JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 96, NO. 87, PAGES 12,099-12,125, JULY 10, 1991

Lg and Rg Waves on the California Regional Networks From the December 23, 1985 Nahanni Earthquake

LISA A. W AID AND THOMAS H. HEATON

U. S. Geological Survey, Pasadena, California

We investigate Lg and Rg propagation in California using the central and southern California regional netwotks. Approximately 550 stations constitute these two short-period networks providing a dense coverage of almost the entire state. The waveforms recorded from the December 23 1985, Nahanni, Canada, earthquake are used to construct three profiles along the propagation path (almost N-S) and three peipCndicular to the propagation path (almost E-W) to look at the nature of propagation of these two types of surface waves. Groups of records from stations in various geologic~ ~d tectonic provinces in California are also examined in order to establish regional charactensttcs of the surface waves. We find that the propagation characteristics of Lg differ from those of Rg across California; Lg waves are apparently more sensitive to crustal heterogeneities. The most striking observations are tile similarity of coda for both the Lg and tile Rg waves within geologic provinces and the mruked difference in coda between regions. These differences are seen in the amp~~~~· coda duration, shape of the energy envelope, frequency content, and sharpness of tile phase In1tiati.on. In general, a decrease in tile Moho depth near tile Pacific Coast is correlated with a decrease in the surface wave amplitude, especially at higher frequencies (0.15-0.2 Hz). Most interesting is the association of the San Andreas fault with abrupt changes in the wave train amplitudes. The surface waves are amplified in the vicinity of tile fault zone and then decrease in amplitude after tile zone is crossed. In the Coast Ranges, amplitudes are low and waveform coherence is poor. The Rg phase dominates the record in the Sierra Nevada, and both surface waves are amplified by the thick sedimentary sequence of the Great Valley.

INTRODUCTION

The Lg and Rg phases were first recognized by Press and Ewing [1952] who described them as channel waves trapped in a granitic low-velocity zone at the base of the crust. The designations "L" and "R" derive from the trans­verse Love wave type particle motion and the retrograde el­liptical Rayleigh wave type motion, respectively, that these two phases exhibit. The "g" signifies the granitic layer in which the waves were once thought to propagate. Knopoff et al. [1973] and Levshin [1973] demonstrated that Press and Ewing's simple model did not adequately explain these ob­servationally complex phases and, furthermore, that a low­velocity zone is not required for the existence of Lg and Rg waves; a simple monotonically increasing shear veloc­ity structure with depth is sufficient. . For purely continental paths, Lg is often the most prominent phase on seismo­grams recorded at regional and intermediate ( < 1000 km) distances within the 0.5- to 5-Hz frequency range. Lg does not, however, propagate across oceanic paths greater than 100 km [Press and Ewing, 1952]. Although the Lg wave has motion primarily in the horizontal plane, the phase also contains substantial components of vertical and radial mo­tion, and thus it can be studied on vertical-component seis­mograms as we have done here. In general, the Lg phase has a group velocity of 3.51 ± 0.07 km/s (about the av­erage crustal shear wave velocity) in the western United States [Press and Ewing, 1952], a period of 0.5-6 s, and

This paper is not subject to U.S. copyright. Published in 1991 by the American Geophysical Union.

Paper number 91ffi00920.

Ol48-0227/9l/9lffi-00920$05.00

exhibits reverse dispersion (higher frequencies have higher velocity) at .d20° from the source. LK has been studied a great deal more than the Rg phase, which has a later arrival time. The Rg phase has a group velocity of 3.05 ± 0.07 km/s (about the average crustal Raleigh wave velocity) in the western United States [Press and Ewing, 1952] and an 8- to 12-s period. Both phases have durations up to several minutes, and Ruzaikin et al. [1977] and Cara et al. [1981] argued that this characteristic can only be reasonably ex­plained by multipathing, scattering, and mode conversions caused by lateral inhomogeneities. The Lg coda is largely affucted by the propagation path and receiver site and is not strongly influenced by the source mechanism [Gregerson, 1984; Campillo, 1986].

The Lg and Rg phases have been successfully modeled as a supetposition of fundamental and higher-mode sur­face waves propagating within the crustal waveguide [e.g., Knopoff et al., 1973; Cara and Minster, 1981; Bouchon, 1982; Herrmann and Kijko, 1983; Kennett, 1985]. Other studies using Lg and Rg include those involving spatial at­tenuation, magnitude scaling and underground nuclear ex­plosion yields [e.g., Nuttli, 1986, 1973; Campillo, 1987; North, 1985; Campillo et al., 1985; Shin and Herrmann, 1987], lateral anisotropy effects [e.g., Kafka and Reiter, 1987; Kennett, 1986], and observational analyses [e.g. Bdth, 1954; Gregerson, 1984; Der et al., 1984; Ruzaikin, 1977; Kafka and Reiter, 1987]. Data analyses are difficult due to the many factors which affect the Lg and Rg phases, and not all studies are in agreement. Most studies, however, do fD,gree that local and regional geology can have a significant effect on the Lg and Rg waveform characteristics [Rivers, 1980; Der et al., 1981; Campillo, 1987].

In this study we use the December 23, 1985, Nahanni, Canada, earthquake as recorded on the central and south-

12,099

12,100 WAU> AND HEATON: Lg AND Rg WA~ IN CAUFORNIA

ern California regional networks. The reasons for choosing this particular event are that it was recorded on most of the stations and that its shallow hypocentral depth, large mag­nitude (Ms 6.9), intermediate distance to California, and purely continental path resulted in good Lg and Rg waves with which to do a comprehensive study. The epicenter is 2267 km away from the northernmost station in the Central California Network and 3098 km away from the southern­most station used in this study (Figure 1 ).

Since all the short-period instruments in the California regional networks are nominally identical, we corrected only for individual station gains (rather than the correcting for the entire instrument response). Consequently, this is not intended to be a detailed quantitative analysis of Lg and Rg; rather it is a descriptive presentation and summary of a large Lg and Rg data set recorded on the short-period networks in California. We attempt to correlate the crustal geology and structure with the various features in the wave trains.

CALIFORNIA REGIONAL NETWORK DATA

The regional networks in California were originally de­signed for the purpose of locating regional earthquakes and

determining magnitudes. The network instruments are all short-period velocity seismometeiS with a natural period of 1-s and, with the exception of a few stations, record only vertical motions. In this study we use only vertical com­ponent seismograms. The Central California Network has about 300 stations, and the Southern California Network has about 250 stations. Both networks are monitored with the California Institute of Technology-U.S. Geological Survey (Caltech-USGS) Seismic Processing System (CUSP) digital recording system [Johnson, 1983; Given et al., 1987]. In ad­dition to digital recording, frequency modulation (FM) ana­log tapes continuously record most stations in the networks as a backup to the real-time processing system. Since the digital recordings do not include the surface wave portion of teleseismic events and often do not trigger on teleseismic events, we obtained the Nahanni data from the FM analog tapes.

We use 15 min of data from approximately 200 stations, most of which are part of the Central California Network. The records are band-pass filtered from 0.01 to 1.0 Hz us­ing a zero-phase Butterworth filter and then decimated to 5 samples/s for use in the study. The lack of complete infor-

PROFILES

41 °

J7•

34° 36°

. .. & • • .. . . ...

\

33° 12J• 122· 121°

50 KM w.w.w

124" 123" 122° 121° 120" 119" 118° 117° 116" 115" 114 "

Fig. 1. Map of California showing profiles presented in this study. A-A', 8-B', and C-C' are parallel to the propagation direction, and D-01

, E-E', and F-F' are perpendicular. The inset in the upper right shows the location of the Nahanni earthquake, indicated by a star. The inset in the lower left shows the profiles across the San Andreas fault. Stations are shown as triangles. Abbreviations are H, Hayward fault; C, Calaveras fault; SAFZ, San Andreas fault zone; R, Rinconada fault.

WAll) AND HEATON: Lg AND Rg WA~ IN CAUFORNIA 12,101

mation about the instruments at the time of the earthquake did not permit us to remove the entire instrument response with confidence; however, we were able to correct for the gain of each instrument. This correction allowed us to ob­tain the nominal instrument response and use relative am­plitude information since all the instrument response curves are nominally the same. We did not carry out an inde­pendent check on the station responses; therefore, although we use the apparent amplitudes as if they were real, some anomalous amplitudes may be the result of incorrect gain in­formation. We make interpretations only in the cases where systematic correlations lead us to believe the amplitudes are reflecting a real variation.

