AD/A-004 825

SOME NOTES ON LONG-PERIOD SIGNALS

Roy Greenfield

Ocean and Atmospheric Science, Incorporated

Prepared for:

Advanced Research Projects Agency Air Force Office of Scientific Research

November 1974

DISTRIBUTED BY:

mi] National Technical Information Service U. S. DEPARTMENT OF COMMERCE

I II II ' _^tuMMM|iaia|Mfck^^|^MUa^M1MMMiH|^MU^gU|ttMHaMMiaiMM|M| - - ■ ^—^M—^—^.

"' ■ ■ mum wm^~~ 1 ■ ■^ —

UNCLASSIFIED

REPORT NUMBER

AFC

SECURITY CLASSIFICATION Of THIS PAGE (Wh*n n«(. «-n/.r..')

REPORT DOCUMENTATION PAGE 2. GOVT ACCESSION NO

4. TITLE (mi Subtlllt)

SOME NOTES ON LONG-PERIOD SIGNALS

7. AuTHORf«J

Roy Greenfield

9. PERFORMING ORGANIZATION NAME AND ADDRESS

Ocean & Atmospheric Science, Inc. 145 Palisade Street, Dobbs Ferry, NY 10522

II CONTROLLING OFFICE NAME AND ADORESS

Advanced Research Projects Agency/NMR 1400 Wilson Boulevard, Arlington, VA 22209

14. MONITORING AGENCY NAME A ADDRE.=.Sfyr 'iHlerenl Irnm CoMfoJI/n« Olllct)

Air Force of Scientific Rescarch/NP 1400 Wilson Boulevard Arlington, VA 22209

KEAD INSTKUCTIONS BEFORE COMPLETING FORM

3 RECIPIENT'S CATALOG NUMBER

5. T/PE OF REPORT ft PERIOD COVERED

Quarterly: July 1, 1974 Sept. 30. 1974

6 PERFORMING ORG. REPORT NUMBER

TR 74-252 B. CONTRACT OR GRANT NUMÖERr.J

F44620-74-C-0066

10. PROGRAM ELEMENT, PROJECT TASK AREA ft WORK UNIT NUMBERS

62701E AO 1827

12. REPORT DATE

November (Ä 1974 '3. NUMBER OF PAGES

43 I'. SECURITY CLASS, (ol (M. repon;

UNCLASSIFIED 15«. OECL ASSIFICATION DOWNGRADING

SCHEDULE

16. DISTRIBUTION STATEMENT (W IM» ftepofi;

Approved for public release; distribution unlimited

17. DISTRIBUTION STATEMENT (ol th, vbttfcl tnlerecf In Ulock 30, II dlllottnl horn Rtport)

18. SUPPLEMENTARY NOTES

Nuclear seismic discrimination; surface ware; ellipticity constraint; rr ixed events.

It. KEY WORDS fConllnua on levtrt» tide II nectfary and Idenllly by block number;

Nuclear seismic discrimination* «urfarc wave; ellipticity constraint;

mixed events. NATIONAL TECHNICAL INFORMATION SERVICE

20 ABSTRACT rConflnu« on ravsrta «/da // nai .

US DApsr1m«nl ol Commerce Spr'ngheld VA. 23151

PRICES SUBJECT TO CHANGE ■■ .nmy anu lUBUlliy Oy (.(IM « miuiurt;

The problem investigated was the one of long-period signals (30-60 seconds) from a single three-component seismometer to determine the Love and Rayleigh components of an explosive event in the presence of noise or an interfering earthquake.

A review of current knowledge concerning long-period noise characteristics and interfering earthquake signals is made. In addition, a review of work concerning; experimental studies ol ellipticity is made. (OVER)

DD , 'XTn 1473 EDITION OF I NOV «S IS OBSOLETE

UNCLASSIFIED SECURITY CLASSIFICATION OF THIS PAGE (man D»(« Enlerad;

m ■■- ■- — - — ■ ■

»v.. . — ..,_..„

UNCLASSIFIED

20. Abstract (continued)

A critique of OAS's proposed ellipticity filter is made and some rec ommendations for improvement arc given. These include the least- squares estimation of the explosion components assuming that the explosion azimuth is known.

/O- UNCLASSIFIED SECURITY CLASSIFICATION OF THIS PAGECWTi«n n«f« F.nlr,, d)

^-^U—,. ._^ - ■ - ■——- - -• - --

mmm^^^*™*~ mpwwwM 1 ■ ■ ■ ■ • ""' i

: l. ! D

i.

I

i, f

OAS OCEAN & ATMOSPHERIC SCIENCE, INC.

145 PALISADE STREET DOBBS FERRY. NEW YORK 10522

914-693-9001

TR 74-252

SOME NOTES ON LONG-PERIOD SIGNALS

by

Dr. Roy Greenfield

Submitted under Contract No. F44620-74-C-0066

AIR FORCE OFFICF OF SCIENÜHC RESEARCH fAF^ri NOTICE 0^ TRANSMITTAL TQ DOC " ' This technical rerorl :,- ■;. ■„ reviewed end is ipprwal ior pri( , . , •..;, i90.12 (;b) Di^r;'-:: ' ■ . led.

D, IV, TAYU I Submitted to: Teclmlcal Information Officer

Air Force Office of Scientific Research/NP 1400 Wilson Boulevard Arlington, Virginia 22209

D D C

^< FFR 10 1975

JlksiHsnTrEili D

:.■ •

November 26, 1974

/A

■ : . ■ ^t- .^ - ■ ■ ■ - ■ M^^MMMMMi^MMteaMaBfl

lU.j I^HlllipiH »li !■• I *li' P^^BWBPPWBWI^^P^^"— '■- ■IPPIWl^^W! ^I^WÜPWP^IWP^P^P-^W^P^BI^W .^^"»^

OAS Form Approved

Budget Bareau No. 22-R0293

OCEAN & ATMOSPHERIC SCIENCE. INC. 145 PALISADE STREET

DOBBS FERRY. NEW YORK 10522 914-693-9001

QUARTERLY TECHNICAL REPORT NO. 2

For Period July 1, 1974 to September 30, 1974

Sponsored by Advanced Research Projects Agency

ARPA Order No. 18 27

ARPA Order Number: Program Code Number! Name of Contractor: Effective Date of Contract: Contract Expiration Date: Amount of Contract Dollars; Contract Number: Principal Investigator

and Phone Number: Program Manager

and Phone Number: Short Title of Work:

1827 4F10 Ocean & Atmospheric Science, Inc. April 1, 1974 March 31, 1975 $128.579 F44620-74-C-C066 Dr. Bernard Harris 914-693-9001 Russell Gershman 914-693-9001 Seismic Verification of Mixed Events

This research was supported by the Advanced Research Projects Agency of the Department of Defense and was monitored by the Air Force Office of Scientific Research under Contract No. F44620-74-C-0C66

/ C

■-■■■ - ■ -- ■■ - - i i -nil mi^^^M—^M«^*

m^^m m» •' ' '■ '"■" ■

TABLE OF CONTENTS

Page

1. 0 Assumptions 1

Z. 0 Literature

2. 1 The Nature of Interference

2. 2 Ellipticity Values and Variations

2

2

16

3. 0 Critique of Present Approach

3. 1 On Spectral Estimation Procedures

3. 2 On the Ellipticity Filter

22

34

4, 0 General Comment 36

Appendix from Newton 38

References 42

ii

rmiiiii M r i — --- - — ■■ ■—■

■nan mi—.

