Between Punta Gorda on the northern California coast and the head of the Gulf of California, 1,350 km to the southeast, lies the active transform boundary that forms the modern San Andreas fault system (fig. 6.1). Dextral motion between the North American and Pacific plates along this system is accommodated within an elongate zone, broadening from about 100 km at its north end to about 300 km in southern California. The San Andreas fault proper hugs the east side of this zone at its south terminus and gradually migrates across the zone, lying on the west edge of the zone at its north terminus. The San Andreas fault system transmits about three-fourths of the relative motion across the plate boundary, as shown by various geologic and geodetic evidence. Much of this motion is stored elastically in the upper crust along the major faults in the system, ultimately to be released in large plate-boundary earthquakes. These large earthquakes and their implications for the mechanics of North
American-Pacific plate interactions are the subject of this chapter.
Figure 6.1 - Seismicity highlights broad and complex zones of active tectonism in the western United States. Dots, earthquakes scaled by magnitude in unit-magnitude steps, smallest for M<4; circles, earthquakes of M>7; red dots, recent events, generally since 1975; pink symbols, historical events. From Engdahl and Rinehart (1988).
Earthquake activity in California and Nevada at the latitude of the San Andreas fault extends well beyond the confines of the San Andreas system (fig. 6.2). In the past century alone, only about half of the M≥ 6 activity has fallen within the San Andreas system; of the rest, half is associated with the western Basin and Range province, and the other half with the Mendocino triple junction and the Gorda plate. Although activity in the latter region reflects the tectonics of the triple junction and the collision of the Gorda plate with the North American plate, seismicity east of the San Andreas system along the east flank of the Sierra Nevada and in the Basin and Range province reflects the incomplete accommodation of plate motion along the San Andreas fault system. A significant proportion of this "missing" motion occurs in the Basin and Range, the seismicity of which plays an integral role in the tectonics of the plate boundary.
Figure 6.2 - Seismicity of California, Nevada, and northern Baja California, 1769-1989. Earthquakes are listed in table 6.1 and plotted by magnitude class.
EARTHQUAKE HISTORY OF THE SAN ANDREAS FAULT SYSTEM
The historical record of major earthquakes affecting California, western Nevada, and northernmost Baja California (table 6.1) includes basic seismologic data on 206 of the largest earthquakes occurring between 1769 and 1989. This catalog lists all known events of M≥6 and includes new and updated information on their locations.
The record of seismicity within the San Andreas fault system and surrounding regions is both geographically and temporally uneven and incomplete before the introduction of practical seismographic instrumentation around the turn of the 20th century. In general, the density and distribution of people who left written accounts of their experiences determines the reliability of the catalog during the preinstrumental period. From the establishment of the Franciscan missions beginning in 1769 until their secularization in the 1830's, detailed accounts of events that damaged the missions are available, and these accounts form the primary source material for earthquakes occurring during this period. Life in California was a constant struggle for survival at that time; posting to a mission evidently was considered a hardship assignment, and so essentially nothing was recorded about events that were only felt, even when they were destructive at nearby missions. After secularization and before
the gold rush, the quality of the record degrades with the cessation of the annual reports of the missions. Other sources of records also are notably weak during the Mexican period, from the early 1830's until 1846.
The discovery of gold in 1848 transformed the written record of earthquakes with the advent of newspapers throughout the gold fields in the Sierran foothills and in the San Francisco Bay region. Printed accounts of earthquakes have been extensively used, notably by Toppozada and others (1988), to quantify the seismicity of California from 1850 onward. They estimated that their historical catalog is probably complete for the San Francisco Bay region and central Sierra Nevada from 1850 on for earthquakes of M≈6. The same level of completeness is not achieved, however, for the San Andreas fault system in southern California until the 1890's. Statewide, the catalog of earthquakes is substantially complete for earthquakes of M≈7 after about 1850 (see Agnew, 1985). The quality of the catalog for central Nevada, where much significant 20th century seismicity has occurred, is less complete. Questions remain today about purported events as late as 1903 in this
region (Slemmons and others, 1959).
Reports of the local effects of earthquakes continue to playa major role in determining the locations and sizes of earthquakes well into the 20th century. The earliest seismographs capable of systematically detecting California and Nevada earthquakes were installed throughout the world by John Milne beginning in 1896. Seismograms from these instruments and their successors provide useful instrumental magnitudes from 1898 onward. However, not until the development of the Wood-Anderson seismograph and its deployment throughout California beginning in 1926 do instrumental measurements fully supplant noninstrumental magnitudes and epicentral locations.
The objective in assembling a single catalog from these many sources, spanning many different types and qualities of information, has been to achieve uniform spatial coverage without sacrificing any events of historical significance. M=6 was chosen as the threshold magnitude because probably all events of this magnitude are known from the instrumental period beginning in 1898, and the preinstrumental record is reasonably complete at this level in some areas for an additional half-century. All earthquakes with at least one reported magnitude of at least 6.0 have been included in the catalog. Because magnitude is an estimated quantity and has some inherent uncertainty, events with reported magnitudes within a few tenths of a unit of 6.0 are also included. In addition to those earthquakes with cataloged magnitudes, original documents for others with reported high intensities or of particular historical significance have been reexamined in an attempt to refine their locations and
A word of introduction should be added about earthquake locations and magnitude scales and their use in this chapter. Earthquakes are complex physical processes generated by sudden slip on faults, and as such they can only be grossly characterized by simple concepts. Two seismologic conventions are in common use for assigning a single geographic coordinate to an earthquake: One measures the center of energy release, frequently as estimated from the intensity distribution for preinstrumental events; the other measures the location of the initial point of rupture, or hypocenter, as determined from seismic traveltime measurements. Either point on the Earth's surface above the hypocenter or the center of the intensity distribution is sometimes referred to as the epicenter, and each type of location appears in table 6.1, with preference given to instrumental epicenters. Fortunately, the geographic differences between these distinct physical measures become significant only for the
largest events, M≈ 7, when viewed at the scale of the entire San Andreas fault system.
Magnitude, as commonly used to compare the sizes of different earthquakes, also represents an extreme simplification of the earthquake process and by itself cannot fully characterize the size of any event. Traditionally, seismologists have developed a suite of magnitude scales, each with its own purpose and range of validity to measure an earthquake. Because no single magnitude scale can be systematically applied to the entire historical record, a summary magnitude, M, is introduced here to facilitate comparisons between events. As described below in the subsection entitled "Quantification of Earthquakes and Magnitude Scales," M is taken as the surface-wave magnitude (MS), when available, and as a modified intensity magnitude (MI) during the preinstrumental era. Generally speaking, M provides a better relative measure of the static, geologic increment of fault slip in the earthquake than it does of the severity of shaking.
The earthquake history of California, western Nevada, and northern Baja California presented here has apparent limitations and can doubtlessly be improved through further research. Nevertheless, it provides a firm observational basis for assessing the tectonic implications of the 2-century-long seismic history, as well as of the prospects for future earthquake activity.
In this section, we briefly discuss some events of particular historical, social, or scientific significance. Although each of the 117 San Andreas fault system events in table 6.1 merits discussion, this task is far beyond the scope of this review, and so the reader is referred to the reports by Richter (1958), Coffman and others (1982), and Townley and Allen (1939) for an introduction to many of these events. Table 6.1 also omits several historically significant events with magnitudes well below the nominal threshold of M=6 adopted here, and so it something less than a complete reference on San Andreas seismicity.
JULY 28, 1769 (M=6)
My major effort in constructing this catalog has gone into identifying and validating all reported events of M≥6. Two conspicuous omissions from table 6.1, events that are commonly mentioned in the literature but that could not be substantiated upon further inspection, should be noted. The first is the 1852 earthquake alleged to have ruptured the Big Pine fault (for example, Jennings, 1975). Toppozada and others (1981) failed to find any evidence supporting the occurrence of a major earthquake at that time in the region. Geologic inspection of the surface trace of the fault by M.M. Clark (oral commun., 1988) similarly failed to provide evidence of any historical activity. The other deleted event appears on the seismicity map by Goter (1988) at lat 35° N., long 125° W., with an epicenter from the catalog of Abe and Noguchi (1983). Although a large (MS=6.8) earthquake certainly took place on March 22, 1902, no evidence has been uncovered to support
a location anywhere on shore in California or, for that matter, in the Western United States. The original location determined by Milne in 1903 placed the event well off the California-Oregon coast at lat 42° N., long 130° W.
The earthquake history of California serendipitously begins with the first overland expedition through the State in 1769. In response to the perceived threat posed by Russian expansion into the northern Pacific and growing British presence in the northwestern Pacific, Spain embarked on the colonization of present-day California through the establishment of a series of Franciscan missions, supported by military garrisons at San Diego and Monterey. In the summer of 1769, Gaspar de Portolá led the first expedition from San Diego to establish a land route to Monterey.
On July 28, while camped along the Santa Ana River, about 50 km southeast of Los Angeles, a sharp earthquake was felt that "* * * lasted about half as long as an Ave Maria" (fig. 6.3). From the diaries of three members of the expedition, we know that earthquakes were felt on nearly a daily basis through August 3, as the party traveled northwestward to near San Gabriel and then westward across Los Angeles to the Pacific. The diary of Fray Juan Crespi (Bolton, 1927) mentions no fewer than a dozen aftershocks, some described as violent. After August 4, no further earthquakes were mentioned as the expedition traveled into the San Fernando Valley and exited to the north.
Figure 6.3 - Early accounts of significant earthquakes reflect the sparse settlement of California in a narrow coastal corridor before the population explosion accompanying the gold rush in 1849. Accounts of the few well-documented events (dates shown) principally derive from mission records at San Diego (SD), San Luis Rey (SLR), San Juan Capistrano (SJC), San Gabriel (SG), San Fernando Rey (SFR), San Buenaventura (SBV), Santa Barbara (SB), Santa Inez (SI), and La Purisima Concepción (LPC), and from the towns of Los Angeles (LA) and Fort Tejon (FT) in southern California. Accounts from the Spanish capital Monterey (M), San Francisco (SF), and San Jose (SJ), as well as mission sources, detail events in north half of the State. Uncertainties in the interpretation of every event before the great earthquake of 1857 (rupture shown; arrows indicate direction of relative movement) are well illustrated by newly uncovered evidence suggesting a San Andreas origin for the December 8, 1812, shock
near Wrightwood (head of connecting arrow), well inland of traditional location along the coastal Kewport-Inglewood fault tail of arrow). Earthquake of December 21, 1812, locates in the Santa Barbara Channel (SBC). Foreshocks of the great earthquake of 1857 locate near Parkfield (P), suggesting unilateral rupture propagation to the southeast.