Since the instrument response has not been entirely re­moved from the data, it is important that we understand what that response is and how it affects the waveforms. Figure 2 represents the theoretical instrument response of the 1-Hz short-period seismometen; with the two most commonly used discriminator types at a variety of attenuation settings. The amplitude response is proportional to w3 below 1 Hz. Although the curves in Figure 2 look different at high fre­quencies for the two discriminaton;, note that the curves are almost identical for frequencies below 1 Hz, which is the region that applies to the data in this study. The effect of this instrument response on our data is that the Lg waves are recorded at a higher amplitude than the longer-period Rg waves, even though the ground displacement from the Rg waves may, in fact, be greater. We cannot be sure that an anomalous amplitude is not due to a deviation in the naturaJ frequency of the instrument, which would particularly affect the low-frequency Rg waves. The two amplitude values for Lg and Rg beside each record in the figures have also been corrected for distance based on ..1-3 for Lg and ..1- 1 for Rg as seen in Figure 11 and is discussed later.

USGS J120

.. 0

........ :::E u .. .......... 0 (/) 1-z ::l 0 u .. ....... 0

(HZ)

The station spacing of the network ranges from about 1 to 50 km, and our profiles have an average station spacing of about 25 km. The length of our profiles varies from about 150 to 350 km. The duration of the Lg and Rg coda varies from region to region, but the typical duration is about 2-3 min for Lg waves and at least 4-5 min for Rg waves. The onset of these phases is emergent in most cases, making it difficult to assign a time to the arrival of the wave group. In this study we assign arrival times as those predicted using the velocities as described by Press and Ewing [1952]. In addition, the complete Rg coda was not on the data tapes we used, and as a result we were unable to determine the length of the Rg wave train with certainty. The Lg waves in this study have a dominant period of about 1.5-3.5 s on the network instruments, and the Rg waves have a dominant period of about 6.5-9 s.

ME'IHOD

We constructed many profiles during the study; however, we present only three N-S profiles and three E-W profiles here based on the best station coverage. All the records shown in the figures have been further band-pass filtered from 1 to 15 s using a zero-phase Butterworth filter in order to minimize noise. The amplitudes shown and referenced are the average Lg amplitude and the average Rg ampli­tude recorded on the short-period instrument and corrected for gain. The average amplitudes were determined using a 150-s window starting approximately 30 s into the wave train for the Lg waves and using a 180-s window starting at the end of the Lg window for the Rg waves. It was found that varying the time windows up to 30 s did not change the amplitudes by more than about 0.02 units. The average of the absolute amplitude of the digital seismogram in the

USGS J101M

.. 0

........ :::E u .. .......... 0 (/) 1-z ::> 0 u ..

........... 0

(HZ)

Fig. 2. Instrument response for the 1-Hz seismometer with a 1120 and a JlOlM discriminator, two of the most commonly used discriminators in the networlc, at various attenuation settings.

12,102 WALD AND HEATON: Lg AND Rg WAVES IN CAUFORNIA

window was taken to determine the average amplitude. It is important to remember that these are narrow-band seis­mograms and not absolute ground motions. The values are only relative amplitudes of the seismograms (corrected for gain) and therefore do not have any units. All records with amplitudes of 2.5 or greater have been normalized to 2.5 in the figures.

The information used to construct the geologic profiles accompanying the waveforms in Figures 3-8 was compiled from Oppenheimer and Eaton [1984], Walter and Mooney [1982], Colburn and Mooney [1986], Eaton [1963], Mavko and Thompson [1983], and Zucca et al. [1986].

In attempting to explain the characteristics of the surface wave coda and the propagation patterns, we have considered the receiver site attributes and the propagation path struc­ture including site geology, crustal thickness, and sediment depth, as well as faults and other abrupt lateral variations. We have also compared the behavior of the Lg wave train with that of the Rg wave train.

PROFILES

The following are descriptions of profiles A-A' through F-F', the locations of which are shown in Figure 1.

A-A'

A-A' shown in Figure 3, is parallel to the propagation direction and diagonally crosses the north Coast Ranges, ter­minating at the Pacific coast. This profile spans a large sec­tion of the Franciscan assemblage including the Central belt (a melange with mostly graywackes and metagraywackes) and the Coastal belt (arkosic sandstones) [Blake and Jones, 1981]. There is no recognizable difference in the wave trains between these two belts, but there is a gradual de­crease in amplitude toward the coastline (Lg decreases from 0.5 to 0.3; Rg decreased from 0.2 to 0.1). This decrease in amplitude for stations on the coastline is seen generally on all stations along the Pacific coast in this study. The onset of Lg is not sharp on most of the records, and the average Lg amplitude is moderate compared to other profiles. The Rg phase is not pronounced in these records. The first 40 s of the Lg wave train includes some 0.15- to 0.2-Hz energy, longer period than the rest of the Lg coda with the exception of KPP and KKP, the two stations farthest inland on this profile.

B-B'

Profile B-B', shown in Figure 4, is also parallel to the di­rection of propagation. It crosses the metamorphic foothill belt of the northern Sierra Nevada [Bateman, 1981], extends through the central Great Valley and the Salinian block of the north Coast Ranges near Monterey Bay [Page, 1981], and terminates at the coastline south of the bay. Unfortu­nately, there are no stations in the Great Valley, and we have no data on the transition from the Sierra Nevada to the Coast Ranges. A remarkable difference in the appear­ance of the wave trains between these two areas can be seen both in the more pronounced Rg waves in the Sierra Nevada (due to less high-frequency energy content) and the slightly higher (0.1 units higher) average Lg peak amplitude in the Coast Ranges east of the San Andreas fault.

The San Andreas and Calaveras fault zones coincide with dramatic changes in the waveforms in this profile. Station

HPH is located in the fault zone and has a very high am­plitude for both Lg and Rg. All the stations west of the fault, on the granitic and metamorphic Salinian block, have a greatly diminished amplitude relative to those on the east side (Franciscan terrane). In this case, the abrupt decrease in amplitudes in coastal areas appears to be related to the fault zone, in contrast to the more gradual decrease in amplitudes seen in profile A-A' toward the coast. Other stations in the area also exhibit this significant change in amplitude.

C-C'

Profile C-C', shown in Figure 5, is the easternmost pro­file along the propagation path. It begins in the foothill belt of the Sierra Nevada, [Bateman, 1981] and spans both the Great Valley and the Salinian block of the south Coast Ranges [Page, 1981 ]. This profile can be separated into three distinct groups based on the appearance of the Lg and Rg waveforms. The. three stations in the Sierra Nevada have relatively sharp Rg arrivals and clear wave trains on­obscured by the higher-frequency energy of the Lg phase. The next group of stations is located on the edge of the Great Valley, and the most noticeable feature of the group is the high amplitude of both surface wave groups. Here, the high-frequency energy of the Lg phase lasts well into the arrival of the Rg phase. The station west of the San Andreas fault zone has a much smaller amplitude for both Lg and Rg waves than stations east of the fault zone.

D-D'

Profile D-D', shown in Figure 6, is the northernmost profile perpendicular to the direction of propagation. It trends from the Pacific coast north of San Francisco (Coastal and Central belts of the Franciscan assemblage) [Blake and Jones, 1981], across the Great Valley to the metamorphic foothill belt of the Sierra Nevada [Bateman, 1981]. The most noticeable difference between the E-W profiles and the N-S profiles is the lack of interstation coherence in the en­ergy groups seen on the E-W profiles in a qualitative sense. This is an indication that heterogeneities in the propagation path produce scattering or changes in the phase velocity [Der et al., 1984]. The records nearest to the coast contain low-frequency energy at the initiation of the Lg wave simi­lar to that on profile A-A'. The records on either side of the Great Valley have anomalously large amplitudes. The Rg wave on the record in the Sierra Nevada dominates much earlier than those in other regions.

E-E'

Profile E-E', shown in Figure 7, trends perpendicular to the propagation direction from the Pacific coast, across San Pablo Bay, the Great Valley, and the metamorphic foothill belt of the Sierra Nevada [Bateman, 1981]. Station MAT (farthest east) is within the granitic batholith of the Sierra Nevada. Three distinct groups of records can be seen on this profile. Starting from the west, the record near the coast has a low amplitude (approximately 0.1 units). Stations CSP, CBW, CRP, and CAC on the broken Franciscan terrane, have high amplitudes(> 0.3 units) and pronounced Rg wave trains. The three records in the Sierra Nevada once again have high average amplitudes and a pronounced Rg wave train.

CD

ci

"' ... ci

.._r tn .. lD N .. L.()

<(

I <(

w _J

l..J.... 0 0::: ::::;;

(.:J 0... ::.::

4000 N

m 2000

0

10

km 20

30

40

~ 0

"' ci ci

"' .... .... "' ci ci

c.. c.. c.. ::.:: ::.:: ::.::

CD

ci CD .... ci

- .... ~ ... ci ci ci ci

.... 0 ... "' N .... N N

ci ci 0 ci

487 Epicentral ~~ ftBifil ::.:: ::.:: ::.:: ::.:: (.:J

Coast Ranges

sedimentary rocks

~ ci

N

ci

<( z (.:J

--ci

"' ci

"' 0 ci

"' ci

s

----Moho

7

Fig. 3. Profile A-A'. The dashed lines represent the approximate Lg and Rg group anival times. The average Lg and Rg amplitudes are shown to the right of each seismogram. The sketch of the geologic cross section shows corresponding structure beneath the stations. Note the difference in scale between topography and subsurface structure. S.Af'Z is San Andreas fault zone.