-"■■" '" ■-■ ■ wm —" 1 JUPBPIPBWI^P»" — '■

1. 0 Assumptions

In this study I generally made the following assumptions and

restrictions on what was considered:

1. We would have available only a single three- component seismometer;

2. We wish to find the frequency domain amplitude of both hove and Rayleigh waves of an explosive signal;

3. The origin time and great circle azimuth to the explosive source is known;

4. Interference may be natural or an interfering earl? quake;

5. The period band of 20 to 60 seconds is of interest, with the primary interest in 40-second periods.

-■■ - _____ _______

2. 0 Literature

2. 1 The Nature of Inter for once

A. The natureil background (excludes identifiable

interfering earthquakes):

1. At the shorter period end (20 to 30 seconds),

the natural noise has two major components. Approximately half

the noise power comes from propagating fundamental Rayleigh

waves generated by ocean wave action. The level of this

noise is variable with meteorological conditions in the oceans.

The level of this component falls rapidly vitb increasing period

from the peak at about 16 seconds. This portion of the noise is

spacially coherent and usually comes from cue or two major

directions at any given time. The power level will be similar

on horizontal and vertical instruments. The power level will

not be decreased by seismometer burial (see Capon, 1969).

The remainder of the voii e is locally generated by atmospheric

loading. Turbulent loading is the most common method of generation,

although occasionally an acoustic gravity wave will significantly add

to the seismic noise. Seismometer burial decreases the noise

level of the turbulent component, but not of the gravity wave

noise. For shallow seismometers the horizontals will have a higher

noise level than the verticals.

i i - ■

2. At the longer periods (30 to 60 seconds), the

ocean generated microseisms are not significant. (Savino, et al. ,

1972b; Capon, 1969; Sorrells and Douse. 197''). The major noise source

is from atmospheric turbulence static generation in the vicinity (apnroxi-

mately 500 in) of the seismometer. This noise has the Tollowing properties:

1. On the vertical seismometer, the minimum power level is at 30 to 40-second period;

2. The power level (at 30 seconds) is much higher (15 dB) on a surface seismometer compared to a deep one during turbulent atmospheric conditions, and abc ut 3 dB higher during calm conditions. Power levels vary by about 15 dB with time;

3. Coherence between verticals is discussed by (Savino, et al, 1972b). The coherence is on the order of 0, 59 for a 1 km separation;

4. (Savino, et al , 1972b, p. 189). There is no sifnificant correlation between the horizontal component with the verticals over the period range of 1 0 to 500 seconds due to the noise being generated by many uncorrelated sources. (I believe that in the 1 0 to 30 second band there will at t:mes be some coherence between horizontals and the vertical. );

5. Noise power levels for OGD are similar on horizontal and verticals (Savino, et al. , 1972b)

f

!■ iH I ^MtM^- —

W^r^^*^mimmrm*^^*^*mm*1^^l^^**'~m ' mmi<miimmM\\t*«wmiW' i' i * m.umvmm "i mw.im«.<•««*-•* ■ ,m,„.,ßm„m ,,,,

B. Earthquake Intorfereace

1. Major Arrivals

The initial,normally discussed, arrivals from

earthquakes consist of the following. No effort is made to discuss

these here since their properties are well known.

a' P and S body waves

b. Love waves (direct path)

c. Rayleigh waves (direct path)

d. major multibounce body waves, PP, SS, PPP, SSS, etc.

e. PL coupled leaky waves.

These arrivals are all arrivals occurring before the Rayleigh

wave arrives.

2. Multipath Surface Waves

a. At 20 seconds

Many multipath arrivals are observed. Some multipath

arrivals have power levels comparable to the main Rayleigh wave train.

Azimuths of multipath arrivals and even the main Rayleigh wave may

differ from the great circle path by as much as 40°. The multipaths

arise because of differences in phase velocity between different areas

of the earth (especially between continents and oceans). These contrasts

cause refraction of waves.

In addition, if a wave hits a boundary at a large angle to

the normal, a significant amount of energy may be reflected. Thus,

multipaths can arise even if source and receiver are on the same

i • I" "J .m^^lWi^WiUBPl^^i^B^W^^w« w ' wm*mm* ■m' . - - i ■ iij IIPIMM U »P— ■•>

continent. The best references for these multipaths are Capon(1970)

and Capoii and Everndcn (1971). The interference of multipath

arrivals is also noted by Newton (1973) and Boore and Toksoz (1969)

(See Table I).

b. At 40 seconds

The properties of these multipath Rayleigh waves are

very important for the present work. The most complete study of

their properties is Capon and Evernden (1971). This work is based

on prc-low-pass filtered data. Table 2 gives a compilation of

their results. Note that the 6dB bandpass of the filter used after the

prefilter was (.019 - .031 Hz) or (52 to 52 second period).

From this table it is observed that the highest power

level is always in the fir^t 200 second window, or in both the first

and second windows. For the first window, the azimuth deviation

(arrival angle difference from the great circle) is 0° with the exception

of a 10 deviation for Event //17. Thus, an assumption, that the

arrival direction of the main 40'second period wave is the great circle

path, is justified.

The power levels of the later arriving multipaths is

generally down lOdB and us-.-ally much more from the main arrival.

The multipath arrivals do have angular deviations of an order of 20°

in some cases.

—^.„^-^—_^.„_^_.______J. J

Table 1: Measured Variations in Azimuth of arrival and Power Levels at Various Times for Events From Capon (1970)

—

lime («■<) Al 40 MC

Axiniuttul

Al J< Mt

tt\ Lltinnj'

At 4" sec

V<. • et

At U H<

.»».i»t

}.\tnt No Al .') «fc At JO fee Al J* <ee At :IJ stc

(dt:) (dft 1

-11

ldt$l Mlb, (db) (db) idbi

i 0 210 0 -5 7 6 0 — 200-400 — 11 11 -12 — 8 2 7

400 000 — — II 10 — — 5 4

coo soo 0 300 ~ö~ "T"

30 __ IS — — 12 5

2 —" — 5 5 — — 200-400 — 0 — — — 2 — — 400 000 — -- -5 -5 — — 0 4

0-200

_- _-• -40__ _.-12 . 0 ~ 4

7... 2.. —.

3 0 — 2CKM00 0 -11 -10 — 3 9 12

400 G00 .— — 2 0 — — 10 11

_r,0o S'M^_ — .17„ )0 — — 23 _ 17

■i 0-2IA) 0 - " 0 — -0 "T" b" — 12

20U-100 — 0 0 — — 4 0 — 400 f.iX) — — 3 5 — — 1 2

COO SiiO 0 200 '

— r- .10 .. 5 ' o' —

4

b e 0 — 20^ 400 — — -8 -9 — — 1 0

403 («) — — -10 -13 — — 9 8

000 soo — -20 -18 — — 21 IS

G o Joo 0 0 3 _. "11 ~9 11 ' — 100 100 — 13 3 3 — 8 5 n 400-000 — — -0 3 1 0

COO-5-00 — — -55 .-0., i-.., 9-

10 0

-... J? 7 0 200 ' "ö 0 -6 —

200 100 — -■ — 0 — — — 2

400 000 — — 24 — — — IG — C00 POO __

o" II

""-3"' 22 _

3 • 2 20

' 0 18

S o 3Ö5 — 200 -100 — — 0 -5 — — 8 4

400 r^w — — — 11 — — ■— 12

coo soo, 0-200

— — — — — — — 9 o~' 0 ' ~4~" — -7 - 1 0 —

200 400 — — -4 -0 — — 10 5

400 C00 — — 12 18 — — 13 14

._i//> soo .._ — 30 . .IS — — IS 17 ... 1Ü 0 200 3 3" — 20 19 —

200 100 3 3 0 — 4 . 0 3 — 400-COO — — -2 -3 — — 2 3

11 „coo soo

O 200 - 5 _-

' -25' —. .

