A Study of the Source Mechanism of the Alaska Earthquake andTsunami of March 27, 1964Part 1. Water Waves!GEORGE PARARAS-CARAYANNISABSTRACT: The geologic history and the general geomorphology of the areaaffected by the March 27, 1964 Alaska earthquake are given. The tsunami-generatingarea is determ ined and the extent of crustal displacement and the limits of theareas of subsidence and uplift, as revealed by geologic evidence, are discussed. Thedimensions of this tsunami-generating area, its volume of crustal displacement,and the energy associated with the tsunami are calculated. Wave activity withinand outside the generating area and the possible generating mechanisms for thetsunami are discussed. A wave refraction diagram of the Alaska tsunami for thenorth Pacific Ocean area is presented in Figure 6.THE ALEUTIAN ISLAND ARC and the AleutianTrench extend for 2800 km from Kamchatkato south-central Alaska along remarkably smoothcurves which are convex toward the south (Fig.1). The Arc forms the Alaska Peninsula and,according to Wilson (1954), intersects, northof Cook Inlet, a second tectonic arc that extendsnorthward from the vicinity of theWrangell Mountains. However, Plafker (1965)regards this second segment as a continuationof the Aleutian Arc. Where the trench impingeson Alaska it loses its identity, althoughan offshore range of seamounts suggests it mayonce have extended around to the south toparallel the continental slope, as postulated byMenard and Dietz (1951). Concavity in theformer shape of the trench on its eastern segmentis also suggested by the sedimentary arcdefined by Wilson (1954), which embracesKodiak Island and the Kenai Peninsula. Asshown by Wilson, such concavity is to be expectedwhere two arcs meet at an acute angle,as is well exemplified where the Aleutian andKuril-Kamchatka arcs intersect. It is also quitepossible that large horizontal movements ofcrustal blocks have helped to change the shapeof the Trench and Arc on their eastern segments.However, no such evidence was foundin a field study following the Good Fridayearthquake (Berg et al., in preparation) .i Hawaii Institute of Geophysics Contribution No .184. Manuscript received June 22, 1966.301The nature of the termination of the easternsegment of the Aleutian Trench is obscuredby thick sediments washed in from the continental shelf against which it abuts offshorefrom Cape Suckling. The sediments are of geosynclinal-dimensionsin the sedimentary arcon Kodiak Island (Menard and Dietz, 1951)and as shown by drilling on the Kenai Peninsula.Woollard et al. (1960) show there isgeophysical evidence for at least 7 km of sedimentsin Cook Inlet, a graben separating theprimary arc from the offshore sedimentary arc.Sediment is about 2 km thick off Kodiak Islandalong the Aleutian Trench, thinning out toabout 0.7 km south of Unimak Island in thedeep water area, according to seismic measurementsby Shor (1962).THE GENERATING AREA OF THE ALASKATSUNAMIAccording to Van Dorn (1964) , the tectonicdislocations associated with the Alaska earthquakeof March 27, 1964 ranged over a distanceof 800 km, from the upper portion ofPrince William Sound to southwest of theTrinity Islands . The dislocations follow a dipolepattern of positive and negative displacementson either side of a zero-line which,intersecting the east coast of Kodiak Island,continues northeast to the western side of PrinceWilliam Sound. There, changing direction, it
