eeeelheenneuhe eie2 ulilllll ueeeeeeeeeeeeeoften very accurate can easily be given forpo, to, and...
TRANSCRIPT
--AISI 532 REGIONAL NETWORK- SEISMICITY OF ASIA, AND
/FREQUENCY-DEPENDENT 0(U) COLUMBIA UNIV NEW YORKD W SIMPSON ET AL 15 FEB 87 AFGL-TR-87-0049
UNCLASSIFIED F19628-85--K-8022 F/G B/ti L
_llllllliollllm
NEEEEEOLINL EIE"2uEEEEEEEEEEEEEEEEElhEENnEuhEUlilllll
ij, | 2
11 1.2 1 16
OTIC FILE COPYAFGL-TR-87-0049
AD-A181 532Regional Network: Seismicity of Asia:and Frequency-Dependent Q
David W. SimpsonPaul G. RichardsArthur Lerner-Lam
DTICTrustees of Columbia University ! "LECTEin the City of New York m9Box 20, Low Memorial LibraryNew York, NY 10027 D
15 February 1987
Final ReportI February 1985-30 September 1986
APPROVED FOR PUBLIC RELEASE; DISTRIBUTION UNLIMITED
AIR FORCE GEOPHYSICS LABORATORYAIR FORCE SYSTEMS COMMANDUNITED STATES AIR FORCEHANSCOM AIR FORCE BASE, MASSACHUSETTS 01731
"This technical report has been reviewed and is approved forpublication"
F E WK0WIT --ZZ- HENRY A/ OSSING /Co t/ract Manager Branch hief
FOR THE COMMANDER
DONALD H. ECKHARDTDivision Director
This report has been reviewed by the ESD Public Affairs Office(PA) and is releasable to the National Technical InformationService (NTIS).
Qualified requestors may obtain additional copies from the DefenseTechnical Information Center. All others should apply to theNational Technical Information Service.
If your address has changed, or if you wish to be removed from the
mailing list, or if the addressee is no longer employed by yourorganization, please notify AFGL/DAA, Hanscom AFB, MA 01731. Thiswill assist us in maintaining a current mailing list.
Do not return copies of this report unless contractual obligationsor notices on a specific document requires that it be returned.
UnclassifiedsEC v CL ASS" A C );~ I~ i-S Pdl.
REPORT DOCUMENTATION PAGEla REPORT SECIR-TY CLASSIFICATION Tb RESTRICTIVE MARKINGS.
Unclassified___________________________2& SECURITY CLASSIFICATION AUTHORITY 3 DISTRIBUTION /AVAILABILITY OF REPORT
Approved for Public Release2b DECLASSIFICATION DOWNGRADING SCHEDULE Distribution Unlimited
4. PERFORMING ORGANIZATION REPORT NUMBER(S) S. MONITORING ORGANIZATION REPORT NUMBER(S)
AFGL-TR-87-0049
6a NAME OF PERFORMING ORGANIZATION 16b. OFFICE SYMBOL 70 NAME OF MON7ORING ORGANIZATIONThe Trustees of Columbia Elf applicable) Air Force Geopiiysic's LahoratoryUnliv. ill the City of NY
6c ADDRESS (City, State, and ZIP Code) 7b ADDRESS (City; State, and ZIP Code)Box 20, Low Memorial Library Hanscom Air Force BaseNew York, NY 10027 Massachusetts 01731-5000
8a. NAME OF FUNDING /SPONSORING T B OFFICE SYMBOL 9 PROCUREMENT INSTRUMENT IDENTIFICATION NUMBERORGANIZATION (if applicable)
Air Force Systems Command ________ F 19628-85-K-00228c ADDRESS (City, State, and ZIP Code) 10. SOURCE OF FUNDING NUMBERS
PROGRAM PROJECT TASK WORK UNITHanscom AFB, MA 01731-5000 ELEMENT NO. NO NO. ACCESSION NO
________________________________ 61101E 5Al0 DA AH11 TITLE (include Security Classification)
Regional Network: Seismicity of Asia: and Frequency-Dependent Q
12 ERONL UTORS)Simpson, David W.; Richards, Paul G.; Lerner-Lam, Arthur
13a. T'VPE OF REPORT 113b, TIME COVERED 114, DATE OF REPORT (Year.MonithDay) 1.PAGE COUNTFinal I FROMZLLa T9 /3Wfl1 6 I 1987 February 15 (s 72
16 SUPPLEMENTARY NOTATION
17. COSATI CODES IS SUBJECT TERMS (Continue on reverse of necessary and identify by block number)FIELD GROUP SUB-GROUP synthetic seismograms, regional seismicity, regional
networks
19 ABSTRACT (Continue on reverse if necessary and identify by block number)This report has three components:
1. *A method is developed for the computation of body-wave synthetics which is suitableto a medium having smooth variation with depth and anelasticity having arbitrary variationwith frequency. The method is applied to an anelastic earth model where Q is assumedto follow a power-law frequency dependence. The outcome is a pulse shape with propertiesquite different from what might be expected if the familiar constant-Q pulse shapes aresurveyed.
2. The seismicity of Soviet Central Asia is studied using Soviet catalogs, and the rela-tionship between topography and seismicity is analyzed using digital terrain data. Theseismicity is concentrated alonig the major top~ographic gradients dividing the distinctphysiographic/tectonic provinces of the region, with relatively low activity in bothhigh plateaus and low basins and 'higher activity along the flanks of mountainous areas.Further, there is a distinct relationiship between seismic activity and active ... .OVER ...
20 DISTRIBUTION /AVAILABILITY OF ABSTRACT 21 ABSTRACT SECURITY CLASSIFICATIONOUNCLASSIFIEOJNLIMITED 0 SAME AS RPT. 0 DTIC USERS Unclassified
22a NAME OF RESPONSIBLE INDIVIDUAL 22b TELEPHONE (include AeCo) 2c. OFFICE SYMBOL-James Lewkowicz ( 7 7-08AFGL/LWI,
DD FORM 1473,84 MAR §3 APR edition may be used until exhausted SECURITY CLASSIFICATION OF THIS PAGEAll other editions are obsolete. U(.-asfe
... CONT...
.;"salt diapirism.
3.- The capabilities of the New York State Spismic Array for studying teleseismic wave-forms and sub-array structure are evaluatedJ4-The array is positioned uniquely with respectto major seismogenic zones to record core phases from earthquakes with magnitudes greaterthan 6.0. Only minor modification of the triggering algorithm is required to lower themagnitude threshold to 5.5. Estimates of inter-station coherence are also obtained,but interpretation of the results in terms of the distribution of crustal and upper-mantlescatterers is difficult due to uncertainties in instrument response. Nevetheless, itappears as if vertically-incident signals are reasonably coherent to distances greaterthan 50 to 60 km, somewhat larger than observed at NORSAR and LASA in the early seventies.
Accesion For
NTrS CRA&IDTIC TAB 0Unannounced QJustification
By ............................ - .....=" Di-t, ib~itio i
Availability Codes
Dist Avail arid I or
CALCULATION OF BODY-WAVE PULSE SHAPES. WITH ALLOWANCE FOR
FREQUENCY-DEPENDENT Q
Paul G. Richards
INTRODUCTION
The computation of body-wave synthetics, in elastic Earth modelsthat vary purely with depth, is now routinely done by several differentmethods. In practice, the resulting synthetics either look the same, ordiffer for reasons that are quite well understood and which indicate whichmethods are fallible in certain situations (Helmberger and Burdick, 1979;Aki and Richards, 1980; Kennett, 1983; Richards, 1985; Chapman andOrcutt, 1985).
But, for computation of pulse shapes in anelastic models, there is aless satisfactory experience, in part because only a limited number ofplausible types of attenuation have been explored to the extent ofincorporation into packages for the computation of synthetic seismograms.The most common type of attenuation incorporated into synthetics is thatcharacterized by constant Q (i.e. a Q that is frequency-independent, butwhich is allowed to vary with depth), which is handled by methodssuggested by Carpenter (1967). But seismic data indicate that in manypractical situations the assumption of a constant Q is demonstrably false.Unfortunately, once the false assumption is dropped, it is not yet clearwhat should replace it.
Our recent work in this area has been based on a method of computingsynthetics that is effective in elastic media for situations wheregeometrical ray theory is accurate. These include the interpretation ofbody-wave waveform data for earthquakes and explosions at distancesfrom about 300 to 850. We first describe this method, which would hardlybe needed (since there are so many other effective methods for evaluatingthe predictions of ray theory in elastic media that smoothly vary withdepth), were it not for the generalization it allows in the case of anelasticmedia - generalizations that other methods usually cannot supply. Thenwe show an application of the method to an anelastic Earth model where Qis assumed to follow a power law frequency dependence. The outcome is apulse shape with properties quite different from what might be expectedif the familiar constant-O pulse shapes are surveyed.
11
Our conclusion is, that a fundamental problem in the interpretation ofbody-wave broadband pulse shapes - namely, the appropriate choice ofattenuation model to explain both spectral content and pulse width, andhow these quantities are affected by propagation - is unresolved. But, theconventional model that uses constant Q is flawed.
