Earthquakes, Quaternary Faults, and Seismic Hazard in California
Total Page：16
File Type：pdf, Size：1020Kb
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 91, NO. B12, PAGES 12,587-12,631, NOVEMBER 10, 1986 Earthquakes,Quaternary Faults, and SeismicHazard in California STEVEN G. WESNOUSKY 1 SeismologicalLaboratory, California Institute of Technology Data describingthe locations,slip rates, and lengthsof Quaternary faults are the primary basisin this work for constructingmaps that characterizeseismic hazard in California. The expectedseismic moment Moe and the strengthof groundshaking resulting from the entire rupture of each mapped fault (or fault segment)are estimatedusing empirical relations between seismic moment Mo, rupture length, sourceto site distance,and strong ground motion•s. Assuminga fault model.whereby the repeattime T of earthquakesoneach fault equals Moe/M•o (where the moment rate Mo g is propor- tional to fault slip rate), it is observed that the moment-frequency distribution of earthquakes predictedfrom the geologicdata agreeswell with the distributiondetermined from a 150-yearhistori- cal record. The agreementis consistentwith the argument that the geologicrecord of Quaternary fault offsetscontains information sufficientto predict the averagespatial and size distribution of earth- quakesthrough time in California. The estimatesof r for each fault are the foundationfor con- structingmaps that depict the average return period of >• 0. lg peakhorizontal ground accelerations, and the horizontal componentsof peak acceleration,peak velocity, and the pseudovelocityresponse (at 1-periodand 5% damping) expectedto occur at the level of 0.1 probability during a 50-year period of time. A map is alsoformulated to showthe probabilitythat >• 0. lg horizontalground accelera- tions will occur during the next 50 years. The maps serveto illustratethe potential value of Quater- nary fault studiesfor assessingseismic hazard. Interpretationof availableslip ratesindicates that the largestand most frequent occurrenceof potentially destructivestrong ground motions are associated principally with the San Andreas, San Jacinto, Calaveras,Hayward, and Ventura Basin fault zones. Other regionsof similarly high hazard may yet remain unrecognized. This inadequacyresults pri- marily from an incompletedata set. Numerous faults, for example, are mapped along the coastal region of northern California and within the Modoc Plateau, but relativelyfew studiesrelating to fault slip rate are reported. A similar problem existsfor other stretchesof coastalCalifornia where marine reflectionstudies provide evidenceof active faulting offshoreyet yield little or no information of fault slip rate. Geologicaland geophysicalfield studiescan work to remove thesedeficiencies. A concerted effort to locate and define ratesof activity on all faults in California is the most promisingmeans to further quantify presentlevels of seismichazard in California. INTRODUCTION describethe averagerates of offsetacross Quaternary faults, The causalrelation betweenearthquakes and faulting was with an aim toward developingmaps that depict long-term well describedsome 100 yearsago by Gilbert [1884] during seismic hazard in California: specifically,the average size his discussion of earthquakes in the Great Basin. Koto and spatialdistribution of earthquakesthrough time and, of [1893] independentlynoted the samecausality during his more practicalconsequence, the expectedoccurrence rate of study of the great 1891 Nobi earthquake of Japan. Later the resultingstrong ground motions. field investigationsof the great 1906 California earthquake The motivation for this study residesin the uncertainties inherent to more conventional forms of seismic hazard by Lawson[1908] and his colleaguesremoved most remain- ing doubt of the close connection between faulting and analysis,which are basedprimarily on earthquakefrequency earthquakes.Moreover, Reid's [1910] geodeticstudy of statistics obtained from historical catalogues of seismicity. The uncertainty in such analyses is large that same earthquakeprovided a physicalmodel to explain when the historical record is too short to define secular rates the occurrenceof earthquakefaulting, the now well-known of seismicity,which is the usual case,particularly when rela- concept of elastic rebound. These observationsprompted Willis [1923] to suggestas earlyas 1923 that a fault map of tively small regionsare considered. In contrast to historical data, the geologicrecord of Quaternary fault offsetscontains California was a good indicator to the sitesof future earth- information on the occurrence of earthquakes through quakes. Since these studies,knowledge of the mechanicsof periods of time many orders longer than the averagerepeat earthquake faulting has continued to grow, moderate to time of large