Topographic Form of the Coast Ranges of the Cascadia Margin in Relation to Coastal Uplift Rates and Plate Subduction

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 99, NO. B6, PAGES 12,245-12,255, JUNE 10, 1994

Topographic form of the Coast Ranges of the Cascadia Margin in relation to coastal uplift rates and plate subduction

Harvey M. Kelsey,• David C. Engebretson,'•Clifton E. Mitchell,3 and Robert L.Ticknor'•

Abstract. The CoastRanges of theCascadia margin are overriding the subducted Juan de Fuca/Gordaplate. We investigatethe extent to whichthe latitudinal trend in averagetopography of the CoastRanges is a functionof the latitudinalchange in attributesrelated to the subduction process.These attributes include the variableage of the subductedslab that underlies the Coast Rangesand average vertical crustal velocities of thewestern margin of theCoast Ranges for two markedlydifferent time periods, the last 45 yearsand the last 100kyr. Thesevertical crustal velocitiesare computedfrom the resurveying of highwaybench marks and from thepresent elevationof shoreplatforms that have been uplifted in thelate Quaternary, respectively. Topographyof theCoast Ranges is in parta functionof theage and bouyancy of theunderlying subductedplate. This is evidentin the fact that the two highesttopographic elements of the CoastRanges, the Klamath Mountains and the Olympic Mountains, are underlain by youngest subductedoceanic crust. The subductedBlanco Fracture Zone in southernmostOregon is responsiblefor an age discontinuityof subductedcrust under the Klamath Mountains. The northernterminus of the topographicallyhigher Klamaths is offsetto the northrelative to the positionof the underlyingBlanco Fracture Zone, the offset being in thedirection of migrationof thefracture zone, as dictated by relativeplate motions. Vertical crustal velocities at thecoast, derivedfrom benchmark surveys, are as much as an orderof magnitudegreater than vertical crustalvelocities derived from uplifted shore platforms. This uplift rate discrepancy indicates thatstrain is accumulatingon theplate margin, to be releasedduring the next interplate earthquake.In a latitudinalsense, average Coast Range topography is relativelyhigh where benchmark-derived, short-term vertical crustal velocites are highest. Because the shore platform verticalcrustal velocites reflect longer-term, permanent uplift, we inferthat a smallpercentage of theinterseismic strain that accumulates as rapid short-term uplift is notrecovered by subduction earthquakesbut rathercontributes to rockuplift of theCoast Ranges. The conjecturethat permanentrock uplift is relatedto interseismicuplift is consistentwith theobservation that thosesegments of the subductionzone subject to greaterinterseismic uplift ratesare at approximatelythe samelatitudes as thosesegments of the the CoastRanges that have higher magnitudesof rock uplift over the long term.

Introduction The topographyof the Coast Rangesis the net result of uplift of rocks due to plate convergenceand erosion by surface The coastal ranges of northern California, Oregon and processes. We hypothesize that latitudinal variations in Washingtonform a belt of elevatedtopography along the Coast Range topographyare a functionof changingphysical northwest coast of the United States. This area includes the attributes of the Cascadia subduction zone from north to south. northern California Coast Ranges, the Klamath Mountains, To test this hypothesisempirically, we compare latitudinal the OregonCoast Ranges, the Willapa Hills, and the Olympic variation in averagetopography to the latitudinalvariation in Mountains(Figure 1). For our purposes,we collectivelycall vertical crustal velocity at the coast, as measuredby uplift these features the Coast Ranges. The Coast Ranges are rates relative to sea level, and to the latitudinal variation in situatedon the leadingedge of the North Americanplate, and the age of the plate being subductedbeneath the CoastRanges. crest elevations are 120-200 km east of the deformation front For our analysis, we present data on latitudinal and of the Cascadia subduction zone. The ranges are thus longitudinal trends in the topographyof the Coast Ranges, overridingthe subductedJuan de Fuca/Gordaplate (Figure1). variation in uplift rate along the westernmargin of the Coast Ranges, and latitudinal variation in the relative elevation of the subductingplate beneath the Coast Ranges, as derived 1Departmentof Geology,Humboldt State University, Arcata, from the relationshipbetween lithospheric subsidence and age California. of oceanic crust. 2Departmentof Geology,Western Washington University, For the coastaluplift rates, the data include surfaceuplift Bellingham. 3Departmentof Geological Sciences, University of Oregon,Eugene. rate for two differenttime periodsusing datums appropriate to thesetime periods:the last 45 yearsof uplift usingresurveyed Copyright1994 by the AmericanGeophysical Union. highway benchmarksand the last 80,000-125,000 years of uplift using uplifted shore platforms. Surface uplift rates Papernumber 93JB03236. derived from both geodeticand shoreplatform surveysare 0148-0227/94/93JB-03236505.00 referencedto eustatic high standsof sea level, where eustatic