"" 13 ~ 'it" ._ «_-

— 200 100 — — -5 -20 — — 4 9

400-600 — — 8 -20 — — 1 2

COO soo 0 200

-_ — — -0 3 r .M

0

12~" IT" ""-n' — -• — 200 400 — 8 -14 50 — 7 0 2

403 COO -28 -2S -17 — 12 7 3 — ßÖOLfiKL -2S_ —___ ... — .)2_ —a- ■ *-

11 0-2(K) -3 -4 — -- 9" 11 — 200 100 .- 0 2 — — G 0 — 400 OiO 30 — 37 -5 14 — 5 1 goo- SOO 0 2QO

41 o

37 -tt 31 15 '11

13 " 7

10 14

U -4 — 7 — 200-lfX) — 21 -7 -2 — 9 i 7

IOO-TIOG — 2«.) 0 „ o — 10 4 0

no-soo — — 39 - 2 — - — 10 4

Tabl e 1: M easured Variations in Azimuth of Arrival and (contd) F ower Levels at Various Times for Events A

F torn Capon (1970)

1 imt («cc) AlimutKal deviations* Powtr Ltvtlst

ivcnl Xo. At Si «c (dft)

At 2:- itc (des)

At 20 KC At W sec (•If«) Mb)

At Ji itc At Hitc (db) idb)

At :o -n. • idb)

15 0-200 0 -4 -9 - u 8 0 _ 200-100 — — 6 -10 - - C 1 400-W)0 — — — 0 — 7 000-SCO — 20 20 - — 14 10

10 0-200 0 0 0 — 4 2 7 200-100 25 13 -5 -10 8 0 2 5 400-COO 25 13 7 -10 12 3 0 5 COO 800

"TRSxl 25 10

25 ' ~' 10

25 r

-10 10 - — 12" "

8 6 ii W

7 17

200-100 — 10 7 5 - 11 7 8 400-COO — — — 5 - — — 0

_fi00 S00 — — — -17 - 3 —~""

8 IS 0-200 0 2 — - C

20/--100 — — 2 3 - - 0 0 400-000 — — 27 7 - — 14 13 COO SOO ■0-2013" *T

— 37 _ _ _ — 17 — 19 0 — 6

200-'00 0 0 2 -0 5 8 7 11 40C COO 2 2 -0 -e i 4 2 9 f^O-800

0 0" - 6 _ -8 — — 1 5

sir- Ü'200 '

. 20..._. _. _ 200-100 0 0 0 0 10 3 0 8 400 000 — — • 0 0 — 9 0 000 S00._. _3j -23 — __ — 10

6 5 . 18_..>

21 0-200 0 -5 — 7 13 200 100 — — -11 -9 — — 3 0 400 GOO — — — — —

fl COO-GOO 0-200 0" 0 3 _' 9 6 "l3"'

. —

200-100 —- — 3 3 — - 0 11 400-COO — — — 10 - — 4 COO-SCO — —

0 10 -

— ' 6 —4 0 8

23 0-2U0 0 0 200 400 — — -29 -29 - — 7 1 400 COO — — -39 -39 - — 4 10 COO SOO — — -30 -39 -

13 7 "' — 13 5~ 0"~

14 24 Ü-200 0 0 7 ~~i

200-400 — — — 15 — 10 400 COO — — — — — COO 800 — — ^—-

~ —' 5 ' ~o — '

25 o äoo 0 -5 200-100 — 10 -12 -19 - 9 0 6 400 COO — — 3 3 - — 6 2

20 COO-SOO

"0-200 0- 0 10 10 -

"-- 14 - 8 13 A—

8

200-400 — 0 0 6 - 10 ' 2 0 400 COO — — — G - It 000-800 — — — — — — —

1 ''Tiiwcr IcvrU in dh rrl.-itivf to maxinium power. ■^ <T AziniuUinl (icvi:inoiw r.rc «lidcicnccs liclv.t-tii iric.isuroil aii)mith of arrival and true azimuth of cricciilei of cvcnl.

- - - ^-————»—. l^^>.«^M^i

f -'""P ■ ■ ' ' «««■■■■IWi

Table 2: Measured Variations in Azimuth of Arrival and Power Levels at Various Times of the 40-sec Period Rayleigh Wave Group for 10 Events- From Capon and Evernden (1971)

lnnl.Vo. T","' MCI

AlimulMI I)rMili.int

i Utg i pi»ct t.nti:

I \h M 0 0 :?0(M(r) IG -13 miuxi -5 -20 COO MM -1 -18 SOO-KX» -1 -I'J

lOOO-IJOO -{ -22 nofr-iKxi H -20 H0O-MO0 ~i -19

2 0-200 0 0 ?00-I(J0 0 -5 400-000 0 -10 f^JO soo -I! -25

' töo-iorto -!! -24 Kxo-iaoo -II -23 I^'X) 1-1(0 -'I -17 1-100-1000 -II ~2I

3 0-200 0

-24

200-^00 0 400-000 0 C0O-G00 -{ -32 NW-IOOO -30 -20

1000-1203 -30 -27 1200-1-100 H -33 1100-1 COO -5 -31

6 0 200 0 0 200-100 0 -9 400-000 20 -9 COO-SOO 20 -8 SOO-1000 20 -8

1000-1200 -11 -11 320O-H00 -11 -13 1403-1000 -1! -19

8 0-200 0 0 200-100 0 -14 400 COO —!1 -30 CO3-S00 -ii -34 S00 1000 --II -30

1000-1200 -II -30 1203-1-100 -II -38 NOO-ICOO -II -40

16 , 0-200 0 0 1 200-400 12 o J 400 C00 -II

* ^ uOO SOO 25 ■i 4 800-1000 -11 -7

1000-1200 H -13 1200-1K)0 H -11 noo-ioaj -I' -8

: similar power

in two windows

, - J

Table 2: Measured Variations ir Azimuth of Arrival and (contd. ) Power Levels at Various Times of the 40-sec

Period Raylcigh Wave Group for 10 Events. From C?pon and Evernden (1971)

Event \n Timt* Aiimuthil Do i-ilion' • Pootr I.cvrl:

i: 0-200 10 0 2()0-t00 10 -6 •100-000 I

1 -10 600 S00 It -17 MO-IOOO -§ -IS

1000-li?(»0 -5 -21 1200 M00 -5 -23 HOO-ICOO -5 — 22

19 0-200 0 0" 0, 200-100 0

400-000 -li -15 COO-SOO -11 -20 800-1000 -17 -21

1000-1200 -II -24 1200-1100 --122 -23 1400*1000 -30 -22

22 0 200 0 0 200^)00 0 -10 400*00 M -22 eoasoo -11 -22 soo-iono -II -23

Moo-taoo -II -29 1200-1100 -li -24 140O-3ÜCO -li -27

25 0-2(xj 0 0 200-100 19 -13 400 C00 H -19 COO-SOO . -J -18 no-iooo -s -15

1000-1200 -{ -18 1200-1100 -§ -IS 1400 1CO0 -\ -10

• Time is mensured relaiivc lo onset of Rnyleigh wave.

f Azinmthnl (icviation? are (llfTcrcnccs between measured aiiminti of arrival and true azruuth of e])!- ccntcr of cvem.

} Power level.« in decibe!» relative to power of ninin proi!i>.

{ I'ower levtl is bolow that of background noise lind no dirccluui of arrival can be asMtmed.

I 1'ower level is above that <A backpruur.d noise but no direction of arrival can be assigned.

