Land-level changes from a late Holocene earthquake in the northern Puget Lowland, Washington

Harvey M. Kelsey Department of Geology, Humboldt State University, Arcata, California 95521, USA Brian Sherrod U.S. Geological Survey, Department of Earth and Space Sciences, University of Washington, Seattle, Washington 98195-1310, USA Samuel Y. Johnson U.S. Geological Survey, Pacific Science Center, 1156 High Street, Santa Cruz, California 95064, USA Shawn V. Dadisman U.S. Geological Survey, 600 4th Street, St. Petersburg, Florida 33701, USA

ABSTRACT show a geomorphic and structural setting sim- An earthquake, probably generated on the southern Whidbey Island fault zone, caused ilar to that in the offshore bathymetry and 1±2 m of ground-surface uplift on central Whidbey Island ϳ2800±3200 yr ago. The cause seismic stratigraphy. Light detection and rang- of the uplift is a fold that grew coseismically above a blind fault that was the earthquake ing (lidar) imagery from central Whidbey Is- source. Both the fault and the fold at the fault's tip are imaged on multichannel seismic land shows a pronounced lowland in which refection pro®les in Puget Sound immediately east of the central Whidbey Island site. Hancock marsh is at the lowest point, situated Uplift is documented through contrasting histories of relative sea level at two coastal on a northeast projection of the syncline (Fig. Similarly, the bathymetric high in Holmes .(2 ؍ marshes on either side of the fault. Late Holocene shallow-crustal earthquakes of Mw 6.5±7 pose substantial seismic hazard to the northern Puget Lowland. Harbor (between shotpoints 700 and 800, Fig. 2) projects to the upland that rises abruptly Keywords: paleoseismology, relative sea level, blind faults, seismic re¯ection, Puget Sound. just to the north of Hancock marsh. The Han- cock marsh lowland is bordered by steep (10 INTRODUCTION (Fig. 1). Taken together, these studies indicate percent) slopes to the north and gradual slopes Late Holocene earthquakes accompanied by that the northern Puget Lowland was host to to the south (Fig. 2), similar to the asymmetry ground-surface displacement indicate a signif- several shallow-crustal, surface-deforming of the folded Pleistocene sediment offshore to icant seismic hazard in the central and south- magnitude 6.5±7 earthquakes in the past 3000 the southeast. ern Puget Lowland of Washington. Earth- yr. Although the Pleistocene section appears to quakes with surface rupture or regional be faulted, no late Holocene fault scarp is ev- coseismic uplift, associated with the Seattle LATE QUATERNARY TECTONIC ident on the lidar image in the area where a and Tacoma fault zones (Fig. 1), have affected DEFORMATION ON CENTRAL fault would propagate to the surface (Fig. 2). the Seattle and Tacoma urban region approx- WHIDBEY ISLAND We infer therefore that if a fault was active in imately three or four times in the late Holo- Multichannel seismic re¯ection data in the late Holocene, surface deformation would cene (Johnson et al., 1999; Brocher et al., Holmes Harbor immediately east of central be folding above a blind fault tip. 2001; Blakely et al., 2002; Nelson et al., Whidbey Island (Fig. 2) prompted us to in- 2003). vestigate the possibility of late Holocene fault- RESEARCH APPROACH In contrast, less is known of the late Ho- related ground-surface deformation. Identi®- If a fold in the study area above a fault tip locene activity of faults in the northern Puget cation of the Pleistocene section is derived has been active in the late Holocene, raising Lowland, a region that has not been subject to from seismic facies analysis, projection from the upland to the north relative to the Hancock damaging earthquakes in historic time (Lud- nearby boreholes (e.g., Standard Engstrom, lowland, then relative sea levels at Hancock Fig. 2), and iterative correlation and mapping marsh should be different from those recorded