These sketchy reports suggest that the explorers traveled near or through the epicentral area of a moderate earthquake (Richter, 1973; Toppozada and others, 1981). Comparisons between the accounts of the aftershocks and more recent events suggest an event of similar size and location to the 1933 Long Beach, 1971 San Fernando, or 1987 Whittier Narrows earthquake. If significance is placed on the absence of aftershocks while crossing the source region of the 1971 San Fernando earthquake, the evidence would seem to favor a source in the Los Angeles Basin. An event on either the San Andreas or San Jacinto faults, some 50 km to the northeast, could conceivably have been the source of the 1769 earthquake. The description of the duration of strong shaking, however, suggests a magnitude more of 5-6 than of 7-8.
DECEMBER 8, 1812 (M=7)
A more distant source would make the long, felt aftershock sequence even more remarkable because it would be well removed from the expedition route.
The first of two significant earthquakes to occur in southern California in 1812 occurred on December 8 and destroyed the church at Mission San Juan Capistrano, killing 40 neophytes (fig. 6.4); damage was also sustained at San Gabriel. The accounts of this earthquake and the later one on December 21 cannot be readily disentangled at San Fernando Rey and at San Buenaventura, considerably complicating the interpretation of this event.
Figure 6.4 - Mission San Juan Capistrano as drawn by Henry Miller in 1856, 43 years after the December 8, 1812, earthquake. Vaulted stone church at right collapsed in that earthquake, killing 40 worshipers. Photograph courtesy of the Bancroft Library.
Analyses of these scanty data by Toppozada and others (1981) and Evernden and Thompson (1985) place the epicenter along the south half of the Newport-Inglewood fault zone (fig. 6.3). This location is somewhat constrained by the interpretation of no damage at Buenaventura during the event. The Los Angeles Star of January 10, 1857, however, stated that the December 8 event severely damaged the church tower (Agnew and Sieh, 1978). The same story attributed the collapse of the stone arch roof of the church at San Juan Capistrano to poor construction, a possibility made credible by the death of the master mason before completion of the church (fig. 6.3; Duncan Agnew, oral commun., 1988).
DECEMBER 21, 1812 (M=7)
Recently, Jacoby and others (1988) proposed that this event ruptured the San Andreas fault at Wrightwood (fig. 6.3), on the basis of dendrochronologic dating of distress to trees growing on the fault trace. Sieh and others (1989) argued that this rupture extended at least 25 km northwestward into the peat bog at Pallet Creek. The fault rupture in this event preserved at Pallet Creek is comparable in size to the rupture formed in the 1857 earthquake.
The preferred location of the December 8, 1812, earthquake on the San Andreas fault as proposed by Jacoby and others appears in table 6.1. A magnitude of about 7 is consistent with the inferred extent of damage. The lateral extent of rupture is unconstrained to the southeast and may well have extended into the San Bernardino Valley. However, the accounts of the earthquake from Indians living in the San Bernardino Valley that were thought to place some constraint on the rupture are now believed to be fictitious (Harley, 1988), leaving Mission San Gabriel, some 40 km from the rupture, as the nearest point of observation.
The second major episode of earthquake activity in 1812 damaged the missions along the Santa Barbara Channel and western Transverse Ranges just 13 days later, on December 21 (fig. 6.3). All investigators place this event in the Santa Barbara Channel and assign a magnitude of about 7 (see Toppozada and others, 1981, and Evernden and Thompson, 1985, for two recent analyses). This sequence appears to have involved two events of comparable magnitude separated in time by about 15 minutes. A vigorous aftershock sequence accompanied the earthquakes and lasted until the end of the year at Mission Santa Barbara and Mission La Purisima Concepci6n. Reports of a tsunami appear to be exaggerated, although some kind of wave activity probably accompanied the earthquake (Toppozada and others, 1981; McCulloch, 1985).
JUNE 10, 1836 (M=6¾)
Little is known about the strong earthquake of June 10, 1836, that struck the then lightly populated San Francisco Bay region. An account of the event, published in the aftermath of the 1868 earthquake, provides the principal rationale for associating this event with the Hayward fault. Louderback (1947) systematically compared the two events and concluded that the 1836 earthquake probably ruptured the Hayward fault. Lindh (1983) proposed that the 1836 event ruptured the north half of the fault, whereas the 1868 event is known to have ruptured the south half, thereby avoiding the paradox of two large events on the same segment separated by a scant 32 years.
JUNE 1838 (M=7)
The pioneering historical work of Louderback (1947) reveals that a major earthquake with probable rupture of the San Andreas fault occurred in June 1838. Documentation of the event is so poor that its date cannot be fixed more precisely than "late June." Louderback concluded that the shock was comparable in magnitude to the 1906 earthquake. Current opinion suggests a smaller event involving only the 60+-km-Iong segment of the fault on the San Francisco peninsula (Working Group on Earthquake Probabilities, 1988).
JANUARY 9, 1857 (M=8¼)
The great Fort Tejon earthquake of January 9, 1857, I ruptured 300 km of the San Andreas fault from near Parkfield to Wrightwood and offset the fault by as much as 9½ m on the Carrizo Plain. The fault rupture and the effects of the earthquake have been extensively studied, notably by Agnew and Sieh (1978) and Sieh (1978b). The epicenter of this event appears to have been at the extreme northwest end of the fault rupture, as determined by the intensity patterns of two M=6 foreshocks centered near Parkfield (Sieh, 1978a). Strong shaking lasted from 1 to 3 minutes, consistent with unilateral rupture propagation to the southeast (fig. 6.3).
OCTOBER 21, 1868 (M=7)
The earthquake caused only two deaths in the sparsely settled southern California region. Damage was most severe along the fault zone; nearly every building sustained damage at Fort Tejon. In Los Angeles, then a city of about 4,000 people located approximately 60 km from the fault, some houses were cracked, but none were severely damaged (Agnew and Sieh, 1978). Modified Mercalli intensities (MMI's) of VII or more occurred in the San Fernando Valley, San Gabriel Valley, and Ventura region.
It is natural to compare the 1857 and 1906 earthquakes, the two greatest earthquakes of the San Andreas fault in historical time. The 1906 fault break was longer, whereas maximum and average surface offsets were larger in 1857. These differences approximately balance each other, and so the seismic moments of the two events are approximately equal. Moment magnitudes computed using comparable data are M=7.8 for the 1857 earthquake and M=7.7 for the 1906 event. A summary magnitude of M=8¼ was assigned by analogy with the 1906 earthquake.
Known as the "great San Francisco earthquake" until 1906, one of California's most destructive earthquakes occurred on October 12, 1868, resulting from slip on the Hayward fault. Heavy damage occurred in communities situated along the fault and in San Francisco and San Jose (fig. 6.5). Sadly, many of the engineering lessons learned from this earthquake and openly discussed at the time, such as the hazards of building on "made ground" reclaimed from the San Francisco Bay or the admonition to "build no more cornices," were long forgotten by the time of the 1906 earthquake.
FEBRUARY 24, 1892 (M=7)
Figure 6.5 - San Francisco Morning Chronicle of October 28, 1868, richly illustrates severe damage sustained by buildings of poor design or located on filled land during earthquake on the Hayward fault. Reduction of this figure from its original publication size has made some type illegible; it is not needed to convey information intended by this illustration. Photograph courtesy of the Bancroft Library.
The strong earthquake of February 24, 1892, located near the United States-Mexican border was assigned to the Agua Caliente fault north of the border by Toppozada and others (1981) and to the Laguna Salada fault in Baja California by Strand (1980). The literature on earthquakes in Baja California contains numerous references to this earthquake as having originated near the Agua Blanca fault, about 100 km southwest of Strand's epicenter (for example, Richter, 1958). The two recent intensity maps clearly rule out this epicenter and place it on the southern section of the Elsinore fault system.
APRIL 19 AND 21, 1892 (M=6½ AND 6¼)
A pair of strong earthquakes rocked the west side of the Sacramento Valley on April 19 and 21, 1892, heavily damaging the towns of Vacaville, Dixon, and Winters. The first shock was stronger and caused heavy damage at Vacaville; the aftershock was more severe at Winters. The earthquakes are reminiscent of the 1983 Coalinga, Calif., earthquake, in that both sequences were positioned along the western margin of the Central Valley. Focal mechanisms of small earthquakes located along this boundary zone show numerous examples of low-angle-thrust focal-mechanism solutions of similar orientation to the Coalinga earthquake, in addition to strike-slip mechanisms (see chap. 5; Wong and others, 1988), suggesting the possibility of a similar mechanism for these 1892 earthquakes.
DECEMBER 25, 1899 (M=6.4)
Heavy damage occurred in the towns of San Jacinto and Hemet, located along the San Jacinto fault, from an earthquake on Christmas Day 1899. Six fatalities were attributed to the earthquake.
APRIL 18, 1906 (M=8¼)
The California earthquake of April 18, 1906, ranks as one of the most significant earthquakes of all time. Today, its importance comes more from the wealth of scientific knowledge derived from it than from its sheer size. Rupturing the northernmost 430 km of the San Andreas fault from northwest of San Juan Bautista to the triple junction at Cape Mendocino (fig. 6.6), the earthquake confounded contemporary geologists with its large, horizontal displacements and great rupture length. Indeed, the significance of the fault and recognition of its large cumulative offset would not be fully appreciated until the advent of plate tectonics more than half a century later. Analysis of the 1906 displacements and strain in the surrounding crust led Reid (1910) to formulate his elastic-rebound theory of the earthquake source, which remains today the principal model of the earthquake cycle.