12,245 12,246 KELSEY ET AL.: TOPOGRAPHIC FORM OF COAST RANGES

50 ø

.:::::::.•:::...... :.:.:.:.::.:.:....:.:.:.:.:. \

JUAN DE FUCA 45 ø PLATE 7

PACIFIC PLATE

40 ø

130 ø

CASCADIA CFW DEFORMATION YB c FRONT

WINDOW I I I I I I

I I I I I I I

126 ø I I

Figure 1. (a) Locationmap of the Pacific Northwestin the contextof plate tectonics;triangles are Cascade volcanoes.(b) Locationmap of the coastalranges situated on the leadingegde of the North Americanplate betweenthe Mendocinoescarpment and the Canadianborder. Trianglesdenote selected Cascade volcanoes. CR, mouthof the ColumbiaRiver; CFW, CapeFoulweather; YB, YaquinaBay; CA, CapeArago; CB, Cape Blanco;CF, Cape Ferrelo;HB, HumboldtBay; CM, Cape Mendocino. (c) Shadingshows the area of the CoastRanges, 65 arc min of longitudewide, for which the digital elevationmodel ETOPO-5 was employed to computeaverage topography. The box depictsthe size of the computationalwindow (65 arc minx 15 arc min) that was employed for each individual calculation of average topography. See text for further explanation. KELSEY ET AL' TOPOGRAPHIC FORM OF COAST RANGES 12,247 sea level is an expression of Earth's geoidal surface. The three adjacentlines of latitude (computationalwindow (Figure geodetically derived uplift is in referenceto present sea level lc) is 65 arc min of longitude x 15 arc min of latitude on a 5 and the shore platform uplift is in reference to the arc min grid), with 13 out of 39 of the grid pointschanging for apppropriatelate Plesitoceneeustatic sea level high stand(see each calculation. Thus each calculationof averagetopography below). Becauserock uplift relative to the geoid is equal to representsan areaof about2400 km2 (27.8 km x 86.6 km). surface uplift minus exhumation [England and Molnar, 1990] Because we will subsequentlycompare uplift rates at the and because there has been negligible erosion of either the coast to trends in averagetopography for the Coast Rangesas shoreplatforms or the highway benchmarks,the coastaluplift a whole, we determined whether north-southtrends in average rates that we discussare both uplift rates of rock relative to the topographyfor the coastal (western) side of the Coast Ranges geoid as well as surfaceuplift rates. Subsequentuse of the term are sinfilar in form to trends of average topography for the "uplift rate" should thus be unambiguous. Using uplift rate Coast Ranges as a whole. Using the same computational data integratedover the last 45 years and that integratedover technique but applied to narrower widths, we determined the last 80,000-125,000 years, we can compare interseismic average topographyfor a 25-arc-min-wide swath on the west uplift with long-term uplift at the coast along the Cascadia side of the Coast Ranges, and we also computed average margin. topography for a nonoverlapping 25-arc-min-wide swath on the eastern side of the Coast Ranges. Comparing the western Average Topography Along the Crest profile to the profile for the Coast Ranges as a whole (Figure of the Coast Ranges 2), the profiles show that first, the highest average topography is between 41 ø and 45 ø, correspondingto the We calculated average topography for the Coast Ranges Klamaths Mountains; second, average topographyfluctuates (Figure 2) using the public domain 5 arc min digital elevation around 200 m between 43.5 and 46ø; and third, a spike in model ETOPO-5 [National GeophysicalData Center, 1988]. average topographyoccurs at 47.5o-48ø , correspondingto the Using ETOPO-5, we calculated a running average for the Olympic Mountains. The two profiles differ considerably topography of the Coast Ranges in a north-south trending along two segmentswhere coastal plains or embaymentsare swath extending from the coast inland for 65 arc min of unusually wide, extending as much as 20 km inland. These longitude(Figure l c). The swathextends between latitudes 41 ø segments include the southern Oregon coastal plain (Cape and 47øN along the trend of the Coast Ranges, for a total Blanco to Coos Bay; latitudes 43ø-43.3ø) and the embayments length of 778 kin. The averagewidth of the swathis 86.6 km of southernWashington (Columbia River, Willapa Bay, Grays (90.9 km at latitude 41 ø and 82.2 km at latitude 47ø). Each Harbor; latitudes 46ø-47ø). With the exception of these individual calculation of average topography at 5-arc min coastallowlands, the westernportion of the Coast Rangeshas latitudinal intervals is the running average of the average latitudinal trends in elevation similar to that of the Coast altitude at 39 different grid points, 13 grid points from each of Range as a whole (Figure 2), albeit the average elevations

1400

1200 - AverageTopography

000 WesternCoast Range EasternCoast Range 800 CombinedCoastRange

600

400

200

_ , ,-, .---

, -200 i i i i i i i i i i i i i i i i i i i i i 41 42 43 44 45 46 47 48 LATITUDE ( DEG NORTH )

Figure 2. Latitudinal variation in average topographyfor the 25-arc-min-wide western portion of the Coast Ranges,for the 25-arc-min-wideeastern portion of the CoastRanges and for the entire 65-arc-min- wide Coast Ranges. All longitudinaldistances are measuredeastward (inland) from the coast. The eastern and westerndata setsdo not use overlappingdata. General topographictrends are similar for the eastern, western, and combined data sets. 12,248 KELSEY ET AL.: TOPOGRAPHIC FORM OF COAST RANGES along the coast are lower because Coast Range drainages plates presently subductingat the Cascadia margin, yields transportsediment to the west. latitudinal estimatesof the relative elevation of the subducting The error limit for the latitudinal variation in average plate at a position beneath the crest of the Coast Ranges topographyof the Coast Ranges (Figure 3) is the standard (Figure 3). error of the mean. Becauseof the runningaverage, sequential The lithospheric subsidence model predicts that rapid standarderrors are correlated,but every third standarderror is density changes occur within the oceanic plate in close independent. We also show an envelopecurve depictingthe proximity to ridges. However, thick accumulationsof young upper bound of elevations used in the calculation at each sedimentsalong the Cascadiamargin and a lack of seismicity latitude ("maximum elevation" in Figure 3). The maximum near the trenchaxis precludedirect measurementof depthsto elevation is the average elevation of the crest of the Coast the top of the subductingJuan de Fuca plate beneaththe Coast Ranges computed as a three-point running average. The Range. We acknowledgeproblems inherent in estimatingslab maximum elevation profile shows the magnitudeof surface morphologynear subductionzones using the simpleage-depth uplift, or uplift of rocks minus exhumation,along the Coast relationshipcited above and etnphasizethat our interpretation Range crest. focuseson the relative subsidenceof the slab due to changesin slab age along the margin. When oceanicplates are loadedby subduction under continental crust, the differential subsidence Overlying TopographyVersus Age of Subducted Juan de Fuca Plate would be enhancedover that causedby an increasein age of the plate alone. This enhancementis because a less dense, Subsidence,or changein elevation, of the oceanfloor as it younger slab will subsideless than an adjacentolder, denser movesaway from the spreadingridge is expressedby slab, if both are loaded by the same overlying continental plate. Thus the differential elevation of two adjacent e(t) = ctm subsidingslabs, as computedby the Parsons-Sclaterrelation in the case of no loading, is a minimum estimate for the [Parsons and Sclater, 1977], where e(t) is the vertical distance amount of differential subsidence that must occur if both slabs in meters that the ocean floor has subsidedin moving away are subsequentlyloaded by an overridingplate. from the spreading ridge and t is the age of the plate in Approximationsfor the age of the subductingplates, needed millions of years. Constantc, the linear slopeof the change- to calculate relative subsidence, are made difficult by a in-elevationversus t •/2 relation, is 350 [Parsonsand Sclater, complexpattern of spreadingalong the Juande Fuca andGorda 1977]. The use of this subsidencemodel, in conjunctionwith ridges. These complications include propagating rifts, an approximateage distributionof the Juande Fucaand Gorda changes in relative motion poles, deformation within the

