Niem and Niem, 1985). These sediment accumulation rates are within the range of typical marine basins at convergent boundaries and represent sedimentation on the continental shelf and slope prior to the creation of the Coast Range and Portland Hills (Schwab, 1976).

Paleogene sedimentary rocks in the Tualatin basin are modeled to be ~5 km thick, producing a −44 mGal gravity low (McPhee et al., 2014). They thin eastward toward the Paleogene coastline, which was near the western Cascades (Fig. 10; Niem et al., 1992b; Retallack et al., 2004; Evarts et al., 2009). The interfingering of marine sedimentary rocks with Cascade arc-derived volcanics suggests deposition prior to the emergence of a subaerial Coast Range ca. 20 Ma (Armentrout, 1983; McKeel, 1984; Stanley, 1991; Niem et al., 1992a; Snavely and Wells, 1996). Our early Miocene to Eocene isochore map shows the Paleogene basin depocenter was >100 km to the west of the modern Cascade arc, with basement elevations increasing to the east (Figs. 10 and 11). This is consistent with Evarts et al. (2009), who suggest that Paleogene and early Miocene sedimentary rocks interfingered with western Cascade volcanics at the Paleogene coastline.

A −28 mGal, northwest-striking gravity low centered over the Portland Hills coincides with 1.5–1.8 km of Paleogene and early Miocene sedimentary rocks bound by the dextral Sylvan-Oatfield and Portland Hills faults (Fig. 10). We interpret this geometry as requiring early normal displacement on these steeply dipping (~70°) faults that offset the basement surface, creating a graben at the present-day location of Portland Hills, into which a thicker package of Paleogene and early Miocene sediments were deposited (Fig. 13). This interpretation is supported by the Barber #1 exploration well, located in the Portland Hills, which penetrates ~2.1 km of volcanic rock, agglomerate, and sands (Newton, 1969). Faulting likely continued during episodic Paleogene north-northwest–directed extension, consistent with Eocene normal faulting in the Mist gas field (Niem and Niem, 1985), and the eruption of Eocene tholeiitic and alkalic basalts where the Coast Range is today (e.g., Tillamook Highlands) and offshore (Wells et al., 1984, 2014; Snavely and Wells, 1991, 1996; Snavely et al., 1993). Rift flank uplift along the edges of the Portland Hills graben (a graben coinciding with the present-day location of the Portland Hills), resulted in up to a couple hundred meters of relief along the northeast margin of the Tualatin basin and southwest margin of the nascent Portland basin (Fig. 13).

Middle Miocene flood basalts of the CRBG inundated the Portland and Tualatin basins following the ancestral Columbia River to the sea (Beeson et al., 1989a; Beeson and Tolan, 1990; Reidel et al., 2013; Wells et al., 2020a). Significant variation in flow thickness is evident in our CRBG (15.5–16.5 Ma) isochore map (Fig. 9), suggesting that there was preexisting topography. Basalt lava flows were deposited onto the Eocene basalt of Waverly Heights, incipient western Cascade arc, Goble Volcanics, Dutch Canyon anticline, and Paleogene to early Miocene sediments (Figs. 9 and 11; Beeson et al., 1989a, 1989b). Our basin model shows that flow paths were influenced by the major northwest-striking fault zones that still dominate the study area (Beeson et al., 1989a; Anderson et al., 2013; Reidel et al., 2013), forming a graben that was filled by CRBG. Previous workers have suggested that an incipient Portland Hills anticline diverted the earliest Grande Ronde Basalt flows of the CRBG (R₁/N₁ magnetic polarity), limiting their extent to the Portland basin (Beeson et al., 1989a; Evarts et al., 2009). However, the discovery of Downey Gulch and China Creek flows of the Grande Ronde Basalt (N₁ magnetic polarity) in the Tualatin basin suggests these early flows inundated the Tualatin basin as well (T.L. Tolan, 2004, written commun.; Dinterman and Duval, 2009; USGS, 2013; Wells et al., 2020a). Our CRBG isochore map, based in part on well data that bottom in the R₂ Wapshilla Ridge unit in the Portland basin (Well MULT 106000) and older N₁ China Creek Member (Well WASH 55816) in the Tualatin basin, suggests the two basins were still connected at ca. 16.5 Ma (USGS, 2013).