As a basic reference about the earthquake and the damage it caused, geologic observations of the fault rupture and shaking effects, and other consequences of the earthquake, Lawson's (1908) report remains the authoritative work, as well as arguably the most important study of a single earthquake. In the public's mind, this earthquake is perhaps remembered most for the fire it spawned in San Francisco, giving it the somewhat misleading appellation of the "San Francisco earthquake" (fig. 6.7). Shaking damage, however, was equally severe in many other places along the fault rupture. The frequently quoted value of 700 deaths caused by the earthquake and fire is now believed to underestimate the total loss of life by a factor of 3 or 4. Most of the fatalities occurred in San Francisco, and 189 were reported elsewhere.
Figure 6.6 - California earthquake of 1906 showing extent of fault rupture along the San Andreas fault, location of epicenter near San Francisco, maximum extent of structural damage, and limits of perception of shock. Modified from Lawson (1908) and Toppozada and Parke (1982).
Figure 6.7 - San Francisco, Calif., on the morning of April 18, 1906. This famous photograph by Arnold Genthe shows Sacramento Street and approaching fire in the distance. Although some buildings sustained heavy damage in the earthquake, this and many other photographs taken of the city before fire swept through show no visible evidence of damage in most structures, Photograph courtesy of the Fine Arts Museums of San Francisco, Achenbach Foundation for Graphic Arts.
At almost precisely 5:12 a.m. local time, a foreshock occurred with sufficient force to be felt widely throughout the San Francisco Bay area. The great earthquake broke loose some 20 to 25 slater, with an epicenter near San Francisco (Bolt, 1968; Boore, 1977). Violent shocks punctuated the strong shaking, which lasted some 45 to 60 s. The earthquake was felt from southern Oregon to south of Los Angeles and inland as far as central Nevada (fig. 6.6). The highest MMI's of VII to IX paralleled the length of the rupture, extending as far as 80 km inland from the fault trace. One important characteristic of the shaking intensity noted in Lawson's (1908) report was the clear correlation of intensity with underlying geologic conditions. Areas situated in sediment-filled valleys sustained stronger shaking than nearby bedrock sites, and the strongest shaking occurred in areas where ground reclaimed from San Francisco Bay failed in the earthquake. Modern seismic-zonation practice accounts for
the differences in seismic hazard posed by varying geologic conditions (see Borcherdt, 1975, and Ziony, 1985, for analyses of the San Francisco Bay and Los Angeles regions, respectively).
The characteristics and amount of surface fault slip in this earthquake varied to a remarkable degree along the length of the rupture. Peak displacements of 6 m were measured near Olema on the Point Reyes peninsula, where the surface trace of the rupture formed a sharp, well-defined break (fig. 6.8). In contrast, the fault break was extremely difficult to recognize along its southern-most 90 km, where the surface offset averaged only about 1½ m or less (see chap. 7).
Figure 6.8 - Trace of 1906 earthquake rupture near point of maximum offset (6 m) near Olema on the Point Reyes peninsula north of San Francisco. Photograph by G.K. Gilbert. View northwestward.
The magnitude of 8.3 commonly quoted for the 1906 earthquake comes from Richter (1958) and, within the precision of reporting, is identical to the 8¼ listed by Gutenberg and Richter (1954). Table 6.1 also lists other magnitudes for this earthquake, derived from recent analyses of both the same data used by Gutenberg and Richter and new data. Strictly speaking, a "Richter magnitude" (ML) for the earthquake cannot be determined because no appropriate seismographs were in operation at the time. Jennings and Kanamori (1979) used related measurements extracted from simple pendulums at Yountville, Calif., and Carson City, Nev., to derive ML=6.9, substantially smaller than the traditionally quoted value. ML which is based on the single largest peak on a seismogram at approximately 1-s period and takes into account neither the duration of the event nor longer period motions, is saturated for this event.
NOVEMBER 21, 1915 (M=7.1)
Geller and Kanamori (1977) used the unpublished worksheets of Gutenberg and Richter to compute a body-wave magnitude of mb=7.4, using the procedure of Gutenberg and Richter (1956). Because long-period (14 s) P-waves were used in this calculation, it cannot be directly compared to the short-period mb values routinely reported today.
Other workers since Gutenberg and Richter have studied the long-period surface waves of the 1906 earthquake and computed Ms values. Bolt (1968) confirmed an Ms of about 8¼, whereas Lienkaemper (1984) found Ms= 8.3 from an analysis of all the records collected by Reid (1910). Lienkaemper's magnitude combined data from both damped and undamped instruments, correcting each for magnification at the appropriate period of motion. Abe (1988), who analyzed only the undamped Milne seismograms, obtained Ms= 7.8, using slightly different procedures and a systematic set of station-magnitude corrections. Also, the four damped seismometers (all in Europe) give Ms=8.1. Longer period (50-100 s) surface waves analyzed by Thatcher (1975) indicate a seismic moment of 4x1027 dyne-cm, equivalent to M=7.7, in agreement with the seismic moment of 5x1027 dyne-cm obtained from geodetic
data, thus giving M=7.8 (Thatcher and Lisowski, 1987). Finally, Toppozada and Parke (1982) assigned an intensity magnitude (MI) of 7.8 on the basis of the total area (48,000 km2;) undergoing shaking of MMI VII or higher.
The "traditional" magnitude of 8¼ is retained here, except where seismic moment is used for quantitative purposes.
The major earthquake of November 21, 1915, triggered a spectacular steam eruption of a mud volcano, creating a 100+-m crater in Volcano Lake, Baja California, near the north terminus of the Cerro Prieto fault. Extensive cracking of the levee around the lake was noted at the time of the shock, but no tectonic ground displacements were found (Seismological Society of America Bulletin, 1916). This event may well be related to the November 29, 1852, earthquake (M=6 ½±), which also triggered a mud-volcano eruption at Volcano Lake that was observed at Fort Yuma, Ariz. Each of these events was probably associated with the Cerro Prieto fault.
APRIL 21, 1918 (M=6.9)
The communities of Hemet and San Jacinto were severely damaged for the second time in 19 years by the large earthquake of April 21, 1918, on the San Jacinto fault. Both the 1899 and 1918 earthquakes produced similar intensity patterns throughout the southern California region, and these two events have been compared to each other. However, surface waves on Milne seismograms at common stations (Victoria, British Colombia, and Toronto, Ontario, Canada; San Fernando, Spain) average 3 times larger for the 1918 earthquake, corresponding to a difference in Ms of ½ unit. As with the 1836 and 1868 earthquakes on the Hayward fault, the relation between the rupture zones in these two events is unclear. Surprisingly, no surface rupture was found for an event of this size, despite a specific search for it.
NOVEMBER 4, 1927 (M=7.3)
The Lompoc earthquake of November 4, 1927, is the largest known event in the San Andreas system west of the San Andreas fault proper. This event produced a tsunami with local runup heights of 1.5 to 1.8 m (McCulloch, 1985). The exact location of the earthquake and its association with any causative structure remain the subject of a spirited debate (Gawthrop, 1978, 1981; Hanks 1979, 1981).
MARCH 11, 1933 (M=6.3)
Rupture of the Newport-Inglewood fault on March 11, 1933, caused major damage and a loss of 115 lives in Long Beach and surrounding parts of the Los Angeles Basin. Structural damage to public schools was particularly serious, and had the event occurred when schools were in session, the calamity would have been far worse. The Field Act, mandating construction standards for schools in California, was enacted as a consequence of the earthquake.
DECEMBER 30 AND 31, 1934 (M=6.5 AND 7.0)
The major sequence that occurred along the Cerro Prieto fault on December 30 and 31, 1934, appears to have ruptured the surface trace of the fault near where it enters the Gulf of California. Aerial photographs of the fault crossing a tidal flat taken in 1935 show very fresh appearing fault morphology; subsequent photographs display a substantially subdued morphology (Kovach and others, 1962).
The Imperial fault was discovered from its 60+- km-long rupture in the Imperial Valley earthquake of May 19, 1940. Faulting was predominantly right-lateral strike slip and attained a peak offset of more than 6 m at the United States-Mexican border (fig. 6.9). The first instrumental measurement of strong ground motion adjacent to a fault rupture was obtained from an accelerograph located about 7 km from the surface trace. This record, which provides clear evidence of irregular seismic-energy release during the course of the event (Trifunac and Brune, 1970), has played a major role in shaping building codes for earthquake-resistant design.
JULY 21, 1952 (M=7.7)
Figure 6.9 - Surface faulting in 1940 Imperial Valley earthquake offset (6 m) regular rows of orange trees. Fault displacement along this section of the Imperial fault was confined to a narrow zone.
The Kern County or Arvin- Tehachapi earthquake of July 21, 1952, ruptured the White Wolf fault in the largest event to strike California since 1906. The earthquake led to 12 fatalities, and 2 more occurred during a large aftershock on August 22. Field studies of the earthquake (Oakeshott, 1955) describe the geologic, seismologic, and engineering aspects of the earthquake. From a tectonic standpoint, this event is notable for its conjugate relation to the San Andreas fault. Left-lateral slip with a significant reverse-slip component occurred on the northeast-striking, south-dipping fault plane.
FEBRUARY 9, 1956 (M=6.8)
More than 19 km of the hitherto-unknown San Miguel fault in Baja California ruptured in the earthquake sequence of February 9, 1956. The fault offset was consistently right lateral and up to the northwest, and attained maximum horizontal and vertical separations of 78 and 91 cm, respectively (Shor and Roberts, 1958). The sequence contained numerous aftershocks, including three of M≥6. About 2 years earlier, a pair of M=6 events that occurred to the south and west of the San Miguel fault may have been associated with the Agua Blanca fault.
APRIL 9, 1968 (M=6.5)
The Borrego Mountain earthquake of April 9, 1968, produced the first documented rupture of the San Jacinto fault system when right-lateral displacements of nearly 0.4 m occurred along a 30-km-Iong segment of the Coyote Creek fault. The U. S. Geological Survey (1972) published a detailed description of the event.