1800

1600

1400

1200

1000

800 Maximum elevation

600

400

200 - ; Average•topography 7 • I I _- Relativeelevation to top of subductedplate based on lithosphericsubsidence -200 ,,, 41 42 43 44 45 46 47 48 LATITUDE ( DEG NORTH ) Figure 3. Latitudinal variation in averagetopography and age of subductedslab along the crest of the CoastRanges. Top curve,maximum average elevation used in eachtopographic calculation; middle curves, averagetopography bounded by the standarderror of the mean;bottom dashed line, relativeelevation of the top of the subductedslab beneaththe crestof the CoastRanges. KELSEY ET AL.: TOPOGRAPHIC FORM OF COAST RANGES 12,249

Gordaplate, and an uncertainhistory for the developmentof The top of the relatively young,more bouyant subducted crust the Blanco Fracture Zone [Wilson, 1986, 1990; Embley and south of the Blanco Fracture Zone has an elevation that is Wilson, 1992]. Age estimates(Table 1) were determinedby about500 m higherrelative to the top of the slab northof the reflectingrecognizable magnetic isochrons from the Pacific Blanco Fracture Zone. The north facing scarp in the plate onto the Juan de Fuca and Gorda plates using the downgoingslab, with relatively higher slab to the south,is magneticisochron map of Atwaterand Severinghaus[1989]. presentlydirectly below the zone of maximum changein A similar result is obtainedthrough an eastwardextrapolation averagetopography along the CoastRange crest (Figure 3). of the isochronpatterns observed immediately to the westof This altitudinal change marks the topographic division the Cascadia deformation front. between the Klamath Mountains to the south and the The ageof the slabbeneath the crestof the CoastRanges is elevationally lower Oregon Coast Ranges to the north. uniformexcept for markedlyyounger slab at the southernend Similarly, the topographicallyhigher Olympic Mountains,at beneaththe Klamath Mountains,which form the Coast Ranges the northern end of the Coast Ranges in Washington, are in southernmostOregon and northernmostCalifornia (Figure underlain by relatively young subductedplate that is, in 3). Youngerslab also underlies the OlympicMountains at the relative elevation, about 300 m higher than older subducted north end of the Coast Range. The age discontinuityin the plate further south. subducted slab beneath the Klamath Mountains is due to subductionof the Blanco Fracture Zone (Figure lb), which GeodeticUplift Rates juxtaposesyoung crust to the southagainst older crust to the north (Table 1). The younger slab beneath the Olympic Differences in the relative heights of bench marks Mountainsreflects subductionof progressivelyyounger plate reoccupiedby successivefirst-order leveling surveys provide a to the north as the Pacific-Juan de Fuca spreading center data set of geodeticallyderived uplift ramsalong the coastfor becomescloser to the leading edge of the North American the past -45 years, from 1941 to the period 1987-1988 plate (Figure lb) (Table 1). (Figure4) [Mitchell et al., 1991, 1992, this issue]. The trends The most abruptlatitudinal change in averagetopography in relative bench mark elevation changesbetween leveling occurs above the subductedBlanco Fracture Zone (Figure 3). surveys(Figure 4) are probablyreal and of tectonicorigin becauseanalysis of tidal recordsalong the samesegment of coastyields the samemagnitudes and trendsin uplift. The Table 1. Age of the SubductedJuan de FucaPlate as tidal gage records and the leveling surveyrecords are two Projectedthrough the Overlying Pacific Plate to the Crest of the CoastRanges independentdata sets; differencing the tidal records and differencing the leveling records yield similar relative differencesin uplift rate betweencoastal localities [Mitchell et Latitude* øN Chron •' Age, Ma al., this issue]. The geodeticallyderived uplift rates are referencedto a 48.0 5/5A 11.0 47.4 5A 12.0 contemporarysea level rise of 1.8 mm/yr [Douglas, 1991], 47.25 5B 15.3 with no correctionfor post glacialrebound. In magnitudeand 47.0 5D 17.86 variability, the uplift rates are severaltimes larger than the 46.75 5D 17.86 predictedmagnitude of present-daypost glacialrebound, and 46.5 5D 17.86 the regional variation of uplift rate is of a much shorter 46.25 5D 17.86 wavelengththan that predictedfor present-daypost glacial Pseudo Fault rebound [Mitchell et al., this issue]. These geodetically 46.0 5C 16.60 deriveduplift ratesare upliftrates of rockrelative to the geoid 45.75 5D 17.86 and thus are directly comparableto the shoreplatform uplift 45.5 5D 17.86 rates, described below. 45.25 5D 17.86 45.0 5D 17.86 44.75 5D 17.86 Uplift Ratesof ShorePlatforms 44.50 5D/5E 18.2 44.25 5D/5E 18.2 Uplift ratesare calculatedfor upliftedshore platforms in the 44.0 5E 18.8 latituderange 42 ø to 45øN. The magnitudeof uplift,relative to 43.75 5E 18.8 presentsea level, wasdetermined for thejunction of the paleo- 43.5 5E 18.8 seacliff and the platform(the shorelineangle). Uncertainties 43.25 5E 18.8 in calculationof the magnitudeof uplift of the shorelineangle 43.0 5E 18.8 are (1) the altitude of the shorelineangle at the time of its 42.75 5E 18.8 formation relative to altitude of the eustatic sea level high 42.5 5E 18.8 stand that eroded the shoreline angle (accordingto Wright 42.25 5E 18.8 [1970] and Trenhaile [1980], the two elevationsare the same, 42.0 5E 18.8 Blanco Fracture Zone + 2 m); (2) altitudeof paleo-sealevel high stand,this altitude 41.75 4 7.41 beinga functionof the ageof the high standand the sealevel 41.5 5E/4 8.5 model employed (see below); (3) present-dayaltitude of shoreline angle, where the error in assignedaltitude is a *Latitudeposition is alongcrest of CoastRanges function of surveying accuracy; we used 7.5 arc min •- Chron on the Juan de Fuca plate determined from topographicmaps for assigningaltitude, where the erroris +6 reflection of chronsfrom Pacific plate to Juan de Fuca plate m, or one half the contour interval; and (4) original seaward acrossthe spreadingcenter. tilt of the platform;this uncertaintyonly appliesin casesof 12,250 KELSEY ET AL.: TOPOGRAPHIC FORM OF COAST RANGES

