The Clearwater River basin, Olympic Peninsula

Holocene strath terraces, climate change, and active tectonics: The Clearwater River basin, Olympic Peninsula, Washington State

Karl W. Wegmann* Division of Geology and Earth Resources, Washington State Department of Natural Resources, Olympia, Washington 98504, USA

Frank J. Pazzaglia† Department of Earth and Environmental Sciences, Lehigh University, Bethlehem, Pennsylvania 18015, USA

ABSTRACT The ⬃400 km2 Clearwater River basin, located on the Pacific flank of the actively uplifting Olympic Mountains of western Washington State, contains a well-preserved flight of Holocene fluvial terraces. We have collected a large data set of numeric ages from these terraces that is used to elucidate the geomorphic, fluvial, active tectonic, and climatic processes that operate at Holocene spatial and temporal scales. Detailed field mapping reveals three prominent Holocene straths and their overlying terrace deposits. Terrace ages fall into three broad ranges: ca. 9000–11 000 yr B.P. (Qt4), 4000–8000 yr B.P. (Qt5), and 0–3000 yr B.P. (Qt6). Terrace deposit stratigraphy, sedimentology, and age distributions allow us to consider two alternative models for their genesis. The favored model states that the terrace ages are coincident with lateral incision of the Clearwater channel, emplacement of the terrace alluvium, and the carving of the straths. Vertical incision of the Clearwater channel was primarily relegated to the brief (⬃1000 yr) intervals when we have no record of terraces. Alternatively, the straths were carved as the channel incised vertically during the brief time periods between dated terrace deposits, and the terrace ages record a subsequent long time of alluviation atop the straths and concomitant termination of vertical incision. In both models, we envision a Clearwater River channel at or near capacity with a temporally variable rate of both lateral and vertical incision. Small deviations from this at-capacity condition are *E-mail: [email protected]. † Corresponding author; e-mail: [email protected].

driven by variations in the liberation and delivery of hillslope sediment to the channel. We consider several causes for variable hillslope sediment flux in this tectonically active setting including Holocene climate change and ground accelerations related to earthquakes. Holocene rates of vertical incision are reconstructed along nearly the entire Clearwater Valley from the wide distribution of dated terraces. Incision rates clearly increase upstream, mimicking a pattern documented for Pleistocene terraces in the same basin; however, the rates are 2–3 times those determined for the Pleistocene terraces. The faster Holocene incision rates may be interpreted in terms of an increase in the rates of rock uplift. However, we favor an alternative explanation in which the Holocene rates represent a channel rapidly reacquiring its stable, graded concavity following protracted periods of time in the Pleistocene when it could not accomplish any vertical incision into tectonically uplifted bedrock because the channel was raised above the bedrock valley bottom by climatically induced alluviation. These results illustrate how, even in tectonically active settings, representative rates of rock uplift inferred from studies of river incision should be integrated over at least one glacial-interglacial cycle. Keywords: geomorphology, Holocene, paleoclimate, radiocarbon dating, terraces. INTRODUCTION River terraces are preserved in a wide range of active tectonic settings and watershed sizes. Mapped, correlated, and dated terraces form the basis of our understanding of how river

systems evolve. In the active tectonic setting, the history of fluvial processes is both a direct consequence of and has a dynamic interaction with rock deformation (Bull and Knuepfer, 1987; Wells et al., 1988; Merritts et al., 1994; Personius 1995; Gardner et al., 1992; Burbank et al., 1996; Maddy, 1997; Pazzaglia et al., 1998; Hancock et al., 1999). A key debate that has overshadowed more than three decades of intense characterization of fluvial systems in tectonically active settings is how tectonic processes can be isolated from the myriad of geomorphic processes that also influence the genesis and subsequent preservation of fluvial stratigraphy (Schumm, 1969; Schumm et al., 1987; Bull, 1990, 1991; Sugai, 1993; Tucker and Slingerland, 1997; Hancock and Anderson, 2001). For example, rapid rates of rock uplift and changes in base level that accompany both co- and interseismic deformation introduce the potential for rivers to incise and abandon their valley bottoms as they seek a new base level of erosion (Gilbert, 1877; Bull, 1991). But the precise manner in which the rivers accomplish that incision and what terrace record will be left behind appears to be dominated by how the watershed responds to changes in climate—most notably, glacial-interglacial–scale climate change (Bull, 1991). Because climate change tends to occur more frequently than secular changes in the rates of rock uplift, actively incising rivers have the potential to leave a relatively high-resolution record of their incision. The rate of fluvial incision is commonly interpreted in terms of the rate of rock uplift with the built-in assumption that terrace formation is a short-term disequilibrium phenomenon oscillating about a longterm equilibrium profile (Mackin, 1948) more or less represented by the physical characteristics of the modern profile (Knox, 1975; Bur-

GSA Bulletin; June 2002; v. 114; no. 6; p. 731–744; 8 figures; 1 table.

For permission to copy, contact [email protected] 䉷 2002 Geological Society of America

731

WEGMANN and PAZZAGLIA

bank et al., 1996; Pazzaglia and Brandon, 2001). With these debates and assumptions in mind, we set out to map, describe, and date a suite of Holocene fluvial terraces preserved

astride an actively uplifting forearc of the Cascadia subduction margin. These terraces preserve a high-resolution record of river incision over relatively short (103 yr) time scales that can be interpreted in terms of subduction-zone

Figure 1. (A) Location map of the Olympic Mountains showing the major tectonic features of the Cascadia margin. HRF—Hurricane Ridge fault. Note that the Olympic Peninsula is the only part of the Coast Ranges to expose the accretionary wedge. The rectangle around the Olympic Peninsula is enlarged in B. (B) Location of the Clearwater drainage (white) with respect to other major rivers on the Olympic Peninsula. Spot elevation is Mount Olympus. (C) Simplified reference map of the Clearwater basin showing place names indicated in the text, spot elevations (m), and numbered locations of radiocarbon samples referenced in Table 1. Location of radiocarbon dates and features mentioned in the text with respect to their distance from the Pacific Coast are indexed to the cross-hatched line with the 0 km and 45 km labels at its ends paralleling the Clearwater channel. The cross hatches are in one kilometer intervals; 5 km cross hatches are darker.

732

and upper-plate earthquakes (Atwater, 1987, 1992; Atwater and Hemphill-Haley, 1997), contemporary surface deformation (Savage et al., 1991; Dragert et al., 1994), changes in Holocene climate (Bond et al., 1997; Gavin et al., 2001), and the liberation of hillslope regolith (Reneau et al., 1989; Reneau and Dietrich, 1991). The motivation for the study comes from ample opportunity for 14C numeric age control, in the form of organic material preserved in the terrace deposits, and the opportunity to compare Holocene-scale to Pleistocene-scale rates of fluvial incision (Pazzaglia and Brandon, 2001) as well as Neogene-scale rates of exhumation (Brandon et al., 1998). We seek to demonstrate the conditions under which Holocene terraces form and establish limits to some of our assumptions as to how these terraces can be used to interpret the nature and rates of active tectonic processes. SETTING The Olympic Mountains of western Washington are the northernmost extension of the Cascadia margin Coast Ranges (Fig. 1A). The Olympics represent a prominent forearc high produced by northeast-directed oblique subduction of the Juan de Fuca plate beneath North America at a present plate-convergence rate of ⬃36 mm/yr (DeMets and Dixon, 1999). With an area of ⬃7700 km2, the Olympic Mountains are the highest part of the Coast Ranges with the high point, Mount Olympus, reaching 2427 m. The range has a general domal shape and a distinct radial drainage pattern with stream headwaters and highest elevations centered on Mount Olympus. Most of the moisture falls on the Pacific-facing, western slopes. Major streams here have decidedly concave-up profiles and knickpoints confined primarily to upstream reaches where channel slopes are very steep. The valleys have developed along Neogene northeast-trending structures interpreted as strike-slip faults, such as the Clearwater River shear zone (Stewart, 1970; Gerstel and Lingley, 2000). The streams have mixed bedrock and alluvial channels, mostly in glacially modified valleys. It is common to find ⬃3 m of alluvium in the channels atop the bedrock thalweg. The Olympic Mountains are cored by the Olympic subduction complex, an Oligocene– Quaternary accretionary wedge up to 30 km thick composed of stratigraphically discontinuous Neogene–Quaternary clastic, marine sedimentary and shallow-intrusive and extrusive, intermediate to mafic igneous rocks. These rocks, exposed in the core of the range, were

Geological Society of America Bulletin, June 2002

CLEARWATER RIVER BASIN, OLYMPIC PENINSULA, WASHINGTON STATE

formerly overlain by a structural lid of Eocene basaltic basement (Crescent Formation) and thin, Neogene forearc-basin sediments of the Coast Range terrane (Fig. 1A). The contact between the Coast Range terrane and core rocks of the accretionary wedge is a thrust fault (Hurricane Ridge fault) that has been folded into a large east-plunging antiform as the forearc (Crescent Formation) was peeled back and eroded during uplift of the wedge in the past 17 m.y. (Tabor and Cady, 1978a, 1978b; Palmer and Lingley, 1989; Brandon and Vance, 1992; Brandon et al., 1998). Zircon and fluorapatite fission-track data from the range show steady exhumation at the rate of ⬃0.75 km/m.y. since ca. 14 Ma (Brandon and Vance, 1992; Brandon et al., 1998) for the central massif and much slower rates for coastal regions. Rates of river incision determined from Pleistocene fluvial terraces corroborate the unroofing rates in that they increase from ⬍0.1 mm/yr at the coast to ⬃0.6 mm/yr in the upper parts of the Clearwater basin, and to numerically extrapolated rates of ⬃1 mm/yr across the central massif (Pazzaglia and Brandon, 2001). Estimates of averaged long-term rates of tectonic exhumation (⬃0.26 km/m.y.) for the entire Olympic Peninsula are consistent with estimates for modern averaged rates of erosion (0.18–0.32 km/m.y.) from the Elwha and Hoh River basins (Nelson, 1986). The similarities in rates of long-term exhumation, fluvial incision, and modern erosion suggests that the Olympic Mountains have been close to an erosional steady state since ca. 14 Ma (Brandon et al., 1998; Pazzaglia and Brandon, 2001). Our studies are focused on the 52-km-long Clearwater River basin (Fig. 1, B and C) that drains 390 km2 in the western Olympic Mountains. The Clearwater River, a mixed bedrock alluvial stream, flows in a west-to-southwest direction, an orientation roughly parallel to the direction of plate convergence and normal to the modern Pacific coastline. Basin elevation ranges from ⬃12 m at the confluence of the Clearwater and Queets Rivers to 1172 m along its eastern divide. The Clearwater River is located between the larger Hoh River to the north and the Queets River to the south, both of which were extensively glaciated (Thackray, 1996). Outwash from the Hoh Valley periodically spilled into the middle and lower Clearwater basin during Pleistocene glacial stages and influenced Pleistocene terrace genesis (Thackray and Pazzaglia, 1994; Pazzaglia and Brandon, 2001; Fig. 1B). The Clearwater basin was likely not glaciated at all during the late Wisconsinan; however, there is evidence for a small tongue of ice extending ⬃8 km