Alaska Earthquake and Tsunami, I-PARARAS-CARAYANNIS 303is greater than the 1906 San Francisco earthquake( 8.3) , and equal to or greater than the1960 Chile earthquake (8.4). The epicenterof the earthquake was at 61.05 °N, 147.7°W(USCGS, 1964) , near the east shore of UnakwikInlet in northern Prince William Sound .Geological investigations have defined theland areas affected by the earthquake. To theeast, the zone of deformation appears to die outbetween the Bering Glacier and Cape Yakataga .The northwestern limit of tectonic changesextends at least to the west side of ShelikofStrait and Cook Inlet (Plafker , 1965) . Thenorth inland limit is known only along thehighway connecting Valdez and Fairbanks; itappears to extend in a northeasterly directionto the vicinity of the Wrangell Mountains, andquite possibly into the Alaska Range.The area of uplift covers about 105,000 kmandextend s from southern Kodiak Island northeastto Prince William Sound . It includes thesouthern and eastern parts of Prince WilliamSound , the coastal area as far east as the BeringGlacier, and the continental shelf and part ofthe slope to a depth contour of approximately200 m. The maximum uplift on land was 10 mat the southwest end of Montague Island, butis suspected to have been considerably more offshore.Uplift also occurred along the extremesoutheastern coasts of Kodiak Island and SitkalidakIsland , and part or all of Sitkinak Island.The maximum measured uplift of SitkalidakIsland was 0.4 m. The estimated uplift of SitkinakIsland was from 0.35 to 0.65 m and possiblyas much as 1.5 m ( Plafker, 1965).The area that subsided included the northernand western parts of Prince William Sound,the western segment of the Chugach Mounta ins,portions of the lowlands north of them, mostof the Ken ai Peninsula, and almost all of theKodiak Island gro up. This area of subsidencecovers approximately 110,000 krn-, and is800 km long and 150 km wide. Plafker (1965)estimates that the volume of crust that has beendepressed below its pre-earthquake level is about115 krn",The seaward limits of the earthquake and thetsunami-generating area were determined bymeans of a series of refraction diagrams basedon Snell's Law of Refraction using the velocityequation for shallow water waves, C = Vgd.Such a method of preparing refracti on diagramshas shown good results, especially if carried outon large-scale charts with detailed bathymetry(Johnson, O'Brien, and Isaacs, 1948) .In constructi ng the refraction diagrams forthe Alaska tsunami, the marigrams of differenttide gauge stations around the Pacific were consultedand the total travel time of the first waveat each station was determined. Then refractio ndiagrams were constructed toward the earthquakearea from each tide gauge station inlengths of time equal to the calculated traveltime for that station. It was assumed that thelast wave front in each refraction diagram wouldcorrespond to a point on the boundary of thegenerating area, and if enough refracted wavefronts from different stations were plotted, anenvelope defining the tsunami-generating areacould be drawn.W ave fronts were refracted from Yakatat,Cape Yakataga, Seward, Uzinki, Kodiak, OldHarbor, Unalaska, Adak , Attu , and Honolulu .The last front of each of the refracted waves isshown by a heavy dashed line in Figure 1. Theseaward boundary of the generating area is nearthe 200-m depth contour which defines the edgeof the continental shelf. Maximum displacementof the ocean floor occurred along the continentalshelf, from an area south east of Kodiak Island,to an area close to Cape St. Elias south of theisland of Kayak (Fig. 1) . Geologic evidence,however, has shown positive land displacementsas far north as Cape Suckling and as far eastas the Bering Glacier. It is quite prob able, therefore,that the tsunami-generating area extendedfarther to the northeast, although waves generatedin such shallow water would reach tidegauges much later and their origin would notbe identifiable.Unfortun ately, this same wave refraction techniquecould not be used to define the northernand western boundaries of the main tsunamigeneratingarea, because conditions in PrinceWilliam Sound and elsewhere along the coastof Alaska were further complicated by localtsunamis, oscillations, and surge . In addition,no tide gauge stations were operating in thearea, and personal accounts were conflicting asto arrival times of the different waves.The northward limit is assumed to be restrictedby the land boundaries, and the western