CRBG flows thin ~60–90 m across the rift flanks of the Portland Hills graben. Within the graben, flows reach thicknesses of up to ~240–300 m (Figs. 9 and 12). These thickness estimates are reasonably constrained by both well and outcrop data (Table S2 [^{footnote 1}]). Previous geologic mapping suggests that flows thinned depositionally onto gravity highs associated with the Eocene basalt of Waverly Heights and the Dutch Canyon anticline prior to inundating the Tualatin basin (Beeson et al., 1989b, 1991), though erosion associated with rift flank uplift was also likely. Later flows encountered less topography as earlier flows filled in preexisting lows, which is reflected in the widespread distribution of ca. 15.5 Ma Sentinel Bluffs flows of the Grande Ronde Basalt during N₂ time (Beeson et al., 1989a).

Our basin model suggests that faults in the Portland and Tualatin basins were structurally inverted following the emplacement of the CRBG in mid-Miocene time, recording a change in stress from transtension to transpression in the forearc. The dextral Sylvan-Oatfield and Portland Hills faults, which bounded the Portland Hills graben during Paleogene to early Miocene time, provided planes of weakness exploited by transpressive stress in midto late Miocene time, resulting in structural inversion (Fig. 13; Sibson, 1985; Sibson et al., 1988; Letouzey et al., 1990). Fault geometries characteristic of inversion are supported by regional aeromagnetic data that suggest the Sylvan-Oatfield and East Bank faults are steeply dipping structures with reverse slip (Blakely et al., 1995) and described by Beeson et al. (1989a) as flower structures. Our CRBG isochore map indicates that structural inversion did not occur until sometime after CRBG was deposited in mid-Miocene time, as CRBG thickens at the present-day location of the Portland Hills (Fig. 9). A change from transtension to transpression in mid- to late Miocene time is consistent with the onset of shortening in Washington State, documented in accelerated uplift of the Washington Cascades (Reiners et al., 2002), north-south shortening along the Seattle fault (ten Brink et al., 2002), and across the Yakima fold and thrust belt (Reidel et al., 1989; McCaffrey et al., 2016).

The east-west–trending Beaverton fault in the southern Tualatin basin also shows evidence of structural inversion in mid- to late Miocene time. Cooper Mountain, a post-CRBG fold with an east-west–trending axis in the hanging wall of the Beaverton fault shows stratigraphic offset on the top CRBG surface consistent with reverse deformation on a fault dipping to the south (Fig. 5). Other, approximately east-west–striking folds in the area (i.e., Parrett Mountain and the Chehalem Mountain uplift) also formed in response to north-south shortening (Beeson and Tolan, 1990). However, our CRBG isochore map shows that CRBG is relatively thick in the hanging wall, suggesting the Beaverton fault was a normal fault in the mid-Miocene (Fig. 9). McPhee et al. (2014) has also suggested that the Beaverton fault is an inversion structure that formed as a normal fault in response to north-northwest extension in Paleogene time. Structural inversion in the Tualatin and Portland basins is also consistent with the history of deformation on the continental shelf in Oregon where normal faults were reactivated as thrust faults during late mid-Miocene transpression (Snavely and Wells, 1996). Our work provides evidence of a major change in the stress regime from transtension in Paleogene time to transpression in Neogene time.