Geodetic data

& onshore projection of Wecoma Tillamook Head Fault BayYaquinaFault

Cape Ferrelo && Cape Blanco (i• CapeArago oo o Horence Marine terrace data & & & &

42 43 44 45 46 Latitude,degrees north Figure 4. Latitudinalvariation in uplift ratesderived from resurveyedbench marks between 1941 and the 1987-1988period, and from sea cliff/platformjunctions of 80-125 ka shoreplatforms. Triangles, resurveyedbench marks; circles, shore platform data. poor exposurewhere platform altitudeis measuredat a point model [Muhs et al., 1992], rather than the New Guinea sea seawardof the shoreline angle, necessitatingcalculation of level high standmodel [Bloomet al., 1974; Chappell and shoreline angle elevation using the original slope of the Shackleton, 1986]. If the New Guinea model is more platform [Bradley and Griggs, 1976]. appropriatefor the west coastof North America,uplift rates Uplift rates can only be determinedat localitieswhere the would be systematicallyhigher for surveyedpoints on the 80 shoreline angle elevation can be located on a terrace of a ka and 105 ka platformsbut not for points on the 125 ka known age. The agesof the late Quaternaryeustatic sea level platform. The maximumincrease in upliftrates would be about high stands are well established at 80, 105, and 125 ka 0.1 mm/yr. Givenall the aboveuncertainties in themagnitude [Mesolella et al., 1969; Bloom et al., 1974; Chappell and of uplift of a shorelineangle formed at 80, 105, or 125 ka, the Shackleton,1986]. The 80 ka and 105 ka wave-cutplatforms maximum error in the uplift rate calculation is about +0.2 were identified based on terrace correlation, using soil mm/yr. development[Bockheim et al., 1992] and altitudinalsurveys, Shore platform uplift rates are for the most part <0.25 to terracesof known age in the Cape Blanco[Kelsey, 1990] mm/yr and have a rangeof-0.04 to 0.87 mm/yr (Figure4). and Cape Arago [Mclnelly and Kelsey, 1990] areas. Marine Abruptchanges in shoreplatform uplift rate,in somecases by terracesediments of known age have age assignmentsbased almostan orderof magnitude,occur at severallocalities along on a radiometricage from U-seriesdating of coral and amino- the Cascadiamargin. All abruptchanges in uplift rate canbe acid racemization-ratiocorrelation ages using molluscs (Muhs relatedto a distinctstructure, a fault or a fold, that displaces et al., 1990). In centralcoastal Oregon, 125 ka platformages the shore platform. Uplift rates abruptly change across were assigned[Ticknor, 1993] eitherbased on terracesequence structuresnear Cape Ferrelo (CF, Figure lb and Figure 4) (the terracein questionbeing the next one altitudinallyabove [Kelsey and Bockheim,1994], Cape Blanco(CB, Figure lb the 105 ka terrace)or basedon correlationwith a 125 ka age- andFigure 4) [Kelsey,1990], CapeArago (CA, Figurelb and assignedwave-cut platform near Yaquina Bay, Oregon Figure4) [Mclnelly and Kelsey,1990] and Newportffaquina [Kennedyet al., 1982]. Bay (YB, Figure lb andFigure 4) [Ticknor,1993]. We usedpaleo-sea level high standelevations as determined The platformoffsets near Newport (Figure 4) are part of a from the coastal California sea level high stand elevation regionally extensive upper plate fault zone at this latitude, KELSEY ET AL.: TOPOGRAPHIC FORM OF COAST RANGES 12,251