from the basin headwaters in early or pre-Wisconsinan time. There are radiocarbon-dead (⬎45 ka) lacustrine deposits overlying a lodgment till at this location (Wegmann, 1999). The lack of an extensive tongue of ice scouring the entire Clearwater Valley has permitted the basin to preserve a fine suite of river terraces that have been mapped, described, and used to infer Pleistocene rates of fluvial incision (Wegmann, 1999; Pazzaglia and Brandon, 2001). Hillslope processes throughout the Olympic Mountains are dominated by bedrock-involved landslides, debris flows, and colluvial creep down the ubiquitous steep hillslopes (Reid and Dunne, 1984; Serdar, 1999; Gerstel, 1999). Debris flows in low-order drainages are very active, but may have been even more prevalent near the Pleistocene-Holocene boundary as suggested by the excavation of colluvial hollows (Reneau et al., 1989). Climatically driven introduction of sediment to the fluvial system from hillslope colluvial hollows has been used to explain the creation of a widespread and well-preserved strath terrace in the Coast Range of Oregon at ca. 8–12 ka (Personius et al., 1993). The lower reaches of the Clearwater basin have been logged for the past seven decades, but in the past 40 yr, rapid clear-cut logging throughout the upper two-thirds of the basin has removed ⬃75% of the original forest cover (J. Cederholm, 2001, personal commun.). One consequence of stream-adjacent logging has been the removal of the natural source of large woody debris (LWD). Abundant LWD in stream channels has been demonstrated to be an effective ephemeral trap for transported alluvium in unlogged, glaciated watersheds (Abbe and Montgomery, 1996a); however, the relatively steeper fluvial gradients of the glacier-free Clearwater Valley have limited the retention of LWD in the main channel with respect to the Queets and Hoh Valleys (J. Cederholm, 2001, personal commun.) Logging activity may also be associated with the widespread accumulation of silt on the flood plains and sand and gravel on elongate midchannelbar islands. The flood plains and bar islands nearly everywhere have a community of similar-aged red alders (Alnus rubra) that core samples reveal to be ⬃30 yr old (Wegmann, 1999). The nearly uniform age of the floodplain alder stands is generally coincident with the onset of intense clear-cutting. FLUVIAL TERRACES River terraces are the geomorphic and sedimentologic expression of form and process


Figure 2. (A) Schematic diagram illustrating the relationships between terraces (t1, t2), straths, the flood plain, and the valley bottom. The correspondence between the width of the flood plain and the width of the valley-bottom strath is corroborated by exposure of the valley-bottom strath in tributary channels, shown entering from the right part of the diagram. (B) Photograph of the Qt5 strath terrace near river kilometer 20 on the Clearwater River. Both the terrace strath and the valley-bottom strath, being carved by the modern channel (beneath water surface), are visible in the photograph.

adjustments in a fluvial system (Schumm et al., 1987). Terraces are unconsolidated, allostratigraphic units with a basal unconformity called a ‘‘strath’’ typically cut across bedrock and a constructional bench-like top called a ‘‘tread’’ (Fig. 2). The terrace deposit between the strath and tread varies in texture, stratification, and thickness. When it is thin (⬍3 m), it represents essentially all of the sediment in transport in a bedrock or mixed bedrock and alluvial channel during bankfull or larger discharges (Wolman and Miller, 1960). In contrast, thicker terrace deposits represent alluvial valley fills that lift the channel off the strath for some period of time (Bull, 1991). For the sake of simplicity, the following discussion will be limited to consideration of incision of mixed bedrock and alluvial channels in which straths typically have low relief and the terrace deposits are relatively thin (Fig. 2B). Such terraces are commonly called ‘‘strath terraces’’ (Bull, 1991). The process of carving a strath, although not well understood, can be readily observed for many bedrock and mixed bedrock and al-

733


luvial channels (Gilbert, 1877; Mackin, 1937). In these settings, the width of the valley bottom roughly corresponds to the limits of lateral channel corrasion (Fig. 2A). Alluvium underlying the flood plain typically is not thicker than ⬃3 m, and during low-flow conditions bedrock exposed in the channel bottom can be observed to project laterally as the base of the flood plain. Tributary streams incised through the flood plain provide additional windows of observation confirming that the bedrock exposed in the active channel continues laterally as the unconformity at the base of the valley bottom (Fig. 2A). In this respect, the strath of a terrace is the paleo–valley-bottom base, and the terrace tread is the constructional top of the (commonly modified) paleo–flood plain (Fig. 2A). Straths represent a time when the channel and the alluvium undergoing transport are in direct communication with the underlying bedrock. Thus, first-order channel characteristics such as gradient and concavity are probably most similar between the modern channel and paleochannels during periods of strath cutting. These are the conditions under which the assumption of using a modern river’s longitudinal profile as a long-term proxy for the equilibrium profile has the greatest chance of being valid (Burbank et al., 1996; Pazzaglia and Brandon, 2001). The assumptions of similar channel gradient and concavity cannot be so readily assumed when considering the terrace treads and the flood plain, especially when the terrace deposits are thick and the flood plain represents the constructional top of a thick valley-bottom fill. Strath terraces are common and typically unpaired in tectonically active areas (Bull, 1991), but they may extend for many kilometers along the length of a valley. It has become clear from diverse tectonically active settings that even though terraces may lie at variable distances above the modern valley bottom, their ages tend to cluster around dates temporally coincident with known climatic changes (Bull and Knuepfer, 1987; Merritts et al., 1994; Burbank et al., 1996; Pazzaglia et al., 1998; Fuller et al., 1998; Hancock et al., 1999). Geomorphologists have long appealed to the impacts of climate change on the fluvial system as the driving mechanism behind those modifications of channel form and process that alternately widen and narrow the valley bottom and ultimately lead to the genesis and preservation of strath terraces (Schumm, 1969; Bull and Knuepfer, 1987; Bull, 1991; Pazzaglia and Gardner, 1993; Meyer et al., 1995; Pazzaglia and Brandon, 2001).

734

METHODS We have carefully mapped strath terraces at a 1:12 000 scale in the Clearwater basin (Wegmann, 1999). The mapping was carried out on foot and by repeated canoe traverses. Minimum strath ages are provided by radiocarbon dates from organic material preserved in the overlying terrace alluvium (Figs. 1C and 2B). We started with the reasonable assumptions that the alluvium overlying the strath is genetically related to the strath carving (Gilbert, 1877; Mackin, 1937; Hancock and Anderson, 2001), that the straths were not subsequently reoccupied by the active channel (Merritts et al., 1994), and that the organic material is rarely recycled from older deposits nor is it representative of the subordinate old-wood component in the basin. We have carefully interpreted our findings in the context of this last assumption because samples of woody material collected from bars in the active channel of the adjacent Queets River (Fig. 1B) have returned radiocarbon ages ranging from modern to ⬎1000 yr (Abbe and Montgomery, 1996b). Multiple ages from a single terrace deposit at a point and along the lateral extent of a mapped terrace provide an estimate of the degree of uncertainty in strath age. Material used for 14C dating consists of detrital charcoal and wood as well as fir cones and needles and deciduous leaf matter. Care was taken to insure that the dated organic samples represented in situ, fluvially deposited materials associated with terrace aggradation, not the subsequent emplacement of organic matter by penetrating tree roots or weathering profiles. We report the 14C ages in both radiocarbon and calibrated calendar years before present (14C yr B.P. and A.D./B.C., respectively). The disparity between radiocarbon and calibrated ages is significant throughout the Holocene, and a single radiocarbon age can return more than one unique calibrated age. Multiple calibrated ages and variable standard errors can be reconciled by a summation of the normal distributions of individual calibrated 14C ages and their associated 1␴ error (Ramsay, 2000). The result is a probability density plot of ages also used in similar studies of abundant radiocarbon dates (Meyer et al., 1995) or related normally distributed age data (Bull and Brandon, 1998). Coring representative red alders (Alnus rubra) at numerous localities provided an age for Qt7 (Wegmann, 1999). We correlate terraces along the Clearwater valley by the map relationships, through deposit sedimentology and stratigraphy, and by considering general weathering characteristics


such as relative soil development that we assume to be primarily influenced by the terrace’s relative age. It is important here to not introduce circularity in our methodology. Terrace correlation on age criteria alone assumes that a given terrace was more or less instantaneously created along its entire extent. Alternatively, a correlation based solely on the map relationships could potentially force terraces of very different age and genesis to be considered the same simply because it is extremely difficult to identify polygenetic straths or major unconformities in terrace deposits of similar sedimentology and texture. Given these concerns, we start with the mapped field relationships and allow the relative and numeric ages to help refine the correlation. The exercise is analogous to using biostratigraphy to transform lithostratigraphic units into chronostratigraphic units where time lines can be extended throughout the rock column. In our case, we wish to use the terraces as time lines along the river. Once properly correlated, we can measure the separation between a given strath level and the modern channel virtually anywhere in the drainage to obtain a point measurement of the rate of fluvial incision. Vertical incision rates are determined from the separation between the dated Holocene straths and the valley-bottom strath, estimated by the low-water-level elevation during summer base-flow discharge. This is an appropriate datum from which to measure strath-height separation because it helps mark the reachlength average elevation of the valley-bottom strath and avoids measurements affected by local relief of the channel bed (Pazzaglia and Brandon, 2001). The separation between the terrace strath and valley-bottom strath was measured directly in the field with a tape measure, accurate to ⫾0.1 m for terrace straths ⬍10 m above the channel or with an altimeter, accurate to ⫾1 m for terrace straths ⬎10 m above the channel. We do not choose a single age for the straths based on numeric age data. Rather, we choose to calculate incision rates at points for which we have an uncalibrated radiocarbon numeric age. The age of the alluvium at that point is assigned to the age of the strath. We use the uncalibrated ages to avoid the complexity of multiple calibrated ages and their associated errors in the rate calculation. This approach allows us to observe regional trends in vertical incision without forcing circular reasoning upon when the strath was cut, when the alluvium was deposited, and when the strath was finally abandoned. Incision-rate errors are derived from the combined effects of 14C age standard errors, fluctuations in the elevation of the sum-