12,104

CD N d ... N d

"<t L() .. lO N .. L()

CD I

CD

w _..J

LL 0 n::: u Q_

J: 0

4000 N

m 2000

km

0

10

20

30

40

WAID AND HEATON: Lg AND Rg WAVES IN CAUFORNIA

.... d

N d

lr

~

N N d .., .., d

lr "'-<(

1

Epicentral

Sierra Nevada Great Valley

metamorphic rocks

- ~ - ~ - - - _S!M!i~~~'l - - - , -rocks

CD

d

N d

-' Q_

J:

.... - "' N ..,

0 0

d d d d d

I<> .., .... .., .., I<> 0 0

d d d d d

.---.. Vl

""'-. E

.:£

L1)

I"') ,.._., ,, ~0'

_J

01

J: ~ u ::::>

Q_ ..., Vl J: ID ID ID

s Coast Ranges

~ cJ ' J

sedimentary rocks

Moho

Fig. 4. Profile 8-81• The dashed lines represent the approximate Lg and Rg group arrival times. The average Lg

and Rg amplitudes are shown to the right of each seismogram. The sketch of the geologic cross section shows corresponding structure beneath the statioos. Note the difference in scale between topography and mbswface structure. SAFZ is San Andreas fault zone, and C is Calaveras fault.

-u I

u U.l _j

: OlD <'1-

d 00 .., ..,_

-N d 00

LL 0 0::: 0....

2678 2728

4000 rN

m 2000

Sierra Nevada

granitic batholith

CD 0 "' "' .... d d d ., .., 0

"' ... "' d d d

2778 2828 2878 ..... Epicentral Distanc~(kmcJ ~ ~ I ~ ~ ~

Great Valley

.. - -li.8_CI!f1J!nJtiY • • - • ' rocks

0 "' 0 It) N "' d d d d

N 0 .... .., ... N N d d d d

2928 2978

w C:: Oct: ~ Vl <t--~ ~ ~~

Coast Ranges

~ (%

I

sedimentary rocks

., 0 d

0

d

I u [lJ

3028

s

km

0

10

20

30

40

----------------Moho

Fig. 5. Profile C-C'. The dashed lines represent the approximate Lg and Rg group arrival times. The average Lg and Rg amplitudes are shown to the right of each seismogram. The sketch of the geologic cross section shows corresponding structure beneath the stations. Note the difference in scale between topography and subsurface structure. SAFZ is San Andreas fault zone.

0 I

0

w _..J

l.J.... 0 0::: D...

4000

m 2000

0

10

km 20

30

40

01 IIDN ~ :! 0 o~ 0 ci cici ci ci ci

~ N.., N ~ 0 "' ci cici ci ci ci

0 50

uu > 0... (.!) O:::l!! U 0 CD (.!)(.!) (.!) (.!) (.!)

w Coast Ranges

~"' c,'llf' I

sedimentary rocks

... N ci

10 N ci

1-:::l!! z

100 Profile

0 ID N ci ci

.., "' N ci ci

150 200 Distance(km)

0::: 0::: LL. J: <( <(

... N ci

01 N ci

250

Great Valley E

Sierra Nevada

'·,sedimentary / metamorphic ---------' rocks

rocks

Moho Fig. 6. Profile D-D'. Lg and Rg group anival times are indicated. The average Lg and Rg amplitudes are shown to the right of each seismogram. The sketch of the geologic cross section shows corresponding structure beneath the stations. Note the difference in scale between topography and subsurface structure. SAFZ is San Andreas fault zone.

"<;j­L{)

r---­N L{)

w I

w w _J

LL 0 0:::: 0....

CD 0 0 CD 0 0

<( f-z

4000 w

Ill ('j

0 Ill ,., 0

a.. Ul u

CD .. .... ('j ('j .. 0 0 0 .. .... ,., ,., N .. 0 0 0

3: a.. u CD a:: <( u u u

Coast Ranges

m 2000

0

10 sedimentary rocks

km 20

30

40

N N 0

~ 0

Great Valley

--- sedimentary , / ..... _ -~

-rocks-

~ .... N

0 0

"' ~ 0 0

Sierra Nevada

metamorphic rocks

~------Moho

E

Fig. 7. Profile E-E'. Lg and Rg group anival times are indicated. The average Lg and Rg amplitudes are shown to the right of each seismogram. The sketch of the geologic cross section shows corresponding structure beneath the stations. Note the difference in scale between topography and subsurface structure. H is Hayward fault, and C is Calaveras fault.

v L()

OC> N L()

LL I

LL

w _j

LL 0 0:::: Q_

ID 0 0 01 0 0

0::: ll.. Ill

4000 w

m 2000

0

N 01 0 ID 0 N ,..,

0 0 0 0

• ~ ID N ,..,

0 0 0 0

:::::E <.:> > 0::: Vl Vl 0::: I Ill Ill Ill ll..

Coast Ranges

~ (/)'If'

j

01 • 0 ,.., Ill 0

Profile Distance(km) :::::E 3: ll..

E

1 0 sedimentary,granitic,and metamorphic rocks

km 20

30

40 Moho

WAU) AND HEATON: Lg AND Rg WAVES IN CAUFORNIA 12,109

F-F'

Profile F-F', shown in Figure 8, is also perpendicular to the propagation direction. It begins at the Pacific coastline and crosses the Salinian block [Page, 1981] and Francis­can assemblage [Blake and Jones, 1981] in the south Coast Ranges just south of Monterey Bay, ending at the west edge of the Great Valley. The San Andreas fault once again sep­arates low amplitudes, associated with the Franciscan ter­rane to the east, from higher amplitudes, associated with the Salinian block, to the west.

lei to the propagation path to investigate the effect of the fault on the surface waves (the locations of these profiles are shown in Figure 1). Most profiles are about 50 km in length. The amplitude values for Lg and Rg beside each record in these profiles have not been corrected for distance; the 50-km range changes the amplitudes by less than about 5%, which is beyond the resolution of the amplitudes in this study. All the profiles show an increase in amplitude (espe­cially Lg) in the vicinity of the San Andreas fault followed by a decrease in amplitude south of the fault (after the fault is crossed from north to south).

Profiles Across the San Alldreas Fault

Figure 9 shows six short approximately N-S profiles that were constructed across the San Andreas fault zone paral-

Figures 1 Oa-1 Oe show the Lg and Rg amplitude changes along these profiles as they cross both the Calaveras fault (in some cases) and the San Andreas fault. Profiles XF2 and XF3 are plotted separately from the others because they

XF1 5:27:54

N CCY~~~~~~~~~~~~W~,MM/WV~~~~\I~I~~~~~~~~'V\1\~VV'~~ -o~ ;:;·-....~

60s

0 35 0.21

J sF 0 NM~r-I>Jr.Mt-Wirw,¥J.N##WNJ~~~M~-\VJIJI!fiM~VIIN~h'IJW"'-'MWWWAJI~~M~V"f'IJVWI~'IoN.""'""""""''+iv-'V'rr'-M,JV> 0 45 0., 2

SAFZ ~ j~1 ~ ~~~~~~~~~~~--l'~tlq~~»M~~~~~~~~~~{J&,'!ft_~~~!Jt 8·~~ 8:H ......... "' 'AN

J SM 2.- m..'MIJI~#..-.N..-'-"""~WWowJ/Ir.Nvf>t-JMW!VNM~fti.M~~WA/!It-M~-..Jv>!\JV'ItjlfMJ""""W\IIIIVV"MII'VV""N'LN'<I/V'o/-~'{"rvv'lf"~ 0 39 0., 6

S JBL N v...,~.,.,..,.,...,.....,.,.._"""""..,.._""""'VW""'""""¥fi'"'V'',...,.fJ."I/V"'rvMJN-J-'V'fvV""'"tN-,I"'J'\r--MIV'--"""""oJ'M"'v'\'lf'W"-""""""'.JW'V"\AM,f\.J""-'vWVV\<"VV"V'-'V 0 2 1 0.1 1

"' <D

"'

XF2 5 27 54

N CMH~~,~~~~~~~~~~NM~~~~~~ -o

esc ;:;·

JST 0

JAL ~ N J"""""""'""""""""'""""VVVV'V""""-J.M./~'IM/'M~~NM~WWI'\J\"'/'1'\~NNV'rllw'-IW'v'l~ JHL ;......_

SAFZ ~ JEC 3 JBZ '-...../

S JPL

-<D

XF3 5·27·54

N cco~~~~~~~~~~wuvwqrwi~IIU~I~W11~/VIIIIIIVUII~/Ul~~~vmu~u\JVII'V~u\rv~~~~~~u~ -oco

CAD0·- ~~~~~~~~~~~~~~~~~~~~~NV~~WW' JCB 0

60s

60s

HGW~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ..-..0

SAFZ ~ ~t~ 3 Ol ~~~~~~~mll•lfl HCO'-...../

S HDL N co "'

Fig. 9. Short (50 km) profiles across the San Andreas fault. Location of the San Andreas fault zone (SAFZ) is shown. Profiles are approximately N-S.