304limit to extend to the west side of ShelikofStrait and Cook Inlet.In estimating the travel time of the tsunami,corrections were made for the delay at the islandof Kodiak in the arrival of the groundshocks from Prince William Sound. These correctionsranged from 1 minute to 6 minutes andwere based on the fact that the Navy W eatherCentr al on the island of Kodiak listed the timeof the principal shock in Prince W illiam Soundas 6 minutes later than the time listed by theU. S. Coast and Geodetic Survey. This wouldimply that the wave front generated on thenortheast side of the disturbance area had a 6minute head start on the wave front generatedsoutheast of Kodiak.The tsunami-generating area covers an area700 km long by 150 km wide, a total of about105,000 krn-, The volume of the uplifted crustalong the continental shelf is about 96 krn". Theenergy associated with the tsunami has beenestimated by Van Dorn ( 1964) to be of theorder of 2.3 X 10 21 ergs. This estimate is basedon the source dimensions of an area 240 nauticalmiles by 100 nautical miles and an upliftof 1.8 m ( 6 ft ) at the northeastern end of thisarea and zero at the southwestern end. Thisestimate, however, is considered low because thegenerating area had dimensions that were largerthan those estimated by Van Dorn.Using our source dimensions, and assumingthat the total energy was equal to the potent ialenergy of the uplifted volume of water, thetotal energy for the tsunami in the Gu lf ofAlaska was calculated as follows:1E t = - pgh 2 A61= - (1.03) ( .980) (1 0 3 ) (10 4 ) ( 1.8 3 2 ) (1.56X 10 7 ) (7 X 10 7 ) = 5.88 X 10 21 ergswhereE, = E" = total energyp = 1.03 g/cm 3 = densityg = 980 cm/sec 2h = height of displacement = 1.83 mA = area1 erg = g cm 2 sec- 2The waves generated in the Gulf of Alaskawere of an unusually long period, on the orderPACIFIC SCIENCE, Vol. XXI, July 1967of an hour or more:' Th eir energy radiation waspreferentially directed toward the southeast andthis is why more damage was done to the N orthAmerican coast than anywhere else east or southof the generating area. This preferential directivityof energy radiation can be attributed tothe orientation of the tectonic displacementsalong the continental shelf of the Gulf ofAlaska, and the long period of the waves canbe related to the long seiche period of theshallow shelf.According to Japanese seismologists (Iida,1958), the generating area of a tsunami roughlycorresponds to the distribution of the majoraftershocks. This appears to be indeed the casein the Gulf of Alaska.There were 52 aftershocks of the Alaskaearthquake. The largest had a magnitud e of 6.7.The aftershocks occurred in an area from about15 km north of Vald ez to about 55 km south ofTrinity Islands, and were heavily concentratedon the northeast and the southwest of the upliftedregion (USCGS, 1964), which also wasthe main tsunami-generating area.The vast area of tectonic movements indicatesthat wave crests were generated along one ormore line sources from the region of maximumupli ft. Thus, the shores of the Kenai Penin sulawere struck within 20 minutes after the start ofthe earthquake, and those of Kodiak Island ,within 34 minutes.Unfortun ately, the violence of the earthquakeleft south-cent ral Alaska without a tide gaugein operation. The only reliable record from thegenerating area is the one that was obtained bypersonnel of the U. S. N avy Fleet W eatherStation at Kodiak; it is shown in Figure 2. Thisrecord has been corrected for the 1.7-m (5 .6-ft)submergence of the area.Outside the immediate generating area, therecord of Cape Yakataga, as constructed fromthe personal account of C. R. Bilderback, a residentof the area, is the next most reliable record.This record is the only one obtained outsidethe generating area that shows an initial dropin the water level (Berg et aI., in preparation ) .Withdrawal of the water immediately followingthe earthquake has been reported from Kayak,Middl eton, and Hinchinbrook islands, as wellas from Rocky Bay and N uka Bay, at the end