RAY CALCULATIONS WITHOUT RAY TRACING
An elementary use of the relationship between travel time (T),distance (X) and intercept time ( r ) concerns fitting a ray between a fixedsource and receiver, and finding the geometrical spreading, withoutactually doing any ray tracing.
Thus, in the frequency domain, the wavefield for a generalized ray ina structure that varies solely with depth can be written
W(X0, v)= J f(p, (p) exp[ico(T(p, In) -pX(p, (p) +pX0 }] dp (1)
where T(p, a)) - pX(p, In) is the vertical integral of vertical slowness alongthe ray path,
T-pX= t f J /[/v 2 -p 2ldz, (2)
ray
in which v(z) is the velocity (complex, if there is attenuation, and alsofrequency-dependent if there is dispersion); and f( p, co) includes theradiation pattern as well as relevant reflection/transmission coefficientsfor the generalized ray of interest. There are minor differences of detailbetween plane stratified and spherically stratified media. Richards (1973)obtained examples of (1) even for rays that have turning points, ratherthan reflection from a specific interface.
In order to obtain the ray-theory approximation for an array ofreceivers, consider the following six steps:
(i) One fixes o) and computes f and '(p) for a set of equally spaced realp-values, in a range of the real ray-parameter axis that is close to whereone expects complex saddle points to lie for the range of distances (i.e. X0 -
values) at which one wishes to know W. Let the increment in p here be Ap.(ii) At fixed X0, one searches through the array of Real[T - pX + pX 0 I to
see at which of the sampled points in p it has a minimum (or, a maximumfor certain kinds of ray). Label this discrete point as Pjs. The complex
2
saddle must presumably lie near this point of the real p-axis. If there is noattenuation, the saddle will lie on the real p-axis if there is a real raybetween source and receiver.
(iii) Find the values (complex, if there is attenuation) of the threeconstants To, po, and DXDP that best fit the sampled phase factor accordingto
T-pX+pXo=T O -(p-po) 2 xDXDP/2 (3)
in the vicinity of Pjs. Note that if one just uses Pjs itself, and the pointsjust before and just after it, closed form expressions that in practice areoften very accurate can easily be given forpo, To, and DXDP. Aki andRichards (1980) introduce the function J(p) = T - pX + pX o (see their page423). In terms of this sampled function,
Jjs-I = J(Pjs-1) = To - (Pjs- Po )2 x DXDP / 2 + Ap. po "DXDP - (Ap) 2. DXDP / 2
is = J(pjs) =To -(pjs-Po) 2 xDXDPI2
Yjs+l =J(Pjs+) = To "(Pjs- Po) 2 x DXDP / 2- Ap .po "DXDP - (Ap) 2 . DXDP / 2.
Therefore, the closed form expressions for the desired constants are
DXDP = [JJs+-2Jjs+Jjs1]/( 4p)2 ,
Po = "Jjs+ 1- Jjs_ 11( 2. Ap. DXDP)
and finally
To = Jjs + (pjs'po) 2 xDXDP/2. (4)
The point of this exercise, of course, is that po is an estimate of theposition of the complex saddle; To is an estimate of the complex traveltime; and DXDP (signifying dX/dp at the saddle) is the relevant constant,complex in general, needed to evaluate geometrical spreading.
(iv) Interpolate from f(Pjs, o) and f( Pjs+l, o) to evaluate f( po, Co).(v) Claim that the saddle point approximation to W is
W(Xo, co) = f(po, w) exp(iaWTo ) x 4 [2n + (ico.DXDP)I. (5)
3
Choose the next value of X., and loop back to the second step to do variousdistances. Choose the next value of frequency o), and loop back to the firststep. This loop is necessary only if Q is frequency-dependent, and/or ifthere is allowance for body-wave dispersion. Also, for some types ofanelasticity, it may be possible to abbreviate this step and estimate moredirectly the global frequency-dependence of the saddle-pointapproximation.
(vi) Finally, one can go to the time domain:
W(X0, t) = (1/2x) J W(X0, co) exp[-i) t] da). (6)
Note that the whole effort is accomplished using real values ofray-parameter, at the time-consuming stage of tabulating f and t atdiscrete p values. Often, the dependence on frequency in (5) is sufficientlysimple that (6) can be evaluated explicitly.
The method allows easily for investigation of differentattenuation-dispersion pairs. This is handled just by insertion of someappropriate rule for evaluating v(z, o)), and then recognizing that r = r( p, 0)).For example, in a constant Q medium in which the method of Carpenter(1967) is used to handle dispersion, the real body-wave velocity ve(z) of anelastic model is replaced by
v(z, 4) = v(z) [ 1 + [ / Q(w) ] [( 1/1c ) In ( cal %)- i / 2] 1 (7)
The method is simple and rapid to execute, and has been found quiteaccurate for 1-D (that is, purely depth-dependent) problems (Richards,1985). A comparison of synthetics produced by this method and Cagniardfirst motion method is given here in Figure 1.
A series of papers by Borcherdt (most recently, Borcherdt et al.,1986) has drawn attention to the need for more careful handling ofattenuation. Specifically he has advocated working with plane waves thatmay propagate in a direction that differs from the direction of most rapidattenuation. (The "direction of propagation" is the direction of most rapidphase increase.) He has shown that such waves have properties that muchconventional analysis (i.e., that based on plane waves propagating in thedirection of maximum attenuation) cannot reproduce. Fortunately, themethod based on (1) through (5), which finds stationary values of theingrand at complex values of horizontal slowness, gets around Borcherdt's
4
valid objections to conventional analysis, yet does so for attenuatingmedia in a way that requires little change from computational experiencewith elastic -media. The subject is further discussed in Richards (1984).Incidentdlly, we note here that the whole procedure has recently beengeneralized so that it carries over to 3-D problems in layered structures,with planar interfaces that can have any strike and dip. In effect one candevelop a generalized ray theory for a medium built up from a stack ofarbitrary wedges, and the geometrical ray approximation can be computedby a method that is an extension of (1) to (7) (Richards - 1987 proposal tothe National Science Foundation).
70 mnp 2.15 mu
m /J lO.45m
'" -2.25m. - 1.07mtL
1.78mu 15s 2.35m.
0.6Mp.
1.7 0mu -2.3Omp. -I.23my.
=15 8=450 8=75°101o(t)'
-' 2.5 x 102'
dyne-cm/sec
2 3 4 5t, seconds
Fig. 1. Each pulse gives the P-wave group (P+pP+sp) at distance 800, and30° off the strike direction, from a point source 15 km deep. Rake is 900,and results are given for three different dips. These are displacements in theupward vertical direction (after a free surface correction). t* is taken as 1second, and the time dependence of the source is given at the bottom of thefigure.
The upper row uses a Cagniard first motion approximation [Langston andHelmberger, 19751. The lower row (which has the same time scale but isdisplayed with a slightly different amplitude scale) uses the complex saddlepoint spectral method reviewed in the text, with model PEM-C and a Q struc-ture that gives t" = 1 at 800.
Numbers displayed are displacement in m. The two methods, each ofwhich is basically geometrical ray theory adapted to an attenuating medium,agree to within about 10%.
_5
EXAMINATION OF BODY-WAVE PULSE SHAPES IN DIFFERENT TYPES OFMEDIA
We have used the method based on (1) through (7) to compare pulseshapes in three different Earth models. We have used a fixedsource-receiver distance, one for which geometrical ray theory is usuallydeemed adequate even for broad-band body wave pulses, so that attentioncan be focussed on the single issue of what effects result from differentchoices about the frequency-dependence of Q.
Thus, in Figure 2 is shown the P-wave pulse shape from an impulsivesource in Earth model SL8, for which Q is constant and quite low in value.Note the charateristic features of this pulse: high frequencies have beenselectively removed, and the pulse shape is broadened with an asymmetricrise and fall. The attenuation-dispersion pair is based on equation (7).
I I I
Figure 2. The P-wave pulse at 480 in Earth model SL8, for animpulsive shallow source. Constant Q, Carpenter law.Tick marks are at 1 s intervals.
6
1 1
In Figure 3 is shown the same pulse, but for Earth model AFL (i.e. fromArchambeau, Flinn and Lambert, 1966). Q is again constant, but issignificai..ly higher than the value for SL8.
Figure 3. The P-wave pulse shape at 480 in Earth model AFL,from a shallow impulsive source. Tick marks, is.
Since both AFL and SL8 are data-driven anelastic Earth models, thefirst based on relatively high frequency signals (about 2 Hz) and thesecond based on relatively low frequencies (about 0.005 Hz), it is clearlyunsatisfactory that each model is presented in terms of afrequency-independent Q structure. That is, Q is constant and has lowvalue in SL8; Q is constant but with a high value in AFL. Varioushigh-frequency phenomena displayed in seismic data are not present inSL8, and low frequency phenomena in seismic data are not present in AFL.
It is therefore appropriate to consider some kind of interpolation forQ across the frequency range whose extremes are represented by thefrequency values typical of the data bases underlying these two Earthmodels. We have done this by assuming a Q that has a power-lawdependence on frequency,
Q(COJ) = Q0 ( 0)/ M 1 -s (8)
for some choice of constant s. 00 is a constant, the value of 0 ( o) atfrequency oo .