win et al., 1991). Northern Puget Lowland from a large suite of U.S. Geological Survey in a coastal marsh north of the projection of faults include the southern Whidbey Island, and industry seismic re¯ection pro®les (John- the fault at the surface. The Crockett marsh Utsalady Point, Strawberry Point, and Devils son et al., 1996, 2001a, 2001b; Rau and John- wetland is ϳ1 km north of the projected fault, Mountain fault zones (Johnson et al., 1996, son, 1999). The seismic data show a promi- and the Hancock marsh wetland is ϳ1km 2001a) (Fig. 1). The southern Whidbey Island nent asymmetric syncline, steeper to the north, south of the projected fault (Figs. 1B and 2). fault zone, although lacking a Holocene sur- developed in Pleistocene sediment. Holocene Barring any Holocene fault displacement be- face trace on Whidbey Island, is a major strata at the top of the section also appear to tween the wetlands, they should have identical basin-bounding structure that has a pro- dip to the south. In the Pleistocene section, the relative sea-level histories because they are nounced gravity and aeromagnetic signature strata thicken in the fold axis and thin on the close (8 km) to each other. Any glacio- and deforms Quaternary sediment visible on limbs, suggesting that the fold has been grow- isostatic in¯uence on relative sea level should offshore seismic re¯ection pro®les (Johnson et ing in the Quaternary. Along the pro®le of the be re¯ected identically in both wetlands, and al., 1996, 2001b; Blakely and Lowe, 2001). section, the modern bathymetry mimics the any tectonic in¯uence on relative sea-level re- The Utsalady Point fault in northern Whidbey underlying structure; a bathymetric high in sulting from strain accumulation and release Island has late Holocene fault scarps, and Holmes Harbor overlies the uplifted northern on the underlying subduction zone similarly trenching investigations indicate one or two fold limb. We infer that the syncline is cut by should result in identical relative sea-level earthquakes in the late Holocene (Johnson et a steep, north-side-up fault on the basis of perturbations at both sites. Therefore, different al., 2003). In this paper we provide data that truncation of seismic re¯ections and inferred relative sea-level histories implicate late Ho- suggest that an earthquake occurred ϳ3000 yr offset of the base of the Quaternary section. locene tectonic displacement caused by fold- ago on a buried fault beneath central Whidbey The deformed Pleistocene marine section is ing above an active fault (Fig. 2). Island, resulting in 1 m or more of vertical, ϳ2±4 km southeast along regional strike from If relative sea-level curves for the two north-side-up displacement. The fault is likely the narrow central neck of Whidbey Island, marshes are not the same, then the divergence part of the southern Whidbey Island fault zone where both island topography and geology of the two curves would indicate the timing

᭧ 2004 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or [email protected]. Geology; June 2004; v. 32; no. 6; p. 469±472; doi: 10.1130/G20361.1; 4 ®gures; Data Repository item 2004081. 469 Figure 1. A: Faults in Puget Lowland of northwest Washington and southern Vancouver Island, modi®ed after Johnson et al. (1996, 2001b). B: Three main fault traces of southern Whidbey Island fault zone, adapted from Johnson et al. (1996, 2001b). Crockett marsh and Hancock marsh straddle trace of northern fault strand. In area of incomplete seismic line coverage in Admiralty Bay, gap is shown in northern strand. U.S. Geological Survey (USGS) 1995 seismic re¯ection line shows fault in Admiralty Bay between Crockett and Hancock marshes (Johnson et al., 2001b). Earlier industry seismic re¯ection line east of USGS line does not show fault in Admiralty Bay (Johnson et al., 1996); however, industry line may not have extended far enough north to cross