"-vtSb&L

Ma*M*_. __ liMMl

r^mmmmmmmmmm^mii^rm\inK u ■ i i nni^w^nmma«

The description of Love wave multipaths at 40 second

period is given by O.pon (1971, L). The Love wave multipaths have

similar properties (azimuth deviations and relative power) as the i

Rayleigh wave multipaths.

3. Late Arriving Earthquake Interference

By late arriving interference, we mean energy

which comes after the Rayleigh wave and* simple multipaths. This inter-

ference includes

a) Surface waver going the long way around the earth or going around more than once,

b) Multibounce body waves,

c) PKPPKP (also called P'P') waves. We will be discussing the roughly 40 second and above period range.

Capon and Evernden (1971) - note that the coda level decays

rapidly to more than 30 dB below the main Rayleigh arrival in

6-7 minutes for some events. For other events, the coda decreases

much more slowly, with coda only 7 to 13 dB below the main

Rayleigh wave power for period of 5,000 seconds or more. Rates of

decay are shown in Figure 0, for the first 26 minutes.

By f-k determinations of azimuth and phase velocity across LASA,

10

- -— ---^- ■ - ■-■■■-.-.- ■■ ----■--,. ^ ■—.^^^ ■ ... ^.. . ^ _ . _,., . .. .. -_^.^ iMafla^|MBH|

'-«" " ' '•>'>' ^m^*mm*mm*m

•40 r

-3?

-?e

" -24 _l D > B 3 a Q

o O- -,0 < O o

-1?

W

X

Oi3

• •

• • •

•

X

• 5 1 X : • i

i ,

C.'-O 135-167 ?0-?5: J5 67 10-IJJ KT-IO 2li-ZC7

TIME Af Ttn O'JStT OF RAYLCIGH WAVE (mn)

(a)

Wolfs X AViRAGf POWtft

fOR DATA ftBOVi Ntnt n vtL

t INO;:AUS DATA KUW (.CISF: LLVtL

« I

-4D

UJ -J? o

4 UJ > < -I LI

-?"

J

l_J

J_

MAXIMUM

"< AVERAGE

I MMMM UM

I I J_ JJ 6 7 too UJ 10 7 iO;i ?5J ^'.7

DIFFLRENCE IN ARRIVAL TIMES Of RAYLEIGH WAVES (ir n)

(b)

Fig. 0: Rate of Decay of Coda Power Level and Performance Characteristic of Proposed Detection Method. From Capon and Evernden (1V71)

11

• ■ »a

Capon and Evernden (1971) identilied waves from the paths a), b), and

c) described above.

Savino et al.(1972 b) p 168

They give the duration of coda from large events (average

body wave magnitude of 6) as approximately 9 hours at Ogdensburg, (OGB),

but note that the predominant periods are 60-180 seconds. Thus,

much of this coda is out of our band. This type of coda dominates

the OGB record approximately 10% of the time. I ^peculate that

with a high pass cutoff at 60 second period, the fraction of time

that the earthquake coda predominates the atmospheric noise would

be greatly reduced.

Savino et ?1. (1974 b, p. 167)

Following Alsop and Brune (1965) it is suggested that multiply

reflected P and S body waves make up a large portion of the late coda.

This agrees with Capon and Evernden (1971). We note that the

polarization (ellipticity) of body waves will be much different from

Rayleigh waves. If the wave arrives at the seismometer as a P wave,

it will be predominantly on the vertical; and if, as an S wave, predom-

inantly on the horizontals. (I have not checked this last statement

carefully for 40 second and longer waves, but I do know of some

literature to look at if you desire) (Fig. 1 gives OGD seismograms

from seveval events)«

12

m ummm i — ——___^—^»—

vmrmmmi^mmmmmmmmm^^K^^nmmn« .n >' > ....■ —.~i..i.i-ii . n » ui m imiiiii « 1 1..1. .n. ■

A P,,,^ BACKGROUND AFTER <MMI SEA EVENT 13 FEB 70

UPt . . . ?000 ' *>*

B KURILE IS1 13 JUN 69

--ce^o'

'^fe^X^^^fe^i^^^^^

^ UPT

•»i ?,'.'»,•»• ■'•'•■••.'.•'■'' '>.''' •'■; '."'■>'.■ •• 1 >•.'■'',:'•. ■'••.■■ '.• ''' '• v ''.'v ;,.■.•'•'■1

' .' . . , .', I 12^0 3iAur, yr. " I .1 ' .'.•.''..

D

1240 31 AUG 70 .1 |l 1 ' •. 1 • I,

PERU-BOLIVIA tft AUOtJ 1 WIN

IH- . ' 1/33 ^ H

•--' ,->- •. . . v _ .•^.••.•'»-■-' % - ,'_•;>• A**..^;^', .«•• y.i, -•-.• .; ' ...-<-.«■

Fig. 1; Comparison of High-Gain Seismograms (A), (B), and (C) at OGD with a relatively low-gain Seismogram (D) from an instru- ment at Palisades, New York, for .he duration of Intermediate and Deep-Focus Earthquakes. Epicentral Data for the Earth- quakes in (A), (B), and (C) are given in Table 3. The Epicentral Data Reported by the USCGS for the Event in (D) are Aug. 15, 1963, 17h25m06s, 13. 8S, 69. 3W, Peru, Bolivia, h = 543 km. The magnitude was estimated at 7 1/4 by the Seismological Laboratory, Pasadena. From Savino et al. 1972

-■!' ■ — ——.-..-—.-

The gea«ral duration of coda at 40-60 second period and it*

prcpertus (i.e.. types of waves) does not seen; to have been

published. Neither Savino et al. or Capon and Evernden have

addressed the very late coda in this band systematically.

Savino et al. (1972 a)

They give results from 8 of the high gain stations. Detecti

ability is discussed, as is noise levels at the various stations. Band

pass filtering U discussed as is polarization filtering.

on

■MBMWMaHB

i ■■■ H * nmmim immm-^mm^m*^^*-^**^*^^*^** ^w^finii^n .■■>»»■■ ■■■■■■■ i^i iw^wnii ., ,.wv.... , - —-

2. 2 Ellipticity Values and Variation8

Boore and Toksbz

Phaf.e difference between vertical and horizontal components

were usually 90 i 10 ( p, 339). This paper also shows theoret-

ical phase between horizontal and verticals for iruUipath interference.

Irterference peaks are usually between 20 and 30 seconds.

Figure 2 shows ellipticity values and gives a means of

estimating scatter as a function of period.

Newton

Figure 3 shows scatter of azimuth of arrival data. f-k analysis

with LAS/, gives about + 5% ('90%' confident" limited to great circle

azimuth. (See also Newton, p. ^3), Using ampiitude ratios of horiz-

ontal power of f-k gives errors in azimuth of about 8° both positive

and negative This work was done, however, on only a few carefully

selected earthquakes, and applied to the main part of the Rayleigh

wave train.

Higher mode Rayleigh waves were also studied. These waves

have higher group velocity than the fundamental mode ar.d arrive

somewhat before the fundamental mode. The eFipticity for these higher

modes differs from the ellipticity of the fundamental mode. (Figure 4,

a, b, and c)

On Figure 5, I have compared the Boore-Toksoz results with

Newton. Note that they differ.

15

- -- -- - -- - - - ■ '

1000

-w « I »"n wm

>

B 0.^00;- a I

| I w 0.700-

I i

o.eoc'

• -:-» • •• • • ; ~i

."00EL M I

10 S 20 2i 30 35" -10" 45 "10 55 PERIOD (sec)

Fig. 2: Measured Ellipticity for Different Events and Different Sites. The Solid Line is the Theoretical Ellipticity Corresponding to Model Ml. From Boore and Toksoz.