FEBRUARY 9, 1971 (M=6.5)
The San Fernando earthquake of February 9, 1971, ranks as one of the most serious California earthquakes in historical time. The event claimed 58 lives and caused more than half a billion dollars in property damage, including the destruction of two hospitals, a freeway interchange, and the Van Norman Dam. The earthquake ruptured north-dipping, high-angle reverse faults beneath the southern margin of the San Gabriel Mountains and broke the surface along a discontinuous, 15-km-Iong zone. Surface displacements averaged about 1 m. Seismograms of the earthquake reveal a steeply dipping deep fault and a more shallowly dipping near-surface fault (Langston, 1978; Heaton, 1982). Numerous publications report on detailed investigations of this event, including the summary report published by the U. S. Geological Survey (1971).
OCTOBER 15, 1979 (M=6.5)
The Imperial fault ruptured for the second time in less than 40 years in a major surface-faulting earthquake on October 15, 1979 (U.S. Geological Survey, 1982). The event broke the north 30 km of the fault, or approximately half the length of the 1940 fault break. However, it was clearly much smaller than the earlier event; maximum surface offsets were well under 1 m, in contrast to 6 m observed in 1940, and the seismic moment was smaller by nearly an order of magnitude. Within the zone of overlapping surface rupture, the two events display nearly identical displacement profiles (Sharp, 1982), suggesting that the 1979 earthquake represents a characteristic rupture of this segment of the fault. Strong-ground-motion records for the 1979 earthquake form an unparalleled suite of near-field recordings and have stimulated numerous investigations into the dynamics of the source.
MAY 2, 1983 (M=6.5)
Our understanding of the nature of the earthquake hazard posed by active faults in the San Andreas fault system was fundamentally altered by the occurrence of the Coalinga earthquake of May 2, 1983, on a low-angle thrust fault deep beneath the western margin of the San Joaquin Valley (Rymer and Ellsworth, 1990). Before this event, it had been thought that the major, seismically active faults in California could be recognized on the basis of their surface exposures and record of late Quaternary activity. However, no surface expression exists for the fault system responsible for either this event or the M=5.9 North Kettleman Hills earthquake of August 4, 1985, that adjoins it to the southeast. Instead of a surface fault, the buried deformation is expressed at the surface by active folds (the Coalinga anticline and the Kettleman Hills) that grew during the earthquakes (Stein and King, 1984).
NOVEMBER 24, 1987 (M=6.6)
The orientation of the fault and the style of movement on it present another major challenge to prevailing models of the San Andreas system, because this earthquake resulted from a release of compressive forces oriented nearly perpendicular to the trace of the San Andreas fault. Accumulating evidence on the orientation of the stress fIeld astride the San Andreas fault suggests that only a small component of the total stress acts to accommodate the plate motion along the San Andreas fault itself (Mount and Suppe, 1987; Zoback and others, 1987).
The Superstition Hills fault ruptured in its entirety on November 24, 1987. The total amount of separation substantially increased by persistent afterslip in the months after the main shock; in fact, the rate of afterslip was so great on the south half of the surface break as to leave open the possibility that all of its displacement occurred as afterslip. The earthquake was preceded by a major fore shock (M=6.2), on a conjugate, northeast-trending, left-lateral strike-slip fault that intersected the Superstition Hills fault at the main-shock epicenter. The surface-faulting pattern of the entire sequence was particularly remarkable for the occurrence of numerous breaks on other conjugate faults in the north quadrant around the main break (see Hanks and Allen, 1989).
OCTOBER 18, 1989 (M=7.1)
In the late afternoon of October 17, 1989, the San Andreas fault ruptured in its first major earthquake since 1906 at 5:04 p.m. P.d.t. (0004 G.m.t. on Oct. 18). Centered along a remote segment of the fault in the southern Santa Cruz Mountains, the Loma Prieta earthquake reruptured the southernmost 40 km of the 1906 fault break, producing the Nation's most costly natural disaster. The earthquake claimed 62 lives and injured an additional 3,757 people. It destroyed 963 homes and damaged more than 18,000 others, displacing 12,000 people from their residences. The combined dollar loss to the private and public sectors exceeded $6 billion (Platker and Galloway, 1989).
Damage in the epicentral region was most severe where the earthquake shaking was compounded by local ground failures, commonly involving landslide movement but also including some fractures of probable tectonic origin; the shaking clearly reactivated some fissures observed in 1906. Primary fault displacement, however, did not reach the surface. In the hard-hit communities of Santa Cruz, Watsonville, and Los Gatos, unreinforced-masonry buildings bore the brunt to the damage, and ground conditions played a significant role in the damage patterns.
The earthquake also caused grave damage and claimed the greatest number of lives far to the north, in San Francisco and Oakland, about 100 km from the epicenter. There, the earthquake selectively destroyed structures known to be at risk or located on poor ground (Platker and Galloway, 1989). The root cause of the devastation in the Marina District of San Francisco (fig. 6.10), as well as at most other sites along the margin of the San Francisco Bay, was liquefaction-induced ground failure. All of these localities sit on land reclaimed from the bay and are underlain by young, water-saturated sedimentary deposits. As we know from the clear lessons of history, provided by the earthquakes of 1865, 1868, and 1906 (Lawson, 1908), such materials perform poorly even under modest levels of earthquake shaking. The collapse of the double-decked section of California Interstate Highway 880 in Oakland (fig. 6.10), where 41 people died, resulted principally from design defects. The section of
the viaduct that collapsed was founded on soft estuarine sedimentary deposits that amplified the strong ground motion; the adjoining section, founded on alluvium, rode through the earthquake.
Figure 6.10 - Damage in October 18, 1989, Lorna Prieta earthquake occurred at distances as far as 100 km from the epicenter in areas underlain by water-saturated, unconsolidated material. A, Liquefaction-induced ground failure in the Marina district of San Francisco (top) was restricted to land reclaimed from the San Francisco Bay. B, In Oakland, the second deck of Interstate Highway 880 collapsed onto the first deck. Here, poor design was the principal culprit, although failed section sits atop estuarine sedimentary deposits that amplified the shaking.
The earthquake broke the San Andreas fault where it makes a conspicuous leftward bend, connecting straighter subparallel segments to the north and south. The fault plane dips 70° SW., and movement in the earthquake involved comparable amounts of right-lateral strike slip and reverse slip, a kinematic response driven by the need to remove material from this compressional fault bend as the Pacific plate moves to the northwest around it. The rupture nucleated at the base of the seismic zone, at 18-km depth, and spread unilaterally upward and bilaterally along strike, filling a conspicuous void in the preevent seismicity. Geodetic data collected immediately after the event suggest an average strike-slip displacement of 1. 6 m and an average reverse-slip displacement of 1.2 m, rising from the hypocenter at 18 km to within 6 km of the surface.
SEISMICITY OF THE WESTERN BASIN AND RANGE PROVINCE
The Loma Prieta earthquake fulfilled a long-term forecast for the rupture of this specific segment of the San Andreas fault (Lindh, 1983; Sykes and Nishenko, 1984; Working Group on California Earthquake Probabilities, 1988). The high earthquake potential assigned to this segment stemmed from its behavior in the 1906 earthquake, when the fault displacement, as measured at the surface, averaged about 1.5 m, far less than the average for the entire rupture. Estimates of the long-term slip rate along this segment of the San Andreas fault suggested that the strain released in the 1906 earthquake would be renewed in 75 to 136 years, implying that another earthquake was possible in the coming decades. With its occurrence, the Lorna Prieta earthquake became the second event in 2 years to fill a recognized seismic gap along the San Andreas; the first was the 1987 Superstition Hills earthquake. The Lorna Prieta earthquake also represents the third historical rupture of this segment of the San
Andreas fault; the first was the October 8, 1865, earthquake, nominally assigned M=6½, which also caused liquefaction-induced ground failure in San Francisco.
The advent of plate tectonics and its application to western North America by Atwater (1970) provided a unifying framework for the contemporary tectonics of the western Basin and Range and its interaction with the San Andreas fault system to the west. Deformation within the province reflects soft coupling of the San Andreas fault system to the North American craton and distribution of the relative plate motion-by mechanisms yet unknown - well over 500 km into the continental interior. The ubiquity of normal- fault-bounded ranges throughout the province tends to belie the underlying nature of present-day deformation within the region. Within historical time, this region has undergone nearly equal proportions of extension on normal faults and dextral shear on strike-slip faults.
The earthquake history of the western Basin and Range province is poorly known before the instrumental period, owing to sparse settlement of this high-desert region. The deficiencies of this record are illustrated by the uncertainties associated with fresh-appearing fault scarps discovered in 1911 near the north end of what would become the rupture zone of the 1954 Fairview Peak earthquake. Upon reviewing the scant historical evidence, Slemmons and others (1959) concluded that these scarps formed about 1903. The absence of an event of sufficient size in the instrumental record, however, suggests that the scarp forming event is older (or substantially smaller than M=6). Current understanding of 19th-century seismicity includes an episode of activity along the California-Nevada State line, including a probable rupture of the Olinghouse fault on December 27, 1869 (Sanders and Slemmons, 1979), although this conclusion was questioned by Toppozada and others (1981).
Surface faulting has accompanied numerous earthquakes in the region, the most significant of which are discussed below. Notable additional surface-faulting events include the M=6.3 Excelsior Mountain, Nev., earthquake of 1934 and the ML=5.6 Fort Sage Mountain earthquake of December 14, 1950, located in northeastern California (Gianella, 1957). Ground rupture may have also accompanied the M=6 earthquake of January 24, 1875 (see Gianella, 1957). If so, this observation would move the epicenter listed in table 6.1 to lat 39 ¾° N., long 120 ½° W.