fault. Yellow rectangle represents area of Figure 2A. Two short white lines (labeled t) show locations of two transects in Figure 3. and magnitude of the vertical displacement pendicular to the barrier sand bar that sepa- across the wetland; cores were taken every 50 caused by folding. In order to determine sea- rates the Puget Sound (Admiralty Bay) from m until the dry upland was reached. level histories at the two marshes, cores were the wetland. Each of the two core transects taken in a transect across each wetland per- starts at the sand bar and proceeds inland GEOLOGIC CROSS SECTIONS AND RELATIVE SEA-LEVEL CURVES At each marsh, the geologic cross section (Fig. 3) indicates relative sea-level rise over time because the sand barrier seaward of each marsh has built upward during the late Holo- cene. As the barriers built upward, peat wet- lands aggraded behind them (Fig. 3). Paleo±sea-level index points de®ne a rela- tive sea-level curve at each site. Sea-level in- dex points are at the contact of the peat with the underlying substrate (Fig. 3), which is beach sand, sandy gravel, or a thin veneer of beach sand over Pleistocene glacial sediment. The sand-peat contact is a paleo±sea-level in- dicator because the underlying sand has ma- rine diatoms and the overlying peat has fresh- water and brackish-water plant macrofossils. The elevation of the sand-peat contact in the modern marsh, based on level surveys, is the same at each marsh within 0.2 m (see Fig. DR11) and is at the approximate level of mean high water (MHW) (Fig. 3). The sea-level curves for the two marshes (Fig. 4A) are based on the age and depth of paleo±sea-level index points relative to mod- Figure 2. A: Light detection and ranging (lidar) image of central Whidbey Island, location of offshore seismic line number P183 (irregular line in lower right with solid dots repre- senting every 100 shotpoints), and projection along regional strike of synclinal fold and 1GSA Data Repository item 2004081, Figure fault (dashed line) from seismic line. SEÐStandard Engstrom borehole (Rau and Johnson, DR1, survey data, and Table DR1, radiocarbon 1999). B: Multichannel seismic re¯ection pro®le in Holmes Harbor; data collected in 1995 ages, is available online at www.geosociety.org/ with processing described in Johnson et al. (2001a). TWTÐtwo-way traveltime. Three sets pubs/ft2004.htm, or on request from editing@ of distinctive solid black shapes track three different re¯ectors along trend of pro®le. C: geosociety.org or Documents Secretary, GSA, P.O. Same as B except uninterpreted seismic pro®le. Box 9140, Boulder, CO 80301-9140, USA.

470 GEOLOGY, June 2004 Figure 4. Top: Relative sea-level curves for Crockett and Hancock marshes. Bottom: Combined relative sea-level curves for Crockett and Hancock marshes, showing reconstruction of single sea-level curve (Hancock curve with superposed, displaced Crockett curve) if there had been no vertical displacement of Crockett vs. Hancock marshes. MLLWÐmean lower low water.

Figure 3. Cross sections of Crockett and Hancock marshes and stratigraphic data from two cores at Crockett marsh. Hancock marsh differs from Crockett marsh in that in Han- cock traverse, herbaceous peat is exposed in intertidal zone seaward of sand-barrier berm. amount of vertical displacement is 1.5 m Ϯ At Hancock, sand-barrier berm is underlain by peat because sand barrier ®rst built upward 0.5 m up to north. and prograded seaward; then, while sand barrier continued to build upward, barrier started to retreat landward, eroding peat that had been previously deposited behind it. At Crockett, Paleoecologic data from cores at the two sand barrier built upward and seaward as relative sea level rose. MHWÐmean high water; marshes provide corroborative evidence for MLLWÐmean lower low water. upward displacement at the Crockett marsh relative to the Hancock marsh ϳ3000 yr ago. At the Crockett marsh, a 