306 PACI FIC SCIENCE, Vol. XXI, July 1967876543en 20::llJ~llJ::E0-I-2-3-4LEGEND" ..... Projected tide after theearthquake--Tide before earthquake-516 17 1827-MARCH19 20 21 22 23 24TIME IN HOURS -ALASKA STANDARD TIMEI 228-MARCHFIG. 3. Marigram of wave activity at the town of Yakutat.downtown section of Valdez, ruining almost allthe merchandise in the stores. These wavescould not have come from the generating areaoutside Prince William Sound because if thiswere so, it would have taken them only 34minutes to reach Valdez. It is more likely, then,that the waves at Valdez arrived in resonanceat high tide, from the immediate area of PortValdez.Maximum positive crustal displacement inPrince William Sound occurred along the northwestcoast of Montague Island and in the areaoffshore. These earth movements caused a gradientin hydrostatic level and the resultingshort-period wave raced through Knight IslandPassage within 10 minutes and on toward ChenegaIsland, inundating the village of Chenegato an elevation of 15.5 m and completely destroyingit. This same wave continued norththrough Knight Island Passage and inundatedPerry and Naked islands, but to lesser heights(Berg et aI., in preparation) .Bathymetric surveys by the USCGS (1964)in the area off Montague Island and at thenorth end of Latouche Island revealed a numberof large submarine slides. It is possible,therefore, that the combination of submarineslides and the tilting of the ocean floor due touplift created the solitary wave reported atChenega village and at Perry and N aked islands.A second wave about 40 m high (125 ft)was reported coming out of the Valdez Narrows
Alaska Earthquake and Tsunami, I-PARARAS-CARAYANNIS8I I I I I I I I I307.,V)76 -543a::1LI2t-1LI:E,I18,,!~~ \~~1. I\1\1\":! ,..' .1\ I \ ~III 1/ \ 1\ ) ... .. .. . .III \ 1 11 1, U \ I \ ...'... .. ;I II! \ !~\II k. ...,./....\.I, !\ fI ··.. \ \/111 .I. I . v - : " \ I \ , , II V\ .v -: , ,' \ I , Io f----- (~ ~-- ~ - -- 1/- \'_1_' --1-I-2-3-4-5 I16 1727-MARCH~ 111i I ~\ \, \!1/ \ / ~ ~~ ~FIG. 4.and spread ing across the Sound (Plafker andMayo, 1965) . This wave was caused by slumpingof the glacial deltas in Port Valdez whichhad been shaken loose by the force of theearthquake.TSUNAMI MECHANISMMost tsunamis result from earthquakes havingfocal depths of less than 60 km. Iida(1958) has derived an empirical relation givingthe maximum focal depth H ( in km ) foran earthquake of magnitude M which has resultedin a detectable tsunami :LEGEND• Observed---Approximate water level- - - --Hypothetical water level:.. .... Projected tide after theearthquake--Tide before earthquakeI I I I I I I I I19 20 21 22 23 24 2 4TIME IN HOURS-ALASKA STANDARD TIME 28-MARCHM = 6.42 + 0.01 H (1)where M is the Richter magnitude given byDiagram of wave activity at Cape Yak ataga.log E(ergs) = 11.8 + 1.5 M (2)The focal depth of the Alaska earthquake wasabout 20 km. This was shallow enough to createtsunami waves even though the epicenter of themain shock was as much as 100 km inland fromthe coast. A number of shallower aftershocksover a large area ranging from HinchinbrookIsland to southeast Kodiak Island indicate thatcrustal movements over a wide area were involved.Undoubtedly these shallow aftershockscreated smaller waves that could not be separated,in the tide gauge records, from reflectionsof the initial tsunami.If the tsunami waves that hit the island ofKodiak were the result of crustal movementsonly, then the first wave could be expected to-