Fig. 8. Profile F-F'. Lg and Rg group anival times are in~cated. The. average Lg and Rg ~plitudes are shown to the right of each seismogram. The sketch of the geologtc cross section shows correspon~g structure beneath the stations. Note the difference in scale between topography and subsurface structure. SAFZ ts San Andreas fault zone.

0 40 0.23

0 64 0.26

0 36 0 , 4

0 31 0.16

0 35 0 13

0 43 0.14 0 69 0 14

0 51 0 12

0 83 0.37

0 35 0.24

0 31 0.1,

0 21 0.09

Hl 8:B 0 25 0.1,

0.27 0 19

12,110 WALD AND HEATON: Lg AND Rg WAVES IN CAUFORNIA

XF4 5·27:54 60s

0 47 0.18

0 56 0.14

0 0 54 0.16 1 14 0.18 HORw· .. ~~~~~~~~~~~ HFH :-'" ~

__....._ 111 0 24

Q 23 0.12 SAFZ --7 HSF 3 "' ~""""Mfvv-..,.tJ,AN.JvwJ.i...I\.JW\fWA"""f/vAI~IMIAIIM!n.IMJIMMIV!MfiriM\IN~M!NIIV\IVI~~ft'VJ\fW~IIJ\fv1MIN-.rr-tv"''//r""",..,....Ntvfi.N-~I.Jl,tl,.,f..J,

HFP'-"

S BSR 0 38 0.17

XFS 5:27 54 60s

0 39 0.15

0 29 0.08

0 54 0.07

0 43 0.19

0 34 0.13

0 17 0.07

XF6 5:27·54 60s

N BAV 1'1 ~...,.....""'""".,._,.........,~~iV\II\INI&"Wiflli'M~f"M({I~\\'tii,N~W'\.,Mt/I.A/'oJ\[V'>j'VV'WWV\NIJ'\/\/V\fW"W\I'IrYIJ'\/"'I\!'-""""'"4fV"""-I'"VV\rvv~/VV"'-f'-'V' 0 31 0 18

BBG'=' ~ o 71 0.24 ()

BBN0

SAFZ --7 Ul

BRV,..... ~-I"""""WV'-.J.M~(WW~\11'\MI'-1/l\~

0 67 0.25

S PLO

Fig. 9. (continued)

have a slightly different amplitude profile. All Lg profiles show an increase in amplitude near the San Andreas fault, and those crossing the Calaveras fault also show an increase in that area. XFl, XF4, XF5, and XF6 (Figure lOa) have a gradual rise in Lg amplitude, peaking at the San Andreas fault, followed by a shaip drop in amplitude after the fault is crossed. XF2 and XF3 (Figure lOb), on the other hand, have high Lg amplitudes at the Calaveras fault that decrease after the fault is crossed, followed by another increase in amplitude that occurs just after the San Andreas fault is crossed

The Rg profiles are somewhat different. XF2 and XF3 (Figure 1 Od) show an increased amplitude at the Calaveras fault, followed by a decrease. The amplitude does not in­crease at the San Andreas fault on these two profiles. XFl, XF4, XF5, and XF6 (Figure lOc), however, do show a grad­ual rise in amplitude approaching the San Andreas fault with a peak at the fault, followed by a shaip decrease in ampli­tude after the fault is crossed. Differing crustal structure along the length of the faults, or perhaps the varying prox­imity of the faults themselves, appears to have an effect on

the amplitude variations of the seismograms. Profiles XF2 and XF3 cross both the Calaveras and the San Andreas faults and look similar. However, XFl, XF4, XF5, and XF6 have a different amplitude variation pattern. These three profiles cross only the San Andreas fault (see Figure 1). XFl crosses the Hayward fault and the southern tip of the San Francisco Bay before reaching the San Andreas fault, so those surface waves may have further complications.

The plot of the average values of all profiles (Figure 1 Oe) shows a rather dramatic increase in amplitude for Lg waves at the approach to the San Andreas fault followed by a sharp decrease and yet another much smaller increase in amplitude at the Rinconada fault. The amplitude variation for Rg waves mirrors that of the Lg waves but is more subtle. Additional data were added at the SW end of profiles XF3, XF4, XF5, and XF6 for this plot in order to obseiVe the amplitude variation at the Rinconada fault.

On a larger scale, Figure 11 represents the average Lg and Rg amplitudes across California from north to south. The 17 data points that make up each curve were derived by calculating the average amplitude of all the seismograms

a a ~

WALD AND HEATON: Lg AND Rg WAVES IN CAUFORNIA

SAfZ Co ~

-o Q) N

o· E~ LO 0 z

Q)

-o ::::10 ........ ~c:i E <(

0> a:: a

"! 0

0 0

12,111

0 0 c:i·_~-3o _______ T2-o-------~1-o----~~-----,,o-------.2o 0-~3-o _______ T2_o _______ ~,-o------or-~----,ro------,2o

Distance(km)

Lg Amplitudes Across SAFZ

b~ Calaveras SAFZ

0 0

d

-o Q) N

0 CD c:i

Q)

-o ::::10 ........ ~c:i E <(

Distance(km)

Rg Amplitudes Across SAFZ

Calaveras SAFZ

c:i_~-30----~_T2-o-------~,-o----~o~-----,,o-------,2o 0 0 c:i·_4-3o ____ .__~-20 _______ T,-o------o~------,ro------,20

Distance(km) Distance(km)

Lg Amplitudes Across SAFZ Rg Amplitudes Across SAFZ

Fig. 10. Lg and Rg amplitudes acims the San AD~.fault Locati~s tJf the profiles XFl throUgh XF6 are shown in Figtue 1. (a) NonnaliZedLg amplitUdes fot profiles XFl, XF4, XFS, &Qd XF6. These profiles (except XFl) c~ss the San Andreas fault only. (b) Noimlillzed Lg ampli~s for ~eli ~ IUid XF3. These profiles cross both die Calaveras fault and the San Andreas fault. (c) Normalized Rg amplitudes ,for pio~es XFl, XF4, XFS, and XF6; These profiles (except XFl) cross the San Andreas fault only. (d) Normalized Rg amplitudes for profiles XF2 and XF3. These profiles cross both the. Calavems fault and the San Andteas fault. (e) Avemge of Lg amplitudes and Rg amplitudes for all profiles. Additional data was added to the SW end in order to extend the profile line across the Rinconada fault.

12,112 WAUl AND HEATON: Lg AND Rg WAVF!S IN CAIJFORNIA

e g

0 co 0

0

"' 0

a. E <(

Calaveras

vo o.f'! ~o Rg v > <(

SAFZ Rincanada

0 0 o_+3-0 ____ .__r2_0 ______ T1-0----~~-----1~0------~20------~3~0------,40

Distance(km)

Average Amplitudes Across SAFZ

Fig. 10. (continued)

0

"' c:i

0

"' 0

a. E <(

~~ ~0 Cll > <(

Lg

0

0+-----~-----,------~-----r----~ 0225 245 265 285 305 325 Distonce(km) *10

Average Amplitudes Across California

Fig. 11. Profile of average Lg and Rg amplitudes across Califor­ilia. The 17 data points wete cletennined by taking 50-km-wide "banda" of California from north to south and calculating the av­erage !IJllplitude from all the recorded seismograms within each R:Bpeetive band. The first and last legs of the curves are dashed because they we~e based on only thtee and four stations, teSpeC­

tively. The smooth dashed curve associated with the Lg data is the predic:tect amplitude decay of 1-Hz Lg waves in the western United States from Der et al. [1984]. The two smooth dashed curves as­sociated with the Rg data are the ..1-UII distance correction for 20-s Rayleigh waves from 0/cal [1989], and the best-fitting curve of ..1-1.

recorded within 50-bn-wide "bands" (equidistant from the Nahanni epicenter) of California from north to south at reg­ular intervals. All bands contain from 5 to 43 stations (av­erage of 19), with the exception of the northernmost and southernmost bands which contain only three and four sta­tions, respectively. The first and last legs of the curves are dashed as a result. The smooth dashed curve associ­ated with the Lg data is the predicted amplitude decay of 1-Hz Lg waves in the western United States after Der et al. [1984] (..:1-3) and seems to fit the data reasonably well. The Rg data set has two curves associated with it in Figure 11. The ..:1-1·66 curve is the theoretical distance correction for 20-s Rayleigh waves in the range 20°-160° as determined by Okal [1989]. This figure depends on factors such as dispersion and attenuation. Although our data set does not cover a large enough distance to resolve an Rg amplitude decay figure with certainty, it appears that the ..:1- 1 curve, rather than the ..:1- 1·66 curve, is the best fitting approximate curve to the data.