Figure 3. Longitudinal valley profiles of the Clearwater River and Holocene terraces. The profiles were produced by orthogonally projecting elevation data to a vertical plane oriented down the middle of the Clearwater Valley. Plot does not show terrace Qt7, which essentially lies on the valley’s long profile, or terrace Qt5/6, which cannot be clearly represented at this scale. mer base-flow water level, and uncertainty in the measurement of the strath height, which includes both errors associated with the physical measuring of the strath height as well as variations in strath height along an outcrop face. The relative percentage of an incision rate error is greatest for the younger terraces and decreases with increasing terrace age and separation above the channel. RESULTS Channel and Longitudinal-Profile Characteristics The lower 28 km of the Clearwater Valley is characterized by a mixed bedrock and alluvial channel that meanders across a 1–2-kmwide, low-relief valley. The alluvium, observed to be mobile during bankfull or higher discharges, has not been observed to be thicker than 2–3 m. Short (⬍100 m) but numerous stretches of bedrock are exposed in the channel bottom. Modern-channel gravel bars are transitory features that have changed significantly in recent time as shown by a comparison of gravel-bar location and size between

1990 aerial photographs and 1997–1998 field observations. The average valley gradient along this lower reach is 0.003 (Fig. 3). The upper reaches (above valley kilometer 28; Fig. 1C) of the Clearwater valley have a dominantly step-pool bedrock channel with short alluvial reaches concentrated at the junction of larger tributaries. Here, the average valley gradient steepens to 0.052 (Fig. 3). The Clearwater channel bed is a flat-bottomed, low-relief unconformity cut across dipping bedrock. In the lower 10–15 km of the river valley (Fig. 1C), the channel bed extends laterally beneath the flood plain to the edges of the valley bottom where it is exposed by tributary streams that have incised through the flood plain (Fig. 2A). Farther upstream, the channel bed does not extend laterally to underlie what is easily recognizable on a topographic map as the valley bottom. Rather, it is clearly inset into a terrace strath (Fig. 2A) that in turn extends to the edges of what appears to be the valley bottom on a 7.5-minute quadrangle map (see Fig. 8, cross section D–D⬘ of Pazzaglia and Brandon, 2001, for an illustration of this relationship). Separation between the channel bed and terrace straths increases


upstream. Downstream they merge into the same unconformity (Fig. 3). Thus, we treat the modern channel bottom as the valley-bottom strath, exerting caution as to how far laterally we can project it under the flood plain, and use it as the datum to which we measure strath separation. The macrorelief on the valley-bottom strath is typically ⬍2 m, although there are some deep pools (⬃4 m), especially in meander-cutbank locations. We have endeavored to avoid these local deeps and focus more on reconstructing the average elevation of the valley-bottom strath for consecutive 500 m reaches of stream. Fluvial terraces are best preserved in the reaches of the river characterized by a broad valley bottom and nonresistant mudstone and siltstone. Narrow bedrock reaches, and locally gorges, tend to correspond to outcrops of resistant medium- to thick-bedded, medium- to coarse-grained litharenite as well as pebble to cobble conglomerate. In contrast, broader valley bottoms, and alluvial reaches, tend to be underlain by nonresistant, fractured, and thinly bedded mudstone and siltstone, as well as pervasively sheared rocks associated with the east-northeast–trending Clearwater River

735


shear zone (Stewart, 1970). Distinct knickpoints in the upper part of the drainage confirm that all are coincident with resistant rock types. A broad convexity in the valley longitudinal profile around valley kilometer 25 (Fig. 1C) appears to be a persistent feature interpreted as being due to a fault or rising anticline (Fig. 3; Pazzaglia and Brandon, 2001). We have observed features on the valleybottom strath that lead us to consider both abrasion of the channel bed by coarse-bedload tools and plucking of weakened channel bedrock along joint and shear planes as likely mechanisms by which the channel incises vertically and laterally. The process of potholing, a common mechanism for channel-bed lowering on many bedrock streams (Whipple et al., 2000), does not appear to play a significant role in the Clearwater Channel, as potholes have been found only in the narrowest bedrock gorges where their formation and preservation are restricted to resistant pebble conglomerates. Terrace Stratigraphy and Map Distribution There are three distinct types of mappable terraces in the Clearwater basin: Fill, fill-cut, and strath (Bull, 1991; Fig. 4). Terraces Qt1, Qt2, and Qt3 (Pazzaglia and Brandon, 2001) are middle and late Pleistocene fill and fill-cut terraces found in mostly the middle to lower reaches of the basin. The thick alluvial fills of these Pleistocene terraces are consistent with major periods of valley aggradation during basin-wide, glacial cycle–driven changes in hydrology and sediment yield (Pazzaglia and Brandon, 2001). Inset below the Pleistocene terraces are several well-preserved Holocene strath terraces (Qt4 through Qt7). These terraces are composed of a basal, 1–3-m-thick axial-channel coarse-gravel facies that is overlain by a 1–3m-thick fine-sand and silt overbank facies (Fig. 2B). The channel gravel facies locally preserves sedimentary structures consistent with lateral accretion processes, such as might be expected for point and transverse bars. These bars and sediment of similar texture are present in the modern channel, so by analogy, we take the coarse-grained facies of the terrace deposit to represent the bedload being transported when the terrace strath was cut. In contrast, the fine-grained facies represents vertical accretion processes atop the flood plain that occur presumably during floods. The Holocene terraces are both paired and unpaired and widely distributed, although there are some channel reaches devoid of ter-

736

Figure 4. Composite cross section representing the terrace stratigraphy and general age ranges in the Clearwater basin. For simplicity we do not show the treads of terrace Qt5/ 6. We adopt the convention of naming terraces in order of increasing age, where ‘‘1’’ is the oldest (highest) terrace in the landscape, and assigning numbers to straths only. In this manner, terrace Qt2 may have more than one tread, which we designate with lower case alpha-numeric subscripts (e.g., Qt2a, Qt2b). The cross section is representative of the terrace deposits on Figure 5 as if they were all projected to the cross section X–Xⴕ. Note that there is no Qt1 deposit in Figure 5; this deposit is shown for relative elevation purposes only. races. In any ⬃500 m reach where the terraces are preserved, we can distinguish three major strath terraces—Qt4, Qt5, and Qt6—each vertically separated by at least 2 m of bedrock (Figs. 4, 5). Especially in the upper middle reaches of the river valley, we locally observe a strath between Qt5 and Qt6 designated Qt5/ 6. The youngest terrace, Qt7, has a more complicated morphology and distribution. In the lower reaches of the river (below valley kilometer 15; Fig. 1C) it is a fill-cut terrace cut into Qt6. Here both Qt6 and Qt7 share the same strath that is also coincident with the channel bed and valley-bottom strath. The treads of both Qt6 and Qt7 have been inundated by floodwaters during large historic floods. In the medial reaches of the valley (between valley kilometers 15 and 30; Figs. 1C and 5), Qt7 occupies its own strath that is coincident with the channel bed, both of which are clearly inset into the Qt6 strath. Here floodwaters inundate the Qt7 tread, but the Qt6 tread shows little evidence of floodwater inundation over the past several centuries. Farther upstream where the valley bottom narrows significantly, we do not recognize a Qt7. All Holocene terraces above Kunamaskt


Creek (valley kilometer 35; location shown in Fig. 1C) are unpaired strath terraces cut either into bedrock or into an older (radiocarbon dead) alluvial-glacial valley fill of middle to late(?) Pleistocene age. Locally, terrace deposits are buried by a tributary’s alluvial-fan deposits, which become more prevalent upstream of Kunamaskt Creek. Small unpaired late Holocene terraces with little lateral continuity may be associated with LWD jams in the upper 10 km of the channel (Fig. 1C). The correlation between these terraces and Qt6 and Qt7 is unclear. The Holocene terraces are best preserved between valley kilometer 15 and 30 (Figs. 1C and 5), and we draw heavily upon this reach to understand the lateral extent and morphology of a given terrace. Here the valley bottom has narrowed to ⬃500 m, but the channel maintains an active meandering pattern. Qt5 tends to equal or exceed the width of the modern valley bottom evident on a 7.5-minute topographic map, whereas Qt6 is wholly confined to (and helps define) the modern valley bottom (Fig. 5). Mapping the terraces through this reach clearly demonstrates that the straths have a broad lateral extent. In some cases we

Figure 5. Map of river terraces in the medial part of the Clearwater Valley (modified from Wegmann, 1999). Letters following terrace names designations: a, b, c, d— treads that share a common strath; ipc—inset paleochannel. X–Xⴕ is the location of the cross section shown in Figure 4.



737


are able to physically trace a strath and its overlying terrace deposit for hundreds of meters of nearly continuous exposure along the river bank. Uniform relative separation between the terrace strath and valley-bottom strath allows us to confidently project across reaches lacking exposure. Holocene terrace deposits atop the straths are of nearly uniform thickness. We could not discern any appreciable thickening or thinning of a given terrace. Rather, changes in thickness of the terrace deposits appear to be more closely related to the local relief on the terrace strath, the terrace location with respect to the valley-bottom edges, and the proximity to significant tributary streams. Terrace deposits are thinnest where exposed near the edge of the valley bottom and thickest immediately downstream of a major tributary. In summary, the geomorphic and sedimentologic characteristics observed in the modern channel—a mostly continuous mantle of uniform-thickness coarse-gravel bars and overbank fine-silt deposits atop a broad, laterally continuous, lowrelief bedrock strath—is precisely what appears to be represented in the Holocene terrace deposits. We therefore assume that the modern channel and its deposits represent valley-bottom conditions throughout the Holocene. Terrace Age and Correlation Terrace ages are provided by 38 radiocarbon dates (35 AMS [accelerator mass spectrometry] and 3 standard beta-decay ages; Table 1). The radiocarbon dates reveal three broad age groupings on the probability density plot (Fig. 6). These three age groupings roughly correspond to three of the four mapped strath terraces Qt4, Qt5, and Qt6. The three age groupings are identified solely upon visual inspection of significant temporal breaks of ⬃600 and 1200 yr where we have no record of terrace alluvium. Figure 6 indicates that the alluvium overlying the Qt4 strath was deposited at ca. 11 000–9000 yr B.P. A paleochannel inset into the Qt4 tread near Crooks Creek (Figs. 1C and 5, deposit Qt4ipc) yields a limiting minimum age for that modified tread of 6040 ⫾ 55 14C yr B.P. (6947–6760 calibrated years before present, 4997–4810 B.C.; Table 1). Alluvium overlying the Qt5 and Qt6 straths was deposited between ca. 8000–4000 yr B.P. and between ca. 3000 yr B.P. and the present, respectively. Confirmation of the close correspondence to map and allostratigraphic determination of the terraces (the foundation of the Qt4–Qt7 terrace stratigraphy) and the three numeric age groupings is borne out in a comparison be-

738

TABLE 1. RADIOCARBON SAMPLES, DATED MATERIAL, 14C, AND CALIBRATED AGES Station Terrace* number

Strath height†

Lab Material** Meth.†† Age ⫾ 1␴§§ number§

1 2 3 3 3 4 4 4 5 6 7 8 8 9

Qt6 Qt6 Qt6 Qt6 Qt6 Qt6 Qt6 Qt6 Qt6 Qt6 Qt6 Qt6 Qt6 Qt6

N.E. N.E. 0.7 ⫾ 0.1 0.7 ⫾ 0.1 0.7 ⫾ 0.1 0.5 ⫾ 0.1 0.5 ⫾ 0.1 0.5 ⫾ 0.1 0.5 ⫾ 0.1 N.E. 0.75 ⫾ 0.1 N.E. N.E. N.E.