which includes a set of west-northwesttrending faults on the Rock uplift of the core rocks of the Olympics since 14 Ma is continental rise and shelf [Goldfinger et al., 1992]. These of the order of 1 km/m.y., as derived from fission track faults project onland to structuresthat offset or fold the shore analyseson zircon [Brandon and Vance, 1992]. At the locus platforms. The platform offset north of Newport (Figure 4), of late Neogenedomal uplift of the Klamaths,the magnitudeof near Cape Foulweather (CFW, Figure lb), is at the site of rock uplift is at least 7 km [Mortimer and Coleman, 1985] and onshoreprojection of the Wecoma Fault of Goldfinger et al. commenced10-5 Ma, yielding rock uplift rates of 0.7-1.4 [1992]. At Yaquina Bay, the 125 ka platform is offset 75 m by km/m.y. the Yaquina Bay Fault (Figure 4) [Ticknor, 1993]; this fault Surface uplift rates in both mountain systemsare less than has the largest componentof vertical slip, 0.6 m/kyr, of any rock uplift rates because of erosion. Differencing the active fault in coastalOregon or Washington. maximum elevation profile from the average topographic profile in Figure 3 (Figure 6), the elevationdifferences for the Olympics and the Klamaths are similar to each other and Discussion significantly greater than elevation differences in the Variation in Average Topography as a Function intervening areas. Assuming all of the Coast Ranges have of Age of the Underlying Subducted Plate been subjectto erosionin the late Neogeneand Quaternaryand thus have the same surface age, surface processeshave Average topographyis highestwhere the relative elevation removedapproximately the samemass per unit area of material of the subducted plate, calculated based on lithospheric from both the Olympics and the Klamaths, despitedifferences subsidence[Parsons and Sclater, 1977], is highest. We infer in age and composition of the rocks in the two regions. from this circumstantialrelationship that uplift of the Coast Further, the mass per unit area of material removed in the Rangesis in part a function of the relative age and densityof intervening Coast Ranges is considerably less (Figure 6). the underlyingsubducted slab of the Juande Fuca/Gordaplate. Though the Klamaths are composedof relatively "old" rocks Further, the high topography of the Klamath Mountains is compared to the Olympics, they are not, relative to the offset to the north, relative to the position of the Blanco Olympics, an old mountain range. Because the mass of Fracture Zone (Figure 3 and Figure 5a versus 5b), and it is material removed by erosion is greater, not less, in the offset in the direction of migration of the fracture zone as Klamaths and the Olympics than in the intervening ranges, dictated by relative plate motions [Atwater, 1970; selective erosion cannot accountfor the high topography. A Engebretson et al., 1985]. We suggest the northward higher rate of rock uplift, not contrastingrock ages or greater movementof the fracturezone servesas a submergedbouyant rock resistanceto erosion, is the major factor influencing the plow dynamically uplifting crust in front of it. Movement is location of maximum topography. at the rate of the margin-parallel component of plate We offer a brief comment on uplift mechanisms, convergenceof the Juan de Fuca plate (North America fixed), recognizing that our data provide insufficient evidence for which is presently 24 _+ 10 mm/yr at latitude 44øN discussing the issue in detail. Though we infer that the [Riddihough, 1984]. McNutt [1983] also related subductionof magnitude of rock uplift in the Klamaths and Olympics is the Blanco Fracture Zone to the elevation of the Klamath similar for the late Neogene, the mechanismsof uplift are Mountainsbased on an observedanomalous isostatic gravity probably different. Brandon and Calderwood [1990] and high associatedwith the Klamath and SiskiyouMountains in Brandon and Vance [1992] suggestthat importantmechanisms northern California. She proposedthat the subductedBlanco of uplift for the Olympic Mountains include accretionof late FractureZone first passedbeneath northern California near the Oligoceneand early Miocene sedimentsto the undersideof the end of the Miocene, accountingfor both the creation of the subductedslab and a possiblearch in the slab, with maximum observed gravity anomaly and for the rapid rise of the curvature under the Olympics. No commensuratelyspecific Klamaths at that time [Diller, 1902; Mortimer and Coleman, mechanismof uplift has been reportedfor the Klamaths. The 1985]. mechanism is likely to be different, however, because the Klamaths lie adjacent to the southernmostCascadia margin, Generation of High Topography: Role of Rock where the slab is not arched, and tens of kilometers of oceanic Uplift Rate Versus Rock Age Versus crust have not been accreted under the Klamaths in the late Differential Erosion Rate Neogene. Any proposeduplift mechanismfor the Klamaths From our data, we infer that the higher parts of the Coast shouldincorporate the differential subsidenceof the subducted Ranges,the Klamath and Olympic Mountains,are high in part slab under the north end of the Klamaths and the fact that the becausethey are dynamicallysupported by youngersubducted underlyingsubducted crust is relatively young. Only the latter oceanic crust. However, what is the relative importance of attribute, presence of young subductedcrust, is common to differential rock uplift, rock age, and selectiveerosion in the both the Olympics and the Klamaths. generationof the areasof highesttopography? The Olympics Finally, the integrateduplift history for the Coast Ranges are composedof much younger and less metamorphosedcrust is not one of significant vertical displacements of the than the Klamaths. The corerocks of the Olympicsare middle Olympics and the Klamaths alone. While the vertical Eocene to early Miocene sedimentsand volcanics [Tabor and displacementsof the Olympics in the last 14 m.y. may have Cady, 1978; Brandon and Vance, 1992], whereas the core been of the order of 12 km [Brandon and Vance, 1992], the rocks of the Klamaths are Paleozoic and Mesozoic Coast Ranges in southern Washington and in Oregon also metamorphosedsediments and plutonic rocks [Irwin, 1966; require significant though more modest vertical Hotz, 1971]. displacements,because these crustal features have positive Estimates of late Neogene and Quaternaryrock uplift for topographyyet are largely composedof Eocene and younger these two ranges are similar within an order of magnitude. marine strata [Walker and Macleod, 1991]. 12,252 KELSEY ET AL.: TOPOGRAPHIC FORM OF COAST RANGES

600 . 600 400• Relativeelevation oftop of 400 20 subductedplate• 200 0 -200 '! ...... • .... , .... , .... , .... -200 41 42 43 44 45 46 47

1000 1000

800 /• Averagetopography 800 (forwidth =.60' of longitude 600 6OO

400 400

200- 200

5 c Resurveyed highway benchmarks E . 3 • 2' •.1- 0

-1

E. 0.5 formed80-120 ka 0.5 • o o

:::3-0.5 ...... , .... , .... -0.5

Latitude, degrees north E t t t tt Cape Yaquina Cape Blanco BayFault Ferrelo Cape onshoreprojection, Arago WecomaFault

Figure 5. Data setsfor platerelief, topography,and vertical crustal velocity, correlated by latitude. (a) Relativeelevation of topof subductedJuan de Fuca/Gordaplates. (b) Averagetopography. (c) Upliftrates from resurveyedbench marks. (d) Uplift ratesfrom shoreplatforms 80-125 ka in age. (e) Locationof coastalsites where abrupt changes in upliftrate occur due to upperplate faulting or folding.