The existence of distinct mid-Miocene to Pleistocene depocenters in the Portland and Tualatin basins suggests they were effectively separated by the Portland Hills during this time, consistent with petrographic analysis of the Tualatin basin fill (Fig. 8; Wilson, 2000). It is likely that basin separation was synchronous with structural inversion in mid- to late Miocene time since the Portland Hills consists entirely of CRBG blanketed by loess of the Missoula floods (Evarts et al., 2009). The location of the Portland basin depocenter during late Neogene time suggests that the East Bank fault is exerting local control on accommodation space. Our 0–15 Ma isochore map shows that post-CRBG basin fill in the Portland basin is ~55 m thicker than in the Tualatin basin, providing further evidence that uplift of the Oregon Coast Range has progressively pushed the forearc basin depocenter eastward through Neogene time (Fig. 12). In the Puget Sound to the north, where lateral separation between the accretionary wedge (Olympics) and magmatic arc (Cascades) is greater, the Seattle basin reaches a maximum depth of ~9 km, nearly ~3 km deeper than the Tualatin basin (Johnson et al., 1994; Symons and Crosson, 1997; Rau and Johnson, 1999; Blakely et al., 2002; ten Brink et al., 2002; McPhee et al., 2014). Most of this difference is attributed to Neogene (<20 Ma) sedimentary rocks in the Seattle basin, which are significantly thicker than contemporaneous rocks in the Portland basin (~3.6 km thick versus ~500 m thick, respectively) (Fig. 13; Johnson et al., 1994). Relatively thin Neo-gene deposits in the Portland basin are not due to low sediment supply as the Columbia River traverses the area. In the Portland basin, a relatively short distance between the Coast Range and Cascades has reduced accommodation space since the Miocene. In the Seattle basin, there is still sufficient separation between the two; thus, accommodation space remains. Accordingly, we suggest that the location of the Oregon Coast Range and Olympic Mountains uplift relative to the magmatic arc has exerted a first-order control on forearc basin accommodation space since Miocene time, not only in the Portland and Tualatin basins, but likely along the entire Cascadia forearc from the southern Willamette Valley to the Puget Sound.

Average sediment accumulation rates across 0–15 Ma strata in the Portland and Tualatin basins are ~30 m/Ma, consistent with late Miocene to late Pliocene rates for the Tualatin basin estimated by Wilson (1997). This is an order of magnitude less than sediment accumulation rates estimated for Paleogene to early Miocene strata and is consistent with the forearc transition from a marine basin in an environment of extension (more accommodation space) to a continental forearc basin currently undergoing shortening (less accommodation space). Because the top of the CRBG in the Portland and Tualatin basins is ~370 m below sea level (Fig. 5), some tectonic subsidence in the forearc lowland has occurred since mid-Miocene time. It may be the result of ongoing flexure of the upper plate from the Cascadia magmatic arc load (e.g., Waltham et al., 2008), or possibly subduction erosion driven by negative volume changes in the subducting slab (Engebretson and Kirby, 1995; Rogers, 2002). The M 7.1 Anchorage earthquake of 2018 was an intermediate-depth (55–75 km) normal faulting event in the downgoing slab that caused subsidence of up to 3 cm in the overlying lowlands surrounding Cook Inlet (He et al., 2020).