AA26659 AA26660 AA26661 B120935 B85435# AA26662 AA26664 AA26665 AA26666 AA26667 AA26668 AA26669 AA26671 AA26672

dw dw dw dc dw dw dw rts-bls dc dc dpd dw dw dw

A A A A B A A A A A A A A A

260 883 431 1670 620 1935 929 373 1328 1120 1130 273 710 105

⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾

42 51 40 30 80 53 41 40 44 44 40 42 43 40

10 11 12 13 14 15 16 17 18 19

Qt6 Qt6 Qt6 Qt5/6 Qt5/6 Qt5/6 Qt5 Qt5 Qt5 Qt5

2.5 ⫾ 0.5 2.7 ⫾ 0.5 N.E. 4 ⫾ 0.5 5 ⫾ 0.5 4.6 ⫾ 0.5 N.E. 3 ⫾ 0.5 6 ⫾ 0.5 5.5 ⫾ 0.5

AA26675 AA26676 B75764# AA26682 AA26684 AA26686 AA26673 AA26677 AA26678 AA26691

dw dc dw dc dc dc dc dc dc dw

A A B A A A A A A A

860 790 2020 2380 2765 2045 4440 3915 4345 4180

⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾

40 40 60 50 50 45 50 45 45 50

19 20 21

Qt5 Qt5 Qt5

8 ⫾ 0.5 AA26695 2 ⫾ 0.5 AA26680 6.75 ⫾ 0.5 AA26683

dc dc dc

A A A

4460 ⫾ 60 4400 ⫾ 60 6260 ⫾ 60

22 23 24

Qt5 Qt5 Qt5

7.5 ⫾ 0.5 AA26694 4 ⫾ 0.5 AA26685 3 ⫾ 0.5 AA26687

dc dc dc

A A A

4925 ⫾ 55 5380 ⫾ 55 6210 ⫾ 60

25 26 27 28 29 30 31 32

Qt5 Qt5 Qt5 Qt4ipc Qt4 Qt4 Qt4 Qt4

dc dc dfcn dc dc dc dc dc

A A B A A A A A

5090 5410 4380 6040 6985 8005 8410 8880

4.15 3 4 4.6 6 5 30 2.4

⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾

0.5 0.5 0.5 0.5 0.5 0.5 5 0.5

AA26688 B120936 B85436# AA26692 AA26690 AA26681 AA26693 AA26689

⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾

75 50 70 55 60 65 100 65

Calibrated age(s) and 1␴ range## 416 (302) 287 909 (787, 775, 774,) 732 514 (506) 473 1570 (1545) 1530*** 660 (635, 605, 560) 535*** 1946 (1878) 1822 923 (907, 859, 830, 811, 795) 788 500 (466) 324 1291 (1274)1188 1062 (1051, 1030, 992) 997 1064 (1053, 1028, 1008) 970 422 (306) 291 673 (663) 652 267 (245, 230, 132, 121, 68, 37, 1) 0††† 790 (758, 751, 742) 729 732 (691) 671 2009 (1954) 1885*** 2452 (2353) 2346 2921 (2852) 2782 2054 (1993) 1933 5260 (5039, 5006, 4993) 4884 4417 (4408) 4260 4967 (4867) 4850 4829 (4815, 4754, 4730, 4718, 4711, 4666, 4659) 4590 5285 (5046) 4970 5048 (4970) 4867 7253 (7227, 7222, 7210, 7165, 7164) 7032 5714 (5649) 5600 6278 (6195) 6003 7234 (7174, 7158, 7113, 7094, 7055, 7053, 7035, 7034) 7008 5921 (5891) 5739 6279 (6204) 6179*** 5034 (4879) 4854*** 6947 (6886, 6818, 6809) 6760 7925 (7790) 7737 9010 (8989, 8824) 8724 9527 (9469, 9446, 9437) 9300 10165 (10144, 10142, 10110, 10080, 10064, 10063, 10028, 10020, 10007,9998, 9920) 9804

Note: Locations of some of these samples are shown in Figure 5. *Qt4ipc is the inset paleochannel into the Qt4 tread at Crooks Creek (Fig. 5). † Height of bedrock strath in meters measured from summer base-flow water level to the base of terrace deposit; N.E.—strath not exposed. § Sample number assigned at dating laboratory: AA—National Science Foundation–University of Arizona (NSFArizona) Accelerator Facility for Isotope Dating; B—Beta Analytic, Inc. # Previously reported in Pazzaglia and Brandon (2001). **Abbreviations: dc—detrital wood; dc—detrital charcoal; rts-bls—rooted tree stump buried by landslide debris; dpd—detrital plant debris; dfcn—detrital fir cones and needles. †† Method—carbon isotopic analytical method; A—accelerator mass spectrometry; B—beta-decay counting. §§ Age ⫾ 1␴ ⫽ conventional radiocarbon age and analytical standard deviation (yr B.P.) as reported by NSFArizona and Beta Analytic, Inc. ## Calibrated calendar ages, given in the following format: minimum of 1␴ calibrated age ranges (calibrated years B.P.) maximum of 1␴ calibrated age ranges, where endpoints of 1␴ calibrated age ranges are calculated from intercepts of (14C age ⫹ 1␴) and (14C age–1␴) with calibration curve (method A of Stuiver et al., 1998). ***Calibrated calendar ages determined by using the method of Stuiver et al. (1993). ††† Represents a ‘‘negative’’ age B.P. (1950).

tween Figure 6 and Table 1. The terrace designation in Table 1 is based upon the allostratigraphic and field-mapping correlation prior to dating. Our map and field determinations of terrace age (Table 1; Fig. 5) do not conflict with the age groupings (Fig. 6). Everywhere, stratigraphically older terraces yield the correspondingly older 14C dates. In this case, the elevation of the terrace strath above the valley-bottom strath is a robust criterion for the


relative age and correlation of the terraces, even though the alluvium mantling a given strath may have a broad age range. The maximum age of any given terrace is likely an overestimation of the true age because of the propensity of old-growth forests to have living trees ⬎500 yr old, as well as the rare possibility that organic material is recycled, or scavenged, during erosion of older terrace deposits. The age-frequency distribu-


tion of a given terrace, such as Qt5, is skewed toward younger ages (Fig. 6). The old tails on these distributions may mean that the strath was cut and alluvium with old wood was deposited immediately preceding strath abandonment. Alternatively, the skewed distributions mean that (1) alluvium overlying the strath truly does have a large age range, and (2) there is more of the younger alluvium preserved at the expense of the older alluvium. Rates and Distribution of Vertical Incision We have extracted a subset of 25 uncalibrated radiocarbon dates to calculate point incision rates along the Clearwater long profile. Rates of vertical incision range from ⬃0.05– 1 mm/yr in the lower third of the basin to ⬃1– 2 mm/yr in the medial third, to ⬃2–3 mm/yr in the upper third (Fig. 7). Although similar in pattern to the incision documented for Pleistocene terraces (Pazzaglia and Brandon, 2001; Table 1), the rate of Holocene incision exceeds that calculated from the Pleistocene terraces by factors of two to three.

Figure 6. Probability density and frequency distribution plots for 38 radiocarbon dates (Table 1) of the Holocene terraces. The plot was produced by calculating calendar ages and their accompanying 1␴ errors and then summing the probabilities using the program OxCal (Ramsay, 2000). Note the brackets beneath the frequency histograms. At the 1␴ confidence level, the groups of ages between ca. 2000–6000 B.C. and A.D. 2000–1000 B.C. statistically coincide with terraces Qt5 and Qt6, respectively. Uncalibrated radiocarbon ages and their respective 1␴ errors are shown by the small data points at the base of the plot for comparison.

DISCUSSION Conceptual Model for Strath-Terrace Genesis The terrace data in the Clearwater basin are particularly well suited for considering various models of strath-terrace genesis in the tectonically active setting. The Clearwater River is incising because the Cascadia forearc high continues to rise. If the rate of river incision perfectly matched that of the uplifting rocks and if the river responded to no other external forces including base-level changes or watershed hydrology, the river might be expected to leave an incomplete, disjointed suite of unpaired strath terraces similar to those envisioned by Bull (1990) or Merritts et al. (1994) and modeled by Hasbargen and Paola (2000). Under these conditions, it is a ‘‘random’’ component of fluvial incision into bedrock that fortuitously leads to the preservation of a strath terrace. Our data are not consistent with this type of strath-terrace record, so we must appeal to other conceptual models for strathterrace genesis that are decidedly more deterministic, such as those modeled by Hancock and Anderson (2001). We quickly dispense with the possibility that the terraces are formed by logjams of LWD (Montgomery et al., 1996; Abbe and Montgomery, 1996a, 1996b). Prior to largescale logging in the upper Clearwater watershed, the main-stem Clearwater channel did