1200

andaverage topographic elevation, I• 1000800 • Differenceinelevation betweencrest elevations r• • 600

400

200

Latitude,degrees north Figure 6. Differencein elevationbetween crest elevations and correspondingaverage topographic elevationfor the CoastRanges, extending from the KlamathMountains on the southto the Olympic Mountains on the north. KELSEY ET AL.: TOPOGRAPHIC FORM OF COAST RANGES 12,253

Average Topography Versus Vertical Crustal Table 2. AverageVertical Velocitiesof CrustMeasured Velocities at the Coast at the Coast,Based on Resurveysof Highway Bench Marksand on UpliftedShore Platforms The two sets of vertical crustal velocities, those derived from resuveyedbench marks and thosederived from uplifted LatitudeRange VuM,* VpLAT,? (VpLAT)(VBM)-1õ shoreplatforms, represent velocities averaged over markedly øN mm/yr mm/yr different lengthsof time (45 yearsversus 80-125 kyr). The crustal velocities derived from bench marks probably record 42 - 42.5 3.9 0.30 0.08 elastic strain accumulationin the interseismicperiod between >42.5 - 43 4.4 0.26 0.06 subductionevents [Weldon, 1991; Mitchell et al., this issue]; >43 - 43.5 3.8 0.23 0.06 thus much of the strain recordedby the uplift of benchmarks >43.5 - 44 2.2 0.03 0.01 >44 - 44.5 0.8 0.05 0.06 (Figure 4) may be recovered in the next subductionzone >44.5 - 45 0.07 0.34 4.9 earthquake. The vertical crustal velocities derived from shoreplatforms are long-termvelocities that integratethe effectsof as many *Averagevertical velocity at coast,derived from bench mark resurveys,1941 to ---1986. as a few hundredelastic strain-accumulation-and-releasecycles ?Averagevertical velocity at coast,derived from 80, 105, on the subductionzone. Long-term vertical velocities of or 125 ka uplifted shoreplatforms. platforms bear a definable relation to averagetopography, {}Ratio of vertical velocities derived from platforms to though smoothcross-latitude trends in the platform vertical vertical velocities derived from bench mark resurveys, velocities are not apparentbecause the localized folds and averagedfor 5 arc min latitudinalincrements. faults cause abrupt changesin platform uplift rate. Where platform vertical velocities are variable but in general the Coast Ranges. The conjecturethat permanentrock uplift relatively high, in the latitude range 420-43.3ø , average is related to interseismic uplift is consistent with the topography is higher (Figure 5). In contrast, negligible observation that those segments of the subduction zone vertical platform velocities between latitudes 43.30-44.6ø correspondto lower averagetopography over the samelatitude subject to greater interseismicuplift rates are latitudinally related to those segments of the Coast Ranges that have range. The segmentof the CoastRange with the lowestand least variable average topographyis identical in latitudinal highermagnitudes of rock uplift over the long term. rangeto that segmentof the coastwhere platform uplift rate is Vertical Crustal Velocity Anomalies in Central uniformlylow (_• 0.1 mm/yr). Coastal Oregon: A Casacadia Segment Boundary? Trends in averagetopography are correlativelatitudinally to trends in vertical crustal velocities derived from bench Centered at 44.6-44.8øN is an anomalous area of the Coast marks. The south-to-north descending trend in these two Ranges where the average topographyis low, bench mark variables is especially notable in Oregon, though the uplift is negligible, and platform uplift rates are in part descending trend of the bench mark velocities between exceptionallyhigh (Figure 5b versus5c versus5d). Bench latitudes 430-45ø is offset to the north by about 1-1.5ø of mark uplift rates reach their lowest magnitudejust north of latitude relative to the descending trend in Coast Range Newport,Oregon (Figure 4). In the sameplace, shore platform elevationfrom 41.5ø44 ø (Figure 5b versus5c). This offset is uplift rates are anomalouslyhigh becauseof late Quaternary in the same direction as that of the north edge of high movementon the YaquinaBay fault and the onshoreextension topographyof the Klamath Mountains,which is offset north of faultingassociated with the Wecomafault of Goldfingeret relativeto the underlyingposition of the BlancoFracture Zone al. [1992] (Figures4, 5d and 5e). (Figure5a versus5b). On the basisof the benchmark uplift rate data, we infer that The magnitudeof the differencein verticalcrustal velocities elastic strain accumulationis not occurringat this time on the derived from the geodetic data and the platform data is upperplate in the YaquinaBay region [Mitchell et al., this significant,being as largeas 4.5 mm/yr (Figure5c versus5d). issue]. The fact that the average topographyof the Coast This differenceleads us to two pointsof conjectureabout the Rangesis at one of its lowest points at this same latitude relationshipbetween recoverable and permanentstrain in the suggests to us that anomalously low rates of strain CoastRanges. First, as notedabove, the differencein vertical accumulation at this latitude may have been the norm crustalvelocities derived from the geodeticdata and from the throughoutthe late Neogene and Quaternary. If so, the platform data signifiesthat shortterm strain accumulationis Yaquina Bay region of the centralOregon coast,with its transient and will probably be released as a megathrust concentrationof upperplate faulting[Golfinger et al., 1992; earthquake. Second,platform uplift rates are a fraction of Ticknor, 1993], may be a diffuse segmentboundary with benchmark uplift rates (column4, Table 2) in all cases,save regardsto subductionzone earthquakes. Persistently low rates the exceptiondiscussed below. If present-daybench mark of strain accumulationmay impede northwardor southward uplift rates are typical of the averageinterseismic uplift rate propagatingplate rupture. On the basisof geologicevidence and if platformuplift rates are typicalof the long-termuplift associated with salt marsh coseismic subsidence, Nelson and rate, then this fraction denotesthe approximatepercent of Personius [1994] also suggestthat a segmentboundary may uplift, which accumulatesbetween subduction events, that is occuralong the Cascadiasubduction zone near 44-45øN. not recoveredby movementon the plate interface during a subductionearthquake. Averagingby 30 arc rain increments Conclusions the vertical velocity data available (Table 2), about 5% of interseismic uplift (range is 1-8%) would not be recovered We cannotdiscount the hypothesisthat the latitudinaltrend coseismicallyand wouldcontribute to permanentrock uplift of in averagetopography of theCoast Ranges is a functionof the 12,254 KELSEY ET AL.: TOPOGRAPHIC FORM OF COAST RANGES