2-cm-thick, clean, ern mean lower low water (MLLW). Accel- NORTH-SIDE-UP DISPLACEMENT ®ne sand layer occurs within the peat se- erator mass spectrometry 14C dating of plant 3000 YR AGO quence in core J ϳ25 cm above the sand-peat macrofossils (mainly seeds: Table DR1; see The two relative sea-level curves are not the contact; the sand layer contained 2870±3130- footnote 1) in the peat immediately above the same. Superposition of the Crockett and Han- yr-old detrital conifer leaves (Fig. 3). The thin contact provides age control. Age ranges in cock curves (Fig. 4B) shows that, for ages sand layer separates an underlying herbaceous Figures 3 and 4 are years before A.D. 1950 older than ϳ3200 yr, equivalent-age sea-level and detrital peat from an overlying peat con- and were converted from the laboratory- index points at Crockett are 1±2 m higher than taining woody detritus from a forested wet- reported 2␴ 14C age ranges by using the meth- at Hancock. Also, there is an ϳ700 yr interval land. Similarly, in adjacent core IJ, 25 m far- od of Stuiver et al. (1998) (Table DR1; see of time, after 3200 yr, when the Crockett site ther north, a contact of the same age as the footnote 1). The height and width of the rect- shows no change in relative sea level, or a sand layer separates an underlying herbaceous angles (data points, Fig. 4) express the mag- relative sea-level fall, whereas there is no pe- peat from an overlying peat containing coarse nitude of the error in locating the relative sea- riod of time at Hancock when relative sea lev- woody detritus (Fig. 3). At the time of depo- level curve in time versus depth space. The el is not rising. The two relative sea-level sition of the thin sand layer, the wetland at width of the rectangle is the age range of the curves in the past 2500 yr are similar within Crockett emerged suddenly, changing from a sand-peat contact (Fig. 3). The height of the limits of sea-level resolution. scrub-shrub wetland to a relatively drier for- rectangle is the uncertainty in elevation of the The apparent upward shift of the Crockett ested wetland at core site J and from a peat sand-peat contact, Ϯ0.3 m, calculated as the relative sea-level curves is caused by upward wetland to a forested wetland at core site IJ square root of the sum of the squares of the vertical displacement at Crockett relative to (Fig. 3). Although not convincing because of three variables that determine elevation rela- Hancock ϳ3200±2800 yr ago (Fig. 4). Recon- limited core observations, the thin sand may tive to MLLW: survey error (Ϯ0.01 m), tidal structing one relative sea-level curve common have been deposited by a tsunami generated measurement at site (Ϯ0.20 m), and variation to both sites requires vertically lowering the by sea¯oor displacement offshore in the Puget in elevation of the modern sand-peat contact Crockett curve onto the Hancock curve by Sound; the same displacement would have (Ϯ0.20 m). 1.0±2.1 m (Fig. 4), which implies that the produced the rapid emergence, and a change

GEOLOGY, June 2004 471 to drier site conditions, observed at the Crock- Pleistocene, and possibly Holocene, strata. structure in Puget Lowland, Washington: Re- ett wetland relative to the Hancock wetland. Folding appears to be at the tip of a high-angle sults from the 1998 seismic hazard investiga- tion in Puget Sound: Journal of Geophysical The drier environment at Crockett persisted fault that displaces the base of Quaternary Research, v. 106, p. 13,541±13,564. for ϳ700±800 yr, on the basis of 14C ages, at sediment. The axis of the syncline, projected Johnson, S.Y., Potter, C.J., Armentrout, J.M., Miller, which time the Crockett marsh again respond- 4 km northwest along regional strike to central J.J., Finn, C., and Weaver, C.S., 1996, The ed to relative sea-level rise through peat ag- Whidbey Island, underlies the Hancock marsh southern Whidbey Island fault: An active structure