Figure 7. Incision rates (left vertical axis) determined from Holocene terraces (solid black circles) and Pleistocene terraces (gray triangles) plotted with respect to distance along the valley’s long profile (right vertical axis). Rates reflect calendar ages, and error bars on the Holocene data are 1␴ standard error. Note that the incision rates for both data sets increase upstream. not contain numerous logjams (J. Cederholm, 2001, personal commun.), contrary to observations in the adjacent Queets River. We have not observed the sedimentary textures or deposit morphologies associated with terraces or ponded deposits behind LWD dams in the Clearwater River. We know of no location in the lower and middle reaches of the basin where a large concentration of jumbled woody debris representing paleo–log jams are preserved. In summary, we do not dispute the presence of log-jam terraces in the prehistoric Clearwater River. But we do question the likelihood of preserving these features as the terraces we have found and mapped. A more likely possibility for Holocene ter-


race genesis is a downstream fall in base level and the subsequent upstream migration of one or more knickpoints. This is an appealing model because it has been used to explain late Pleistocene and Holocene terraces described for small, steep mountainous drainages of the Oregon and northern California Coast Ranges that empty directly into the sea (Seidl and Dietrich, 1992; Merritts et al., 1994). But for the Clearwater River, we can find no compelling evidence that the mapped terraces project upstream into knickpoints or knickzones (we use the term knickzone to describe kilometer-scale channel reaches where the change in channel gradient is expressed across a broad convexity; see Zaprowski et al., 2001) (Fig. 3). Nor

739


can we demonstrate that a given terrace strath, tread, or terrace alluvium systematically youngs upstream as it should if it is related to a headward-propagating knickpoint. All of the terraces investigated in this study are Holocene in age, so we cannot call upon dramatic falls in sea level to even initiate the upstream march of a knickpoint, unless the knickpoints were created prior to the Holocene during Pleistocene eustatic lows. If that was the case, the long time periods represented by the relatively slowly retreating knickpoints should be immediately apparent in time-transgressive terrace deposits, which we do not observe. Furthermore, the slow steady rise of Holocene sea level has produced a relative base level rise for the Clearwater River that should be accommodated by channel stability or aggradation rather than incision, which we also do not observe (Bull, 1991). It is possible that relative falls in base level are created by a rate of rock uplift at the coast that has outstripped the corresponding rise of sea level in the Holocene. But more rapid fluvial incision upstream rather than downstream, preservation of Pleistocene deposits at sea level (Pazzaglia and Brandon, 2001), and low amounts of unroofed section at the coast (Brandon et al., 1998) together do not favor this possibility. Instantaneous falls in base level attributed to rupture on faults remains a option, and there are some field relationships near valley kilometer 25 supportive of active faulting (Figs. 1C and 5); however, the general continuity of the straths and uniform thickness of terrace deposits make it difficult for us to link genesis to such poorly defined faults (Wegmann, 1999). The apparent lack of obvious base-level control on terrace genesis brings us to consider a river with a variable rate of both lateral and vertical incision modulated by changes in watershed hydrology and sediment yield (Meyer et al., 1995). This general model is supported by the map distribution of terrace deposits (Fig. 5), their stratigraphy and morphology (Fig. 4), and numeric age (Fig. 6, Table 1). The model holds that the Clearwater River alternates between phases dominated by lateral incision, valley-bottom widening, and the carving of straths balanced by phases of vertical incision, valley-bottom narrowing, and limited strath cutting (Fig. 8). We consider the possibility that the strath and alluvial deposit are genetically related, that is, the ages of the terrace alluvium estimate the times of enhanced lateral corrasion and valley-bottom widening (Gilbert, 1877; Mackin, 1937; Fig. 8, case A). For example, a lateral corrasion phase represented by the

740

Figure 8. Conceptual models for strath cutting. White bars indicate alluviation (sediment aggradation atop the strath), black bars indicate valley-bottom narrowing and vertical incision, and gray bars indicate valley-bottom widening, strath cutting, and abandonment of some of the alluvium performing the incision atop the strath. (A) Terrace alluvium and strath cutting are contemporaneous and span the dates collected from the alluvium atop a strath. Temporal gaps represent times of valley-bottom narrowing and vertical incision when the likelihood of strath preservation is very low. (B) The complete termination of vertical incision as the straths are buried by a (relatively) thick alluvial deposit. Deposition of the terrace alluvium is not related to the strath cutting. Straths are cut rapidly by vertical and lateral processes during the temporal gaps, and incision ceases during the time periods represented by dated alluvium. Qt5 terrace would span ⬃4000 yr. During that time, the channel would sweep laterally, carve the strath, and temporarily store alluvium in the flood plain. Progressive lateral sweeps of the channel would tend to remove previously deposited alluvium, leaving behind new, younger alluvium in its place. Assuming that the alluvial thickness above the valley-bottom strath is equal to the active layer as we infer to be the case during periods of strath cutting, there should be a disproportionately large amount of alluvium in the flood plain that immediately precedes strath abandonment and only a fragmented record of older alluvium marking the onset and duration of strath cutting. The skewed age distributions support this idea: the highest frequency of ages are skewed toward the younger side of the distribution (Figs. 6 and 8). The time periods marked by the complete lack of any dated terrace alluvium indicate one of two possibilities. First, these may simply be times when the river was not carving laterally and leaving an alluvial record atop straths, but rather was narrowing its valley bottom, as the river mainly incises vertically. Alternatively, the river always maintained a component of lateral corrasion and formed terraces, but these terraces were subsequently destroyed by a younger, major valley-bottom widening event. In this case, a younger terrace is inset into the valley at the removal and expense of one or more older terraces. The very fragmentary record of Qt4 may be the result of this process. The pro-


longed period of valley-bottom widening and strath cutting represented by Qt5 may have simply cut out most of the Qt4 record. An alternative way of interpreting Figures 6 and 8 is to consider that there are discrete phases of strath cutting, followed by alluvial aggradation above the strath, and then vertical incision (Fig. 8, case B). This is the same model generally applicable to terraces produced during glacial-interglacial changes (Bull, 1991) and one that appears to match the data for the Pleistocene terraces in the Clearwater basin (Pazzaglia and Brandon, 2001). In this case, the straths would be cut during the temporal breaks in the summed probability distribution (Figs. 6 and 8, case B). In this model, the times of valley-bottom widening are relatively short, and the times of alluvial aggradation atop the straths are long, spanning the time represented by the range of radiocarbon ages. At one locality we may have insights into the time span represented by alluviation atop the strath. Three samples collected in a vertical profile in Qt6 returned ages of 1935 ⫾ 53 14C yr B.P. (A.D. 0–130; AA26662) for the base of the channel-gravel facies, 929 ⫾ 41 14C yr B.P. (A.D. 1030–1160; AA26664) in the middle of the channel-gravel facies, and 373 ⫾ 40 14C yr B.P. (A.D. 1440– 1640; AA26665) (outer ring of a rooted tree stump) for the overbank facies. We have no way to unequivocally interpret these stratigraphically consistent ages in the context of old wood endemic to the setting and stress that


these data simply provide some typically difficult-to-obtain insight. Taken at face value, they would suggest a phase of Qt6 terrace formation over an approximate 2000-yr time span. An important limitation on these two alternatives concepts of when the straths are cut lies in the rates of lateral corrasion the channel might achieve in carving out a strath. By ‘‘lateral corrasion’’ we refer to any process that causes the channel to sweep laterally. These processes include lateral translation of the channel to increase the meander amplitude, sometimes called meander swing, as well as the down-valley sweep of meander loops. In Figure 5, both the width and meander wavelength for the Qt5 strath is ⬃500 m. If the strath and terrace deposit are genetically related and if the Qt5 alluvium spans ⬃4000 yr, then the channel is able to sweep laterally at a rate of 12.5 cm/yr. This is a relatively fast rate for a bedrock-erosion process, not to mention the fact that ⬃2 vertical meters of rock must also be consumed during the lateral incision to inset Qt5 below Qt4. If, on the other hand, the straths are cut in the short time periods between the dated alluvium, the rate of lateral corrasion needed to carve out the Qt5 strath is ⬃83 cm/yr, approaching what may be an unreasonably rapid rate for a bedrock-erosion process. Nevertheless, published rates of meander swing and sweep of semiarid alluvial channels are as high as 400 and 800 cm/yr, respectively (Wells et al., 1982). Where lateral incision involves the carving of bedrock, the rates are almost certainly slower. The simplest way to maximize the time necessary to laterally carve out wide straths is to assume that the strath cutting and overlying alluvium are genetically related and contemporaneous as we have alluded to throughout this paper (Fig. 8, case A). In this case, the age of the youngest alluvium closely dates abandonment of the strath. When mapped correctly, the strath can be used to measure the rate of vertical incision anywhere along the river where the strath is preserved, but may be lacking a numeric age date in the overlying alluvium. We propose that the Clearwater channel, currently in a mode of minor valley widening and alluviation, ‘‘normally’’ rests at or slightly below capacity. A decrease in sediment flux to the channel will drop the channel below capacity and foster valley-bottom narrowing and vertical incision in the manner envisioned by Meyer et al. (1995) and Sklar and Dietrich (1998) (Fig. 8). Valley-bottom narrowing, essentially a reduction in the meander amplitude, allows a river with a decreased amount of abrasive tools to attain the gradient or con-

cavity necessary to maintain incision opposing the rate of rock uplift. Indeed, the alluvium is critical to both lateral and vertical abrasion processes (Foley, 1980; Slingerland et al., 1997; Sklar and Dietrich, 1998). In contrast, terraces record a time of increased sediment flux from hillslopes throughout the basin when the channel is pushed slightly over capacity, the meander amplitude increases, the process of lateral corrasion is maximized, and the valley bottom widens. Possible Causes of the Variation in Sediment Flux to the Channel The Holocene terraces along the Clearwater River suggest a delicate balancing act between the delivery of sediment from the hillslopes and corresponding channel response. The variations in hillslope sediment flux cannot be so great as to completely cause the valley to aggrade, as the terraces older than Qt4 indicate happened in the late Pleistocene. The terracegenesis model calls for an investigation of possible triggers for altering the hillslope sediment flux. Hillslopes that have been destabilized by a change in climate and vegetation are the most likely, often-cited sources of a temporally variable sediment flux (Reneau et al., 1989, 1990; Reneau and Dietrich, 1991; Bull, 1991; Meyer et al., 1995; Pederson et al., 2000, 2001). Several studies in Cascadia appeal to this model to explain the widespread preservation of a ca. 10 ka strath terrace formed during the stripping of hillslope sediment at the late Pleistocene–Holocene climatic boundary (Merritts et al., 1994; Personius, 1995; Massong, 1998). Qt4 may represent the late Pleistocene–Holocene climate transition terrace in the Clearwater basin. The explanation as to why this terrace is spatially insignificant in the Clearwater basin when compared to other similarly sized basins in Cascadia is not clear. It is possible that valley-bottom widening and strath cutting during the middle Holocene, leading to the subsequent genesis of Qt5, removed most of the evidence for the earlier late Pleistocene to Holocene transition terrace. Global (Bond et al., 1997) and local (Huesser, 1974, (1977, (1978, (1983; Worona and Whitlock, 1995; Gavin et al., 2001) late Pleistocene and Holocene paleoclimate reconstructions provide permissive, but by no means unequivocal explanations for variable hillslope sediment fluxes in the Clearwater basin. For example, Bond et al. (1997) argued for climatic oscillations at ca. 8500, 5000–3000 and 2000–100 yr B.P. and Pisias et al. (2001) recognize similar millenial scale climatic vari-