latitudinal change in attributes that are related to the variables that spatially correlate with differential uplift of subduction process. On the basis of data and discussion rock along the north-southtrend of the Coast Ranges. From above, we make the following conclusions. Elevation of the this we infer that the variable density of the subductingplate average topographyof the Coast Rangesappears to be in part affects the pattern of strain buildup and releasealong the plate a function of the age and density of the subductedplate that boundary, which in turn is manifest by changes in underlies the Coast Ranges. Where the plate is young and topographicform along the trendof the CoastRanges. relatively bouyant, the surface of the overlying Coast Range is higher. Acknowledgments. This researchpartially supportedby NSF grant Underneath the Coast Ranges, a discontinuityin age of the 8903470 to H. Kelsey and J. Bockheim. Engebretsonacknowledges subducted slab occurs at the Blanco Fracture Zone. Because supportfrom a Joint Venture (JOVE) grant from NASA. Mitchell topographyis sensitive to the age and relative bouyancyof acknowledgeshelpful consultationwith L. Riggersand E. Balazsof the National GeodeticSurvey and funding from the National Earthquake the underlyingslab and the Blanco FractureZone is migrating HazardReduction Program of the USGS. We thankR. J. WeldonII for along the margin at the speed of the margin-parallel helpful discussionand M.T. Brandon,P. Bodin, and D. Merritts for componentof convergenceof the Juan de Fuca plate, the locus thoughtfulreviews. of the greatest amount of rock uplift in the southernCoast Ranges is migrating north with time. Latitudinal trends in average topography are similar to References latitudinal trends in vertical crustal velocity for the last 45 Atwater,T., Implicationsof plate tectonicsfor the Cenozoicevolution years, a period of interseismic strain accumulation in of North America, Geol. Soc. Am. Bull., 81, 3513-3536, 1970. Cascadia. A marked descentto the north in both topography Atwater,T. and J. Severinghaus,Tectonic maps of the Pacific,in The and vertical crustal velocity, based on resurveyed bench Geologyof North America,vol. N, The EasternPacific Oceanand marks, is evident in the data. The descendingtrend in vertical Hawaii, Plate 3B, GeologicalSociety of America,Boulder, Colo., crustal velocity is offset by about one-and-halfdegrees of 1989o latitude to the north relative to the topographictrend, the Bloom, A., W.S. Broecker,J. Chappell,R.K. Matthews,and K.J. offset also being in the direction of the margin-parallel Mesolella, Quaternarysealevel fluctuations on a tectoniccoast: new component of plate convergence. 230Th/234Udates from the Huon Peninsula, New Guinea, Quat. Res., 4, 185-205, 1974. For the latitude range 42o-45ø (a 340-km segmentof the Bockheim,J.G., H.M. Kelsey,and J.G. Marshall,Soil development, Cascadiamargin), vertical crustalvelocity data are available relativedating and correlationof late Quaternarymarine terraces in for 80-125 ka shoreplatforms. For this segmentof the Coast southwesternOregon, Quat. Res., 37, 60-74, 1992. Ranges, average topographyis relatively low where vertical Bradley, W.C., and G.B. Griggs, Form, genesis,and deformationof crustal velocities are low, and average topographyis higher centralCalifornia wave-cut platforms, Geol. Soc.Am. Bull., 87, 433- where crustal velocities at the coast are variable but on the 449, 1976. average significantly higher. In general, averagetopography Brandon,M.T. and A.R. Calderwood,High-pressure metamorphismand is also higher where faulting and folding of shoreplatforms is uplift of the Olympicsubduction complex, Geology, 18, 1252-1255, most evident. 1990. Differential erosion in responseto rock units of varying Brandon, M.T., and J.A. Vance, Tectonic evolution of the Cenozoic resistancecannot account for the major differencesin average Olympic subductioncomplex, Washington state, as deducedfrom fission track ages for detrital zircons, Am. J. Sci., 292, 565-636, Coast Range topography. Rather, the Coast Ranges have 1992. been subjectto different amountsof rock uplift alongthe trend Chappell,J., and N.J. Shackleton,Oxygen isotopes and sealevel,Nature, of the subductionmargin. The mechanismsof uplift for the 234, 137-140, 1986. topographically highest portions of the Coast Ranges, the Diller, J.S., Topographicdevelopment of the KlamathMountains, U.S. Olympicsand Klamaths,are not the same;but the presenceof Geol.Surv. Bull., 196, 60 pp., 1902. young, relatively bouyant subductedcrust under both of these Douglas,B.C., Global sea level rise, J. Geophys.Res., 96, 6981-6992, regions is an integral componentof the uplift mechanismsin 1991. both of these areas. Embley,R.W., and D.S. Wilson, Morphologyof the Blancotransform Differential total rock uplift best correlatesto age of the fault zone- NE Pacific:Implications for its tectonicevolution, Mar. underlyingplate but also correlatesto vertical crustalvelocity Geophys.Res., 14, 24-45, 1992. Engebretson,D.C., A. Cox, and R.G. Gordon,Relative motions between reflecting interseismicstrain accumulation. Topographically oceanicand continentalplates in the Paciifcbasin, Geol. Soc.Am. higher portionsof the Coast Rangesspatially correlate with Spec.Pap. 206, 59 pp., 1985. segmentsof the subductionzone where elastic strain buildup England, P., and P. Molnar, Surface uplift, uplift of rocks, and presentlyis greatest. We infer that a small percentageof the exhumationof rocks,Geology, 18, 1173-1177,1990. interseismicstrain that accumulatesas rapid short-termuplift Goldfinger,C., L.D. Kulm, R.S. Yeats,B. Applegate,M.E. MacKayand is not recoveredduring subductionearthquakes. This residual G.F. Moore, Transverse structural trends along the Oregon strain contributes to permanent rock uplift of the Coast convergentmargin: Implicationsfor Cascadiaearthquake potential Ranges, with commensurately more rock uplift where andcmstal rotations, Geology, 20, 141-144, 1992. interseismicstrain is greatest. Coast Range topographicform Hotz, P.E., Plutonic rocks of the Klamath Mountains, California and may in part be dependenton the persistanceof this relation for Oregon,U.S. Geol. Surv.Prof. Pap., 684B, 20 pp., 1971. Irwin, W.P., Geologyof the KlamathMountains Province, Bull. Calif. million of years. Div. Mines Geol., 190, 19-38, 1966. In summary,topographic form of the CoastRanges could be Kelsey, H.M., Late Quaternarydeformation of marine terraceson the in large part a function of the subductionprocess. Variations Cascadiasubduction zone near Cape Blanco,Oregon, Tectonics, 9, in the age of the subductingplate under the CoastRanges and 983-1014, 1990. the varying character of interseismicstrain accumulationand Kelsey, H.M. and J.G. Bockheim,Coastal landscape evolution as a coseismicstrain release along the Cascadiamargin are both function of eustasy and surface uplift rate, Cascadiamargin, KELSEY ET AL.: TOPOGRAPHIC FORM OF COAST RANGES 12,255