in the Puget Lowland, Washington: gradation. This ϳ750 yr interval is roughly coastal wetland. On the basis of relative sea- Geological Society of America Bulletin, equivalent to the time interval when relative level analysis of Hancock marsh in compari- v. 108, p. 334±354. sea level at Crockett was static or fell slightly son to Crockett marsh, a coastal wetland ϳ8 Johnson, S.Y., Dadisman, S.V., Childs, J.R., and (Fig. 4). km northwest across the projected trace of the Stanley, W.D., 1999, Active tectonics of Hancock marsh shows a profound change buried fault, Crockett marsh underwent a 1±2 the Seattle fault and central Puget Sound, WashingtonÐImplications for earthquake haz- in the diatom assemblages deposited ϳ3000 m abrupt uplift relative to sea level 2800± ards: Geological Society of America Bulletin, yr ago (352 cm depth, core R, Fig. 3), which 3200 yr ago at a time of unchanging relative v. 111, p. 1042±1053. is synchronous, within limits of the calibrated sea level at Hancock marsh. Because there is Johnson, S.Y., Dadisman, S.V., Mosher, D.C., radiocarbon ages, with the time of abrupt rel- no late Holocene fault scarp between the two Blakely, R.J., and Childs, J.R., 2001a, Active tectonics of the Devils Mountain fault and re- ative sea-level fall at Crockett marsh. Prior to marshes, we infer that the uplift is a result of lated structures, northern Puget Lowland and 3000 yr ago, the Hancock diatom assemblage coseismic reactivation of the buried fault and eastern Strait of Juan de Fuca region, Paci®c was dominated by the intertidal diatom ¯ora overlying fold evident in the re¯ection pro- northwest: U.S. Geological Survey Profes- Paralia sulcata and Trachyneis aspera.At ®les. The fault beneath Holmes Harbor is sional Paper 1643, 45 p. ϳ3000 yr ago, the assemblage became dom- within the mapped area of the southern Whid- Johnson, S.Y., Mosher, D.C., Dadisman, S.V., Childs, J.R., and Rhea, S.B., 2001b, Tertiary inated by Fragilaria construens, a cosmopol- bey Island fault zone; this fault zone is the and Quaternary structures of the eastern Juan itan diatom ¯ora that characteristically pio- likely source of the earthquake 2800±3200 yr de Fuca Strait: Point map, in Mosher, D.C., neers disturbed wetland sites. The change to ago. Our research demonstrates the utility of and Johnson, S.Y., eds., Neotectonics of the Fragilaria construens does not require a relative sea-level investigations to identify ac- eastern Strait of Juan de Fuca: A digital geo- logical and geophysical atlas: Geological So- change in relative sea level, but it does re¯ect tive blind faults. ciety of Canada Open File 3931, CD-ROM. a sustained change in site conditions, such as The earthquake ϳ3000 yr ago in central Johnson, S.Y., Nelson, A.R., Personius, S.F., Wells, a change in vascular ¯ora or slightly reduced Whidbey Island and the one to two late Ho- R.E., Kelsey, H.M., Sherrod, B.L., Okumura, salinity (Krammer and Lange-Bertalot, 1991). locene earthquakes related to the Utsalady K., Koehler, R., Witter, R.C., and Bradley, L., fault on northern Whidbey Island indicate that 2003, Evidence for one or two late Holocene earthquakes on the Utsalady Point fault, north- DISCUSSION the northern Puget Lowland is vulnerable to ern Puget Lowland, Washington: Geological The best explanation for the dissimilarity of surface-deforming, shallow earthquakes ac- Society of America Abstracts with Programs, relative sea-level curves is vertical tectonic companied by strong ground motions. Such v. 34, no. 7, p. 479. earthquakes pose signi®cant hazard to the 85- Krammer, K., and Lange-Bertalot, H., 1991, Bacil- displacement on a blind fault between the two lariophyceae: 3. Teil: Centrales, Fragilari- wetlands, a fault inferred to be part of the km-long northern Washington urbanized aceae, Eunotiaceae, in Ettl, H., et al., eds., southern Whidbey Island fault zone. The sur- coastal corridor extending