ability in the Pacific Northwest. We note that these dates are roughly coincident with the gaps in our terrace ages. Unfortunately, this favorable coincidence is obscured by detailed, local palynological studies that seem to indicate a major climatic shift to drier conditions at the late Pleistocene–Holocene boundary followed by a gradual shift to more wet (maritime?) modern conditions beginning ⬃3000 yr ago. The latter shift is not particularly abrupt, highly variable, and evidence for it is locally absent from a core taken at Yahoo Lake (Fig. 1C), within the Clearwater basin (Brubaker, 1998, personal commun.; Gavin et al., 2001). Nevertheless, late Pleistocene and Holocene climate change is a favored explanation for triggering the episodic liberation of sediment from hillslopes throughout the Coast Ranges (Reneau et al., 1989, (1990; Reneau and Dietrich, 1991; Personius et al., 1993). In particular, Reneau et al. (1989) and Reneau and Dietrich (1991) documented a ca. 7–4 ka evacuation of colluvial hollows that is more or less coincident with the Qt5 terrace alluvium. These studies suggest that the residence time for regolith on Olympic Peninsula hillslopes is about 5–6 k.y. and its liberation is primarily limited by the rate of its in-situ production working in concert with climate change, rather than the magnitude of climate change alone. The landscape must be first primed and ready to respond before a trigger can cause sediment to be liberated from the hillslopes and delivered to the channel. Any discussion of potential triggers that might liberate sediment from hillslopes leads us to consider the role of earthquakes (Keefer, 1984; Shimazu and Oguchi, 1996; Hovius et al., 1999; Hovius, 2000; Densmore and Hovius, 2000) in this seismically active setting. We are drawn to the large body of research that has demonstrated the recurrence (⬃300– 500 yr) of large-magnitude (M 8 or greater) earthquakes associated with slip on the subducting Juan de Fuca slab (Atwater and Hemphill-Haley, 1997). Also potentially important are the earthquakes that originate in the upper plate, such as those attributed to the Seattle fault rupture ⬃1000 yr ago (Bucknam et al., 1992; Johnson et al., 1994, 1999). The geomorphic events triggered include tsunamis and coastal uplift and/or subsidence in Puget Sound (Atwater and Moore, 1992; Bucknam et al., 1992), large-scale rock avalanches on the Olympic Peninsula (Schuster et al., 1992), and major landslides into Lakes Washington and Samamish (Karlin and Abella, 1992; Jacoby et al., 1992; Logan et al., 1998). Although we are hesitant to place too much emphasis on age coincidence, the liberation of

741


sediment during this, or similar upper-plate earthquakes, may have contributed to the alluvium that was ultimately incorporated into Qt6. If having to wait until a landscape is primed with enough regolith is the true ratelimiting factor (Reneau and Dietrich, 1991), then earthquakes are as viable a trigger as climate change. Furthermore, it may be that different components of a given landscape respond to different sediment-liberating triggers; for example, colluvial hollows may preferentially respond to climatically induced changes in hillslope vegetative communities, whereas large-scale deep-seated bedrock landslides may respond more frequently to seismic triggers. In both cases (climate or earthquakes), it is the same general coincidence of ages between the trigger and the response that appeals to the geomorphologist. The earthquake trigger explanation suffers from the fact that earthquakes are ubiquitous throughout the Holocene; it would be virtually impossible with the range of radiocarbon dates to convincingly appeal to a single earthquake as the cause of a river terrace. Likewise, the climate argument suffers from an underwhelming documentation of Holocene vegetative response within the Clearwater basin, a response that we might expect if climate change is forcing hillslope-sediment liberation, as is documented for periods of colluvial-hollow excavation in central California (Reneau et al., 1990). Comparison of Holocene to Pleistocene Rates of Incision The rates of river incision determined from Holocene terraces increase in the upstream direction, similar to the rates determined from Pleistocene terraces (Pazzaglia and Brandon, 2001; Fig. 7). However, the Holocene incision rates are everywhere more rapid than their corresponding Pleistocene incision rates. Some studies have pointed out the apparent bias that may result from measuring rapid geologic and geomorphic rates with data collected over a short rather than long time span (Gardner et al., 1987; Mills, 2000). Mills (2000) has suggested that an increase in the rates of river incision is real and can be directly attributed to the fluvial response to a progressively cooler climate characterized by glacial-interglacial cycles. The terraces of the Clearwater basin do not readily lend themselves to that particular explanation because they cover such a limited time span across the late Pleistocene and Holocene. One way to reconcile the differences in incision rate without having to invoke recent increases in the rate of rock uplift is to consider

742

the general differences between the Holocene and Pleistocene Clearwater River. With respect to the Pleistocene river, the Holocene Clearwater River is a stream that is mostly at capacity. It is well equipped with just the right amount of abrasive tools to accomplish incision of its valley bottom in the face of tectonic processes that continue to advect rocks upward through its channel bed. Subtle deviations from this condition foster the variations in the degree of accompanying lateral incision that make and preserve strath terraces. In contrast, the Pleistocene Clearwater River was a stream that, during significant periods of time, was above capacity. During these periods, the channel was lifted completely off the strath, vertical incision ceased because the river and its abrasive tools were no longer in contact with the bedrock of the valley bottom, and the stream flowed as an alluvial channel. There was no headway made on consuming the rock uplifted beneath the channel bed by tectonic processes when the river was in this aggradational phase. The river profile gradient was certainly steepened, and its concavity may have been reduced as uplift continued beneath the valley bottom. It is only during interglacial periods like the Holocene, with at- or nearcapacity sediment-transport conditions, that the long-profile gradient can be reduced and concavity possibly increased as the river makes up for time lost to vertical incision during the alluvial phase. During the Holocene, the Clearwater River is again carving into the bedrock at rates equal to or faster than the rate of rock uplift (Figs. 7, 8). These arguments suggest that geomorphic and active tectonic investigations can only extract meaningful rates of rock uplift from incising streams if the measured incision is integrated across at least one glacial-interglacial cycle. CONCLUSIONS The ⬃400 km2 Clearwater basin, located on the western flank of the actively uplifting Olympic Mountains, contains a well-preserved flight of Holocene fluvial terraces. The Clearwater River is a mixed bedrock and alluvial stream that becomes progressively more influenced by bedrock in the steep, upper onethird of its drainage. Field mapping demonstrates three major and two minor Holocene strath terraces within the lower two-thirds of the 52-km-long river. The three major Holocene terraces are characterized by nearly horizontal straths carved across steeply dipping bedrock; the straths are similar in width to, or wider than, the 0.25–0.5-km-wide modern valley bottom. The terrace deposits atop


these straths consist of ⬃2 m of coarse sand and gravel conformably overlain by ⬃1 m of fine-sand and silt overbank deposits. This sedimentology is consistent with that observed for sediment in the active Clearwater channel and on the flood plain. The terraces are laterally continuous for hundreds of meters. Field relationships could not convincingly determine that the terraces converge or project up stream into channel knickpoints, nor could they argue for down-stream thickening of terrace deposits against dams of LWD. When converted to calendar ages and accounting for 1␴ errors, the terrace ages, determined by 38 14C dates (Table 1), fall into three main groups of ca. 11 000–9000 yr B.P., 8000–4000 yr B.P., and 3000–0 yr B.P., corresponding to terraces Qt4, Qt5, and Qt6, respectively (Fig. 6). These thin Holocene strath terraces stand in marked contrast to thick Pleistocene fill terraces that indicate that the Clearwater Valley was alternately backfilled and then evacuated of thick alluvial sediments. The map relationships, sedimentology, and ages of the terraces allow us to summarize the following interpretations: 1. The Holocene fluvial-terrace stratigraphy of the Clearwater River basin in western Washington State is somewhat different from other, similarly studied basins in Cascadia. Namely, similar studies (e.g., Personius et al., 1993; Merritts et al., 1994; Massong, 1998) have emphasized the presence of a widespread late Pleistocene–early Holocene terrace genetically linked to the well-documented climate change across that boundary. Qt4 in the Clearwater basin is likely representative of the late Pleistocene–Holocene terrace, but its preservation is much subordinate to the middle and late Holocene terraces Qt5 and Qt6. The lack of a well-preserved and widespread late Pleistocene–early Holocene terrace may indicate different climatic conditions for northern Cascadia at this time with respect to Oregon or California, or it may suggest that lateral incision and valley widening related to the Qt5 event were efficient at removing the older terrace. 2. The favored model for genesis of Holocene-scale terraces within the Clearwater basin involves a channel with a temporally variable rate of both lateral and vertical incision. The terraces represent times of enhanced lateral incision, valley-bottom widening, and the carving of straths accomplished by the thin alluvial deposits preserved atop the straths for ⬎2–4000 yr (Fig. 8, option A). The frequency distributions of the ages of the terrace deposits overlying a given strath support this idea because the distributions are skewed toward the


younger extreme of the entire age range (Figs. 6, 8). Valley-bottom narrowing and rapid vertical incision into bedrock were accomplished during the relatively brief (⬃1000 yr) time periods between the carving of the major straths. 3. We envision the Holocene Clearwater River as operating at or near capacity. The alternations in rates of vertical and lateral incision may be linked to temporal variations in the liberation and delivery of hillslope sediment to the Clearwater channel that affect the at-capacity condition. Times of increased hillslope sediment flux favor lateral incision, especially if those times coincided with stable, steady discharges like those envisioned for the valley-bottom widening cycles described in Meyer et al. (1995). Similarly and following the same analogy, a reduction in the sediment yield, especially in concert with a flashier discharge, fosters valley-bottom narrowing and vertical incision. Holocene climate change is an unknown factor in driving these changes in hillslope sediment yield, in part because of the lack of convincing, local, high-resolution proxy records. We appeal to the possibility that ground accelerations associated with earthquakes could be a viable alternative mechanism for liberating hillslope sediment when slope-moisture conditions were favorable to slope failure and/or when the slopes were primed with a sufficient thickness of regolith awaiting a trigger to liberate it. In a way analogous to how frequent large fires in the intermontane west are only effective at destabilizing hillslopes if they happen to occur during relatively dry climates with highly seasonal precipitation (Meyer et al., 1995), the earthquakes might only be effective when some other condition, like soil moisture and/ or adequate regolith thickness, is present. 4. Rates of vertical fluvial incision, determined from Holocene terraces, are similar to the rates determined from Pleistocene terraces (Pazzaglia and Brandon, 2001) in that they are relatively low downstream and progressively increase in an upstream direction, subparallel to the direction of plate convergence (Fig. 7). However, the Holocene incision rates are everywhere more rapid (2–3 times) than their corresponding Pleistocene incision rates of ⬍0.1 m/k.y. at the coast to 0.9 m/k.y. in the upper Clearwater basin; the Pleistocene incision rates more closely approximate the longterm rates of orogen unroofing determined from thermochronology (Brandon et al., 1998). We favor an explanation in which the Holocene rates represent a channel rapidly reacquiring its stable, graded concavity following protracted periods of time in the Pleistocene when it could not accomplish any