We interpret the eastward shift of the Tualatin basin depocenter from the late Eocene to Miocene time as the result of Coast Range uplift. This is consistent with the introduction of marginal marine and continental sedimentary rock facies in the Astoria basin during early Miocene time (Niem et al., 1992a). Coast Range uplift also resulted in the deformation of Eocene to Oligocene marine strata forming a broad arch prior to CRBG emplacement (Wells et al., 1984; Parker, 1990; Werner, 1990). We suggest Neogene uplift of the Coast Range is the result of subducting progressively younger, more buoyant oceanic lithosphere, following the argument of Rogers (2002). Plate motion models (e.g., Engebretson et al., 1985; Severinghaus and Atwater, 1990; Matthews et al., 2016) show 30-m.y.-old lithosphere entering the Cascadia subduction zone ~30–35 m.y. ago, while today, the age of the incoming plate is 10 Ma (Wilson, 2002). Age-depth relations for oceanic lithosphere (e.g., Niedzielski et al., 2016), show the average depth below sea level for 30-m.y.-old lithosphere is ~4.1 km, while the average depth for 10-m.y.-old ocean lithosphere is ~3 km depth. The decreasing age of the incoming plate could potentially lift the Oregon forearc as much as 1 km. There are other processes that may have contributed to uplift of the Coast Range, including underplating of accreted sediments (Brandon et al., 1998; Calvert et al., 2011), extra buoyant asthenosphere (Bodmer et al., 2019), or permanent deformation driven by megathrust shear stresses (Dielforder et al., 2020), but the timing and amount of potential uplift from age-related slab buoyancy suggest it may play a long-term role in Coast Range uplift.

Increased coupling with the obliquely subducting, more buoyant plate may also have contributed to the shift from transtensional to transpressive deformation in mid- to late Miocene time and set the stage for continued clockwise rotation and northward migration of the forearc, deformation observed today in the Portland and Tualatin basins (Wilson, 1997; Wells et al., 1998; McCaffrey et al., 2007, 2013; Evarts et al., 2009; McPhee et al., 2014). The present-day, transpressive deformation of the basins modifies their geometry and provides potential seismic sources around the basin margins. An improved understanding of the geometry and fill of the Portland and Tualatin basins presented here will be useful to continuing ground motion studies of the lowland (Frankel et al., 2018; Wirth et al., 2019).

Cascadia is one of a few subduction zones in which a coast range separates an inland sea from the forearc basins offshore. We show that although the Portland basin is separated from the Tualatin basin by the Portland Hills, analysis of an inversion of gravity data suggests that the two were connected as one continuous basin in the Paleogene, prior to CRBG deposition. During this time, it was an ocean basin with unlimited accommodation space. An eastward shift of the forearc basin depocenter over Neogene time is likely caused by uplift of the Coast Range to the west, restricting the basin and reducing accommodation space. Development of the subaerial Coast Range is the result of progressive subduction of increasingly younger and thus, more buoyant plate, consistent with the observation that similar margins seem to occur where young, warm slab is being subducted.

Local thickening of Paleogene sedimentary rocks and CRBG flows over a gravity low coincident with the NW-striking Portland Hills anticline suggest that it was a graben until mid- to late Miocene time. Neogene dextral transpression in the forearc structurally inverted the Sylvan-Oatfield and Portland Hills normal faults as high-angle, dextral-reverse faults, creating the Portland Hills anticline and effectively separating the Portland and Tualatin basins. This episode of structural inversion resulted from a regional change in stress from transtension to transpression. Decreased accommodation space due to Neogene uplift of the Oregon Coast Range and a change in regional stress caused a ten-fold decrease in sediment accumulation rates across the Portland and Tualatin basins, as they went from being part of a much larger ocean basin (marine sediments) to the restricted continental basin seen today (fluvial sediments). This insight improves our understanding of Coast Range development, forearc migration, and Cascade magmatism as it is recorded in the stratigraphy of the Portland and Tualatin basins. Transpressional oblique-slip faulting continues to play a role in deforming the region as the forearc undergoes clockwise rotation and northward migration, creating evident hazard for the Portland metropolitan and surrounding areas.

Educational software licenses were provided by Esri™ (desktop.arcgis.com/en/), Kingdom™ (https://ihsmarkit.com/products/kingdom-seismic-geological-interpretation-software.html), and Petrosys Pty Ltd (petrosys.com.au). This research was supported by the Department of Energy's Office of Energy Efficiency and Renewable Energy (EERE) office through the Portland Deep Direct-Use Thermal Energy Storage (DDU-TES) Feasibility Study, grant #DE-EE0008104. Publication of this article in an open access journal was funded by the Portland State University Library's Open Access Fund.