Lg AND Rg WAVES 1N DIFFERENT REGIONS

We have taken seven groups of stations from different geologic areas (see Figure 12, after the 1983 map of 0. P. Jenkins as presented by Hinds [1952]), including the Kla­math Mountains, the Sierra Nevada, the Cascade Range, the north and south Coast Ranges, the Great Valley, and the Mammoth area, in order to investigate the characteristics of Lg and Rg coda within each area and to compare the similarities and differences in the wave trains between ar­eas. Again, only relative amplitude ratios between regions are infonnative without the removal of the entire instrument response. Figures 13 and 14 provide a summary of the dis­tribution of Lg and Rg amplitudes in California, and Figure 15 shows the characteristic codas from each of the regions.

42°

41 °

40°

39°

38°

37 °

36°

35°

34"

33"

Geologic Regions

(1

KLAMATH ~-··f.,\ODOC MTNS."·· ... ~-PLATEA.U

··!"". . BASIN & RANGE

< .f\\ \ ...•.... : ··.;

~ · .. . ~ ·· .. "=. ~ · .. ~··... 1-. ···· .. . ·. ~..... ·.. <'2.. . ~ ··.. 1'o "t- . 1' . . '( \

.... <').. ~-\ 't·· .. ·.; ~ ....

·:: ..... : :··

•o'!:>"-

...

BASIN &"RANGE

..

~

50 KM ....... 1

25° 124" 123° 122° 121" 120° 119" 118" 117" 116" 115°

1140

Fig. 12. Geologic regions of California (after 1938 map of 0. P.

Jenkins. as presented by Hinds, [1952}.

42°

41 °

40°

39°

I 38 ° -1

37 °

36°

35"

34"

33"

Lg Amplitudes

"\._·''"jo," ,~x· ·~II

~-·~ ··::

...

.. : .... LON Q-0.2

• HIOHO.S+

50 KM ....... 125" 124" 123" 122" 121" 120" 119" 118" 117" 116" 115"

114

"

Fig. 13. Average Lg amplitudes in California. Average amplitudes were determined by filtering a 150-s time window of the Lg wave train with a two-pass zero-phase Butterworth filter from 1 to 15 s. The unstippled areas are those with inadequate information to de-

termine an average amplitude.

~ ~

~ ?! J;

~ l; ~

i 2 0

'

~ -::;:;

12,114 Rg Amplitudes

42°

41 °

40°

39°

38°

37°

36°

35°

34°

33°

50 KM .......

...

~.).'.} .. : LOW 0-0.1

• HIQH0.2+

125° 124° 123° 122° 121° 120° 119° 118° 117° 116° 115" 114"

Fig. 14. Average Rg amplitudes in California. Average amplitudes were determined by filtering a 180-s time window of the Lg wave train with a two-pass zero-phase Butterworth filter from 5 to 15 s. The unstippled areas are those with inadequate information to determine an average amplitude.

Descriptions and groups of waveforms from each area can be found in the Appendix.

DISCUSSION

Figure 15 is a comparison of the typical wave trains from each of the areas we discussed. An overall examination of the data reveals that the Lg and Rg waveform varies signif­icantly throughout California, and the two types of surface waves are affected differently by crustal heterogeneities and changes in Moho depth. In general, Lg is more responsive to crustal features than is Rg, which is to be expected since Lg has shorter wavelengths. This information is imp<•rtant when using both Lg and Rg to map crustal featmes.

Another general observation is the higher interstation co­herence in the N-S profiles (parallel to propagation) than ih the E-W profiles (perpendicular to propagation). Mrazek et al. [1980] and Der et al. [1984] noted this same phe­nomenon in their data. This indicates fmward scattering of energy from crustal heterogeneities [Der et al., 1984]. Der et al. were able to rule out lateral anisotropy in the geo­logical structure under the array used in their study as the cause of this coherence pattern. However, using only one event, we were unable to do that for this study. In general, interstation incoherency of waveforms is associated with ar-

eas of complex lithology and crustal structure such as the Coast Ranges, particularly south of San Francisco, and in the northern Sierra Nevada.

The thick sediments in the Great Valley seem to have a dramatic influence on the surface waves. The waveforms are amplified on profile C-C' and on the Great Valley records (Figures 5 and AS). The stations on the east side of the val­ley have a slightly higher amplitude than the surrounding area, but those stations on the west side of the valley have significantly larger amplitudes. This may indicate that the degree of amplification is dependent on the depth of the sed­iments, since the sediments are much thicker on the west side of the valley. Alternatively, the greater amplification of Lg and Rg in the west may be a result of the greater distance travelled by the waves in the thick sediments in the basin. This problem cannot be resolved without the addition of data from sources located at a range of azimuths. The am­plification effect of the basin, in general, is consistent with Campillo's [1987] results from modeling of Lg waves in a sedimentary basin. On the other hand, Gregerson [1984] found that the 8-10 km of sediments in the Norwegian­Danish basin has almost no effect on the Lg wave.

The most interesting phenomenon that we see on the pro­files is the change in the Lg and Rg wave trains in the vicinity of the San Andreas fault. This effect is most ev-

WALD AND HEATON: Lg AND Rg WAVES IN CALIFORNIA 12,115

REGIO~~AL COMPARISON OF LG AND RG WAVES 60s

KRP 2342

LHK 2429

KFP

2511 N. Coast

AAR 2559 Sierra Nevada

NHM 2684

MLK 2757

PMG 2990 S. Coast

0.31 0 18

0.39 0.48

0.20 0.26

0.05 0.09

~------------------Rg------------------~

Fig. 15. Typical wave train from each geographical area studied. The epicentral distance of each station is shown below the station name. The average Lg and Rg antplitudes are shown to the right of each seismogram. The records are lined up by Lg arrival time, and the Lg and Rg wave trains are labeled.

ident on profiles B-B' and E-E1 (Figures 4 and 7), but it appears on almost all profiles that cross the fault zone. The amplitudes of both Lg and Rg increase in the vicin­ity of the fault (although the increase is more pronounced in the Lg waves) and then decrease after crossing the fault zone. The short profiles across the fault zone parallel to the propagation direction (Figure 9), and the amplitude versus. distance plots associated with the profiles (Figures 10a-10e) further illustrate this phenomenon. The amplitude variation is convincing and leads us to conclude that the San Andreas fault (or the juxtaposition of different rock types on opposite sides of the fault) indeed alters the Lg and Rg waves.

Evidence of abrupt changes at the San Andreas fault in­clude the following: Zandt [1981] showed that there is a linear low velocity zone subparallel to the San Andreas fault in the upper mantle. Walter and Mooney [1982] showed that the velocity structure and Moho depth on opposite sides of the San Andreas fault are different, with the Moho being shallower west of the fault. Oppenheimer and Eaton [1984] interpret the relatively early Pn residuals west of the San Andreas fault in their study as an indication of a lateral velocity contrast across the fault rather than a v:triation in crustal thickness. Nevertheless, their final results indicate that the Moho depth decreases gradually from about 39 km in the Sierra Nevada foothills to 22-24 km along the Pacific coast. The question of whether the abrupt change coinci­dent with the San Andreas fault is a result of a change in the Moho depth or and change in rock types, or perhaps both, remains unresolved. However, we can compare our data with results of several theoretical studies involving Lg waves and changes in Moho depth. Let us consider both the possibility of a gradual decrease in the Moho and a sharp step in the Moho at the San Andreas fault.

If we assume that the Moho gradually decreases in depth toward the continent-ocean boundary, we have a model sim­ilar to Kennett's [1986] and Regan and Harkrider's [1989]

transitional model. These two studies consider only Lg waves, and an important difference is that the waves in their models travel through the transition region normal to the structure. The waves travelling from the source in our study are arriving obliquely at the continent-ocean bound­ary. Kennett [1986] points out that obliquely travelling waves will show reduced effects of the boundary but may undergo other distortions since the waves spend a longer time in the transition zone. In any case, both Kennett [1986] and Regan and Harkrider [ 1989] predict that a thinning crust at a continent-ocean boundary will produce a concentrated waveguide and leakage of energy out of the bottom of the crust, particularly at higher frequencies. This should show up on a network of stations as an increased amplitude in the direction of the thinning crust due to the concentrated energy, followed by a decrease in amplitude from the con­tinued leakage of energy into the mantle. The increase in amplitude of the Lg and Rg waves at the San Andreas fault occurs over a narrow area and is focused specifically at the fault, which leads us to presume that the amplification prob­ably does not correspond to the concentration of energy that the two previously mentioned studies predict. The decrease in amplitude west of the fault may, however, correspond to a leakage of energy into the mantle due to the thinning crust.