vertical incision into bedrock because the channel was lifted off the bedrock-valley bottom by climatically induced alluviation. These results illustrate how even in tectonically active settings, representative rates of rock uplift inferred from studies of river incision should be temporally integrated over at least one glacial-interglacial cycle. ACKNOWLEDGMENTS The research was funded by National Science Foundation grant EAR-9736748 (to Pazzaglia). We thank the State of Washington, Department of Natural Resources–Olympic region, especially Rick Cahill, and Olympic National Park personnel for access and support. Logistical field support was provided by Brian Atwater and the U.S. Geological Survey. We thank Glenn Thackray, Wendy Gerstel, Jonathan Barbour, Eric McDonald, John Gosse, Mark Brandon, William Lingley, Jeff Cederholm, Alex Wegmann, Emily Repanich, Andrew Zachery, and James Walton, for in-field discussions of these observations. David Montgomery, Andrew Megis, and Ellen Wohl provided helpful reviews on an early version of this manuscript. Gregory Hancock and Andrew Megis provided formal reviews; we are grateful to all who provided necessary criticism and helpful insight. This research was completed when both of us were at the Department of Earth and Planetary Sciences, University of New Mexico. We thank our former institution for logistical and financial support toward the completion of this project. REFERENCES CITED Abbe, T.B., and Montgomery, D.R., 1996a, Large woody debris jams, channel hydraulics and habitat formation in large rivers: Regulated rivers: Research and Management, v. 12, p. 201–221. Abbe, T.B., and Montgomery, D.R., 1996b, Floodplain and terrace formation in forested landscapes [abs.]: Geological Society of America Abstracts with Programs, v. 28, no. 5, p. 41. Atwater, B.F., 1987, Evidence for great Holocene earthquakes along the outer coast of Washington State: Science, v. 236, p. 942–944. Atwater, B.F., 1992, Geologic evidence for earthquakes during the past 2000 years along the Copalis River, southern coastal Washington: Journal of Geophysical Research, v. 97, p. 1901–1919. Atwater, B.F., and Hemphill-Haley, E., 1997, Recurrence intervals for great earthquakes of the past 3,500 years at northeastern Willapa Bay, Washington: U.S. Geological Survey Professional Paper 1576, 108 p. Atwater, B.F., and Moore, A.L., 1992, A tsunami about 1000 years ago in Puget Sound, Washington: Science, v. 258, p. 1614–1617. Bond, G., Showers, W.J., Cheseby, M., Lotti, R., Almasi, P., deMenocal, P., Priore, P., Cullen, H., Hajdas, I., and Bonani, G., 1997, A pervasive millennial-scale cycle in North Atlantic Holocene and glacial climates: Science, v. 278, p. 1257–1266. Brandon, M.T., and Vance, J.A., 1992, Tectonic evolution of the Cenozoic Olympic subduction complex, Washington State, as deduced from fission track ages for detrital zircons: American Journal of Science, v. 292, p. 565–636. Brandon, M.T., Roden-Tice, M.K., and Garver, J.I., 1998, Late Cenozoic exhumation of the Cascadia accretionary wedge in the Olympic Mountains, northwest Washington State: Geological Society of America Bulletin, v. 110, p. 985–1009. Bucknam, R.C., Hemphill-Haley, E., and Leopold, E.B., 1992, Abrupt uplift within the past 1700 years at


southern Puget Sound, Washington: Science, v. 258, p. 1611–1614. Bull, W.B., 1990, Stream-terrace genesis: Implications for soil development: Geomorphology, v. 3, p. 351–367. Bull, W.B., 1991, Geomorphic response to climatic change: New York, Oxford University Press, 326 p. Bull, W.B., and Brandon, M.T., 1998, Lichen dating of earthquake-generated regional rockfall events, Southern Alps, New Zealand: Geological Society of America Bulletin, v. 110, p. 60–84. Bull, W.B., and Knuepfer, P.L.K., 1987, Adjustments by the Charwell River, New Zealand to uplift and climate changes: Geomorphology, v. 1, p. 15–32. Burbank, D.W., Leland, J., Fielding, E., Anderson, R.S., Brozovic, N., Reid, M.R., and Duncan, C., 1996, Bedrock incision, rock uplift and threshold hillslopes in the northwestern Himalayas: Nature, v. 379, p. 505–510. DeMets, C., and Dixon, T., 1999, New kinematic models for Pacific–North America motion from 3 Ma to present; I, Evidence for steady motion and biases in the NUVEL-1A model: Geophysical Journal International, v. 101, p. 425–478. Densmore, A.L., and Hovius, N., 2000, Topographic fingerprints of bedrock landslides: Geology, v. 28, p. 371–374. Dragert, H., Hyndman, R.D., Rogers, G.C., and Wang, K., 1994, Current deformation and the width of the seismogenic zone of the northern Cascadia subduction thrust: Journal of Geophysical Research—Solid Earth, v. 99, p. 653–668. Foley, M.G., 1980, Bedrock incision by streams: Geological Society of America Bulletin, Part II, v. 91, p. 2189–2213. Fuller, I.C., Macklin, M.G., Lewin, J., Passmore, D.G., and Wintle, A.G., 1998, River response to high-frequency climate oscillations in southern Europe over the past 200 k.y.: Geology, v. 26, p. 275–278. Gardner, T.W., Jorgensen, D.W., Shuman, C., and Lemieux, C.R., 1987, Geomorphic and tectonic process rates: Effects of measured time interval: Geology, v. 15, p. 259–261. Gardner, T.W., Verdonck, D., Pinter, N.M., Slingerland, R., Furlong, K.P., Bullard, T.F., and Wells, S.G., 1992, Quaternary uplift astride the aseismic Cocos Ridge, Pacific coast of Costa Rica: Geological Society of America Bulletin, v. 104, p. 219–232. Gavin, D.G., McLachlan, J.S., Brubaker, L.B., and Young, K.A., 2001, Postglacial history of subalpine forests, Olympic Peninsula, Washington, USA: Holocene, v. 11, p. 177–188. Gerstel, W.J., 1999, Deep-seated landslide inventory of the west-central Olympic Peninsula: Washington Division of Geology and Earth Resources, Open File Report 99-2, 36 p. Gerstel, W.J., and Lingley, W.S., Jr., compilers, 2000, Geologic map of the Forks 1:100,000 Quadrangle, Washington: Washington Division of Geology and Earth Resources Open File Report 2000-4, 36 p., scale 1:100 000, 1 plate. Gilbert, G.K., 1877, Geology of the Henry Mountains (Utah): Washington D.C., U.S. Government Printing Office, U.S. Geographical and Geological Survey of the Rocky Mountains Region, 170 p. Hancock, G.S., Anderson, R.S., Chadwick, O.A., and Finkle, R.C., 1999, Dating fluvial terraces with 10Be and 26 Al profiles: Application to the Wind River, Wyoming: Geomorphology, v. 27, p. 41–60. Hancock, G.S., and Anderson, R.S., 2001, Numerical modeling of strath terrace sequence formation in response to climate oscillation: Eos Trans: American Geophysical Union, Fall Meeting Supplement, Abstract H42D-0390, v. 82. Hasbargen, L.E., and Paola, C., 2000, Landscape instability in an experimental drainage basin: Geology, v. 28, p. 1067–1070. Hovius, N., 2000, Macro scale process systems of mountain belt erosion, in Summerfield, M.A., ed., Geomorphology and global tectonics: New York, John Wiley and Sons, p. 77–105. Hovius, N., Stark, C.P., Tutton, M.A., and Abbott, L.D., 1999, Landslide-driven drainage network evolution in

743

WEGMANN and PAZZAGLIA a presteady-state mountain belt; Finisterre Mountains, Papua New Guinea: Geology, v. 26, p. 1071–1074. Huesser, C.J., 1974, Quaternary vegetation, climate, and glaciation of the Hoh River Valley, Washington: Geological Society of American Bulletin, v. 85, p. 1547–1560. Huesser, C.J., 1977, Quaternary palynology of the Pacific slope of Washington: Quaternary Research, v. 8, p. 282–306. Huesser, C.J., 1978, Palynology of Quaternary deposits of the lower Bogachiel River area, Olympic Peninsula, Washington: Canadian Journal of Earth Sciences, v. 15, p. 1568–1578. Huesser, C.J., 1983, Vegetational history of the northwestern United States including Alaska, in Porter, S.C., ed., Late Quaternary environments of the United States, Volume 1: Minneapolis, University of Minnesota Press, p. 239–258. Jacoby, G.C., Williams, P.L., and Buckley, B.M., 1992, Tree ring correlation between prehistoric landslides and abrupt tectonic events in Seattle, Washington: Science, v. 258, p. 1621–1623. Johnson, S.Y., Potter, C.J., and Armentrout, J.M., 1994, Origin and evolution of the Seattle fault and Seattle basin, Washington: Geology, v. 22, p. 71–74. Johnson, S.Y., Dadisman, S.V., Childs, J.R., Stanley, W.D., 1999, Active tectonics of the Seattle fault and central Puget Sound, Washington—Implications for earthquake hazards: Geological Society of America Bulletin, v. 111, p. 1042–1053, 1 loose insert. Karlin, R.E., and Abella, S.E.B., 1992, Paleoearthquakes in the Puget Sound region recorded in sediments from Lake Washington, U.S.A.: Science, v. 258, p. 1617–1620. Keefer, D.K., 1984, Landslides caused by earthquakes: Geological Society of America Bulletin, v. 95, p. 406–421. Knox, J.C., 1975, Concept of the graded stream, in Melhorn, W.N., and Flemal, R.C., eds., Theories of landform development: State University of New York at Binghamton, Publications in Geomorphology, Proceedings of the 6th Geomorphology Symposium, p. 169–198. Logan, R.L., Schuster, R.L., Pringle, P.T., Walsh, T.J., and Palmer, S.P., 1998, Radiocarbon ages of probable coseismic features from the Olympic Peninsula and Lake Sammamish, Washington: Washington Geology, v. 26, no. 2/3, p. 59–67. Mackin, J.H., 1937, Erosional history of the Big Horn basin, Wyoming: Geological Society of America Bulletin, v. 48, p. 813–893. Mackin, J.H., 1948, Concept of the graded river: Geological Society of America Bulletin, v. 59, p. 463–512. Maddy, D., 1997, Uplift-driven valley incision and river terrace formation in southern England: Journal of Quaternary Science, v. 12, p. 539–545. Massong, T.M., 1998, Influence of lithology, sediment supply, and log jams on the distribution of bedrock and alluvial channels [M.S. thesis]: Seattle, University of Washington, 62 p. Merritts, D., Vincent, K., and Wohl, E.T., 1994, Long river profiles, tectonism, and eustasy: A guide to interpreting fluvial terraces: Journal of Geophysical Research, v. 99, B7, p. 14 31–14 050. Meyer, G.A., Wells, S.G., and Jull, A.J.T., 1995, Fire and alluvial chronology in Yellowstone National Park: Climate and intrinsic controls on Holocene geomorphic processes: Geological Society of America Bulletin, v. 107, p. 1211–1230. Mills, H.H., 2000, Apparent increasing rates of stream incision in the Eastern United States during the late Cenozoic: Geology, v. 28, p. 955–957. Montgomery, D.R., Abbe, T.B., Buffington, J.M., Peterson, N.P., Schmidt, K.M., and Stock, J.D., 1996, Distribution of bedrock and alluvial channels in forested mountain drainage basins: Nature, v. 381, p. 587–589. Nelson, L.M., 1986, Fluvial sediment in the Hoh River, in Lum, W.E., II, Reconnaissance of the water resources of the Hoh Indian Reservation and the Hoh River basin, Washington: U.S. Geological Survey Water Resources Investigations Report WRI-85–4018, p. 18–19. Palmer, S.P., and Lingley, W.S., 1989, An assessment of the