We find from causality rules and the evaluation of Hilbert transforms
7
aim~
that the appropriate replacement for -equation (7) is now
v(z, O) =Ve(z) { 1 + [1 /2Q(co) [((Q(co)IQ/ )- 1) tan (sir/2) - ii}. (9)
Consideration of the difference in frequency ranges over which SL8and AFL were generated, and the values of the two separate but constant Qvalues in each model, leads one to a value of about 0.7 for the constant sthat appears in equation (8). Thus, Q has about a 0.3 power-lawdependence on frequency, to fit the two kinds of data base underlyingthese models.
Figure 4. The P-wave at 480 in an Earth model with power-lawQ, having approximately the Q of SL8 at frequenciesfor which the attenuation in SL8 was determined, andthe Q of AFL at frequencies for which the attenuationof AFL was determined. Impulsive shallow source.Tick marks, at is intervals.
The outcome of computing a pulse shape based on this Q( w) mightreasonably be expected to be a pulse somewhat intermediate betweenthose of Figures 2 and 3, since we have merely found a way to interpolatefor Q across a broad band of frequencies that is representative of SL8 atone end, and AFL at the other. But the actual outcome is different. It isshown in Figure 4, and one sees a pulse shape that has an even shorter risetime than does that of Figure 3 for AFL. The reason turns out not to behard to find. It is, that at frequencies even higher than those whichcharaterize the data base for AFL, equation (8) gives a higher value of 0than the constant Q value of AFL itself. Since it is the highest frequencies
8
k 4
that determine the rise time, this latter feature will, in thefrequency-dependent Q model, be reflective of the assumptions built into(8) concerning the very-high frequency dependence of Q. In this case, Q isunbounded as frequency rises, which presumably is why the rise time is soshort in Figure 4.
We therefore reach two conclusions. First, that a constant Qattenuation model is unsatisfactory . (This has long been known: such amodel does not explain observations across the observed range offrequencies.) Second, that its replacement is not obvious. Artificialfeatures, associated with attenuation laws applied outside the range offrequencies for which there is a good data base, can in the time domain bedrawn into the appearance of a pulse shape computed within the frequencyband at which data is acquired. Another example of this general problemwas given by Richards (1985) in connection with an attenuation lawappropriate for a spectrum of relaxation mechanisms.
A good research plan to resolve this issue is to see if high-qualitybroad-band pulse shape data can be acquired, for example from RSTNbody-wave data of deep earthquakes observed teleseismically. It may thenbe possible to obtain the attenuation-dispersion pair directly from thedata. We note a first attempt in this direction (Choy and Cormier, 1986).It is difficult to achieve results, because of the need to remove sourceeffects, and effects of heterogeneity of structure at depth (e.g. thedowngoing slab, which is always present as a cause of the earthquakeactivity itself). However, we are confident that an observational programwill indeed result in suggestions for the attenuation-dispersion pair thatis appropriate in computation of synthetics, and that will narrow therange of current trade-offs between source and propagation phenomena.
References.
Aki, K., and P.G. Richards, Quantitative Seismology - Theory and Methods,vol. I, W.H. Freeman and Co., San Francisco, 1980.
Archambeau, C.B., E.A. Flinn, and D.G. Lambert, Fine structure of the uppermantle, J. Geophys. Res., 74, 5825-5866, 1966.
Borcherdt, R.D., G. Glassmoyer, and L. Wennerberg, Influence of weldedboundaries in anelastic media on energy flow, andcharacteristics of P, S-I, and S-I waves: observational
9
mlp
evidence for inhomogeneous body waves in low-loss solids,J. Geophys. Res., 91, 11503-11518, 1986.
Carpenter, E.W., Teleseismic signals calculated for underground,underwater, and atmospheric explosions, Geophysics, 32,17-32, 1967.
Chapman, C.H., and J.A. Orcutt, The computation of body wave syntheticseismograms in laterally homogeneous media, Rev. Geophys.,23, 105-163, 1985.
Choy, G.L., and V.F. Cormier, Direct measurement of the Mantle attenuationoperator for Broadband P and S waveforms, J. Geophys. Res., 91,7326-7342, 1986.
Helmberger, D.V., and L.J. Burdick, Synthetic seismograms, Ann. Rev. EarthPlanet. Sci., 34, 417-442, 1979.
Kennett, B.L.N., Seismic Wave Propagation in Stratified Media, CambridgeUniversity Press, 1983.
Langston, C.A., and D.V. Helmberger, A procedure for modeling shallowdislocation sources, Geoph. J. Roy. astr. Soc., 42, 117-130,1975.
Richards, P.G., Calculation of body waves, for caustics and tunnelling incore phases, Geoph. J. Roy. astr. Soc., 35, 243-264, 1973.
Richards, P.G., On wavefronts and interfaces in anelastic media, Bulletin ofthe Seismological Society of America, 74, 2157-2165, 1984.
Richards, P.G., Seismic wave propagation effects -- development of theoryand numerical modelling, Chapter in DARPA commemorativevolume "The VELA Program: A Twenty-Five Year Review ofBasic Research", ed. A.U. Kerr, pp. 183-251, 1985.
10
1111 1 N ! 11 6 J I!I Iiii 1
SEISMICITY OF CENTRAL ASIA
D.W. Simpson
Soviet Earthquake Catalogs
The Institute of Physics of the Earth, Moscow issues an annual catalog of earthquakes in theUSSR (Zemletriaseniia v SSSR, hereafter ZSSSR). These annual volumes contain articles andcatalogs for different subregions of the USSR, based on seismic regionalization. The largest of thesecatalogs is for Central Asia, the active region along the USSR's southern border.
The Central Asia region includes the territory of the republics of Tadjikistan, Kazkakhstan,Uzbekistan, Kirghizia and Turkmenia (Figure 1). The main physiographic divisions included are theNorthern and Southern Tien Shan and Pamir Mountains, Tadjik Depression, Kyzul Kum desert andHindu Kush (Figure 2b). The area covered by the ZSSSR catalog extends beyond the borders of theUSSR into Iran, Afghanistan and China to the south. Coverage of these border regions becomes lesscomplete, however, because of the lack of stations in southern azimuths.
Comoarison of ISC and ZSSSR catalogs
The increased coverage with regional and local stations results in a Soviet catalog that has a lowerdetection threshold and higher resolution than that of teleseismic catalogs such as the InternationalSeismological Center (ISC). Figure 3 shows the annual and cumulative numbers of earthquakesreported by JSC and ZSSSR for the period 1964 - 1979 and maps derived from the two catalogs areshown in Figures 4 and 5.
The availability of a detailed catalog for Central Asia, compiled entirely from local data sourcesprovides an interesting opportunity to examine the completeness and accuracy of the ISC catalog.Unfortunately, the ZSSSR and ISC catalogs are not completely independent, since some of the Sovietdata are provided to and used by ISC. Only a subset of stations are reported to ISC however, so thatthe ZSSSR catalog can be assumed to be more accurate and, because of the increased density ofstations, it is certainly more complete.
We have used an algorithm based on nearness in space and time to identify entries for commonevents in the two catalogs. Of the 3720 events reported by ISC for 1964-1978, 3602 are found inZSSSR. Most of the events reported by ISC but not found in ZSSSR are either of low magnitude andlocated near the southern limits of the Soviet catalog or are poorly determined locations (i.e. reportedby less than 5 stations). The only well located events above magnitude 5.0 reported only by ISCwithin the area well covered by ZSSSR are events identified as peaceful nuclear explosions.
For those events that are common to both catalogs, we have determined the differences in locationparameters (dx - longitude, dy - latitude, dr - epicenter (='4(dx 2+dy2), dz - depth, dt - origin time anddm - magnitude).
Epicentral location - Map views of the differences in location( dx, dy), as a function of numberof stations used by ISC, are shown in Figure 6. The origin is the ZSSSR location for each event withpoints placed at the end of the vector connecting the ZSSSR and ISC locations. The difference inepicentral location (dr) is shown in Figure 7. For locations determined with less than 10 stations,differences of over 200 km are observed. The difference in location decreases as the number ofstations used by ISC increases (usually a direct function of the magnitude). The differences in locationare smaller in the EW direction and there is an absolute bias to the north, both most likely resultingfrom the paucity of ISC reports from stations to the north. For events for which more than 50 stations
11
are available in the ISC solution, the standard deviation in the difference between the ISC and ZSSSRlocation drops to about +12 km, with the ISC locations biased about 10 km to the ESE of those givenby ZSSSR.
Depth and origin time - Figures 8 and 9 show the differences in depth (dz) and origin time (dt) asa function of the number of stations used in the ISC location. While dr (Figure 7) shows the gradualdecrease with number of stations, both dz and dt show broad and biased (>0) distributions evenbeyond 100 stations.The bias towards late origin times and greater depths apparent in Figures 8 and 9results from the well-known trade-off between these two parameters in the earthquake locationprocess. Many of the catalogs entries report no depths or standard depths of 0 km (ZSSSR) or 33 km(ISC). Figures 10 and 11 remove these "unknown" depths and attempt to classify the remaining datain terms of events which we can be confident are actually deep or shallow and those in which one ofthe catalogs is in error. In Figure 10 we assume that those events for which both ISC and ZSSSRreport depths greater than 60 km are actually deep. In this case the distribution of dt is symmetricalabout zero. We assume that events that are assigned shallow depths by both catalogs (but not 0 km inZSSSR or 33 in ISC) are actually shallow. Figures 11 and 12 show that for these shallow events thereis a clear tendency for the ISC origin times to be later (average of 3 s), which is compensated for(Figure 12) by the ISC depths being deeper.