Now if we assume that there is a sharp decrease in crustal thickness at the San Andreas fault, we have a model similar to Regan and Harkrider's [1989] step transition. This model predicts a sharp increase in amplitude approaching the step (a sharp decrease in crustal thickness) followed by a sharp decrease in amplitude. The increase in amplitude is due to energy reflected from the transition boundary. This model seems to fit our data better.

The average Lg amplitudes from north to south to Cal­ifornia (Figure 11) seem to agree well with Der et al.' s [1984] predicted amplitude decay based on ..1- 3 • The Rx

12,116 WAID AND IIBATON: Lg AND Rg WAVES IN CALIFORNIA

decay appears to be lower than the predicted theoretical 20-s Rayleigh wave decay of ..:1-1.66 from the work of Okal [1989].

Characteristics associated with waveforms are consistent within particular regions. Regional patterns have been seen in other studies as well. Gregersen [1984] noticed a distinct regional pattern in Lg wave propagation near thl" Nortl1 Sea and attributed it to path propagation differences. Ru:(}ikin et al. [1977] also saw regional Lg patterns in different parts of Asia. The areas exhibiting a large peak amplitude include a part of the Klamath Mountains and Cascade Mountains, the Great Valley, the Sierra Nevada, and a small area near the San Francisco Bay. Oearly, the large amplitude is due to more than one common factor among these areas. Am­plification of seismic waves has been seen many times in sedimentary basins, so it is reasonable to asswne that the thick layer of low-velocity sediments is the reason for the high amplitude in the Great Valley. The north Coast Ranges have low-amplitude or nonexistent Rg wave trains, while the south Coast Ranges have consistently low amplitude Rg wave trains.

The variation of group and phase velocity is also mean­ingful. We attempted to analyze the group velocity of both the Lg and the Rg waves, but the peak envelope was diffi­cult to trace across the profiles and led to dubious results. Wave train length, the shape of the energy envelope, the frequency content, and the nature of the onset of the phases also reflect regional variations.

The north Coast Ranges have an Lg onset characterized by about 40 s of 0.15- to 0.2-Hz energy followed by the more typical Lg short-period energy for the duration of the wave train. This area also displays a gradual Rg onset if Rg is present at all. The south Coast Ranges, on the other hand, have a sharper Lg onset and more Rg content. The records from stations in the Cascade Range have a relatively sharp Rg onset. In the Sierra Nevada, Lg has a gradual onset and a short duration resulting in records dominated by a clear longer-period Rg wave train. These regional variations suggest that it may be possible to map the extent of different geological or structural areas based on the Lg and Rg waveform patterns.

CONCLUSIONS

The enormous amount of short-period data from the Cali­fornia regional networks can be used not only for earthquake locations and magnitudes but also for waveform studies. These short-period regional seismograms can be a valuable source of information with a variety of applications. We constructed profiles of Lg and Rg waveforms both paral­lel and perpendicular to the direction of propagation and observed groups of records from different regions in Cali­fornia. Lg and Rg phases react differently to various crustal changes and heterogeneities in California, and in general, Lg is more sensitive to the propagation path than Rg. Surface wave train characteristics including amplitudes, coda dura­tion, shape of the energy envelope, frequency content, and sharpness of the phase initiation clearly vary with differing geological or structural areas. The amplitudes are low and the coherence poor in the Coast Ranges. Both Lg and Rg are amplified by travel through the Great Valley, and the Sierra Nevada foothills exhibit clear, dominant Rg waves in the record. The decreasing Moho depth near the Pacific coast may have an influence on the amplitudes of the Lg

and Rg phases. The most interesting observation is that the San Andreas fault appears to correspond with amplified Lg and Rg codas in the vicinity of the fault one and decreased amplitudes after the fault zone is crossed.

APPENDIX

Zucco et al. [1986] described the Klamath Mountains in the northwest corner of California as an imbricated stack of oceanic rock layers as intetpreted from the many thin layers with several low-velocity zones. The records from this re­gion (Figure Al) have fairly high amplitude Lg waves (0.4 average) and moderate amplitude Rg waves (0.2 average). Neither Lg nor Rg has a sharp onset.

The Cascade Mountains form a suture between the Kla­math Mountains to the west and the Modoc Plateau to the east Zucco et al., 1986]. Zucca et al. describe the crust beneath the Cascade Range as similar to that beneath the Klamath Mountain but composed of thicker layers without any low velocity zones. The most noticeable characteristic of the wave trains in this region, seen in Figure A2, is their large amplitudes, particularly for Lg waves with an average of 0.8 units. The Rg amplitudes average about 0.3 units. The most interesting feature in this group of records is seen in the records from LHE and LMP, located on the north flank of Mount Shasta volcano. The Lg and Rg codas ex­hibit ringing and do not show any distinct arrivals. Station OSU, located on the north flank of the Sutter Buttes volcano in the Great Valley, also exhibits this same ringing. These three records do not provide sufficient data from which to draw a conclusion about the characteristics of Lg and Rg co­das from stations located on the flanks of volcanoes, but the correlation between the volcanoes and the ringing is curious nonetheless.

The north Coast Ranges span the northern California coastline, including the Franciscan rocks of the Coastal belt, Central belt, and the Yolla Bolly belt [Blake and Jones, 1981]. The stations chosen for this group (shown in Figure A3) exhibit about 40 s of longer-period energy (0.15-0.2 Hz) at the initiation of the Lg wave on all but KCR and KKP at the north end of the group of stations. These two stations are the two farthest inland in this selected group. The Lg amplitudes are moderate with an average of 0.3 units, and the same is true for the Rg waves with an aver­age amplitude of 0.1 units. The five northernmost stations, however, have higher than average Lg amplitudes for this region. The high-frequency Lg energy groups are not co­herent from record to record, but some of the longer-period Rg energy shows coherence between records.

The south Coast Ranges span the southern California coastline from the Monterey Bay area south to the Trans­verse Ranges. They are composed of the plutonic and meta­motphic rocks of the Salinian block, bordered on the SW by the Sur-Nacimiento fault zone and the Franciscan assem­blage, and on the NE by the San Andreas fault and, once again, the Franciscan rocks. The group of stations shown in Figure A4 extends from the Monterey Bay area to just north of the Transverse Ranges. The Lg waveforms are par­ticularly varied in amplitude, yet there are a few coherent energy group arrivals on some of the records. The amrli­tudes appear to be higher for stations located on the granitic Salinian block than those located on the Franciscan terrane, although there are exceptions. Station PMG, for example, has a low amplitude, yet it is underlain by the Salinian

42 ° t-

• KSX

• KRM

• KRP

41 ° t- j

124°

KLAMATH MOUNTAINS 5:25:54

~ ~ KSX 2267

J KRM 2302

KRP 2342

LBK • LBK 2354

• LHM LHM

• LBG 2372

• LHO LBG 2383

LHO 2396

• LBP l LBP 2438

t ~ 123' 122' Lg

Fig. A 1. Records from the Klamath Mountains. Approximate initiation of Lg and Rg is shown. The epicentral distance of each station is shown below the station name. The average Lg and Rg amplitudes are shown to the right of each seismogram.

60s ______,

0 27 0.09

0 58 0.18

0.71 0.24 ~ > ~

0 35 0.19 = 0 52 0.28 ~

?: ~

0 34 0.15 > z 0

0.45 0.20 ~

0 28 0.14

~ ~ z (l

~

~ >

.!'-'

::::i

41 ° 1--

40 ° 1--

I

ALHE ~ A LMP

A LBF ·:

CASCADE RANGE 5:25:54 60s

LHE ~~IIIINV11WUUIIAI/MIIUI\MKIIII..IIIMNIIIUin/IM/Ifti.MIIftii\A/~I'.M.W/'\JW..AMMJ.v.r..AIWMII.fv.M.w~"'"III\II11AII.AAtv. 1 07 0.25 2292

LMP ~~,~m~~~rn~m~~"'~~~~~ 1.58 0.42 2308

-LBF

0.75 0.18 2329

A LHC

LHC 1.28 0.42

2387

A LR_i LHK LRD

20 KM l.wwl.w.l

I

,.,.,0

A OCH

A OGO

2426

LHK 2429

-OCH 2489

OGO 2515

~

Fig. A2. Records from the Cascade Range. Approximate initiation of Lg and Rg is shown. The epicentral distance of each station is shown below the station name. The average Lg and Rg amplitudes are shown to the right of each seismogram.