744

oil and gas potential of the Washington outer continental shelf: Washington Department of Natural Resources Report WSG 89-2, 83 p. Pazzaglia, F.J., and Brandon, M.T., 2001 A fluvial record of rock uplift and shortening across the Cascadia forearc high: American Journal of Science, v. 301, p. 385–431. Pazzaglia, F.J., and Gardner, T.W., 1993, Fluvial terraces of the lower Susquehanna River: Geomorphology, v. 8, p. 83–113. Pazzaglia, F.J., Gardner, T.W., and Merritts, D., 1998, Bedrock fluvial incision and longitudinal profile development over geologic time scales determined by fluvial terraces, in Wohl, E., and Tinkler, K., eds., Bedrock channels: American Geophysical Union Geophysical Monograph Series, v. 107, p. 207–235. Pederson, J., Pazzaglia, F.J., and Smith, G.A., 2000, Ancient hillslope deposits: Missing links in the study of climate controls on sedimentation: Geology, v. 28, p. 27–30. Pederson, J., Smith, G.A., and Pazzaglia, F.J., 2001, Comparing the modern, Quaternary, and Neogene records of climate-controlled hillslope sedimentation in southeast Nevada: Geological Society of America Bulletin, v. 113, p. 305–319. Personius, S.F., 1995, Late Quaternary stream incision and uplift in the forearc of the Cascadia subduction zone, western Oregon: Journal of Geophysical Research, v. 100B, p. 20 193–20 210. Personius, S.F., Kelsey, H.M., Grabau, P.C., 1993, Evidence for regional stream aggradation in the central Oregon Coast Range during the Pleistocene–Holocene transition: Quaternary Research, v. 40, p. 297–308. Pisias, N.G., Mix, A.C., and Heusser, L., 2001, Millennial scale climate variability of the northeast Pacific Ocean and northwest North America based on radiolaria and pollen: Quaternary Science Reviews, v. 20, p. 1561–1576. Ramsay, B., 2000, OxCal version 3.5: Oxford, UK, University of Oxford, downloaded in January 2001, available from http://www.rlaha.ox.ac.uk/orau/index.htm. Reid, L.M., and Dunne, T., 1984, Sediment production from forest road surfaces: Water Resources Research, v. 20, p. 1753–1761. Reneau, S., and Dietrich, W.E., 1991, Erosion rates in the southern Oregon Coast Ranges: Evidence for equilibrium between hillslope erosion and sediment yield: Earth Surface Processes and Landforms, v. 16, p. 307–322. Reneau, S.L., Dietrich, W.E., Rubin, M., Donahue, D.J., and Jull, T.A.J., 1989, Analysis of hillslope erosion rates using dated colluvial deposits: Journal of Geology, v. 97, p. 45–63. Reneau, S.L., Dietrich, W.E., Donahue, D.J., Jull, A.J.T., and Rubin, M., 1990, Late Quaternary history of colluvial deposition and erosion in hollows, central California Coast Ranges: Geological Society of America Bulletin, v. 102, p. 969–982. Savage, J.C., Lisowski, M., and Prescott, W.H., 1991, Strain accumulation in western Washington: Journal of Geophysical Research, v. 96B, p. 14 493–14 507. Schumm, S.A., 1969, River metamorphosis: Proceedings of the American Society of Civil Engineers, Journal of the Hydraulics Division, v. 95, p. 255–273. Schumm, S.A., Mosley, M.P., and Weaver, W.E., 1987, Experimental fluvial geomorphology: New York, Wiley Interscience, 413 p. Schuster, R.L., Logan, R.L, and Pringle, P.T., 1992, Prehistoric rock avalanches in the Olympic Mountains, Washington: Science, v. 258, p. 1620–1621. Seidl, M.A., and Dietrich, W.E., 1992, The problem of channel erosion into bedrock: Catena, supplement 23, p. 101–124. Serdar, C.F., 1999, Description, analysis and impacts of the Grouse Creek landslide, Jefferson County, Washington, 1997–98 [Master’s thesis]: Olympia, Washington, Evergreen State College, 171 p. Shimazu, H., and Oguchi, T., 1996, River processes after rapid valley-filling due to large landslides: GeoJournal, v. 38, p. 334–339. Sklar, L., and Dietrich, W.E., 1998, River longitudinal pro-


files and bedrock incision models: Stream power and the influence of sediment supply, in Wohl, E., and Tinkler, K., eds., Rivers over rock: Fluvial processes in bedrock channels: American Geophysical Union Geophysical Monograph Series, v. 107, p. 237–260. Slingerland, R., Willett, S.D., and Henessey, H.L., 1997, A new fluvial bedrock erosion model based on the workenergy principle [abs.]: Eos (Transactions, American Geophysical Union), v. 78, no. 46, p. 299–300. Stewart, R.J., 1970, Petrology, metamorphism, and structural relations of graywackes in the western Olympic Peninsula, Washington [Ph.D. thesis]: Stanford, California, Stanford University, 129 p. Stuiver, M., and Reimer, P.J., 1993, Extended 14C data base and revised CALIB 3.0 14C age calibration program: Radiocarbon, v. 35, no. 1, p. 215. Stuiver, M., Reimer, P.J., Bard, E., Beck, J.W., Burr, G.S., Hughen, K.A., McCormac, G., van der Plicht, J., and Spurk, M., 1998, INTCAL98 radiocarbon age calibration, 24-000-0 cal BP: Radiocarbon, v. 40, no. 3, p. 1041–1083. Sugai, T., 1993, River terrace development by concurrent fluvial processes and climate change: Geomorphology, v. 6, p. 243–252. Tabor, R.W., and Cady, W.M., 1978a, The structure of the Olympic Mountains, Washington—Analysis of a subduction zone: U.S. Geological Survey Professional Paper 1033, 38 p. Tabor, R.W., and Cady, W.M., 1978b, Geologic map of the Olympic Peninsula: U.S. Geological Survey Miscellaneous Investigations Map I-994, 2 sheets, scale 1: 125 000. Thackray, G.D., 1996, Glaciation and neotectonic deformation on the western Olympic Peninsula, Washington [Ph.D. thesis]: Seattle, Washington, 139 p. Thackray, G.D., and Pazzaglia, F.J., 1994, Quaternary stratigraphy, tectonic geomorphology, and fluvial evolution of the western Olympic Peninsula, Washington, in Swanson, D.A., and Haugerud, R.A., eds., Geologic field trips in the Pacific Northwest, 1994: Seattle, Washington, University of Washington Department of Geological Sciences, p. 2A-1–2A-30. Tucker, G.E., and Slingerland, R.L., 1997, Drainage basin response to climate change: Water Resources Research, v. 33, p. 2031–2047. Wegmann, K.W., 1999, Late Quaternary fluvial and tectonic evolution of the Clearwater River basin, western Olympic Mountains, Washington State [M.S. thesis]: Albuquerque, University of New Mexico, 217 p. Wells, S.G., Bullard, T.F., Condit, C.D., Jercinovic, M., Lozinsky, R.P., and Rose, D.F., 1982, Geomorphic processes on the alluvial valley floor of the Rio Puerco: Albuquerque, New Mexico Geological Society Guidebook 33, p. 45–47. Wells, S.G., Bullard, T.F., Menges, C.M., Drake, P.G., Karas, P.A., Kelson, K.I., Ritter, J.B., and Wesling, J.R., 1988, Regional variations in tectonic geomorphology along a segmented convergent plate boundary, Pacific Coast of Costa Rica: Geomorphology, v. 1, p. 239–265. Whipple, K.X, Hancock, G.S., and Anderson, R.S., 2000, Processes of river incision into bedrock: Implications for the dynamics of landscape evolution: Geological Society of America Bulletin, v. 112, p. 490–503. Wolman, M.G., and Miller, J.P., 1960, Magnitude and frequency of forces in geomorphic processes: Journal of Geology, v. 68, p. 54–74. Worona, M.A., and Whitlock, C., 1995, Late Quaternary vegetation and climate history near Little Lake, central Coast Range, Oregon: Geological Society of America Bulletin, v. 107, p. 867–876. Zaprowski, B.J., Evenson, E.B., Pazzaglia, F.J., and Epstein, J.B., 2001, Knickzone propagation in the Black Hills and northern High Plains: A different perspecive on the late Cenozoic exhumation of the laramide Rocky Mountains: Geology, v. 29, no. 6, p. 547–550. MANUSCRIPT RECEIVED BY THE SOCIETY MARCH 30, 2001 REVISED MANUSCRIPT RECEIVED JANUARY 18, 2002 MANUSCRIPT ACCEPTED FEBRUARY 5, 2002 Printed in the USA