16

- - — - — - — - ■ - - ^ J

'■ ■ I

I

I I t

% ■p 3 ^ M K J O > Ü •r'

O M i o M Ü i x: u C) 4J c u 3 Ü >-< 1 ^ p •H Ü o N

>w M n ^ •H 10 0) ■o W ^1

o u H rH u nj •H

-C Ä U 4J -P :J 3 ■u i f; d

■H H CJ N N M < fU tn

1.0

0.5 -

0.0

5.0

4.6

4.2

3.8

3.4

3.0

25

20

15

10

5

-10

Greenland Sea earthauakc

Tonga Island earthquake

OQ o a - c.

O D ^ C

O

O

ED a a

a □

^—+ o a ~ o,

O B - 0

o a

N-s

-^ O D o

10

_ 0 a

<? o

B8 oe .. .0

0 1 CCS o o —-—^

— E33 •S o

1

CD D

1

I- 0°

20 30 «10 Period (sec)

50

^

(a)

e (b)

N-S

E-W-

■ []

O 0

5 (c)

60

Fig. 3: Results of the f-k Analysis of Signals from the Tonga Island and Greenland Sea Earthquakes. (a) Normalized Amplitude Spectra (b) Phase Velocities. (c) Propagation Directions. (From Newton, 1973, p. 109)

17

MOT ^m

1 1 II 1 i i i

(/) •H w in >i -H w

_ ■-< to -H fj in >, n - o

00 i-; -H - < 10 >, rj 10 fj H r-<

>i C 10 rj < <si < (0 rj r-i n u r. *-■> r5 U rj ,K' c -> ^ I d A;

_ ,Q >1*.4 | o y r< — ^IM t^ 10 N rj —

W >, M 1 O O n ^^ • ( ,0 ^1 rj ri

O V) rj 'J u 0 dfkjo M w nj mil

_ 3 1 1 I I o tr i i VX> X* 0) o -IJ ,y A^ a o U d d ^.; A; d 3 ri d d < <i <^ < o o1 t-1 y 3

& .'■: o1 c o -u 4J ^; ^ tr. u i-i -P v

_ tJ d d ui M •H C) O d a <^ «^w o

IT) K o ü

d d O O O 'Ü r)

•H t/j lo c: t: •»-• d d t 1 4J ß 'O 'O r-H .H r/ «o i* r: io io S •^o H d d H H p .

_ +) rH r-t <: ß (1 d d rH

<2: tt O

1 O O f., Di B '0 ci o c c n J 03D ! «H VH VH ü 0 j . 2 O O H t < I

D O 0<]< <^ Ü o

O <3 CC^ o n

• O <] D G<!oo D h D o or c<i) 005

~ Q o O 0 yO 0 IN

Oo^ o/ < O 0

p / / o o Go or

I" 1 1 i ' I i i o r-i

V 0)

o

v IN

H I

00 «

o I ?

i^TDTqclTTia

18

m

t

CO

4) > U t) in

o 05 S

•-• fa

Si a s o CO *

10 ^ 0 0

go 5J

n)

- ■ ■ -- -■ ■-■- - "

■ ' ' 1 '■ ' ■"■

(0 •H U)

■J> >1 •H rH W (C >i C

rH »0 I"! r; — « AC

10 •H W >i

u

10

I I

IT fj

ß a —

I

«o JO p to

>1 10 n

.1111

0)

Pi 0

u

o

Q)

a

"0

rH

rH c

I

C) y. n p

£ M r3 0)

rH n

rH t:

M

to tn

o o Ü Ü tn H

o o<]<1

4iD< II

I 'I

o W OHM'

SrH s

rH rH H

•O »O 'O OOO S fi 8

I I

<]|<fo<i

0< 4n: I

tan <30j i

©JOdtiCD O

/l C4 oo

o 00

o I*«

_ o

o

Ü 0) 10

o in TJ a

■ri H w

o

o

1 IO 1 1 1 1 r o

«* CM O CO vo • • • • • r H 1 1 ? o

A:iTD-nd-ni3

19

flj u u ti

2

<;

4)

^ r^- o ^» w -O Ü d

o tn J~»

o is 0)

u ■ i-4

5 B o

w ^

0 . I-H

4> H

i ^ 01 ^ O O (X h go .5.5

• r4

—„-., -

WM 11

1 1 1 i 1 1 1 i

-

u W an n<l3 <xco

1 (northeast)

2 (southeast)

3 (southwest)

4 (northwest)

_ M oo

QQ

ector

ector

ector

ector

w w w in

tM w LASA

LASA

LASA

LASA

o

1*9 Q

a<JO>

"*

CM

> Q 0

er >IJIQ CdEM

§ wrrrj

o Q u w ^CTCM i^a u CQ a

<> 0 0 ct«> ü] a a n

OO u 0<i> BM> □ D a

CM

D QraUr.tJ <]t>nO

o in

io

o

U

in

O in -H n M

0)

o

CQ

a D <3

a)

i

«1 o o u I

3

<

<

■ o S

Q

o

I to

d |

i O to

^3 4)

o a c

• H

4)

0)

o 4)

4) >

•■-<

U u

a •

g 1. »J .ü er (u PL. 4J ^

•a ^

| en 0

IU TJ ■n B O nJ

21 m V-4

(N ni d

rC 9 60 5

• H O S H • • u

o ■*

CN! • 00

• H

PM

.^ • ^-. CO

4)

y •■-I

d o W 2

o 00

o o o o

20

■ - ■ - - • - -■ -— --

■ww- —■— "" ■ " -

TTrrr

::(:: ;::l.:

;::l

::|-

%

^^

.::;•• ;::ui;^

--i-^-rr

Üiiiiii

•t:

:.'t:ji

-TTT-

:;.:!::;

.:;i:. :••!::

;: 1

•—XT^TJ-

:ti'::

;i.:::t;;-T rfe:

LÜ1

rn'n!

:|;:: ;:-t: ..i. . r:

r-rr iilkilii::!

_ ....

TTT—r;rv J.

A:

Rp ^ te"'"" ^'": -^*~^ ■j.;:■<■:■ :;JL{ """

rt- ■ K

i.^f:;

NE

^ p •■-I nJ

Is W N

o

B o « 0 e

iwl § o I § ^i z u I rt •-, ■•-,

JJ§|T -i_ —^ _

:l:---

x—L„ :-u4-i^i^a^Ii.Li-jl^jfctLll^^ZILlLiJx-^^-__li.l_

o • r-l M

T-—r--~l--r

... ... ..

|r::;::;h:\:v Wffi

'^

•: r

:; .

•!

!

^ppdinnf -•?/-

mmmmj

m ■■•■ - m m-.m - m.mm .... .. L-

■

3. 0 Critique of Present Approach

3' 1 On Spectral Estimation Procedures

The direct part of the Rayleigh wave (excluding

multipath arrivals) is in the frequency domain

(V/CM ) r

where C C w) is phase velocity, Jk

Y ' is range

~£

1*1 /c

and the arrival time of a given frequency is given by

where fj (J) is the group velocity, given by ^/£uJ

For the 40 to 60 second period range {J (S) is fairly

constant. Thus, energy arrives over a fairly short period of time.

Figures 6 and 7 show LASA LPZ seismograms after filtering by

the filter shown in Figure 8. For the event of Figure 6. the energy

is contained in a 200 second window. Thus, a 200 second window, as

presently being used by OAS. is sufficient for this event. However,

it remains to develop a method to select th2 proper time interval.