0.58 0.38

0.77 0.28

0.33 0.17

0.23 0.17

N

;;

~ > ~ = ~ ?! ct' > z t::l

~ ~ ~ 2!

~ 0

"' z >

40°

39°

38 °

NORTH COAST RANGES 5:26·54 60s

~N~MIVIJ\MIM~foiiWIWI~I~Ni1h\I\\J~\f'vwAMVlrll\~ 0.58 0.24

~I~III/J'Wflv1I"NNIMNI\\IiiiiVMI\i'\INIIIJ~fr.ll~~l,.ll o ss o.2o

~WVVI/rlNIJliWINW~IV'IW\II.fiW\h'tlu'l/hiii/V\JV'IINIJ\/If\'v\/l~tl o.ss o 1s

\~Mf\/\MWt/11\I'W.f.J.JUt+-.IIP/\NIMWoJNI~/~~~~ o 3s o.1s

11rA./"\fNV'MI'\/\t'W~~ 0 31 0.18

~IN o.22o1o

1\~n/\1\~ 016 011

NJ.N'W""""/fJV-.N"-..!"-""Nir"-"""'""""rvt-"""'J\"""'VM~/1~~ 0 18 0.11

f'vJIAI\..I"IAf/~ 0 15 0.21

0 09 0.09

f ,._ ~ 0.06 0.07

Lg R'g

Fig. A3. Records from the north Coast Ranges. Approximate initiation of Lg and Rg is shown. The epicentral distance of each station is shown below the station name. The average Lg and Rg amplitudes are shown to the right of each seismogram.

~ ~

r ?:: ~ ~ c ~

~ ~ 2!

I

;:::;

::0

36 °

35°

SOUTH COAST RANGES 5:27:54 60s

0.21 0.13

\~/\AAfl./\rv.l"'f\.rv\/1.11/1/\.,._ 0.15 0.19

R~

Fig. A4. Records from the south Coast Ranges. Approximate initiation of Lg and Rg is shown. The epicentral distance of each station is shown below the station name. The average Lg and Rg amplitudes are shown to the right of each seismogram.

0.06 0.05

0.28 0.10

0.11 0.20

,~,..,..- 0.21 0.19

0.10 0.08

0.11 0.10

-!"-' ~

~ ~

I ?: ~

~ ;: ~ ; z

I

39°

38°

37°

36°

122" 121 Q

GREAT VALLEY 5:27:54 60s

osu ~Mr1.Wj~MIIIU111VRIIU~IIII"WlftMIAIVoiiAII.I'I~MIIfvvi.MMHn,.w...v,ARMAMI\AMMM,.,AiviA~~WAA"" o 68 0.14 2559

GAR ~V~~~'{YtJifi!VVVW~V~~ o.32 o.31 2588

AFR 2610

NHM 2684

MYL 2774

PWM 2879

PHB 2900

PKE 2922

120"

Fig. A5. Records "from the Great Valley. Approximate initiation of Lg and Rg is shown. The epicentral distance of each station is shown below the station name. The average Lg and Rg amplitudes are shown to the right of each seismogram.

0.31 0.20

0 48 0.47

0 29 0.19

0.36 0.39

\~IIJ-t..rb. 0.37 0.31

0.30 0.42

~ > :z: c = ~ ?: tt' ~ c ;: ~ ~ :z g .., 0

~

-N -N

39 °

38°

37°

36°

& AVRA AFH & AFO

A ASM

& MRF

& MAT

& MNP

\ \

~

'1 ~~ I~

SIERRA NEVADA

AAR 2559

AFH 2593

AVR 2588

AFD 2597

ASM 2611

5:26.54 60s

NAAJ\f.. 0 26 0 21

\Nv.{\,-,1\~ 025 017

0 20 0.18

0 09 0.06

0 27 0 26

MRF 2679

~~~~~~--~~~~~~~~~4v~~~~~~~~~~\M~~~~~~~vvifV~VYV\/\rV'~~~vv--~~~ o1o 0 13

MAT 2723

MNP 2777

WLH 2928

WOF 2993

\1\J\~f\p. 0 17 0.25

0 14 0 26

0 13 0.40

0 07 0.13

121 ° 120° 119° 118° L1 ~

Fig. A6. Records from the Sierra Nevada. Approximate initiation of Lg and Rg is shown. The epicentral distance of each station is shown below the station name. The average Lg and Rg amplitudes are shown to the right of each seismogram.

N

N N

~ ~

I ?: ~

~ ;: ~ ~ z

~ ~

A MD

40'

5 KM I I I I I

1 1 9 °

~ M.KMfMMCS

A MDR A MGP

A M~L M~ MCM

• CL

50'

MAMMOTH

MDC 2743

MLK 2757

MEM 2756

MCS 2758

MDR 2760

5 27:54 60s

~~JV~rv'-'\tv~~--~~~~~~~~~~v- o14 o1s

0 27 0 .32

/\/\~~ 029 OJ2

1f~OJO 042

0 50 O.JS

MGP 2761

~~~~~~~~~~tN~~~~~~~M'~~~~v~~~~~vvvvvv~~~~~~~~~~~ os1 o.J7

MSL 2762

MLC 2763

MCM 2763

MCL 2765

t Lg ~

INvV\JVVVYIJ\J'w-J-V""...fo,,.-vv-- 0 1 4 0 25

V ~''iY"""'fVIJV\f\j-'VVV'VI!'V~Nv'V\-""'""'V\J'-""'-"""'V'v-~r--- 0 12 0.2J

\/I.Atv>IIJ'.f\1\A~ o 2J o.41

!'fJW'rJ\{\N\Af\N'W\(\f'J\Mf'WN\.NV\~ 0 27 0 31

Fig. A7. Records from the Mammoth area. Approximate initiation 0f Lg and Rg is shown. The epicentral distance of each station is shown below the station name. The average Lg and Rg amplitudes are shown to the right of each seismogram.

~ ~

r ?: ~

~ ;: ~ ~ z

I

!" N ....,

12,124 WAUJ AND HEATON: Lg AND Rg WAVF!3 IN CAUFORNIA

block, and PBI has a high amplitude and is located on the Franciscan assemblage. Unlike the north Coast Ranges, Rg appears to have a more consistent average amplitude of 0.2 units and, in general, the same characteristic wave train.

The Great Valley of central California is a northwest trending asymmetric basin (deepest in the west) 700-km­long and 100-km-wide filled with as much as 6-9 km of sedi­ment overlying a crystalline basement [Hwang and Mooney, 1986; Colburn and Mooney, 1986]. Unfortunately, very few stations are located in the valley. Most of the stations in this group are located around the perimeter of the valley, with the exception of OSU on the flank of Sutter Buttes volcano in the center of the northern valley (see Figure AS). Sta­tion OSU, in fact, produced a record distinct from all the others with a long duration of high-frequency ringing and a low Rg amplitude. This record looks much like the records from stations LHE and LMP (Figure A2), also located on the flank of the Mount Shasta volcano.

The valley sediments have a dramatic effect on the am­plitude of the surface waves travelling through it, producing some of the largest surface waves (next to those in the Cas­cades) observed in this study. The Rg amplitude averages 0.3 units and the Lg averages 0.3 units. The three southern­most stations have the longest travel path through the valley and, correspondingly, the highest amplitudes.

The Sierra Nevada batholith trends parallel to the Great Valley. The Moho dips steeply from 27 km under the Great Valley to a depth of about 40 km under the Sierra Nevada [Eaton, 1963; Prodehl, 1979; Mavko and Thompson, 1983; Oppenheimer and Eaton, 1984]. The western foothills of the range are metamorphic, and this is where most of the stations in this study are located Figure A6 shows that the Lg and Rg energy is coherent between stations in this area, and the amplitude tends to increase to the south, although there is much variation from station to station. The dura­tion of the high-amplitude Lg wave train is relatively short compared to the duration in other areas, and the records are largely dominated by the Rg wave train. Most of the stations in the north are located in a complex geologic zone with a network of north-south and northwest-southeast trending faults. The amplitudes in this area of the Sierra Nevada are moderate.

The densest group of stations in this study is in the Mam­moth area shown in Figure A 7. Even though the wave­forms from these stations are very coherent (due to the closely spaced stations), the amplitudes vary considerably over short distances. The Rg amplitudes differ by as much as a factor of 2, but the Lg amplitudes differ by as much as a factor of 4. The amplitudes seem to gradually increase to the south, and then there is a sharp drop-off in the ampli­tudes within a 3-km distance. Both the Lg and Rg arrivals are distinct on the record, and the Rg wave train is more consistently dominant on these records than in any other area

Acknowledgments. We would like to acknowledge Bob Somera at the USGS in Menlo Park for digitizing the FM analog tape and Jerry Eaton, also in Menlo Park, for providing gain informa­tion for the Central California Netwoik stations. We would also like to thank Toshiro Tanimoto and Dave Harlaider of Caltech for providing helpful comments and suggestions. David Hill, Rufus Catchings, David Wald, and two anonymous reviewers provided reviews which greatly improved this document.