MARCH 26, 1872 (M=7.6)
The town of Lone Pine, Calif., was virtually leveled when the entire 100 to 110-km length of the Owens Valley fault ruptured on March 26, 1872, in one of the largest earthquakes in U. S. history. This fault, which lies in the middle of Owens Valley, is distinct from the normal faults bounding the front of the Sierra Nevada to the west. Considerable confusion has existed in the literature regarding the style of faulting in the 1872 earthquake, including interpretations of right-lateral, left-lateral, and normal-fault movement. A recent study of the earthquake offsets by Beanland and Clark (in press) unambiguously demonstrates that fault movement was predominately right-lateral strike slip, with an average horizontal displacement of 6 m (fig. 6.10). The vertical offsets were clearly smaller and averaged about 1 m down to the east. Beanland and Clark estimate a moment magnitude of M=7.5-7.7. Faulting in 1872 largely reactivated earlier Holocene scarps, as recognized by G.K.
Gilbert when he visited the area in 1883.
The event was felt throughout most of California and Nevada, and as far east as Salt Lake City, Utah. Adobe and brick buildings in Owens Valley sustained the brunt of the damage. Minor damage also occurred in the San Joaquin and Sacramento Valleys on the opposite side of the Sierra Nevada, at distances of as far as 400 km. In Yosemite Valley, John Muir witnessed a spectacular rockfall triggered by the earthquake. As severe as the ground shaking must have been, it was noted in the Inyo, Calif., Independent of April 6, 1982, "* * * that not a person would have been killed or hurt had their houses all been made of wood." It is of some historical interest that the first long-term earthquake forecast, made by G.K. Gilbert in 1883 to the citizens of Salt Lake City, was based in part on his observations of the 1872 earthquake. In it, he noted that the rebuilding of Independence with wood-frame buildings was an extravagance, because this great shock had relieved the accumulated
strain, and so many generations would pass before conditions would permit another similar shock to occur (Gilbert, 1884):
The old maxim, "Lightning never strikes in the same spot twice" is unsound in theory and false in fact; but something similar might truly be said about earthquakes. The spot which is the focus of an earthquake (of the type here discussed [1872 Owens Valley]) is thereby exempted for a long time.
Many comparisons have been drawn between the Owens Valley earthquake and the great San Andreas earthquakes of 1857 and 1906. The size of the regions shaken in all three events are comparable, as are the maximum fault displacements. The two San Andreas events have significantly longer rupture lengths, and their seismic moments are larger by a factor of 2 to 3. Whether or not any or all of these earthquakes can be classified as "great" earthquakes becomes a question of semantics. All of them can be classified as great on the basis of their rupture lengths of 100 km or more (Kanamori, 1977), but they all have seismic moments more than 100 times smaller than the largest known earthquakes, such as the M=9.2 Alaska earthquake of 1964. Practically speaking, these events are among the largest known strike-slip events, and they must be close to the size of the largest possible strike-slip events along the San Andreas fault system.
OCTOBER 3, 1915 (M=7.3)
The 1915 Pleasant Valley, Nev., earthquake of October 15, 1915, created a series of spectacular normal-fault scarps in the central Nevada seismic zone of the Basin and Range province (figs. 6.11, 6.12). Four major scarps formed during the earthquake, with an aggregate length of 59 km, and reruptured Holocene scarps at the base of the mountain blocks (Wallace, 1984). Fault movement in the earthquake appears to have been purely dip slip and averaged about 2 m on the steeply dipping fault plane. The earthquake was felt from western Utah to the Pacific coast and from northeastern Oregon to the United States-Mexican border. Instrumental measures of the magnitude range from 7.3 to 7 ¾ and exceed the moment magnitude of 7.2 derived from field measurements (M0=6.1x1027 dyne-cm).
Figure 6.11 - California-Nevada region, showing locations of major historical earthquakes in the western Basin and Range province, 1857-1989. Focal mechanisms of five largest events in lower-hemisphere projection show compressional quadrant shaded and indicate significant shear as well as extensional strain in province. Seismic gaps (labeled) are potential loci of future major earthquake activity (Wallace, 1984).
Figure 6.12 - Fault trace of 1915 Pleasant Valley, Nev., earthquake remains clearly visible in this photograph by R.E. Wallace more than 60 years after event (Wallace, 1984).
The Pleasant Valley earthquake lies at the north end of a 500-km-long belt of historical surface-faulting earthquakes within the central Nevada seismic zone and Owens Valley fault system. The four major earthquake sequences in this zone since 1872 leave two conspicuous seismic gaps that have been discussed as the potential loci of future major earthquake activity (fig. 6.11; Wallace, 1984).
DECEMBER 21, 1932 (M=7.2)
The second historical surface-faulting event in the central Nevada seismic zone on December 21, 1932, produced a discontinuous zone of surface faulting and fissures in the valleys west and north of Cedar Mountain (Gianella and Callaghan, 1934). Within the 60-km-long, north-northwest-trending zone where faulting was observed, most breaks struck east of north and showed clear evidence of right-lateral displacements (fig. 6.11).
JULY 6, 1954 (M=6.6), AND AUGUST 24, 1954 (M=6.8)
The Rainbow Mountain earthquakes of July 6 and August 24, 1954, produced a zone of east-facing normal-fault scarps along the base of Rainbow Mountain, extending northward into the Carson Sink. The July 6 event produced an 18-km-Iong surface rupture at the south end of this zone striking N. 15° E., with maximum displacements of about 30 cm. The August 24 shock extended the zone by 22 km in a N. 20° E. direction, with as much as 75 cm of normal-fault slip. Tocher (1956) noted that the displacement on the northern part of the July 6 break approximately doubled in amplitude between July 16 and September 9; the timing of the additional slip could not be determined.
DECEMBER 16, 1954 (M=7.1 AND 6.8)
The Dixie V alley- Fairview Peak earthquakes of December 16, 1954, produced a 90-km-long zone of right lateral oblique and normal faulting in the central Nevada seismic zone (fig. 6.11; Slemmons, 1957). The first shock, which occurred east of Fairview Peak, produced lateral displacements of more than 4 m and vertical displacements of as much as 3 m. Faulting along this 50-km-Iong zone was predominantly down to the east opposite Fairview Peak and changed polarity to the north. The second shock, which occurred 4 minutes later, had an epicenter on the east side of Dixie Valley in a left-stepping echelon arrangement with the earlier event. Normal-fault scarps formed along a 40-km-Iong zone at the base of the Stillwater Range some 20 km west of the Rainbow Mountain faulting. Vertical displacements exceeded 2 m, and consistent strike-slip displacements were absent.
JULY 21, 1986 (M=6.2)
The Chalfant Valley earthquake of July 21, 1986, is the largest event to date in a series of 33 earthquakes of ML≥5 to occur since 1978 in the White Mountain seismic gap (Savage and Cockerham, 1987). Other principal events in this series include the May 25-27, 1980, Mammoth Lakes earthquakes (M=6.1, 5.9, 5.8, 6.0) and the November 23, 1984, Round Valley earthquake (M=5.7). The series of shocks is of interest not only because of its wide geographic distribution in the White Mountain seismic gap but also because of the contemporaneous uplift of Long Valley caldera (Hill and others, 1985). The Chalfant Valley earthquake created a 10+- km-Iong zone of fractures with as much as a few centimeters of dextral slip on the frontal-fault zone of the White Mountains (Lienkaemper and others, 1987). The earthquake focal mechanism and aftershock distribution show that the predominately dextral strike-slip displacement associated with this event occurred on a
west-dipping fault plane that projects upward to meet the surface break.
SEISMICITY OF THE MENDOCINO TRIPLE JUNCTION AND THE GORDA PLATE
The San Andreas fault terminates at its north end in a transform/transform/trench triple junction just seaward of Punta Gorda. Major earthquake activity lies along the Mendocino Fracture Zone, where it is an active transform fault, and to the north within the Gorda plate, which is undergoing intense internal deformation. The Wadati-Benioff zone is well defined to a depth of 30 km and can be traced eastward to a depth of more than 80 km (see fig. 5.5; Walter, 1986). Strong earthquakes within the Gorda plate locate off shore and span the position of the megathrust; these events appear to lie entirely within the oceanic lithosphere. The 1980 Eureka earthquake, for example, ruptured the Gorda plate from the landward to the seaward side of the megathrust. Despite the high level of seismicity, underthrusting events are rare.
NOVEMBER 23, 1873 (M=6¾)
The severe earthquake of November 23, 1873, was felt from San Francisco to Portland, Oreg.; it inflicted the heaviest damage to Crescent City, Calif., and surrounding communities in the Klamath Mountains. The macroseismic epicenter near the California-Oregon State line and probably inland of the coastline is unique within both the historical and instrumental records.
APRIL 16, 1899 (M=7)
Little is known about the large earthquake of April 16, 1899, with an epicenter seaward of Eureka, where it was described as "one of the severest shocks of earthquake ever experienced." Toppozada and others (1981) corrected the origin time of this event and assigned a nearshore epicenter and an MI of 5.7. The earthquake was assigned an epicenter in the Gulf of Alaska by Milne (1901) on the basis of the traveltime of the maximum amplitude from the five reporting stations; however, a California location satisfies his data equally well. The absence of significant damage along the coast suggests an epicenter well out to sea. An instrumental magnitude (MS) of 7.0 is derived from the surface-wave amplitudes reported by Milne (see Abe and Noguchi, 1983).
JANUARY 31, 1922 (M=7.3)
The intensity pattern of the large earthquake of January 31, 1922, is generally similar to that of the 1899 event. This event was well recorded throughout the world.
JANUARY 22, 1923 (M=7.2)
The earthquake of January 22, 1923, strongly shook the Cape Mendocino region and toppled many chimneys in the area. This earthquake was probably associated with the Mendocino Fracture Zone.
DECEMBER 21, 1954 (M=6.6)
The strong earthquake of December 21, 1954, apparently was located in the crust of the North American plate above the descending Gorda plate. The relocation of this event by Smith and Knapp (1980) suggests a possible association with the active Mad River fault zone. One fatality is attributed to the earthquake.
NOVEMBER 8, 1980 (M=7.2)
The Eureka earthquake of November 8, 1980, resulted from 100-km-Iong, left-lateral strike-slip rupture of the Gorda plate along a northeast-striking fault (see fig. 5.5). Aftershocks of the earthquake extended from within 30 km of the coastline southwestward to the Mendocino Fracture Zone. The focal mechanism of the earthquake is thus conjugate to the San Andreas, with its tension axis aligned in the downdip direction. This event argues for high rates of internal deformation within the subducting oceanic lithosphere and against the extension of San Andreas-style faulting northward of the triple junction.