308 PACIFIC SCIENCE, Vol. XXI, July 1967be the highest, at least within the generatingarea. At Uzinki, Kodiak City, W omen's Bay,and elsewhere on the island of Kodiak, however,the third and fourth waves were the highest.A theory of generation from a singlepattern of crustal deformation is therefore notsatisfactory here. Such factors as reflection fromcoastal boundaries , wave interaction, and resonanceshould be taken into consideration.Slumps or avalanches, similar to the ones thatoccurred in Prince William Sound , are usuallylocalized; they can produce no large tsunamisthat would travel across wide portions of theocean. According to Wiegel (195 4), not morethan 2% of the potential energy of a fallin g orsliding body is converted into wave energy. InPrince William Sound, however, slumping andsliding when added to tectonic movements createdtsunami waves of very large energy, buttheir effect was catastrophic only locally; verylittle of the energy escaped the Sound .SUMMARY AND CONCLUSIONSThe Alaska earthquake of March 27, 1964affected an area of approximately 215,000 krn-,extending from the W rangell Mountains at thenortheast to the Trinity Islands in the southwest,and from the west side of Shelikof Straitand Cook Inlet east to the vicinity of the BeringGlacier.....'110"Geologic evidence has revealed a dipole patternof positive and negative tectonic movementsresulting from this earthquake. The areaof subsidence covers approximately 110,000 km 2and the volume of crust that has been depressedbelow its pre-earthquake level is about 115 km".The area of uplift covers about 105,000 km 2and includes the southern and eastern parts ofPrince William Sound, the coastal area as fareast as the Bering Glacier, and a great part ofthe continental shelf and slope bordering theGulf of Alaska.The seaward limits of the area affected bythe Alaska earthquake and the tsunami-generatingarea were determined by means of a seriesof wave refraction diagrams as shown in Figure5, based on Snell's Law of Refract ion. Thetsunami-generating area covers 140,000 km 2and includes the whole of the region of upliftand part of the region of subsidence. It extendsfrom the Trinity Islands to the Bering Glacierand includes Shelikof Strait, Cook Inle t, andthe continental shelf bordering the Gulf ofAlaska to a depth of approximately 200 m. Thetotal volume of displaced material in the tsunami-generatingarea was estimated to be 120krn", and the energy associated with the tsunamiwas calculated to be in the order of 6 X 10 21ergs.As a result of this work the following conclusionsare drawn:ee·I--1----I--~___1---_1_---___t_;~~""--'----_j_-__:;~~M1r+t__\HHf--4-4---I_i'i-____i".ATTU l~:;,........"..,...,...FIG. 5. Di agram of wave front's ief~acted toward the earthquake area from Attu Island (dashed line),Adak Island (solid line) , and ,Unalaska Island (d otted line) .
Alaska Earthquake and Tsunami, I-PARARAS-CARAYANNIS 3091. Two main tsunami-generating areas canbe distinguished: one along the continentalshelf bordering the Gulf of Alaska; the otherin Prince William Sound .2. The main generating area in the Gulf ofAlaska roughly corresponds to the geographicdistribution of the major aftershocks.3, The energy of the tsunamis gener ated inPrince William Sound was expended inside theSound ; not much energy escaped this closedregion.4. The long period of the waves generatedin the Gulf of Alaska is related to the longseiche period of the shallow shelf.5. The preferential radiation of energy towardthe southeast is attributed to the orientationof the tectonic displacements along thecontinental shelf of the Gulf of Alaska.6. The waves arriving at Cape Yakataga hadtheir origin in .the shallow coastal area near theBering Glacier, whereas the waves arriving atYakatat traveled through the deeper waters.BERINGAf TU l"..