earthquakeson individual faults and orders of large earthquakeshave consistentlybeen observedto rup- magnitude greater than periods covered by historical ture along mapped faults, and faults that break Quaternary depos{tsare now commonly accepted as sources of seismic records. Observations such as these have recently led a number of investigatorsto argue that geologicinformation hazard. The purpose of this work is to use our current of fault offset rates may be used to reduce uncertainties understanding of fault mechanics to interpret data that associatedwith estimates of long-term seismicity and, in turn, seismic hazard [e.g., Allen, 1975; Anderson, 1979; Molnar, 1979]. The following exerciseis thus an attempt at 1Now at Tennessee Earthquake Imormation Center, Memphis placing this idea into a quantitative framework for analysis State University. of seismic hazard in California. The approach I will take Copyright 1986 by the American GeophysicalUnion. follows that developedin a recent study of seismichazard in Japan [Wesnouskyet al., 1984]. The underlyingpremise of Paper number 5B5769. that work was that moderate to large earthquakes occur on 0148-02;•7/86/005B-5769505.00 mappable Quaternary faults and that the occurrencerate of 12,587 12,588 WESNOUSKY'.SEISMIC HAZARD INCALIFORNIA ß MODOC PLATEAU m•(, 195L•.•.• •x, CALIFORNIAEA R T H Q U A K E S Cape Mendocino 194• Pt. Arena SAN sA FRANC•,SCO FRANClSC MONTEREY MONTEREY PARKFIELD 'EHACHAPI MOJAVE •947(6.2)• Pac/Hc SANTAMARIA GABRIEL DESERTSAN BERNARDINO Ocean Pt.Arg MOUNTAINS •ENTURABASIN SANTA PalosVerdes SANDIEGO SANDIEGO Fig.1. (a)Major active faults, physiographic features, and (b) locations of historically recorded moderate to large earthquakesthathave occun-ed in California. Faults that have produced coseismic surface ruptures are stippled. Onshoreand offshoreevents of M>•6.2 are listedin Tables1 and 3, respectively. earthquakeson eachfault is proportionalto the Quaternary seismicityto ratesof seismicityaveraged over longer periods fault slip rate. of geologictime hasnow been observed for the regions It is the parameterof seismicmoment Mo that allowsthe encompassingsouthern California [Anderson, 1979], Utah quantitativeassessment of seismicityrates from fault slip [Doserand Smith,1982], and intraplate Japan [ Wesnousky rate data [Brune,1968]. The seismicmoment Mo is a more et al., 1982],as well as a numberof plateboundaries fundamentalmeasure of eaChquakestrength than is earth- [Brune,1968; Davies and Brune,1971]. The observed quakemagnitude M and,like M, maybe measureddirectly coincidencein seismicityrates over these broad regions pro- from a seismogram.Mo, however,may be relateddi,rectly videssupport for the suggestion[Allen, 1975; Anderson, to physicalparameters that describethe earthquakesource, 1979]that the geologic record of Quaternaryfault offsets is whereasM canonly be indirectlyrelated through its empiri- well suitedfor estimatinglong-term rates of seismicityand cal relationwith Mo [e.g.Hanks and Kanamori,1979]. henceseismic hazard in regionscharacterized by shallow Specifically,seismic moment Mo is definedto equalI4. UA, seismicity,Quaternary faulting, and a shortrecorded his- whereg is the shearmodulus, A is the faultarea, and U is tory. Californiapossesses each of theseattributes. Let me the averageslip occurringacross the fault [e.g.,Aki and thenbriefly review the availabledata concerning seismicity, Richards,1980]. Thus,for a fault on whicha numbern of faulting,and fault slip rates in California. earthquakeshave occurred during a periodof time T•, the DATA AND OBSERVATIONS rate of seismicityon the fault may be characterizedby the seismic moment rate Earthquakesand Faulting Earthquakesin California,and in otherregions of similar tectonicsty]v, generally rupture along faults that show evi- dence of prior Quaternarymovements [Allen, 1975; Further,if a geologicmarker of ageT2 is displacedacross Matsuda, 1977]. It is alsogenerally recognized in Califor- the faultby distanceU, the lo.ng-termrate of seismicityon nia that the occurrenceof earthquakesis limited primarily thefault may be ,expressedasMo--gA (U/T2) [Brune,1968]. to theupper 15 km of theearth's crust [e.g., Sibson, 1982]. This assumesthat fault displacementsare not the resultof Theepicentral distribution of moderateto largeearthquakes aseismiccreep but ratherare the cumulativeresult of many that have occurredin Californiaduring historical time is earthquakesthrough geologic time. Usingthe conceptof picturedon a simplifiedfault map of Californiain Figure1. seismicmoment, a near coincidenceof historicalrates of Onshoreearthquakes of M >/ 6.2 arefurther listed in Table WESNOUSKY.'SEISMIC HAZARD IN CALIFORNIA 12,589 TABLE 1. Earthquakes Onshore California Mappable Surface Date Location Magnitude* Faultt Rupturet References April9, 1857 SanAndreas 7.9õ