Magnitude - The Soviet classification of earthquake size is energy class, k, where k is the log ofenergy release in joules. Rautian (1960) gives the following relationship , based on an analysis ofdeterminations of both magnitude and energy class for regional and local earthquakes:
m = (k-4)/1.8or k =1.8m+4
Figure 13 shows the relationship between k as reported in the ZSSSR catalog and mb for thoseevents with magnitude reported by ISC. The relationship found:
k = 1.93m + 3.29is similar to that reported by Rautian.
SEISMICITY AND TOPOGRAPHY
Many of the topographic features of Central Asia can be discerned in the patterns of seismicityfrom the Soviet Catalog in Figure 2 and 5. Among the more prominent features:
a) There is intense activity along the southern front of the northern and southern TienShan. This zone, the Gissar Kokshal seismic zone, is the site of the largest earthquakes in Central Asia(Leith and Simpson, 1986a) and marks the southern edge of what can be clearly identified as part ofthe "original" Asian plate.
b) Realtively low levels of seismicity are found within the high mountainous areas. TheGissar and Kokshal Ranges of the Tien Shan are flanked on the south and north by high seismicity,but the higher elevations are much less active. Relatively low levels of activity in regions of highelevaiton are also found in the Talas Fergana Ranges and the highest section of the Pamir.
c) Relatively low levels of seismicity are found within the major intermountain basinsand depressions. Most prominent of these is the Issyk Kul Basin, where high activity along the rangesbordering the basin clearly outlines the relatively low seismicity of the basin itself. The Fergana Valleyis similar but less obvious. The Tadjk Depression and Tarim Basin have high activity on their northernedges, near the Gissar Kokshal zone, but are less active in their interiors.
The seismicity is thus concentrated along the major topographic gradients dividing the distinctphysiographic/tectonic provinces of the region, with relatively low activity in both high plateaus andlow basins and higher activity along the flanks of the mountainous areas.
12
I.i. ..
Digital Terrain Data
To study the relationship between topography and seismicity in more detail, we have obtaineddigital terrain data for Central Asia from the Defense Mapping Agency (DMA) topographic database.Topographic elevations are provided at 3" intervals (1200 points per degree or approximately 100 mspacing).
We have carried out preliminay analysis of segments of these data making extensive use of theInternational Imaging Systems (I-S) image processor at Lamont Doherty. The original data areprovided in blocks of 1 x 1' (1200 x 1200 points). While that resolution has been useful in somedetailed studies (e.g. the Kulyab salt dome study described below or as part of our work on inducedseismicity at Nurek reservoir), the volume of data is too large for synoptic studies of large areas and itis necessary to severely decimate the data. The image processing system provides a convenientenvironment for reading the data, creating and examining full resolution images, decimating intosmaller panels and creating mosaics of larger map areas. The image processor can also be used toapply different enhancement techniques to the topographic images, combine the topography with other(eg. epicenter) data and show the results in glowing color. In this report we limit ourselves to lowresolution greyscale versions of these images.
Tadiik Depression and Southern Tien Shan
An example of an image created from the terrain data is shown in Figure 14. This area includes theSouthern Tien Shan (Gissar and Zeravshan Ranges), Tadjik Depression and western Pamir. This isthe most active part of Soviet Central Asia and has been extensively studied by Soviet groups at theTadjik Institute of Seismoresistant Construction and Seismology (TISSS) in Dushanbe and theComplex Seismological Expedition (KSE) in Garm, an outpost of the Insititute of the Physics of theEarth, Moscow. It is also the site of our earlier work on induced seismicity at Nurek Reservoir(Simpson and Negmatullaev 81, Keith et al 82, Leith et a 81) and regional tectonics (Kristy andSimpson 1980, Leith and Alvarez 1985, Leith and Simpson 1986a,b). Figure 15 identifies in moredetail some of the geographic features in a subsection of Figure 14 near Nurek reservoir.
Figure 16 represents topography with intensity directly proportional to elevation. In Figure 17 afilter has been used to locally enhance the contrast within different topographic provinces. Thus therelationship between intensity and elevation is no longer linear or spatialy constant, but many of thesubtle topographic features are more clearly distinguished. In Figure 18 a simple shaded relief maphas been produced by taking the first spatial derivative of the elevation in the NW-SE direction, givingthe impression of illumination from the NW. This filter accentuates NE-SW striking features (eg.ridges in the Tadjik Depression). A similar image, with illumination from the NE (Figure 19)accentuates features perpendicular to those in Figure 18.
One of the first order features obvious in Figures 16-19 is the contrast between the topographicstyle in the northern half of the image and that in the south. The northern segment, with higher andsmoother average topography, a general EW fabric and deeply incised rivers is composed primarily ofPaleozoic granites of the Southern Tien Shan. The southern part, with pronounced, sharp ridgesoriented more NE-SW, shows the results of recent folding and thrusting of the young foreland foldand thrust belt of the Tadjik Depression (Leith and Alvarez, 1985). The seismicity of this area (Figure20) and its relationship to the major physiographic provinces is described in detail by Leith andSimpson (1986a). Three clear types of seismicity are found in the shallow earthquakes of the region(in addition to the intermediate depth activity of the Hindu Kush). The infrequent, large earthquakes(up to M=8+) are confined to the Gissar Kokshal zone, along the front of the Southern Tien Shan, butthat zone has a relatively low level of microearthquake activity. The greatest concentration ofseismicity, in terms of numbers, is within the sediments of the Tadjik Depression, but there are fewevents here with magnitude greater than 5.5. Beneath the sediments, the third type of seismicity is thatwithin the basement itself. The activity in the Depression is especially high in its northeastern end near
13
Garm, in the area of greatest convergence between the Pamir and Tien Shan. In this area, thesediments have been thrust up and out of the Depression itself, onto the Gissar Kokshal zone,producing a complex three dimensional distribution of seismicity in which all three types of activityoverlap (Leith and Simpson, 1986a).
Salt Domes and Seismicity near Kulvab. Tadlikistan
A strong concentration of shallow seismicity (up to magnitude 5) stands out in the relatively lowactivity of the southern Tadjik Depression near the town of Kulyab (Figure 20). There are also largesalt domes exposed at the surface in this same region. We have used Landsat and topographic data inconjunction with the seismicity catalog to investigate the correlation between the seismicity and the saltdomes (Leith and Simpson, 1986b).
Figure 21 shows a black and white version of the Landsat image for the Kulyab area. Areas ofhigh infrared return (vegetation) have been set black in this image so that the dark regions in valleysare irrigated farm land. The two light colored circular features are salt domes. Figure 22 shows aperspective view (from the SW) derived by creating a 3D image from the digital terrain data and addingtexture from the Landsat image. The major dome to the south (Mumin dome) is devoid of vegetationand stands 870 meters above the surrounding plain. The high topography beyond the Mumin dome isassociated with the Sari-Chashma salt diapir, a buried structure identified in geophysical surveys.Figure 23 shows the original and shaded relief versions of the topographic data with seismicitysuperimposed. Most of the cluster of seismicity in the southeastern corner of the image followed am=5.2 earthquake on April 2, 1973.
It is our conclusion (Leith and Simpson, 1986b) that the seismicity is clearly related to the saltdoming. The highest level of current activity is not located at one of the domes which has pierced thesurface, but is near the buried Sari-Chashma diapir, a buried structure which appears to be activelyrising. Because of the lack of depth control in the Soviet hypocentral locations, it is not clear whetherthe earthquakes result from increased stress within the sediments above the dome structure or arerelated to fracturing near or within the salt layer at depth. To our knowledge, this is the only reportedcase of seismicity associated with active salt doming. More detail is to be found in Leith and Simspon(1986b, included as appendix to this report).
14
IV 10
cr z
-' . '
w CD
cnE
co ~V
1 4.)
Sol~ v 4,,
lipb
, -43A
o0 C000
-O 4
CC))
0 0
LO U.
HIVHA HJd HEW1N
17
%0 -
L) ~V3 ~
4/ 4-4 *v -
xU,
x x
x0
I E-A -
xU
Jy x 'k
x x)
x x a xx IK x
le. x
x Lo
x0x0
xx0(0
x x)
18x
Figue 5 picntermapof earthquakes in ZSSSR Cataog 1964-1977
Figl3t 5 Eicentr Go
00
00
0 0
00
00
0 O00
80
0 cX.,
0
0
0, 0 0 (
0 0
00
0 00 00~ 080
0c 00 Yb 0
z* ++ + t
00
EU E
00 0c C
+ 4- -'
00 0b _ _ _ __ _ _
+ 0
0z
0 u. 0 (A I
+ 01r0'
000 04
200
(10
* '
00
+ 0u
* .Z
.4*
CC,,
(WM) )
214
I .0
Un z
UN
v0
0
00
0
(AO DSO
IIm Cu
22-
10
xx 0
x 0 X 0XX 0
XX 0
x CIS
x U') 0zG
XX .tx ~ V.