The spatial distribution of large earthquakes during the past 2 centuries defines the San Andreas fault system as a 100- to 300-km-wide zone containing numerous active faults in addition to the San Andreas fault proper (fig. 6.1). Except for the two largest events, the great 1857 and 1906 earthquakes that together ruptured two-thirds of the total length of the San Andreas fault, large earthquakes are conspicuously absent along the master fault itself. Although these two great earthquakes account for half of the seismic-strain release since 1769, most of the rest occurs on other, smaller elements of the fault system. Major historical events on these secondary faults, such as the 1927 Lompoc and 1952 Kern County earthquakes, serve to define the boundaries of the San Andreas system. Their mechanisms differ significantly from right-lateral strike slip parallel to the plate-motion vector and illustrate the diversity and complexity of seismic-strain release within the plate-boundary
Over the timespan of the historical catalog, the most enduring characteristic of the earthquake distribution may be the spatial clustering of activity at specific localities along the plate boundary. Notable hotspots include the Cerro Prieto, Imperial, San Jacinto, and Calaveras faults, all of which are major branches of the San Andreas fault, and the Parkfield segment of the San Andreas fault in the transition zone between the 1857 rupture and the 150-km-Iong central, creeping segment of the fault. In each of these areas, the seismic activity coincides with these high-slip-rate faults (1-3.5 cm/yr),
and in some places it clearly represents recurrent rupture of the same segment of the fault. At greater distances from the San Andreas fault, the historical events (or sequences) tend to represent isolated occurrences on slower moving faults. Thus, although the overall seismicity spans the broad plate-boundary zone, seismic-strain release over the past 2 centuries correlates with the local rate of fault movement.
In general, the locations of historical earthquakes resemble the overall distribution of microearthquake activity, despite more than six orders of magnitude difference in average seismic moment (fig. 6.13; see chap. 5). One important difference between the distribution of large and small earthquakes is the virtual absence of smaller events along the San Andreas fault segments that ruptured in 1857 and 1906. Similarly, seismic activity is distinctly absent on the potentially dangerous segment between the 1857 break and the Imperial Valley. Except for the central, creeping segment, where numerous small events occur, the San Andreas fault is almost completely aseismic during the long intervals between its rupture in major earthquakes (see figs. 5.6, 5.9).
Figure 6.13 - Distribution of large and small earthquakes along the San Andreas fault system. In general, modern instrumental data (C; see chap. 5) portray same pattern of activity seen in large earthquakes from preinstrumental (A) and instrumental (B) eras. Some areas characterized by high levels of microearthquakes, such as well-defined faults east of northern section of the San Andreas fault (red line), have not produced significant earthquakes in historical time and so are considered probable sites of future activity.
This inverse correlation between the source regions of large earthquakes and the distribution of smaller events can also be observed for smaller main shocks. Recent studies of the rupture dynamics of M=6 events occurring within seismically active regions indicate that the rupture zones of these events are similarly aseismic, with smaller events occurring predominantly outside the slip surface, even during the aftershock sequence (Reasenberg and Ellsworth, 1982; Hartzell and Heaton, 1986; Mendoza and Hartzell, 1988). Thus, the sites of future large earthquakes cannot be identified on the basis of minor seismicity alone.
RATE OF SEISMICITY
The average rate of earthquake activity within the San Andreas system can be estimated from the Gutenberg-Richter frequency-magnitude relation log N=a-bM, where N is the cumulative number of events of magnitude equal to or greater than M during a given time period. For the 77 events along the fault system with summary magnitudes M≥6 since 1852, this relation well describes the population with a=5¾ and b=1 (fig. 6.14). Comparable results are obtained for subsets of M=6 events, such as the instrumental period (1898-1989).
Figure 6.14 - Annual frequency of earthquakes of magnitude ≥M as derived from historical and modern instrumental catalogs. Gutenberg-Richter frequency-magnitude relation, log N=a-bM with a=5¾/yr and b=1, describes observed distribution of earthquakes of M≥6 within the San Andreas fault system during 138-yr interval from 1852 to 1989. Also shown are annual frequency of M≥6 events from the broader Pacific-North American plate boundary, including the San Andreas fault system and the western Basin and Range province, and of M=2-4.5 events in both regions during 1980-86.
It is useful to compare these results from the historical record with the frequency-magnitude relation determined from systematic microearthquake observations. If the historical rate of activity applies today and the frequency-magnitude relation for microearthquakes (M≥3) is described by the same relation, then about 5,600 M≥2 events should be observed each year. This prediction exceeds the number of events observed during the interval 1980-87 by about a factor of 2 (see chap. 5) and suggests that a somewhat smaller value of b = 0.93 may be more appropriate for the extended magnitude range.
PARADOX OF THE MISSING PLATE MOTION
For the catalog as a whole, the rate of earthquake occurrence is well described by a Poisson process, in which the probability of finding one or more events in any interval of t years is P= 1-e-λt, where λ is the average rate of earthquake occurrence. It follows from the observed frequency-magnitude relation that the odds of having at least one M≥6 event per year are 0.43. There is also an even chance of at least one M≥6 event within any 15-month interval, one M≥7 within any 12½-year interval, or one M≥8 within any 125-year interval.
The rate of earthquake activity along the plate boundary can also be usefully compared with plate-motion estimates derived from plate-tectonic theory. Current estimates of the relative velocity across the North American- Pacific plate boundary, determined from the spreading rate in the Gulf of California of 5 cm/yr (DeMets and others, 1987), imply an annual seismic-moment rate of 2x1026 dyne-cm/yr for a 10-km-thick brittle crust, equivalent to a single M=6.8 earthquake. Earthquakes of this size occur far less often, and the principal seismic contribution to the plate motion comes from infrequent large events. The erroneous notion that the smaller events substantially contribute to the total is demonstrably false, as shown by summing the contributions of all the earthquakes below some magnitude. The innumerable events of M≤6 occurring each year contribute less than 10 percent to the total seismic-strain release.
Within the San Andreas fault system, the total seismic-moment release since 1852 corresponds to 70 percent of the total North American-Pacific plate motion predicted by plate-tectonic models. This proportion is somewhat inflated because not all of the earthquakes act to transmit slip along the plate boundary; for example, the 1952 Kern County earthquake, the third largest in historic time, directly accommodated little plate-parallel motion. Although aseismic displacements account for some of the deficit, notably along the central, creeping section of the San Andreas fault, deformation occurring elsewhere, notably within the Basin and Range, contributes substantially to the relative motion between the North American and Pacific plates.
The discrepancy between plate-tectonic estimates of relative motion across the North American-Pacific plate boundary and seismic estimates also holds for geologic and geodetic estimates of motion along the San Andreas fault system. The explanation of this apparent paradox is thought to include deformation within the Basin and Range province in western Nevada and eastern California (Atwater, 1970), which has been the locus of major seismic activity in historical time, including the third largest event, in 1872, and 3 of the 11 M≈7 events in the 20th century.
EARTHQUAKE RECURRENCE AND CHARACTERISTIC EARTHQUAKES
It has long been recognized that the Basin and Range province has undergone substantial extension during the Cenozoic and is presently opening in a N. 60° W. direction (Zoback and Zoback, 1980). Historical seismicity partly agrees with this geologically derived pattern; however, it also indicates a significant component of dextral shear in nearly every well-studied historical event (Shawe, 1965; Doser, 1986). Because the geologic expression of strike- slip displacement is much more difficult to recognize and quantify than vertical slip, a major question is raised about the significance of the historic seismicity for the total strain within the western Basin and Range.
Since the 1872 earthquake, the net seismic strain within the Basin and Range province can be estimated by summing the contributions of individual events. The net shear strain thus determined indicates nearly equal components of extensional strain in a N. 60° W. direction and dextral shear trending N. 10° W. The resulting average-motion vector nearly coincides with the orientation of the San Andreas fault, and the lateral slip largely balances the coastward expansion of the province that results from extension alone. If both the rate and style of historical faulting accurately portray the long-term deformation within the region, they diminish the discrepancy between the predicted and observed rates of motion across the North American-Pacific plate boundary.
Over geologic time, the net displacement across a fault accumulates through the action of countless individual slip events. Measured over many displacement cycles, the average interval between events must equal the average event displacement divided by the remote slip rate. First principles, however, provide little guidance as to the properties of the recurrence, which might range from a totally random distribution of events in both space and time to identical earthquakes repeating at fixed intervals. If recurrence is essentially random, then long-term seismic hazard is described by the Poisson rate of activity, as discussed above. Greater regularities and systematics in recurrence imply that useful time-dependent forecasts of future activity can be derived from knowledge of the past behavior of the fault system.
Results for San Andreas earthquakes have played a central role in establishing the existence of broad regularities in the recurrence process. At Parkfield, the San Andreas fault has ruptured six times since 1857 in M≈6 events with highly repeatable characteristics every 22±6 years. The latest three events, in 1922, 1934, and 1966, for which instrumental records exist, are virtually identical (fig. 6.15; Bakun and McEvilly, 1984). Amplitude data from Milne seismographs uncovered in the preparation of table 6.1 show that the 1901 and 1922 events produced the same surface-wave amplitudes on common stations, strengthening earlier speculations that all the 20th-century events are similar. Intensity data for the 1881 event (Toppozada and others, 1981) and for foreshocks to the great 1857 earthquake (Sieh, 1978b) place them along the Parkfield segment as well. These regularities in the size, location, and timing of all known events at Parkfield led Bakun and Lindh (1985) to
propose a specific recurrence model for Parkfield earthquakes. On the basis of this model, the next in the series of characteristic events is anticipated before 1993, and its forecast represents the first formally endorsed earthquake prediction in the United States.
Figure 6.15 - Surface waves of 1922, 1934, and 1966 Parkfield, Calif., earthquakes as recorded on same seismograph in DeBilt, The Netherlands (DBN; EW, east-west; NS, north-south). These nearly identical waveforms and amplitudes led Bakun and McEvilly (1984) to propose recurrent rupture of same segment of the San Andreas fault as mechanism of Parkfield earthquakes.