southemOregon, Geol. Soc. Am. Bull., 106, 840-854, 1994. Earthquake Hazards in the Pacific Northwest, edited by A.M. Kennedy,G.L., K.R. Lajoie,and J.F. Wehmiller, Aminostratigraphy and Rodgers,W.J. Kockelmen,G. Priestand T.J. Walsh, U.S. Geol. Surv. faunal correlations of late Qu•.Lemarymarine terraces, Pacific Prof. Pap., in press,1994. Coast,USA, Nature, 299, 545-547, 1982. Parsons,B., and J.G. Sclater,An analysisof the variationof oceanfloor Mclnelly, G.W., and H.M. Kelsey, Late Quaternary tectonic bathymetryand heat flow with age, J. Geophys.Res., 82, 803-838, deformationin the Cape Arago-Bandonregion of coastalOregon as 1977. deducedfrom wave-cutplatforms, J. Geophys.Res., 95, 6699-6713, Riddihough,R., Recentmovements of the Juande Fucaplate system,J. 1990. Geophys.Res. 89, 6980-6994, 1984. McNutt, M., Influenceof plate subductionon isostaticcompensation in Tabor,R.W., and W.M. Cady, Geologicmap of the OlympicPeninsula, northem Califomia, Tectonics,2,399-415, 1983. scale1:125,000, U.S. Geol. Surv.Map, 1-994, 2 sheets, 1978. Mesolella, K.J., R.K. Matthews, W.S. Broecker,and D.L. Thurber, The Ticknor, R.L., Late Quaternary crustal defomaation on the central astronomicaltheory of climaticchange: Barbados data, J. Geol., 77, Oregon coast as deducedfrom uplifted wave-cut platforms,M.S. 250-274, 1969. thesis,70 pp., West.Wash. Univ., Bellingham,1993. Mitchell, C. E., R.J. Weldon II, and P. Vincent, Active uplift of the Trenhaile, A.S., Shore platforms: a neglectedcoastal feature, Prog. Pacific northwestmargin (abstract),Eos Trans.AGU, 72 (44), Fall Phys. Geog., 4, 1-23, 1980. Meetingsuppl., 314, 1991 Walker, G.W., and N.S. Macleod, Geologicmap of Oregon,1:500,000, Mitchell, C.E., R.J. Weldon II, and H.L. Pittock, Pacific northwest U.S. Geol. Surv., Reston, Va., 1991. contemporarycoastal uplift profile with confidencelimits and a Wells, F.G., and D.L. Peck, Geologicmap of Oregonwest of the 121st revisedrate of sealevel changeat Astoria,OR (abstract),Eos Trans. meridian, scale 1:500,000, U.S. Geol. Surv.Map, 1-325, 1961. AGU, 73 (43), Fall Meetingsuppl., 119, 1992. Weldon, R.J. II, Active tectonic studiesin the United States, U.S. Natl. Mitchell, C. E., P. Vincent, R.J. Weldon II, and M.A. Richards, Present- Rep. Int. Union Geol. Geophys.1987-1990, Rev. Geophys.,29, 890- day vertical defomaationof the Cascadiamargin, Pacific northwest, 906, 1991. United States,J. Geophys.Res., this issue. Wilson, D.S., A kinematic model for the Gorda deformation zone as a Mortimer, N., and R.G. Coleman, A Neogene structuraldome in the diffuse southemboundary of the Juan de Fuca plate, J. Geophys. KlamathMountains, Califomia and Oregon,Geology, 13, 253-256, Res., 91, 10259-10269, 1986. 1985. Wilson, D.S., Kinematicsof overlappingrift propagationwith cyclicrift Muhs, D.R., H. M. Kelsey, G. H. Miller, G. L. Kennedy, J. F. Whelan, failure, Earth Planet. Sci. Lett., 96, 384-392, 1990. and G.W. Mclnelly, Age estimates and uplift rates for late Wright. L.W., Variation in the level of the cliff/shoreplatform junction Pleistocene marine terraces: southem Oregon portion of the alongthe southcoast of GreatBritain, Mar. Geol., 9, 347-353, 1970. Cascadiaforearc, J. Geophys.Res., 95, 6685-6698, 1990. Muhs, D.R., T.K. Rockwell, and G.L. Kennedy,Late Quatemaryuplift D.C. Engebretson and R.L. Ticknor, Department of Geology, rates of marine terraces on the Pacific coast of North America, WestemWashington University, Bellingham, WA 98225. (Engebretson, southemOregon to Baja Califomia Sur, Quat. Int., 15/16, 121-133, e-mail: [email protected]. wwu.edu) 1992. H.M. Kelsey, Departmentof Geology,Humboldt State University, National GeophysicalData Center, ETOPO-5 bathymetry/topography Arcata, CA 95521. (e-mail: [email protected]) data, Data Announce. 88-MG-02, Nat. Oceanic and Atmos. Admin., C.E. Mitchell, Departmentof GeologicalSciecnes, University of Washington,D.C., 1988. Oregon,Eugene, OR 94703. Nelson,A.R., and S.F. Personius,The potentialfor greatearthquakes in Oregon and Washington-An overview of recent coastal geologic studies and their bearing on segmentationof Holocene ruptures, (ReceivedApril 5, 1993;revised October 4, 1993; central Cascadia subduction zone, in Assessing and Reducing acceptedNovember 15, 1993.)