XX X Ux U-
x X XX0
x X __
AL K 0X XX xX 0 U
K XX~
K X X
X8X
X X KX
,XX
4..~ ~~~~ ~ X r K k,,, ,K %~U
0W 0 0
(DS 4p
4 x 423
'6'x 0x t
A
00-%
> x
zZzw x 0,
xi
W Mx
x LL
>U 0
x0
0 0
(xBS C)p
a-4
ilk .x'
0~ Xx
x 0VX
V4.0 cx0
XX 0
X
Xx
Xxc~
x 0
X~ XX ~m
N NI x
(xS MX
x x 25
___ ___ ___ __ ___ ___ ___ __ 00
ol o
U
a~ 0 VV). >
VfU
V)
U0
00
C~U0 0 0 oCNJ CU
oas 4264
Cfl*1A
riiA
0) 5r
It
-v0 0
27
lib
Figure 14 Topographic map of the Tadjik Depression and Southern Tien Shan created from DMAdigital terrain data. Blanks in upper left and lower right are missing data. Geographic locations areindicated for comparison with other maps. Shading is for illumination from the north.
28
A0
4 .
00
04
0vf
4. J0 .
00
0 = L.
29~
*Figure 16 Raw digital terrain data with intensity proportional to elevation.
30
50
ON,
I''M " I
.449 " .
0 .. ...........*........
C..........
.~. ... ....'c. .o.
Alk 10-
I'WW'
f: 10
A .. &
wooV.
V'YA 4 .. . ~ ~*a~v ,k4
m X "rq 2 :4
g's.' .V.A .. ~ VV.V& 4 .
-V."' **xg*~ g. J.
-F..
I
... ... ... .. ... ... ..
. . . . . . ....- .... .
..., .
.,
Figure 19 Shaded relief map with illumination from the northeast. Comparison with Figures 14 and18, especially of river valleys in the northeast segment, shows how different gradient determinations(illumination directions) enhance features at different azimuths.
3 3
-.4
O
'0
-5;;D
00
06
0~ 0
0in0 0j0 0
00
0 0 00~~w0 Q0C
O0 0 00.0
AIL Cob 0 ME aON 0
0 00 0
9%0 .--
-, ~. -, -. -r ~ C~ -% -, -" -. - -~ -, -,
I
I
6
g
I
I
Figure 22 Perspective view of the area show in Figure 21. Digital terrain data have been used tocreate a pespective image (viewed from the southwest) and the Landsat image has been registered withit to create surface texturing. The salt dome in the lower right stands more than 800 m above thesurrounding valley.
36
69030 70000,
tPUSHION( 9' 00,k.DOME j
.4" o 111 Kulyab. -SARTIS
DOME~SAYAT
,DEPOSIT I MUMIN *Sai-s ADOME Choshmo
0 10 20kmI I ~ 0dzh F? 3.
q. 34'
Figure 23 (top) Raw digital terrain data for the Kulyab area. (middle) Shaded relief map with
seismicity superimposed. (bottom) Location map (from Leith and Simpson, 1986).37
REFERENCES
Keith, C., D. W. Simpson, and 0. Soboleva, Induced seismicity and deformation style at Nurekreservoir, Tadjik, SSR, J. Gphys. Re., 87, 4609-4624, 1982.
Kristy, M., and D. W. Simpson, Seismicity changes preceding two recent Central Asianearthquakes, J. Gephys. Res., U5, 4829-4837, 1980.
Leith, W., and W. Alvarez, Structure of the Vakhsh fold-and-thrust belt, Tadjik SSR: GeologicMapping on a Landsat image base, Geol. Soc. Am. Bull., 96, 875-885, 1985.
Leith, W., and D. W. Simpson, Seismic domains within the Gissar Kokshal seismic zone, SovietCentral Asia, J. bohys. Res., 91, 689-699, 1986a.
Leith, W., and D. W. Simpson, Earthquakes related to active salt doming near Kulyab,Tadjikistan, USSR, Geophys. Res. Lett., .3, 1019-1022, 1986b.
Leith, W., D. W. Simpson, and W. Alvarez, Structure and permeability: Geologic controls oninduced seismicity at Nurek reservoir, Tadjikistan, USSR, Geology, 9, 440-444, 1981.
Rautian, T. G., Earthquake energy (in Russian), in Methods of Detailed Investigations ofSeismicity U. V. Riznishenko, ed., Academy of Sciences USSR, Moscow, pp. 75-113,1960.
Simpson, D. W., and S. K. Negmatullaev, Induced seismicity at Nurek reservoir, Tadjikisan,USSR, Bull. Seismol. Soc. Am., 71, 1561-1586, 1981.
38
1 ' f r . m . ° . , a . " " ' o " ' b ' "" " % " - ° l " . - , " , % " d " , • "
GEOPHYSICAL RESEARCH LETTERS, VOL. 13, NO. 10, PAGES 1019-1022, OCTOBER 1986
EARTHQUAKES RELATED TO ACTIVE SALT DOMINGNEAR KULYAB, TADJIKISTAN, USSR
William Leith and David W. Simpson
Tle Iainont-I)oherty (eological Observatory of Columbia i University
Abstract. In the region near Kulyab, Tadjikistan, ity constructed. Also, as potential sites for hazardoushundreds of shallow earthquakes with magnitudes waste or other liquid storage, they may affect thegreater than 2Y have been reported in Soviet yearly short-term permeability of the cavity walls.catalogs since 1961. This area appears as a well- There is an extensive literature on salt diapirismdefined cluster of activity, distant from the line of [e.g., Lerche, 19861. Salt domes are produced by theepicenters that defines the Gissar-Kokshal seismic post-depositional flow of bedded salt upward throughzone. to the north of the Pamir ranges. The geology the overlying rock formations, because of the gravityof this region is dominated by the presence of (density) instability inherent in buried salt. It is gen-numerous salt domes, surrounded by Neogene and erally considered necessary for salt dome regions toQuaternary continental deposits. The spatial relation- have undergone some degree of tectonism (associatedship between these earthquakes and the salt domes with sedimentary loading) in order for diapirs tostuggests that the two phenomena may be related, develop. Therefore, the process of salt dome forma-.Moderate earthquakes (M > 5) occurred in 1972 and tion is unstable and episodic and the rates of salt197:1, and intensities of surface shaking greater than dome uplift are highly variable, depending on bothMM=6 were reported from earthquakes in 1937, 1952, the tectonics of the region and the state of maturity1969. 1972, 1973 and 1978. The earthquake on 2 April of the diapir [Jackson and Talbot, 1986].1973 and its aftershocks were located in a region Salt is an extremely weak rock, and deforms due-where no salt domes have been mapped at the sur- tilely at geological strain rates [Carter and Hansen,face. However, a buried salt diapir has been mapped 19831. Salt areas are generally not associated withat depth by geophysical means. These earthquakes earthquakes, and to date there is no literature associ-may result from active salt diapirism at depth. The ating earthquakes with active salt doming. Wheremechanism for producing this seismicity could be studies have been made of the seismicity of salt domeeither the active fracturing of the cap rock by the ris- areas in East Texas and the Gulf Coast [Dorman anding diapir, or the concentration of tectonic stresses in others, 1977; Racine and Klouda, 19791, no definitivethe thinned section above and adjacent to the diapir. correlations between salt structures and earthquakesThe salt-related earthquakes may produce lower fre- have been made. Nevertheless, sounds of salt move-quency radiation than other events of the same size. ment hdve been detected in salt dome regions in Iran
r[Kent, 19791, and rock bursts have occurred in under-Introduction ground salt mines [see Baar, 19771. Those salt dome
Salt deposits, and specifically salt domes, are areas where detailed seismicity studies have beencurrently of special interest as possible sites for made (east Texas, Louisiana, Oklahoma) are alsodecoupling of underground nuclear explosions [llan- areas where the domes are mature [Jackson and Tal-non, 19851, or for the underground storage of radioac- hot, 19861 and tectonic strain rates are low. Ourtive nuclear waste, petroleum or gas condensate study, from an area that is currently deforming at a
[C.:(hen, 1977; Borg, 1983[. The relatively structure- relatively high rate, therefore contrasts with studiesless, homogeneous nature of salt formations, coin- of the seismicity of salt dome areas in the U.S.bined with the high solubility of rock salt, lend these This letter describes the geological and seismologi-formations to the mining of the large stable cavities cal attributes of the Kulyab salt dome area, in thethat would be required to decouple a small under- southwestern Tadjik SSR. The seismicity that marksground niicloar explosion. In contrast with bedded this area stands out as a cluster of epicenters distinct'4alt deposits, salt ,hstif are ,'onsidered ideal cavity from the matin trend of the Gissar-Kokshal seismicsit,.s becaue of their larKe vertical dimensions and zone (Figure 1), 50-IlK) km north of Kulyab [I1eithrelatively pur. salt content. In terms of explosion and Simpson, 1088J.decoupling and the detection of clandestine nuclearweapons tests, earthquakes that are spatially associ- Geologyated with salt domes pose the obvious problem of The Soviet republic of Tadjikistan is marked by andiscrimination. In terms of engineering activities, active tectonics and a high rate of natural seismicity.either on the surface or within the salt dome, earth- As part of the India-Asia collision, the Jurassic andquakes re.present a haz.ard to the stability o>f any ray- younger sediments that fill the sedimentary basin of
Copyright 1986 by the American Geophysical Union. the Tadjik Depression, a relatively low region thatforms most of southwest Tadjikistan (Figure 2), have
Paper number 6L7011. been deformed by folding and thrusting in the zone0094-8276/86/006L-7011$$03.00 between the 'amir and Tien Shan ranges.