Geodetic analysis of the strain released in the 1966 earthquake and its subsequent reaccumulation led Segall and Harris (1987; see Harris and Segall, 1987) to identify the zone where strain accumulates and is released, the "Parkfield asperity," as the center of the 1966 rupture zone. This zone of strain accumulation appears to be effectively locked during the interseismic period and corresponds to the center of the 1966 aftershock zone (Eaton and others, 1970) between about 4- and 10-km depth. The significantly fewer events in this part of the aftershock zone than in its periphery suggests that Parkfield earthquakes occur when this locked zone suddenly releases. Aftershocks appear to result from transfer of stress to the perimeter of the asperity.
THE SEISMIC CYCLE
This same pattern of concentrated coseismic slip occupying a quiet region within the overall aftershock distribution characterizes several recent, well-studied events (Mendoza and Hartzell, 1988), three of which, the Coyote Lake earthquake (Aug. 6, 1979), the Imperial Valley earthquake (Oct. 15, 1979), and the Morgan Hill earthquake (Apr. 24, 1984), all have probable antecedents within the historical record. Reasenberg and Ellsworth (1982) identified the June 20, 1897, earthquake as a predecessor to the 1979 event and noted that the 82-year interval between events equaled the 1.2 m of coseismic slip determined by Liu and HeImberger (1983) divided by the long-term slip rate of 1.5 cm/yr for the Calaveras fault. Similarly, the 73- year interval between the July 11, 1911, event and the 1984 Morgan Hill earthquake (Bakun and others, 1984) well predicts the 0.8 to 1.0 m of maximum coseismic slip determined by Hartzell and Heaton (1986). The 1979 Imperial Valley earthquake is more complex
because it reruptured only the northern 30 km of the May 19, 1940, fault break. Again, both the time interval between events and the fault-slip rate compare favorably with the fault slip at depth, as determined from seismograms (Hartzell and Heaton, 1983; Archuleta, 1984). Earlier ruptures of this or other segments of the Imperial fault may well be in the historical record, possibly including the April 19, 1906, event, which occurred the afternoon of the great 1906 earthquake in northern California.
Similar observations of recurrent faulting in events with characteristic magnitudes and locations from around the world (Nishenko and Buland, 1987) suggest a simple, first-order model for seismic potential. In this model, the future behavior of a specific segment of a fault can be forecast from knowledge of the size of past earthquakes, the timing and amount of slip in the latest event, and the long-term rate of fault movement (Lindh, 1983; Sykes and Nishenko, 1984). Accordingly, the probability of an event on a recently ruptured fault segment is low until the elastic strain rebuilds, which may be estimated from the geologic slip rate. As the strain rebuilds, the probability of another earthquake increases. Empirically, the time intervals between successive ruptures of a specific fault segment define a bell-shaped distribution that may be used to estimate the odds of the next event within some future time interval, given that it has not yet occurred.
Probabilities for large earthquakes along the major branches of the San Andreas fault derived from this methodology differ markedly from Poisson estimates (Working Group on California Earthquake Probabilities, 1988). For example, the chance of a repetition of the great 1906 earthquake within the next 30 years (1988-2018) is less than 0.1. In contrast, the chance of an M=7½-8 earthquake on the southern section of the San Andreas fault is 0.6. When the Working Group's report was written, the southernmost part of the 1906 fault break was assigned the highest chance of failure of any segment of the north half of the San Andreas fault. Now that it has ruptured in the October 18, 1989, Loma Prieta earthquake, the probability of another rupture will be small for several decades. A clearer understanding of past seismicity can only help to improve and refine estimates of future seismicity.
An important implication of the characteristic-earthquake model is the existence of a repetitive cycle of strain accumulation and release (Fedotov, 1968). Mogi (1981) suggested the existence of definite stages in the cycle, including a low level of seismicity in the first part of the cycle once aftershock of the latest event subside, a rise in regional activity as strain accumulates, and ultimately the occurrence of another earthquake with its attendant fore shocks and aftershocks, which initiates the next cycle.
The long-term seismicity within the San Andreas fault system displays these characteristics along the rupture zone of the great 1906 earthquake (figs. 6.16, 6.17; Ellsworth and others, 1981). Activity was relatively high during the 19th century, as becomes particularly apparent after 1850, when the record is virtually complete. After the great 1906 earthquake, the level of seismicity changed drastically, and moderate events essentially ceased for 50 years. Since the mid-1950's, the activity level has increased and begun to approach the 19th-century level (Tocher, 1959). This change in activity associated with the 1906 earthquake has been noted many times (for example, Gutenberg and Richter, 1954), and it is an open question whether it represents a premonitory increase (Toppozada and others, 1988) or whether the long quiescent period since 1906 is the essential feature (Ellsworth and others, 1981).
Figure 6.16 - Seismicity of the San Francisco Bay region in quarter-century epochs. Activity was high in the region during at least a half-century before 1906 earthquake and drastically declined afterward for the next half-century. Since the mid-1950's, activity has begun to approach levels last seen in the 19th century. However, both geologic and geodetic evidence suggest that the next great earthquake will not occur for a century or more.
Figure 6.17 - Space-time diagram of seismicity since 1850 along the San Andreas fault system between head of the Gulf of California (A) and Punta Gorda (A'). Change in seismicity rate along northern section ofthe San Andreas fault associated with 1906 earthquake (fig. 6.13) may also be tentatively identified along 1857 earthquake rupture. Persistent activity characterizes south third of the plate boundary since 1890, spanning the entire interval of reliable earthquake reporting in this region.
Comparable variations in seismicity appear to be present in southern California, although the historical record there is less reliable until about 1890. Along the rupture zone of the great 1857 earthquake, the available data suggest a similar period of low activity for several decades after the event (fig. 6.17). Farther south, along the section of the fault that has not ruptured in 3 centuries, the activity level since at least the 1880's is reminiscent of the activity in the San Francisco Bay region before the 1906 earthquake (fig. 6.18). As a potential long-term indicator of high seismic potential, the seismicity surrounding the dormant southern section of the San Andreas fault agrees with independent estimates of long-term potential derived from paleoseismology.
FUTURE USES OF EARTHQUAKE HISTORY
Figure 6.18 - Where will the next great earthquake strike along the San Andreas fault system? Numerous lines of evidence point to its long-dormant southernmost segment (B) as having the highest potential. Large earthquake activity in this region shares many similarities with activity in the San Francisco Bay region before 1906 earthquake (A). In both cases, absence of activity directly on the San Andreas fault is pronounced, and a high regional level of activity is concentrated along other major branches of the fault system.
At this stage in our understanding of the San Andreas fault system, seismicity is still best described as a random process over time, with a highly clustered spatial distribution. There are, however, tantalizing hints of underlying regularities, such as those in the characteristic earthquakes at Parkfield, or in the striking changes in seismicity associated with the 1906 earthquake. The next generation of refinements to this history will assuredly make comparable contributions by reducing the uncertainty in earthquake locations and magnitudes. Modern seismologic methods for extracting new information on the mechanisms of earthquakes have already proved practical for many events from the early instrumental period. Systematic treatment of the full instrumental catalog with these methods will provide a new basis for understanding the tectonics of the plate boundary and the mechanics of earthquakes.
CATALOG OF MAJOR EARTHQUAKES, 1769-1989
The publication of Edward S. Holden's catalog of Pacific coast earthquakes in 1898 represented the first systematic scientific inquiry into the seismic history of California and surrounding regions. This catalog, and its extension by McAdie (1907), formed the primary basis for the monumental catalog of Townley and Allen (1939) covering the years 1769-1928. These catalogs provide detailed descriptive accounts of virtually all the earthquakes that are now known from this period, and all subsequent analyses of seismicity up to the modern instrumental period build on these foundations.
QUANTIFICATION OF EARTHQUAKES AND MAGNITUDE SCALES
Recent studies of preinstrumental seismicity have focused on quantification of the historical record. The catalog presented here relies heavily on the research of Tousson Toppozada and his associates (Toppozada and others, 1981; Toppozada and Parke, 1982), who developed extensive new information on seismic intensities from newspaper accounts and other original sources, and determined locations and magnitudes from the resulting isoseismal maps. In addition, several special studies of important events by other workers have contributed to the catalog.
The development of practical seismographic instrumentation around the turn of the 20th century led to the rapid growth of seismologic data, particularly for those events large enough to register at teleseismic distances on the early instruments. The publication of the Circulars of the Seismological Committee of the British Association for the Advancement of Science (1899-1912) and their continuation as the International Seismological Summary from 1913 on indicate a detection threshold of about M≈6 for the Western United States as early as 1898. Data from these and other sources enabled Gutenberg and Richter (1954) to systematically catalog seismicity from 1904 onward.
Modern seismographic instrumentation first installed in California in 1910 ushered in the era of earthquake observation at regional distances. The Bulletins of the Seismographic Stations of the University of California, Berkeley, from 1910 to the present form the principal source for events in northern California and adjoining areas. Routine epicentral determinations and magnitude assignments for earthquakes in the southern California region date from 1932 and are taken from the catalog of the Seismological Laboratory of the California Institute of Technology, Pasadena. Additional instrumental results come from various other sources, chiefly the U. S. Geological Survey and the University of Nevada, Reno.
The resulting catalog of major earthquakes in California, western Nevada, and northernmost Baja California from 1769 to 1989 (table 6.1) contains 206 entries. This catalog omits several earthquakes listed in earlier catalogs where this or other recent studies have failed to corroborate previous interpretation as significant events or even, in some cases, their occurrence.
Physical measures of the complex mechanical event producing the earthquake take many forms, including the dimensions of the faulted region, the amount of slip, and the strength of the radiated elastic waves. To relate the characteristics of one event to another, the observed quantities must generally be summarized through the use of either an empirical relation, such as magnitude, or a quantity derived from a physical model, such as seismic moment. Both procedures have their place, and the choice of one metric over another depends principally on the purposes of the comparison and the availability of common data.