from Everett, Swasser¯ora von Mitteleuropa, Volume 2/3: face deformation is a fold above the buried Washington, north to the Canadian border. Stuttgart, Jena, Gustav Fischer Verlag, 576 p. fault tip; the seismic re¯ection pro®le 2±4 km Ludwin, R.S., Weaver, C.S., and Crosson, R.S., 1991, Seismicity of Washington and Oregon, to the southeast along regional strike images ACKNOWLEDGMENTS B. Atwater and A.R. Nelson provided guidance in Slemmons, D.B., et al., eds., Neotectonics this fold (Fig. 2). The buried fault is situated and support for ®eld work and age determinations of North America: Boulder, Colorado, Geolog- within the approximately located northern in early stages of this project. Field assistance was ical Society of America Decade of North strand of the southern Whidbey Island fault provided by K. Burrell, C. Johnson, and K. Kelsey. American Geology Decade Map Volume, The manuscript was reviewed by A.L. Bloom, R.J. p. 77±98. zone (Figs. 1 and 2). Because the fault is bur- Nelson, A.R., Johnson, S.Y., Kelsey, H.M., Wells, ied, the fault trend, dip, and sense of slip are Blakely, T.M. Brocher, C.S. Weaver, and R.E. Wells. Access to Hancock marsh was provided by the Na- R.E., Sherrod, B.L., Pezzopane, S.K., Bradley, unclear. We infer from the asymmetry of the val Air Station Whidbey Island and access to L., Koehler, R.D., and Bucknam, R.C., 2003, folded section (Fig. 2) that the fault dips Crockett marsh was provided by Washington State Late Holocene earthquakes on the Toe Jam steeply to the north. If the fault dips 60Њ and Parks. This work was funded by National Earth- Hill fault, Seattle fault zone, Bainbridge Is- land, Washington: Geological Society of accommodated 1±2 m of vertical displace- quake Hazards Reduction Program award 00HQGR0067. America Bulletin, v. 115, p. 1388±1403. ment, then the net slip on the fault was a min- Rau, W.W., and Johnson, S.Y., 1999, Well stratig- imum of ϳ1.2±2.3 m. Net slip could have ex- REFERENCES CITED raphy and correlations, western Washington ceeded 2.3 m if there was a component of Blakely, R.J., and Lowe, C., 2001, Aeromagnetic and northwest Oregon: U.S. Geological Sur- anomalies of the eastern Juan de Fuca strait vey Map I-2621, 3 sheets, 31 p. strike slip. Empirical scaling relationships of Stuiver, M., Reimer, P.J., Bard, E., Beck, J.W., Burr, region, in Mosher, D.C., and Johnson, S.Y., vertical displacement to Mw for historic strike- G.S., Hughen, K.A., Kromer, B., McCormac, eds., Neotectonics of the eastern Strait of Juan G., Plicht, J.V., and Spurk, M., 1998, INT- slip and reverse faults (Wells and Copper- de Fuca: A digital geological and geophysical CAL98 radiocarbon age calibration, 24,000±0 smith, 1994) suggest an earthquake of Mw ϭ atlas: Geological Society of Canada Open-File cal BP: Radiocarbon, v. 40, p. 1041±1084. 3931, CD-ROM. 6.5±7.0. Wells, D.L., and Coppersmith, K.J., 1994, New em- Blakely, R.J., Wells, R.E., Weaver, C.S., and John- pirical relations among magnitude, rupture son, S.Y., 2002, Location, structure and seis- length, rupture width, rupture area and surface CONCLUSION micity of the Seattle fault zone, Washington: About 2800±3200 yr ago, a shallow-crustal displacement: Seismological Society of Amer- Evidence from aeromagnetic anomalies, geo- ica Bulletin, v. 84, p. 974±1002. earthquake of probable Mw ϭ 6.5±7 deformed logic mapping and seismic re¯ection data: the ground surface of central Whidbey Island Geological Society of America Bulletin, Manuscript received 25 November 2003 v. 114, p. 167±177. in the northern Puget Lowland. Multichannel Revised manuscript received 4 February 2004 Brocher, T.M., Parsons, T., Blakely, R.J., Christen- Manuscript accepted 17 February 2004 seismic re¯ection pro®les in Holmes Harbor sen, N.I., Fisher, M.A., Wells, R.E., and the east of central Whidbey Island show folded SHIPS Working Group, 2001, Upper crustal Printed in USA

472 GEOLOGY, June 2004