39
1020 Lecith and %impson: Salt Dome Parthquakes
AAyr398 O.
0 so
69*30 70*00'38*
68' 690 700 71' iPUSHION( fa00'*,,.DOME J 0 Klya
Fig. 1. Shallow e-arthquakes for the area of ~~Klasouthwestern Tadjikistan from 1070-1980. Seisnmicity X _SARTIS
is most intense within the frontal (northern) portions SV DOME
of the Vakhsh fold/thrust belt and along a portion of *1pEPOSIT ~ MUMIN *Sor-the lDarvaz fault. The cluster of seismicity in the ' ,~DOME Chashmalower portion of the figure (arrow) occurs near the ... ,=
salt (jolles mapped in Fig. 2, and well away from the a '0 2 0 dh
belt of epicenters that miarks the fold/thrust belt. A / 44detail of this area is shown in Fig. 3. Fig. 3. (a) Earthquake epicenters fromn th,- Soviet
Catalogs for the Kulyab region, mapped on shadedThroughout the Depression, Jurassic evaporites form digital terrain data. (b) Sketch shows salt deposits,the base of the sections exposed along thrust faults at major faults and geography. There is a close spatialthe surface, and the overall structure of the Depres- correlation between some groups of earthquakes andsion suggests that these basal evaporittes form the mapped salt domes. The cluster of earthquakes in thedecollement across which the thrusts art displaced lower right is associated with the buried Sari1feith and Alvarez, 18861. B~asin formation in mid- Chashma salt diapir (detail in Fig. 4). Location accii-Jurassic time was followed by the deposition of large racy for most events is listed as ± 10 kmn.thicknesses of evaporites. Since the end of the Juras-sic, shallow-water marine sedimentation in theDepression proceeded at a relatively slow rate, as the 19851. Adjacent to the Pamir range in the region ofbasin subsided following the rifting event [Leith, Kul yab. this was followed in about mhiddle Oligocene
time by the deposition of mnore than x ki of Ncogenie68* 70 iolasse ini a north-south trendingK trough of thick
T I anid, now, gently-folded rocks (the, Kulyab synclino-rium). rhis rapid loading of the .lurwssic-l'aleogenesection has provided the drive for the salt migration.
39 he Kulyab area, along the western limb of thePoll.- - Kulyab synclinormn. is marked by the ouitcrop of
- several large bodies of Jujrassic-Lower (retaccous salt__ [Luc hni kov, 18821. Some are obvious domnes (igure
3), whIile others iark the traces of iI ap ped h rust~ 2, faults. Several of the domeits have p'romi nenlt topio-
,,9"2PA M 1,R graphic relief. For exam ple, ait lthe Mu mu dome,
~$ . ,,/,W south of Kulyab (see Figures 3, -1), the cent ral saltTADJIK %0 SAL DOMES stock rises more than 870 rn above the surrounding
plain. This feeds an hiblate glacial salt flow aboveOEPRE SIONfolded andt faultedi kipper Tert iarv strata Sadi kov.
Fig. 2. (;eneralized crustal structure andi geology of G;eophysicAl studies and xploratory Irilling t hat
tenorthern Tadjik Depression. Salt domes are have accompanied the scarch for oil and g;Ls in the
Depression, 14(-50 krm from the So)viet-Afghan border, of salt at (i( 1 Ith [I.urh ni kov, WS92: .Nzintiov andol itIur,
wihfollows the Py-Andiu rivo-r (dash-i1ot line). I8JS2I, and havc einabled th lmnappinig of a /on(- ofDepth to haserneut. beneath teMesoizoie ai e overprossursud frusiu It lo ivst of K ulyal,zoic seiments of the ladjik D~epression is cointouredl [Kakun sftz(V and Vak htikov, 114751j. A well drilled inin 3 km intervals (thin lines-), the northern lorne in Figuire 3i ("Sarti,") encountered
40
Leith and Simpson: Salt Dome Earthquakes 1021
lion of the Vakhsh fold/thrust belt, and is, in fact,comparable to that of the reservoir-induced seismicityat Nurek [Simpson and Negmatullaev, 1981]. Inducedearthquakes are triggered by effective stress changesresulting from changes in the subsurface pore pres-sure regime [Simpson, 19861, and relatively high b-values are typical of induced seismic activity [Guptaand Rastogi, 1976]. We suggest that the high b-valuethat we have determined for the Kulyab region maybe "normal" for that area, because o the existence ofthe naturally high formation pressures noted above.The high b-values may thus be a fluid pressure effectthat distinguishes this region from other areas inwestern Tadjikistan.
The Kulyab area has produced several earthquakeswith magnitudes of about 5 in this century [Gubin,1960), including those in 1937, 1952, 1959, 1972, 1973and 1978. From these, the April 2nd, 1973 earthquake(M- 5.2) is the best studied to date. Kon'kov 119741combined data from the Kulyab seismic station withifuaerscis iuai ifdata to rioate the epicciiter of themaiul .shock, which liem 15, km NNW of thle epicenter
Fig. 4. Kpicenter Croom the April '2nd, 1973 earth- reported in ZSSSI{, at. a depth of 12 ki. It lies uear
quake and its aftershocks, mapped on a Landsat the northern edge of the buried Sari-Chashma diapir,image of the Kulyab area. The location of a mapped 10 km SE of Kulyab. The sequence is plotted withsubsurface diapir (dashed line) is from Kon'kov geologic and geographic data in Figure 4. The earth-1974). This earthquake sequence apparently quakes apparently occurred along the northern
occurred at shallow depth along the northern margin margin of the Sari-Chashma diapir, at depths from 0
of the diapir. to 10 kin. The main event (April 2, OOhO2m43s) wasincluded in a spectral study of Central Asian earth-
pressures of 1.6 and 1.8 times hydrostatic in Turonian quakes by Zapol'skiy and Loginova [1984; figure 21,which suggests that this event produced lower fre-
age strata at depths of 2642 and 2695 m, respectively. quency seismic waves than other earthquakes of com-In general, the westernmost (up-structure) fields are parable size. Low frequency microearthquakes nearoverpressured (up to twice hydrostatic) while those to Nurek reservoir may also be associated with saltthe east are not. deposits, but more work needs to be done to deter-
Seismicity mine the cause of the spectral anomaly.