~ 1'~~' Ws ' , « r >AL E U T \ A ~ •• ;10 ~A LAS KA, ..• .~ ..o-... AN DRE AN OF IS .7. In Prince William Sound two major tsunamiswere distinguished: one had its originnear the west coast of Montague Island, theother originated in the Port of Valdez.8. Two types of tsunami-generating mechanismswere associated with the Alaska earthquake: (a) waves generated directly by tectonicmovements of the sea floor, and (b) wavesgenerated indirectly from landslides, mudflows,and slumping of alluvial deposits.9. In Prince W illiam Sound both generationmechanisms were evident , while in the generatingarea along the Gulf of Alaska, the generatedtsunami was the direct result of tectonic movements.ACKNOWLEDGMENTSThe work on which this paper is based wassupp orted in part by the National Science Foundationunder the United States-Japan programfor cooperative research in the Pacific, throughgrant N o. GF-1 53, and in part by the Officeo'?~
310of Naval Research through contract Nom3748(03) .I am particularly indebted to D . C. Cox,W . M . Adams, and G. P. Woollard for theiradvice and constructive criticism.I would also like to acknowledge with appreciationthe advice, comments, and suggestionsgiven to me by A. S. Furumoto, K. Kajiura,G. W . Groves, H. G . Loomis, G. R. Miller,and Robert Harvey.I also thank the members of the UnitedStates-Japan Cooperative Field Survey (E . Berg,D . C. Cox, A. S. Furumoto, K. Kajiura, H. Kawasumi,and E. Shima) for permission to usedata from their report prior to publication.REFERENCESBERG, E., D. C. Cox, A. S. FURUMOTO, K.KA]IURA, H. KAWASUMI, and E. SHIMA. (Inpreparation.) Field Survey of the Tsunami of27 March 1964 in Alaska. Hawaii Inst. Geophys.Rept. Series.BROWN, D. L. 1964. Tsunamic Activity Accompanyingthe Alaskan Earthquake of 27 March1964. U. S. Army Engr. Dist., Anchorage,Alaska. 20 pp.GRANTZ, A., G. PLAFKER, and R. KACHADOORIAN. 1964 . Alaska's Good Friday EarthquakeMarch 27, 1964: A Preliminary GeologicEvaluation. U. S. Geol. Surv. Circ. 491.35 pp.IIDA, K. 1958. Magnitude and energy of earthquakesaccompanied by tsunami, and tsunamienergy. J. Earth Sci., Nagoya Univ. 6 :10 i 112 .JOHNSON, J. W ., P. O. O'BRIEN, and J. D .ISAACS. 1948. Graphical construction of waverefraction diagrams. H . O. Publ. No. 605 .PACIFIC SCIENCE, Vol. XXI, July 1967MENARD, H. W . 1964. Marine Geology of thePacific. McGraw-Hill Book Co., New York.Pp. 97-116.--- and R. S. DIETZ. 195 1. Submarine geologyof the Gulf of Alaska . Bull. Geol. Soc.Am. 62:239-253.PLAFKER, G . 1965. Tectonic deformation associatedwith the 1964 Alaska earthquake. Science148 :1675-1687.--- and L. R. MAYO. 1965. Tectonic Deformation,Subaqueous Slides and DestructiveW aves Associated with the Alaskan March27, 1964 Earthquake: An Interim GeologicEvaluation. U . S. Geol. Surv., Open File Rept.19 pp.SHOR, G . G., JR. 1962 . Seismic refraction studiesoff the coast of Alaska : 1956-57. Bull.Geol. Soc. Am . 52 :37-57.U . S. COAST AND GEODETIC SURVEY. 1964.Preliminary Report, Prince William Sound,Alaskan Earthquakes; March-April 1964.83 pp.VAN DORN, G. W. 1964. Source mechanismof the tsunami of March 28, 1964 in Alaska .Chap. 10. In: Proc. N inth Conference onCoastal Engineering, Am . Soc. Civil Engr.,pp . 166-190.WIEGEL, R. L. 1954. Laboratory studies ofgravity waves generated by the movement ofa submerged body. Univ. Calif. Inst. Engr.Res., Ser. 3, Issue 362.WILSON, J. T. 1954. The development andstructure of the crust. Chap. 4. In : Gerard P.Kuiper, ed., The Solar System, Vol. 2. TheEarth as a Planet. Univ. of Chicago Press .WOOLLARD, G. P., N . A. OSTENSO, E. THIEL,and W. E. BONINI. 1960. Gravity anomalies,crustal structure, and geology in Alaska. J.Geophys. Res. 65 :1021-1037.