THE RICHTER SCALE (ML)
Because no single procedure for determining magnitude can be applied to the entire historical record, the catalog must be quantified by using various magnitude scales. Each scale is briefly described below to define its origin and to clarify its relation to the other scales. I emphasize that each scale has a particular range of validity and that different magnitude scales will, in general, yield slightly different values for the same event. Such differences in magnitude seem to provide a never-ending source of interest and controversy for the news media, who commonly lump all scales together under the heading of "Richter scale." To the seismologist, such differences are neither surprising nor controversial and can, in fact, provide information on the underlying physical processes of the earthquake source.
I also emphasize that intensity scales characterize the effects of the earthquake at a particular location and are not magnitude scales. Strictly speaking, intensity values (or, for that matter, instrumentally measured values of ground-motion parameters) describe the vibratory motions that are the actual earthquake as observed at a particular location, whereas magnitude values describe the faulting event that generates the earthquake.
The original magnitude scale of Richter (1935) was introduced for the purpose of providing an objective measure of the energy of each earthquake in the initial listing of earthquakes in the southern California region compiled by the Seismological Laboratory in Pasadena. Rather than attempting to measure the energy of the earthquake source directly, he chose to construct an empirical scale derived from a simple measure of the complex seismic waveform. Using only the maximum excursion of the seismogram as measured on a single type of instrument, the Wood-Anderson seismograph, he defined the local magnitude of an earthquake as
ML = log10 A - log10 A0 (Δ) ,
where the empirical function A0 depends only on the epicentral distance of the station, Δ . The zero point was arbitrarily set by Richter to avoid negative magnitudes in the course of routine work. Use of common logarithms means that two earthquakes located at the same distance from a station and having peak amplitudes differing by a factor of 10 will differ by 1 magnitude unit. In practice, readings from all observing stations are averaged after adjustment with station-specific corrections to obtain the ML value. Although Richter (1935) predicted that the local-magnitude scale "cannot hold to any high accuracy," history has proved it to be a powerful quantitative tool for ordering the relative sizes of earthquakes.
SURFACE-WAVE MAGNITUDE (Ms) AND BODY-WAVE MAGNITUDE (mb)
Several points about ML should be emphasized. First, it is strictly defined only for the southern California region, although its applicability to coastal central and northern California has since been shown. Recent studies of the A0 curve suggest that it will require revision and regionalization. Second, because ML has no actual physical units associated with it, other empirical magnitude scales may be freely adjusted to coincide with it. The local-magnitude scale has, in fact, been used as the basis for establishing essentially all other magnitude scales. Finally, because ML is derived from measurements taken from a single, band-limited seismograph, ML values saturate once an earthquake becomes large enough. Thus, the "correct" Richter magnitude ML=6.9 for the great 1906 earthquake obtained by Jennings and Kanamori (1979) reflects the amplitude of seismic waves at periods
near 1 s but not the total energy of this earthquake. Uniformly valid characterization of the "size" of an earthquake requires use of magnitude scales based on longer-period measures of the event.
The successful development of the local-magnitude scale encouraged Gutenberg and Richter to develop magnitude scales based on teleseismic observations of earthquakes. Two scales were developed, one based on surface waves, MS, and one on body waves, mb.
SEISMIC INTENSITY AND EARTHQUAKE MAGNITUDE (MI)
Surface waves with a period near 20 s generally produce the largest amplitudes on a standard long-period seismograph, and so the amplitude of these waves is used to determine MS, using an equation similar to that used for ML.
The body-wave magnitude, mb, which was developed specifically to treat deep-focus earthquakes, presents yet another alternative scale for magnitude determination. Although it presently is the most commonly reported teleseismic magnitude, current practice in its determination differs from that employed by Gutenberg, and so it is omitted from table 6.1. As a short-period magnitude, modern mb values measure the same part of the earthquake energy spectrum as ML.
The magnitudes listed by Gutenberg and Richter (1954) that appear in table 6.1 as MG-R are essentially MS according to Geller and Kanamori (1977); magnitudes attributed to Richter (1958) are based on ML or MS.
Useful estimates of MS can be obtained from many different types of long-period seismographs, including the undamped instruments deployed by Milne beginning in 1897. Abe and Noguchi (1983) constructed estimates of MS from Milne seismograms to resolve a longstanding controversy concerning an apparent peak in global seismicity between 1904 and 1912. Abe (1988) later used the Milne data to determine magnitudes for smaller earthquakes in California between 1898 and 1912. His procedures have been used to compute MS for additional California events occurring between 1898 and 1934, which are listed in table 6.1.
Before the development of seismographs in the late 19th century, descriptions of the effects of earthquakes provided the only means for assessing earthquake size in all but the rare cases where surface faulting was well described. A robust method for relating the area undergoing shaking of a given intensity or greater to ML was developed by Toppozada (1975) for California and western Nevada. Using these relations, Toppozada and others (1981, 1982) successfully assigned an intensity magnitude, MI, to many earthquakes. The isoseismal maps developed in the course of their research also generally provide our best estimates of epicentrallocations. New MI values have been determined for several events, using the same procedures as part of this study.
SEISMIC MOMENT (M0), RADIATED ENERGY, AND MOMENT MAGNITUDE (M)
Magnitude scales based on finite-bandwidth seismographs approach a maximum near which events of clearly different size or energy are indistinguishable. Saturation of ML is apparent for both the 1906 and 1952 earthquakes listed in table 6.1. Recent work by Hutton and Boore (1987) suggests that the local-magnitude scale may begin to saturate at about ML=6. Such saturation, which is understood to arise from the scaling law of the seismic spectrum (Aki, 1967), occurs when the peak of the energy spectrum lies below the frequency range of the Wood-Anderson seismograph.
SUMMARY MAGNITUDE (M)
By using the well-known properties of the seismic spectrum, magnitude scales can be constructed with uniform validity. One such scale, Mw, proposed by Kanamori (1977) is based on the seismic energy radiated in the form of elastic waves by the source. Another nearly equivalent magnitude scale, M, the moment magnitude, is based on the seismic moment, M0=μAu (Aki, 1966), where A is the area of the earthquake rupture surface, u is the average fault displacement, and μ is the shear modulus of the crustal volume containing the fault. Hanks and Kanamori (1979) took advantage of the nearly identical relations between M0 and both ML and MS to define M=(2/3)log10 M0-10.7, where M0 is measured in dyne-centimeters.
These two magnitude scales, though closely related, are not identical. Singh and Havskov (1980) showed that Mw=(2/3)(log10 M0+ log10 Δσ/μ-12.1), where Δσ is the stress drop. Earthquake stress drops generally fall in a narrow range over the entire magnitude spectrum, and so with Δσ/μ≈10-4 (Kanamori, 1977), Mw=M. One advantage to M for the purpose at hand is its dependence on only the static fault offset and rupture area, which can be determined for the 1857 and 1872 earthquakes.
To construct a single, summary-magnitude scale, M, to characterize the relative size of all the events listed in table 6.1, I use each of the scales described above, being careful to consider such factors as the historical period and event location, as well as the quality of individual determinations. Where choices between several magnitude estimates are possible, the summary magnitude, M, is weighted toward long-period estimates of magnitude. Specifically, MS and MG-R are selected when judged reliable (110 events). Many local magnitudes have thus been superseded by surface-wave magnitudes; this effect is most noticeable for the largest events, where saturation of ML becomes important. ML is the principal contributor to 20 summary magnitudes, half of which also have reported M values that agree well. For all but two events before 1898 (1857 and 1872) and for two 20th-century events, M is
based on MI. In effect, the summary magnitude is an intensity magnitude before 1898 and a teleseismic surface-wave magnitude thereafter.
If M is to be uniformly validity across the entire timespan of the catalog, MI must be an unbiased estimator of MS or MG-R. To test this absence of bias, the correlation between MI and the two surface-wave magnitudes has been examined for 23 events with reliable MS or MG-R estimates, and an MI value determined from the isoseismal maps of Toppozada and others (1981) and Toppozada and Parke (1982). This comparison shows that although the two magnitude scales are well correlated, MI systematically underestimates MS and MG-R by 0.3±0.3 units. If the sample is restricted to MS≤6.5 (n=18), the bias is 0.2±0.3 units. To further investigate this apparent bias, ML was compared with MI for 10 common events, for which the bias was
0.25±0.2 units. As a final check, the difference between MS or MG-R and MI for the 12 events listed in table 6.1 also used by Toppozada (1975) to develop MI relations was found to be 0.10±0.19 units.
On the basis of these results, the summary magnitudes from MI values of Toppozada and others (1981) have been adjusted upward by 0.15 units and then rounded to the nearest quarter magnitude unit. Thus, events of MI=5. 7 become M=5¾, and those of MI=5.8 become M=6. No magnitude adjustment exceeded a quarter unit. MI values from other sources have simply been rounded to the nearest quarter unit, because they average 0.2 units higher than the values of Toppozada and others (1981) and Toppozada and Parke (1982), where comparisons can be made. Summary magnitudes for events before 1850 have not been adjusted upward, owing to the imprecision of the original estimates.
Earthquake origin times, magnitudes, and locations before 1990 principally derived from interpretations of felt reports by Toppozada and others (1981); after 1900, data principally derived from Gutenberg and Richter (1954) and bulletins of the California Institute of Technology, University of California, Berkeley, University of Nevada, Reno, and U.S. Geological Survey. See text for discussion of summary magnitude M. Other magnitude scales are ML, local-magnitude scale of Richter (1935); MG-R, magnitudes from Gutenberg and Richter (1954), generally based on 20-s surface waves; MS, 20-s surface-wave magnitude; MI magnitude estimated from felt area at various intensity levels; and M, moment magnitude, defined as M=(2/3)log10 M0-10.7, where M0 is in dyne-centimeters-parenthetical values based on seismogram envelope. MS values before 1935 generally
derived from undamped Milne seismographs, using the formula of Abe (1988)-parenthetical values based on one or two amplitudes; MS values since 1968 measured from vertical seismograms. Absence of reported amplitudes on Milne seismographs (*) suggests MS<6.
Table 6.1 - Major California and Nevada earthquakes, 1979-1989