DiscussionEpicenters of earthquakes in the Kulyab area from
1964-1980, from the yearly catalogs, Zemletracenniye Salt diapirism in active tectonic regions can bev SSSR (hereafter, 'ZSSSR catalogs'), are shown in associated with very high rates of deformation. It isFigure :3, along with shaded digital topography. therefore more appropriate to compare the dynamicsThere are 352 earthquakes li.Led in the catalogs of the KulyaL salt domes with those of the active salt(many duplicate locations do not appear in the fig- dome area of southern Iran [Ala, 19741, rather thanure), all with magnitudes greater than about 2.5. All the more stable region of extensive salt domes alongearthquakes in this area are shallow (< 20 kin), and the U.S. coast of the Gulf of Mexico. For example, athere is a clear association of some events with the detailed study of salt dynamics has been made for themapped salt domes. The large cluster of activity in Kuh-e-Namak intrusion in Iran ITalbot and Jarvis,the lower right corner of the figure occurred following 19841. This salt stock maintains the highest knownthe M=5.2 earthquake on 02 April 1973 (hereafter, natural rate of salt exti usion, estimated by Talbot-1973 earthquake"). This sequence is associated with and Jarvis [1984] to average 170 mm/yr (or about Ia diapir at depth, discussed below. kin/6000 yr). This represents an extremely high
(A Soviet regional catalog lists some smaller events strain rate in the salt stock, and a significant addi-for the Kulyab area (to M < 2) but is incomplete. A tional shear stress in the adjacent country rock [).map showing some smaller events was published in Davis, pers. comm., 19861.the 1979 ZSS.";I{ catalog (p. 30 (verleaf), but we can- Because of the very active nature of the tectonics ofnot specify location accuracy and salt structures are Tadjikistan, we suspect that the diapirs near Kulyabnot located on the map). reflect the same high degree of activity as the Iranian
For the region of Figure 3, the b-value is approxi- intrusions. For comparison, the Mumin dome (whichnmately 1.1-1.2 for earthquakes from 1964-1980. This is more than 7 km wide and rises more than 970 mis relativoly high in comparison with the western por- above the adjacent alluvial plain; see Fig. 4) is
41
1022 Leith and Simpson: Salt Dome Earthquakes
approximately the same size as the Kuh-e-Namak Gupta, H-. K. and B. K. Rastogi, Dams; and Earth-intrusion, and the domes are both in similar arid quakes, 229 pp., Elsevier, Amsterdam, 1976.environments. The Mumin dome must, therefore, Hannon, W. J., Seismic verification of a comprehen-maintain a rate or salt extrusion of the order of the sive test ban, .5cienee 227, 251-257, 1985.Kuh-e-Narnak, in order to maintain comparable Jackson, M. P. A. and C. J. Talbot, External shapes,topography. tinrortinately, data on the deep struc- strain rates and dynamics of salt structures, Hallture of the Sari Chashma diapir or surface strains in Geol. Soc Am 97, 305-323, 19H6.the local area of the diapir are unavailable. Thus, we Kalomazov, R. U. and M. A. Vakhtikov, Appearanceare currently not able to determine the amount or and Nature of anomalously high formation pres-character of the deformation that is associated with sures in the Kulyab megasyneline of the Tadjikthe growth of this salt body. Depression, Neftegazouaya (Geoloqiya ( eofizika I0,
Elsewhere in the Tadjik Depression, shallow seismi- 3-6, 1975.city is largely confined to the sedimentary section Kent, P. E., The emergent salt plugs of southernabove the salt-layer decollement [Leith and Simpson, Iran, J. Petrol. Geol. 2, 117-144, 1979.19861, and salt is generally thought-too weak to hold Kon'kov, A. A., The Kulyab earthquake of 2 April,the large elastic strains that would be necessary to 1973,in Zemletryacenni'i v S.' SR v 1973 (odu, editedproduce a moderate-size earthquake. Thus, we by l.V. Gorbunova and others, p. 88-93, Nauka,suspect that the 1974 earthquake sequence reflects Moscow, 1978.fracturing of the country rock above or adjacent to Leith, W., A Mid-Mesozoic extension across Centralthe Sari Chashma diapir, and not the seismic release Asia?, Nature 813, 587-570, 1985.of stored elastic strains within the salt itself. How- Leith, W. and W. Alvarez, Structure of the Vakhshever, depths and focal mechanisms of earthquakes in fold-and-thrust belt, Tadjik SSR (Reply), Bullthe Kulyab area are poorly constrained, and we must Geo! %Soc A4m. 97, 906-908, 1986.await further data before speculating on the mechan- Leith, W. and D. W. Simpson, Seismic domainsics of the derorrnation. within the Gimiar Kokxhal seisutic zonev, Soviet cen-
tral Asia, .1 Geenphys~ llrq '11, OU-600, 1 996.Ar knou'ledgrin ents. We thank L,. Sykes, 1). D~avis Lerche, I., ed., lieiie of SallI lPomie a~ond Sa'ell 'lectui-
and J. Rachlin for reviewing the manuscript, and J. ics, Academic, in press, 1988.Lewkowicz and DARPA for providing the digital ter- Luchinkov, V. S., Upper Jurassic halogen deposits ofrain data for this region. This work was supported by southeast Central Asia, Trudy Inst Geo! t eo/i:DARPA/AFGL under contract F19628-85-K-0022 and 5.15, 19-33, 1982.by USGS contract G14080001-1168. Lamont-Doherty Racine, D. and P. Klouda, Seismicity of the salt areasGeological Observatory contribution 4041. of Texas, Louisiana, Oklahoma and Kansas, Tech
Report AL-79-3, 27 pp., Teledyne Geotech, Alexan-References dria, 1979.
Sadikov, T. S., Halogenic formations of southwestAla, M. H., Salt diapirism in southern Iran, Amn. Tadjikistan, in Structure and Co)ndition of Formation
A4ssoc. Petrol Geol. Bull 58, 1758-1770, 1978. of S alt Formations, edited by A. L.. Yanshin, pp.Azimov, P. K., Ye. V. Lebzi and Yu. M. Ovchinnikov, 137-143, Nauka, Novosibirsk, 1991.
1982, Deep structure and problems of completion of Simpson, D. W. and S. K. Negniatullacv, Inducedthe sub-salt sediments of southwest Tadjikistan, seismicity at Nurek reservoir, Tadjikistan, USSR,Neftegazoi'aya Geol. t Geofizz 3, 23-27. Bull. Sets Soc Am. 71, 1561-1586, 1981.
Baar, C. A., Applied S'alt Rork Mechanics 1. The in situ Simpson, ID. W., Triggered earthquakes, AIny HrBehavior of .5alt Rocks, 294 pp., Elsevier, Amster- Earth Planet Sec 14, 21-42, 1986.dam, 1977. Talbot, C. .1. and R. 11. .Jarvis, Age, budget and
Borg, I., Peaceful nuclear explosions in Soviet gas dynamics of an active salt extrusion in Iran, Jcondensate fields, in Energy and Technol Reu , pp. Struct. Geo! 6, 521-533, 1984..304 aoskyK.K,&'C.M Oio,-37, Lawrence Livermore Nat. Lab., 1983. Zao'ky . . n .M.Ioioa eatures of
Carter, N. L. and F. D. flansen, Creep of rocksalt, strong, shallow-focused earthuquakes in TadjItkistan.Tertonophy.sics 9"', 275-333, 1983. P'hyseics of te .' ild hForlh p' '1~.305-310, N84,
Cohen, It. L... The disposal of radioactive wastes from /emIrtrytirenniye Sr , 1064- 198() Naiika. Moscow.fission renctors, Sc;i Arn m216, 21-31, 1977.
Dorman, J., G. V. Latham, J. L. Worzel and D.Duimas, Geophysical investigations of salt at the William Leith and D~avid W. Simpson, Lamont-University of Texas geophysics laboratory, in Proc. Doherty Geol. Ohs., palisades, N.Y. 1096-4.('on! 'alt Domne Utiiation and Environmental Con-.siderathons, Inst. Environ. Sci. Louis. 3t. Univ., p.333-351, 1977.
Cmibin, I. Ye., Patterns of Seismic .Actiity in Tadjikis- (Received July 3, 1986fan, 450 pp., Acad. Sci. Tadjikistan, Dushanbe, revised August 22, 19861960 accepted August 25, 1986)
Potential Uses of the New York State Seismic Array for Teleseismic and Regional
Studies
Arthur L. Lemer-LamLamont-Doherty Geological Observatory, Columbia University
1. Introduction
Apart from their utility to the study of regional seismicity, the use of regional seismicnetworks for the study of the properties of teleseismic travel times and waveforms is wellestablished. Studies of the crust and upper mantle subjacent to array installations have been done atLASA and NORSAR [e.g. Aki, 1973; Aki et al., 1977] and SCARLET [e.g. Humphreys et al.,
1984). Arrays have also been useful for the study of distal structure. Bungum and Capon [ 19741
used LASA recordings to measure apparent back azimuths of Rayleigh waves. Engdahl [19681used LASA records to identify and study PKiKP and PmKP arrivals. Husebye et al. [1976] usedNORSAR to measure the properties of PKIKP precursors, and Chang and Cleary [1978, 1981]
and Doornbos [19801 used arrays to measure the properties of PKKP. NORSAR has also beenused to study the nature of P-diffracted and its decay into the core shadow [Doornbos and Mondt,
1979]. The recently operational NARS network [Dost et al., 19841 is being used to assemble abroad range of observations of fundamental and higher-mode surface waveform characteristics andto study the properties of phases converted at upper-mantle discontinuities [Paulssen, 1985; Dost,1986]. In this report, we explore the utility of the New York State Seismic Array (NYSSA),
operated by Lamont-Doherty, for studies of regional variability of structure and the properties of
the core-mantle boundary and D".
Waveform studies using arrays offer the advantage of signal enhancement by assuming (withsome justification) that the background noise is uncorrelated at frequencies of interest. Variousstacking and beaming procedures have been designed to optimize the response of the array to theparticular signal being evaluated [e.g. Capon et al., 1967; Nolet and Panza, 1976; Husebye et al.,1976; Ingate et al., 19851. Such procedures depend critically upon the physical design of the array,that is, the disposition of the instruments. While these procedures elevate the accuracy of various
sorts of signal measurements, of somewhat more importance for global studies is the position ofthe array relative to seismogenic regions [Gee ct al., 19851. NYSSA is situated ideally with respectto the seismogenic zones of the southwestern Pacific for the study of the core phases belonging tothe PKP group. In addition to their instrinsic interest, PKP phases arrive at the array with nearlyvertical incidence, and provide a reasonable input signal with which near-receiver transfer function
can be evaluated, and small-scale regional structure can be estimated. Of importance to the accurateinterpretation of these arrivals is a measure of the coherence of the signal, and the response of thearray to incoming signals of near vertical incidence. In this report, we assess the utility of using
NYSSA for the analysis of these phases. Initial estimates of spatial coherence over a restricted
* 43
80 8 7'70720 70*
STATIONS OPERATED BY LAMONT-DOHERTY010-, 45'
/ *MSAINDV 4
ADNN)A
PNY A42
PTNWN H94'
ki A
44L D 4
portion of the array will be presented. These measurements are important to a thorough
understanding of higher-frequency regional wave propagation, a program of interest to DARPA.
Future DARPA-funded research at LDGO will be concerned with the robust estimation of spatial
coherence using NYSSA and specially installed sub-arrays, and the implications for crustal andmantle heterogneity beneath the array.