Overlapping windows, starting every 30 seconds would

probably assure one window covered the arrival. Perhaps, the

criteria of the window with the most energy in the explosive event

22

- - — - - - — MBMUMMH

■Ill uiiwwa I 11 vm 11 II^II i" it m*^*tm^mm*mnK^i*mm^^mm ***m^m***^^^mqm*m

■4->

C

>

!i -

O <M n ro

O

O ^- 03

-o "

CD TJ Iß

5; :>

— ai I . jm o

CJ cd ro u. -o

ID t in —

- o

<

s W

5 « t i u o

V >

•v o u o

B G >

o lü

H a

Ü

V J H

d o

q «> «I u y i B a o

do

23

■- - ■■ —' -- --

Change

* * * r f\

r*

AO

n

"^V \J WW***^ J.o.

"•^VVi/'c-v— 08 19

' I 1 -I 1 1 1 1 I I I I I t ■ I ^V OB.24 oe 29 06 54

k /fxAc Coe*^

FJ ^/Ww^- -ovw^*^'^ -vvw^ ^ft

Note: M *"Xf^V^^*AfTfWl>-V^- Gain

AO

T

- Ji^- ' ' 'oeLo' ' ' 'oeLo'' ' ' ' 09,U ' (b)

Fig. 7: Prefiltered Wave Forms of Event 19 of October 15, 1967, near the Coast of Nicaragua, Showing Initial Rayleigh-Wave Group Arrival and Multiply Reflected P Waves. From Capon and Evernder. (1971)

24

I ■ IIMII^—i«^——

^■' ■ 1

"' — ■ II — I ■ 1 ■ -•-'■■-"—'"

PERIOD (j«c)

(0)

-* ■ I ■ \ —y'^j^V,^ m mm

I L J I I L J 1 1 I ! L

TIME (mm)

(b)

Fig. 8: Frequency Response and Impulse Response of of Fifth Order Low-Pass Chebyshev Filter. From Capon and Evernden (1971) Pre Low- Pass Filter.

25

- -- -■ ■ -

WW«i ^m^umm^mmm—^^**m*m^~—i

azimuth direction (en the rotated horizontal) could be used to

determine which of the windows actually contains the arrival.

If the window used does not contain most of the energy in

the arrival, I would expect an error in the measured phase spectrum.

Figure 7 shows an event (#19) with a Rayleigh wave of 300 seconds

length. From Table 2, we see that, for three (Events 3, 16, and 19)

of the ten studied, the power levels were similar for the first two

200 second windows. For these events, a longer window should be

used.

For the 20-40 second band, the group velocity increases

with period. This portion of the Rayleigh wave train Lf highly

dispersed and typically arrives over a 400-600 second period (the

arrivals are extended even more ir time by multipaths). Thus, for

this period range no single 200 second window will contain all

frequencies. Results of Capon and Evernden (1971, p. 819) ind

Newton (1973) demonstrate tlvs point. Also see Table I and records

shown in Figures 9-14.

I have not presently developed this approach, but rather than

using spectral analysis, the bank of bandpass filters approach with

cross correlations between components to separate the Ravleigh waves

of two events might be considered.

For the LASA (25 second Peak) seismometers, 20 second

period energy leaked through the digital filters into the 40 second

26

■ - ^J^__

mr^^i^mmm

in

rö

O

f6

i I \\

I

■■•;

I

■ 1 l

I

m o

o o ro 0)

ro O

1

f T ! M t

o

r? x

I !

ro LLI

y

1

UJ UJ

11 I I

1

f

f

CJ

i

o m cvr

- m

ro

O

~ rO

in O K)

O O ro S v-'

I E > W

W ■ u %

^ ^ r-

t • o ^ c ^ 5 T3

E E ^ 5 w Si

ft A

V > o X 0 u H 2 k

o^

I I J 1 1 I

ro

I ! 9 CO ü

(VJ O

O in rO

< m

ro O

27

u ■pr^^wwfpp

■

I t.

f I I ! I ! f

0 c « < «D o

r'' n

f

i (0

4

s o

o K)

s <rt N

O

. o ,^ _ M

IÖ v» o

jL "^r o *-•> — TT ( r_7 ro m *— • * O f 41

C 1 O 0) >

S" K> w . [

o Mt: 0 «J -^

: in 9 £| S | u~

M 0 0 -o c

(i, ctf 4) d X) ■ 0) u ö

n >w i: O

^ t- w B ri 5 •* 9 O

o - fi« g

— N I ■ s « ., 0 .'- _ ■ -c s ^ 1 i H 5 h

• • ' ,''

IO o

'r^* ■—I _ - ■ M 1

' 'Z^* o M s ••H x tu o

■

1

O

o i _ o

K> n o

28

^MMHM MMMMMI

■«'■■>■ ! ' wmfmtvmn^p^m

S ava^ble copy,

* • '• '■..■** **/"v ..•'*.*,■, VA--*-'.^- •'"t -SV

C2 —^j^wJ^/'&^:Z:<ll;:^:y*

„V. ". V.. * -w . - «M

w^^Wi—IXDWP i-^v* i

~***fv*-*Jffrt~*+r,-**»\*j**f MN

- ^>SY^^>^AVAV^VY^.V*"M%*^SV O ■ * ^ ^ ■ ■» » *■■**

^<»^NVV/A*V.V-S^VWV^VI'^VI'^^"',ft*rf^■ -A. *^ ■"■"^ "^ ■ '^

-r HlfllMrtfW^fTftt*>*~*i- ■■■- -^—

D2 H^M»..! ^^jtW^Mr^M^ <*1" ' -~~vy~~w, ■jg.*. .i^V^V/'^'*^^'""

r • jybJ» V^—.^.V/A'' '' WV''," V1»',',''' V.'' yjM(|J(MMM|^H"l''**-,,",—,',l**"w*^*"'****"

I 1 .L J L 1 J L 06 05 0815 Öili ÖFZÖ 06 25 06 JO 06 J5 06« 06 45 06 50 09 55 09 00

Rl P4

Fig 11: The Long-Period Wave Forme for Event 19, the October 15, 1967, Event near the Coast of Nicaragua. From Capon and Evernden (1971)

29

■-■•■• --■- ■ -■ -■--—■-—"—

r Periods (sec) at peak of Gaussian filter transfer functions.

f Relative amplitude scales 256 UO) ^1 .£./ ,«WJ

128 (13)

85 (33) ^N^^^^-vV^^**- :^^ycVfV:^ i

64 (68) IM Ml rM A/1

—-^AA/y\/lywvK.

V'"»-«---*^

.^■yu»»»-^-

^' ■■^w^^MiWuvV\AM/Vw\W wv si/yAiyv»***^*-—j-

, 37(188) Mi—^

. 32(492)

'Klljjjflli^ ^^■^^S•VSA^V^'*^**^^I|^*

?8(708) ~^v

iilliii ., '' f ; ' ''!' ' ., —••■/■A/rvv^Aift/'/;-w-;-vr;V'^—•'■>■•—•■

.26(665)

.23(834)

'»»lillllll'

%l—^——<-.

21(911)

«y^A^^^..»..^,..sVt',Z. .——... 'y/yi'--— riV.V"'">V.v.

20(740) "fl>"l.ll.

"•iiiiiiii' 18(364)

;fv ^*^*^^l*^^^**^^^^^^^AVlV^SVVSVA^^

■ * i *' * m************* ,*i*tHWiiiM0tt*i*m

17(123)

16(101)

<^^^

nllM H 11'I llh |i1|i| H^l , |MlJ (1j;,, •■:,,,,, f,,,n.n

Mil

pirn M|||||llli IIIIIK1'' '"''""

Fig. 12: Narrow-bandpass Filter Traces cf the Vertical Component, Mid-Atlantic Ridge Earthquake. From Newton, 1973.