Bath, M., The elastic waves Lg and Rg along Euroasiatic paths, Ark. Geofys., 2, 295-324, 1954.

Bateman, P. C., Geologic and geophysical constraints on models for the origin of the Sierra Nevada bathlith, California, in The Geotectonic Development of California, Rubey Vol./, edited by W. G. Ernst, pp. 71-86, Prentice-Hall, Englewood Cliffs, N.J., 1981.

Blake, M. C., Jr., and D. L. Jones, The Fransiscan assemblage and related rocks in northern California: A reintexpretation, in The Geotectonic Development of California, Rubey Vol./, edited by W. G. Ernst, pp. 306-328, Prentice-Hall, Englewood Cliffs, N.J., 1981.

Bouch on, M., The complete synthesis of seismic crustal phases at regional distances, J. Geoplrys. Res., 87, 1735-1741, 1982.

Campillo, M., Sismogrammes synthetiques dans les milieux elas­tiques heterogenes, developpement methodologique et applica­tions, Thesis of Doctoral es Sciences Physiques (in French), Univ. of Grenoble, Grenoble, France, 1986.

Campillo, M., Lg wave propagation in a laterally varying crust and the distribution of the apparent quality factor in central France, J. Geophys. Res., 92, 12,604-12,612, 1987.

Campillo, M., J. L. Plantet, and M. Bouchon, Frequency dependent attenuation in the crust beneath central France from Lg waves: data analysis and numerical modeling, Bull. Seismol. Soc. Am., 75, 1395-1411, 1985.

Cara, M., and J. B. Minster, Multi-mode analysis of Rayleigh­type Lg. Part 1. Theory and applicability of the method, Bull. Seismol. Soc. Am., 7/, 973-984, 1981.

Cara, M., J. B. Minster, and R. LeBras, Multi-mode analysis of Rayleigh-type Lg. Part 2. Application to southern California and the northwestern Sierra Nevada, Bull. Seismol. Soc. Am., 71, 985-1()()2, 1981.

Colburn, R. H., and W. D. Mooney, Two-dimensional velocity structure along the synclinal axis of the Great Valley, California, Bull. Seismol. Soc. Am., 76, 1305-1322, 1986.

Der, Z. A., A. O'Donnell, and P. J. Klouda, An investigation of attenuation, scattering and site effects on regional phases, Rep. VSC-TR-81-11, Teledyne Geotech, Alexandria, Va., 1981.

Der, Z. A., M. E. Marshall, A. O'Donnell, and T. W. McElfresh, Spatial coherence structure and attenuation of the Lg phase, site effects, and the interpretation of the Lg coda, Bull. Seismol. Soc. Am., 74, 1125-1147, 1984.

Eaton, J. P., Crustal structure from San Fransisco, California, to Eureka, Nevada, from seismic-refraction measurements, J. Geo­phys. Res., 68, 5789-5806, 1963.

Given, D. D., L. K. Hutton, and L. M. Jones, The Southern Cali­fornia Netwodc Bulletin, July-December, 1986, U.S. Geol. Sun•. Open File Rep. 87-488, 1987.

Gregersen, S., Lg-wave propagation and crustal s1111ctnre diffet­ences near Denmark and the North Sea, Geoplzys. J R. Astron. Soc., 79, 217-234, 1984.

Hernnann, R B. and A. Kijko, Modeling some empirical Lg rela­tions, Bull. Seismol. Soc. Am., 73, 157-171, 1983.

Hinds, N. E. A., Evolution of the California Landscape, Bull. Calif. Dil'. Mines Geol., /58, 1952.

Hwang, L. J., and W. D. Mooney, Velocity and Q structure of the Great Valley, California, based on synthetic seismogram mod­eling of seismic refraction data, Bull. Seismol. Soc. Am., 76, 1053-1067, 1986.

Johnson, C. E., CUSP - Automated processing and management for large regional networks, Eos Trans. AGU, 62, 957, 1983.

Kafka, A. L., and E. C. Reiter, Dispersion of Rg waves in south­eastern Maine: evidence for lateral anisotropy in the shallow crust, Bull. Seismol. Soc. Am., 77, 925-941, 1987.

Kennett, B. L. N., On regional S, Bull. Seismol. Soc. Am., 75, 1077-1086, 1985.

Kennett, B. L. N., Lg waves and structural boundaries, Bull. Seis­mol. Soc. Am., 76, 1133-1141, 1986.

WALD AND HEATON: Lg AND Rg WAVF!! IN CALIFORNIA 12,125

Knopoff, L., F. Schwab, and E. Kausel, Interpretation of Lg, Geo­phys. J. R. Astron. Soc., 33, 389-404, 1973.

Levshin, A. L., Suiface and Channel Seismic Waves, (in Russian), Nauka, Moscow, 1973.

Mavko, B. B., and G. A. Thompson, Crustal and upper mantle structure of the northern and central Sierra Nevada, J. Geophys. Res., 88, 5874-5892, 1983.

Mooney, W. D., and R. H. Colburn, A seismic-refraction profile across the San Andreas, Sargent, and Calaveras faults, west­central California, Bull. Seismol. Soc. Am., 75, 175-191, 1985.

Mrazek, C. P., Z. A. Der, B. W. Blllker, and A. O'Donnell, Seismic array design for regional phases, in Studies of Seismic Wave Characteristics at Regional Distances, Rep. AL-80-1, Teledyne Geotech, Alexandria, Va., 1980.

North, R. G., Instrumental distortion of m,Lg magnitudes, Eos Trans. AGU, 66, 953-954, 1985.

Nuttli, 0. W., Seismic wave attenuation and magnitude relations for eastern North America, J. Geophys. Res., 78, 5212.-5218, 1973.

Nuttli, 0. W., Lg magnitudes of selected east Kazakhstan un­derground explosions, Bull. Seismol. Soc. Am., 76, 1241-1251, 1986.

Okal, E. A., A theoretical discussion of time domain magnitudes: The Prague Formula forM. and the mautle magnitude Mm. J. Geophys. Res., 94, 4194-4204, 1989.

Oppenheimer, D. H., and J. P. Eaton, Moho orientation beneath central California from regional earthquake travel times, J. Geo­phys. Res., 89, 10267-10282, 1984.

Page, B. M., The southern Coast Ranges, in The Geotectonic De­velopment of California, Rubey Vol. /, edited by W. G. Ernst, pp. 329-417, Prentice-Hall, Englewood Cliffs, N.J., 1981.

Press, F., and M. Ewing, Two slow surface waves across North America, Bull. Seismol. Soc. Am., 42, 219-228, 1952.

Prodehl, C., Crustal structure of the western United States, U.S. Geol. Surv. Prof. Pap., 1034, 1979.

Regan, J., and D. G. Hlllkrider, Numerical modelling of SH Lg waves in and near continental margins, Geophys. J. 1nt., 98, 107-130, 1989.

Rivers, D. W., Effect of crustal structure on Lg, Rep. SDAC-TR-80-Th, Teledyne Geotech, Alexandria, Va., 1980.

Ruzaikin, A. L, I. L. Nersesov, aud V.I. Khanturin, Propagation of Lg and lateral variations in crustal structure in Asia, J. Geoplrys. Res., 82, 307-316, 1977.

Shin, T. C. and R. B. Hermtann, Lg attenuation and somce studtes using 1982 Miramichi data, Bull Seismol. Sot ttm . ;:-;, 3R4-397, 1987.

Walter, A. W., and W. D. Mooney, Crustal structure of the Dia­blo and Gabilan Ranges, central California: a reinterpretation of existing data, Bull. Seismol. Soc. Am., 72, 1567-1590, 1982.

Wetmiller, R. J., R. B. Homer, H. S. Hasegawa, R. G. North, M. Lamontagne, D. H. Weichert, and S. G. Evans, An analysis of the 1985 Nahanni earthquakes, Bull. Seismol. Soc. Am., 78, 590-616, 1988.

Zandt, G., Seismic images of the deep structure of the San Andreas fault system, central Coast Ranges, California, J. Geophys. Res., 86, 5039-5052, 1981.

Zucca, J. I., G. S. Puis, B. Milkereit, W. D. Mooney, and R. D. Catchings, Crustal structure of northeastern California, J. Geophys. Res., 91, 7359-7382, 1986.

T.H. Heaton and L.A. Wald, U.S. Geological Survey, 525 South Wilson Avenue, Pasadena, CA 91106.

(Received May 25, 1990; revised March 11, 1991;

accepted March 22, 1991.)