30

- - ■ - ■MMMT^MHMU

w/m —-

rfunctioniSCC) at Peak 0f Gaussian filter transfer

j— Relative amplitude scales

256_(10) f\ ~

v ^_^

■•-^~V>r'v»'> —■ ^■ww»* 64 (68) A

51(140 . ni-,- - > .\-A*<

- - - . y JvW

37(188) /■■^/•MAAA.-WüVvW: —**4lUN^%]!fMM ^Jifi^te/w

.32(492) „iii.

»iHfiilHi 28(708) „.illi.

1—'—' ' ' ■ '**** ■•'■•-^'—•-'-^VvV.r-'-r-- J^<W'^/.Vi«//.A ********** K.

26(665) .illllllMii. ——i— ^^w^w^,—,——^^. r7yj:j>V;.wAv^*M.i^wMv^w*wi'—- ■itiiin 23(834) ..iiMHHiiii...

«■VA'iVi^«*'

21(911) ***•***-■ ■ *■ —<W^m^m»»WWWMW»«

20(740) «...

''Hiiiiiin" ■.■»ANSV/fnVi'jViSvw.% ■«.

^vAVWrt ...HIM'"»!»... ■ t^^>^^ A^M^^A^^^^^vfc^yy^yi^ivyvvvy^g^iii^v^^», >

^MiH'HMiHti...

"'"lllllMl!!«'

18(634)

"^"^■■''■7"^"lllll|l||l|l",

.im:

Fig. 13: Narrow-bandpass Filter Traces of the Radial Componer.t Seismogram Containing the Fundamenta1 Mode Rayleigh Wave from the Mid-Atlantic Ridge Earthquake.

L 31

- --■■

•"-■■" J—"-^ -

11. —Periods (sec) at peak of Gaussian filter transfer

functions.

r Relative amplitude scales , •,

128 (13)A A AW ^

83 (33)

64 (68)

^T^y^-^tA/A/V/t^u^y/vy vAutii^/^/^yy/v,,«^«. .. i.

' M ! i > A A ft A A A j ittrr^^r^TM'Vv'Tvv''"*'1'11' ■ ' •^\AAA./v.|vv»».'> ■—

*—»'WV^-^'^A/^,«».. *mAfiff>f>0t »»«^■■» ■■.■»■<i».,«^.t,».«.«»i.»>....t<i.i. j.

28(708) ^•■^•v»^v»s«^^t««wVv',A'V»'^»V«VWV'**»

26(665) i_rj ' ^^*^*^'*^»»N^^'^vv^^\^^'t^■^y/l^/yyyy^^//<i.^<.svl«v^vvl^vl> «»w*» ■ ■ i

.23(834) ^ -t f i - . «

21(911)

20(740)

18(364)

L17(l?3) V.^ ai—wan ^vw

,16(101)

■^- 'Virt^^VA^^^^*»-*» - ..^wo^^vwMVWSVWWWW

*»«^*^*.Wn<w»VMSVWW^wiw»wVVVWV*i»%* '■*■*-■ * mtM

—— OwWJWWai ^wMWlWWWIIWIWWlWMi * i * i * i — wnKWXHi— i i

'■■^^t>ffiV.W^AW^^^ **■ ■ ■■——^—i

^^jjjjjjj||j|iiiri^iMi||i^ '

.ViWAW-

Fig. 14: Narrow-bandpass Filter Traces of the Transverse Component Seismogram Containing the Fundamental Mode Rayleigh Wave from the Mid-Atlantic Ridge Earthquake.

32

MMHBMMMMMMMIi

. ■■»■(■I VMWIIW wniwv .

spectral estimates used for the high rssolution f-k analysis. This

occurs because the 20 second power level is much above the 40

second power on the LASA seismometer outputs. To eliminate this

problem, Capon and Evernden (1971) used a low pass Chebyshev

filter. (See Figure 8). The spectral leakage problem will probabi/

not be a problem with our high gain, longer period seismometers.

33

■ - ■ i. ^ _ *M. -- ---

Mil I« I» .'U • lll»l " ■IP1..I..I ,., .......L<. ■. UM II

3. 2 On the Ellipticity Filter

Newton (1973) (copy of appendix enclosed) tried a method of

determining the arrival azimuth based on the assumption of a 90°'

phase shift between the radial and vertical signals. The method was to

find the direction at which the phase shift was 9C . The method worked

for simulated data. However, it did not do well on real signals.

o Newton attributed the failure to an extreme sensitivity to the 90

phase shift assumption. He points out that, if the ellipticity is

changing with frequency, the "ellipticity" of a finit bandwidth signal

o will not have a 90 phase shift. He also did an experiment in which

artificial data was used. When the radial data was delayed by

one second, the phase shifts which resulted led to large errors in

the azimuth determined by the algorithm.

The difficulty Newton had in getting the azimuth for a single

Rayleigh wave may indicate that we will have difficulty in the case of

the arrival of two Rayleigh waves. If we do have difficulties, one

thing that may help is to assume that the explosion signal is coming

along the great circle direction. (This assumption is valid for most events

at 40 second periods and longer; this was discussed in a previous

section of this report) The five other quantities (complex amplitudes

of the two signals and azimuth of interfering events) could be found by

least squares. This procecu 'e would perhaps be more robust than

the solution for all 6 quantities. This approach is supported by

34

■ - -■ - ■- ■ ■ ■ -■ ■ ■■ ■ ■■--■- ■ --■ _■ .. ^ . — .,■..■. .. . ■ ...-..-...— ^~

.

Newton (p. 93); he states that the most correct estimate of the radial

power was obtained by using the great circle azimuth.

It appears that multiple reflected P and S body waves will

also cause difficulty. It may be worth while to consider the effect

of these waves on the filter.

35

. - -'-•■■- —-■■--■ -.. -- — -■'■ ——■— . - -. ■-^—J.—.-... ... - -- ■' • ■■ • ■~,-' J

j

4. 0 General Comment

I believe that there are limits to the signal - to interference

gain - we can get with a single seismometer. In the course of our

work on the single seismometer processing, I believe we should

assemble data on signal properties and the three-component coherencies

of signals. If we can, we should consider processing mrfhods for

small arrays, with perhaps 4 high gain, 3-component seismometers.

If the 40 second band pass is useful, then I am sure arrays will

eventually be used.

I think that 40 second periods may be of great use in source

estimation. This is because lateral refractions are not important

at this frequency. Thus, geometrical spreading and focusing (McGarr.

1969; McGarr, 1972; Aki et al. 1972) which is so se-ere at 20 second

period will not be as important at 40 seconds. Therefore, amplitude

estimates at 40 seconds should be more reliable.

I have seme information at hand on explosion spectra but

have not included it at this time.

36

.—: ■- -■ -— ■■■- -■■ - -- - ■■-— II .1 ■■ ^ -...,■■.. ■ — --. ■■ ■ --■■ ^*.^..^^

I Appendix

(From Newton 1973)

■V

■■ ■ ' liirimni ii I -..■*-'■- -' . .. - ■ -■■■■■-- ..■,.... ..- - -. ^ -■■■ ' - - --■■-■ ■■ - ■■■-- - ■ ■

11- Appendix E

NULL METHOD FOR DETERMINING RAYLEIGH WAVE PROPAGATION DIRECTIONS

The fundamental and higher modes of Rayleigh waves are

characterized by the elliptical particle motion where the

vertical and horizontal components have a constant phase dif-

ference of T:/2. If the motion is retrograde the radial

phase leads and for prograde motion the vertical phase leads,

and the transition from one to the other occurs at a node of

either of the components.

If w(t) denotes a vertical component seismogram then

the analytical signal is the complex representation of wCt),

defined by S(t) = w(t)-iw(t)/ where w(t) is the Hilbert

transform of w(t). w(t) is referred to as the "quadrature"

signal and v(t) as the "cophase" signal, (see Bracewell,

1965). The cophase and quadrature signals are also charac-

terized by a TT/2 phase difference, and the quadrature phase

leads.

Now consider the horizontal component seismograms h!

and h z, which are recorded by the orthogonally oriented

instruments. If e1 is the angle between the installed azimuth

for hi and the propagation direction, 9 , for a Rayleigh

wave, then the radial, u(t), and transverse, v(t), seismo-

grams are

I cose*

-sinB'

u

v L

sinO'

cosO1 Lh 2J

38

- '■- "- - ■ — ■ -■ ■ ■ - ■- MMtMrtOMMM

i .

Assuming that only Rayloigh rvjtion is present on u and w

then

u-iu _ U

w-iw

Therefore

lg|. |U| ±iTr/2 lwle ^l|l •

RcalAHliE = uw+uw = 0

fw-iwl w2+w2

which gives the null relation

uw+uw = 0.

Making the substitutions

and

u = hj cos e'+ha sin 6'

u = hi cos S'+hz sin 9'/

we can solve for 9'

6' = -tan -i (h w+h w)

i i

(hjw+h zw)

This algorithm was tested using synthetic signals, the

radial component having the same spectrum and group delays

as the vertical but a constant phase difference of TT/2, and

the transverse component was sinusoidal with slowly changing

period and with maximum amplitude approximately the same as

the other components. Except for the leading and trailing

edges of the Rayleigh wave window the computed 9' was zero.

39

L ■ ■

When the vertical component was shifted by one second with

respect to the horizontal components 0' was seldom zero and

only for a short window where the signal amplitude was the

largest did 9* usually fall within 10 degrees of zero. This

means that digitized seismograms must have nearly perfect

synchronization between componrnts, i.e., instrument group

delays must be uniform and sampling must be synchronous.

Another pitfall for this algorithm occurs when the el-

liptic! ty is changing. This effect can be seen graphically

by drawing two vectors having a given amplitude ratio and

phase separation and then drawing two vectors with phases

ir/2 greater than the first two and with a different ampli-

tude ratio. Although the TT/2 phase difference exists

between the pairs, it does not exist between the resultant

vectors. It was thought that this problem should vanish

when band filtering \s used because the ellipticities for

realistic earth structures are slowly varying with frequency.

However, the complete failure for the method when applied to

high signal-to-noise seismograms prompted a test of this

assumption.

Using the cophase and quadrature signals of a vertical

component seismogram as the vertical and radial test signal

components, it was found that uw + uw did indeed vanish.

When broad-band Gaussian-shaped filters were applied to the

two comp' .ents using slightly different center frequencies,

uw + uw did not vanish. It was concluded that slight depar-

tures from a v/2 phase difference between components, such

40

' -■- -' — - —■- - - i m iniiir—- ■■- —- - - ■■ - - - — _. .. - -...- . it,tttmtmmmitmt m i \ i

as would appear with even small amounts of noise and when

the ellipticity is frequency dependent, arc sufficient to

render useless the null method for determining Rayleigh vi

propagation directions.

41

-- - ■ -Mh^MMlMiÜMiil^imiMMfrlMh^M^^MiMtM^M^^imiiat—. ■ .i... - .. - .. - . . ■■ .^ . ■^—.-. .- .. .■- ^

i. REFERENCES

Note:

1.

2.

3.

4.

5.

6.

7.

8.

10.

11.

All these references are not used in the text but are relevant.

Aki, K., Mendihuren, J.A., and Tsai, Y. : "Reply," J. Geophys. Res., 77: 3827-3830, 1972.

Canitez, N. and Toksoz, M. N. : "Focal Mechanism at Source Depth of Earthquakes from Body and Surface Wave Data," BSSA, 61: 1369-1379, 1971.

Capon, J. : "Analysis of Rayleigh-Wave Multipath Pro- pagation at LASA," BSSA, 60: 1701-1732, 1970.

Capon, J. : "Investigation of Long-Period Noise at the Large Aperture Seismic Array," J, Geophys. Res. 74- 3182-3:94, 1969.

Capon, J. : "Analysis of Microseismic Noise at LASA NOSAR and ALPA, " Geophys. J. , 35: 39-54, 1973,

Capon, J. : "Comparison of Love and Rayleigh Wave Multipath Propagation at LASA," BSSA, 61: 1327-1344 1971.

Capon, J. and Evernden, J. F. : "Detection of Interfering Rayleigh Waves at LASA," BSSA, 61: 807-850, 1971.

Derr, J.S.: "Discrimination of Earthquakes and Explos- ions by the Rayleigh-Wave Spectral Ratio," BSSA 60- 1653-1668, 1970.

Douse, E.J. and Sorrells. G.G.: "Prediction of Pressure Generated Earth Motion Using Optimum Filters." Teledyne Inc. . Technical Report 74-6, Geotech Division, Garland, Texas, August 1, 1974.

Douglas, A. et al. : "The Analysis of Surface Wave Spectra Using a Reciprocity Theorem for Surface Waves," Geophvs J. of P. Astr. Soc, 23: 207-223, 1971.

Larson, R. J. et al. : "Correlation of Winds or Geographic Features with Production of Certain Infrasonic Signals in the Atmosphere," Geophys, J., 26: 210-214, 1971

42

MaMMtftaMMBMBBM

REFERENCES 'contd. )

12. Marshall, P. D. : "Aspects of the Spectral Differences Between Earthquakes and Underground Explosions, " Geophys. J. R. Ast. Soc., 20: 397-416, 1970.

!'• McGarr, A. : "Comments on Some Papers Concerning Amplitudes of Seismic Surface Waves, " J. Geophys. Res. , 77: 3823-3826, 1972.

14' McGarr, A.: "Amplitude Variations of Rayleigh Waves, Horizontal Refraction," Bull. Seis. Soc. Am., 59: 1307- 1334, 1969.

15- Savino, J. M. , McCamy, K. , and Hide, G. : "Structures in Earth Noise Beyond Twenty Seconds - A Window for Earthquakes," BSSA, 62: 17l-176, 1972b.

16. Savino, J. M. , Murphy, A. J. , Ryan, J.M.W-, Tatham, R. , Sykes, L. R. , Choy, G. L. , and McCamy, K. : "Results from High-Gain, Long Period Seismograph Experiments," Geophys. J. R. Astr. Soc, 31: 179-203, 1972a.

17. Sorrells, G.G., McDonald, J.A., Dev, Z.A. , and Herrin, E. : "Earth Motion Caused by Local Atmospheric Pressure Changes," Geophyc. J., 26, 1971.

18- Sorrells, G.G. and Douse, E. J. : "Studies of Long-Period Noise Generated by Atmospheric Pressure Changes," Teled/ne, Inc., Technical Report 74-5, Geotech Division, Garland, Texas, August 1, 1974.

19. Sorrells, G.G. and Goforth, T.T.: "Low Frequency Earth Motion Generated by Slow-Propagating, Partially Organized Pressure Fields," Bull. Seis. Soc. Am., 63: 1583-1601, 1973.

20' Sorrells, G.G.: "A Preliminary Investigation into the Rel- ationship between Long Period Seismic Noise and Local Fluctuations in the Atmospheric Pressure Field," Geophys. J., 26: 71-82, 1973.

21' Ziolkowski, A.: "Prediction and Suppression of Long-Period, Non-Propagating Seismic Noise," Bull. Seis. Soc. Am., 63: 937-958, 1973.

43

• mal