2. Network configuration
The New York State Seismic Array (NYSSA) was developed and installed byLamont-Doherty Geological Observatory (LDGO) to monitor seismicity in the northeastern United
States. The current configuration, shown in Figure 1, comprises 27 stations partitioned into four
sub-regional networks in northern New York State, the Adirondacks, southeastern New York andnorthern New Jersey, and western New York. Station sites were chosen to provide optimal array
characteristics for local and regional seismic monitoring experiments. The dotted line surrounds theregion in which there is nearly complete detection of events above magnitude 2.0. Sensors have
fundamental periods of one or two seconds; together with associated electronics, the total systemresponse is flat to acceleration between .1 and 1 Hz and flat to velocity between 1 and 10 Hz(Figure 2). System output is digitized using a 12-bit A/D with 100 Hz sampling, giving a usableresponse to 30 Hz. The pre-event memory is variable, and can be adjusted to a maximum of 68
seconds.
NYSSA Nominal Response to Displacement
101
* 100
00-
M 10.1
~ o 2
0-2
' 100000
10 -3
10.1 100 101 102
Frequency (Hz)
Figure 2: Nominal NYSSA instrument displacement response, after Lee and Stewart[1981].
The network operates in triggered mode using a short-term average - long-term average
algorithm tuned to seismicity levels in the Northeast. Detected events are transmitted over leased
45
telephone lines to LDGO using analog telemetry. Although the triggering algorithm is designed to
detect local and regional events reliably, teleseismic arrivals, particularly those having higher
frequency content, will trigger the network and be recorded. Since the network trigger must betuned to avoid saturation on small local events, the effective magnitude threshold for teleseisms at
core-study distances is greater than about mb = 6.0. Selected analog recordings have been made of
events with body-wave magnitudes as low as 5.2 - 5.5, however, with sufficient "eyeball"
signal-to-noise levels to be useful for core-phase analysis. To trigger on these events, and to avoid
saturation by small local events, the triggering algorithm must be modified to make near real-time
estimates of incoming signal azimuth and incidence angle.
3. Comparison of NYSSA and NORSAR for PKP analysis
The New York State Seismic Array (NYSSA) was established principally to monitor regional
seismicity in the northeastern United States. Teleseismic triggers typically have been used toprovide external checks on instrument polarities. NORSAR, on the other hand, was established
principally as a network with a well-defined array response to incoming teleseismic arrivals in orderto resolve questions associated with nuclear event detection and discrimination, though it hasproven very successful in addressing basic earth structure problems [e.g. Husebye et al., 1976; Aki
et al., 1977; Doornbos and Mondt, 1979].
NYSSA and NORSAR Deltas for 1985
10-
8 50 running mean. °.0
C 6-0, -0- NYSSA
-6- NORSARC 4-
0
2
0'110 120 130 140 150 160
Delta (0)
Figure 3: Histogram of expected arrivals per sensor at NYSSA (squares) and NORSAR(triangles) for Pacific Rim earthquakes with mb > 5.5. A 50 running mean has beenapplied to simulate wider aperture. Seismicity is taken from 1985 PDE catlaogs.
Figure 3 shows a comparison of expected annual arrivals (per sensor) at NYSSA and
NORSAR as a function of epicentral distance from Pacific rim events with mb greater than 5.5,
using events in the 1985 PDE catalog. The number of arrivals integrated over the distance range
110' - 1600 is about the same for both arrays, but the distributions are different. The NYSSA array
has a peak in the arrival histogram between 1170 and about 1300, with roughly double the number
46
of expected arrivals at NORSAR, while NORSAR should see about 50% more arrivals bewteen130' and 1400 than NYSSA. (Indeed, the study of DF-precursors by Husebye et al. [19761 usedevents concentrated around 131' and 136', corresponding roughly to the peaks in the NORSARhistogram in Figure 3.) Overall the NYSSA histogram is smoother, reflecting a more favorabledistribution of event geometry and a wider aperture. NORSAR, however, should have greaterdetection capabilities due to its more controlled array response [Dahle et al., 1975], a factor whichwe will investigate for NYSSA. The total azimuthal sampling is comparable between the twoarrays, though of course, the patches of the core sampled by each array are different. Our pointhere is not to focus on dissimilar aspects of NYSSA and NORSAR, but merely to demonstrate thatboth NORSAR and NYSSA are potentially capable of illuminating equivalent portions of P'branches. Examples of DF-precursor waveforms from NORSAR are published in Husebye et al.[1976]. Examples of waveforms from NYSSA are shown in the next section.
4. Initial observations of PKP
We have examined 22 teleseismic triggers from events in the southwest Pacific and Indonesiacovering an epicentral range from 1110 to 148'. The network triggers on P'DF or P'BC for theseevents, and there is enough pre-event memory in the standard configuration to obtain goodrecordings of DF-precursors or the AB - BC - DF crossovers. At these azimuths, the networkaperture is about 5'. Examples of three record sections exhibiting DF-precursors are shown inFigures 4, 5, and 6. Figure 7 shows a record section illuminating the BC - DF crossover.
Event 860115.2017, shown in Figure 4, is one of the closest events for which we haveobserved unambiguous precursors (1190 - 1240). (To our knowledge, these are among the closestobservations of DF-precursors yet collected.) DF-precursors are well-observed on stations beyond1220 (these are Adirondack sites) with impulsive onset and a ringing behavior of nearly constantamplitude up to the DF arrival. Total length of the precursive wavetrain is 11 - 12 seconds. Noiselevels at the stations vary, and the precursor is not detected easily at BING, WND, or SANY,although there is some indication of anomalous energy preceding DF by about 6 - 7 seconds atDHN (119.80). Figures 5 and 6 display events 840806.1201 and 850111.1441, respectively, each
illuminating the same portion of the P' travel-time curve (1320 - 137'). Precursive arrivals to DFare clear in Figure 5, but the wavetrain is longer (13 - 16 seconds) and more emergent than inFigure 4, with amplitudes increasing toward the DF arrival (compare, for example, WPNY inFigures 4 and 5). These events are at two different azimuths, however, and it is not yet certainwhether we are observing source effects, source site effects, or diffcrenccs in the characteristics ofD" or the CMB between the two paths. Observations of short-period P at closer distances canassist in the interpretation of this difference. We note, however, that the onset times are
qualitatively consistent with the theoretical least times given in Haddon and Cleary [ 1974, Figure61. We contrast 840806.1201 in Figure 5 with 850111.1441 in Figure 6, where precursivcarrivals, if present at all, are of much smaller amplitude relative to DF. Again, e cannot tell
4 7
CL
Z l 10 , 4h.'
IJ A Q*~~~~~~~l S4d w /'~f~i~jt
NOd - -
c'
C.
NHO P.^) 'ooU
-,
C31
(: as/wj 0k(0 0G )/a3Ljejsj() - wil
48
I., (4)
In-
ZIN34
0)0II) it
C
co CDCl
GUU)
49)
-
j I
C.)
1 41
S )
0 E.C
CY
oc z
m CD co C.
CC.)
5"50
~LZM
21 'AD
QE .
Q00
r-,
U
EO >C
L- E
all .. a.El....4.. ~ n 2... .i
Rild~c\ u~pfy~
~~11
o wl CJ -u oc E
- -- - -.- CD
I ~ s/' PeeTues1-e~
U 51
without further study whether we are observing the effects of source complexity or CMB
heterogeneity (the DF waveforms in Figure 6 are much simpler than those in Figure 5).
Figure 7 displays a record section collected just beyond the B-cusp at about 1430. The
dominant arrival is P'BC, with P'DF appearing only as a small, low-amplitude arrival beyond 145'.
The travel-time curves shown are calculated for the JB model, with a constant time offset of 2.1 s
added in order to line up P'DF with the data. The P'DF and P'BC travel times appear to be well fit by
the offset model at distances greater than 1460, but for distances closer than the BC - DF crossover,
the P'BC arrivals are much earlier (about 2 seconds) than predicted.
Figures 4 - 7 are given as examples of the data quality and the detail available when station
sites are densely distributed at appropriate epicentral distances. A more comprehensive analysis of
these data is underway at LDGO, and will be the subject of another publication.
In the discussion that follows, we use the PKPDF waveform from event 840806.1201 to
estimate signal coherences.
840806.1201
i 1.0
0.80N
X 0.6- - PALPALZ
c 0.4
o 0.2C)
0.0 4 • '-
0 2 4 6 8 10Frequency (Hz)
Figure 8: Amplitude coherence between two different sensors situaled on die same pier,
for one-minute signal duration, using the PKP-DF trigger for event 840906.1201.Cross-spectrum and auto-spectra estimates are smoothed with a .5 Hz frequency boxcar.Coherence is high out to 3 Hz, and acceptable to 5 Hz. Lower values of coherence at lowfrequencies are due to instrumental drift.
5. Coherence estimates
For a period of time, LDGO operated two sensors (Geospace IIS-10 and a Ranger, both
with natural periods of I s) on the same pier at PAL, using amplifiers and filters having
characteristics identical to the rest of the network. Figure 8 shows their normalized coherence
52
IT '!1"lr'f
(smoothed over .5 Hz spectral window) from DC to 10 Hz for the PKPDF signal produced by840806.1201. The coherence is near unity between .5 and 2.5 Hz, and maintains a level of about
.6 to 5 Hz. There is some incoherency below .5 Hz which is noticeable in the time series aslong-period drift. This is easy to spot upon examination, however, and is not a factor in our
interpretation of the waveforms. Evidently, there are no pathologic instabilities in the electronics
which might impact signal coherence below 2 Hz. In what follows, we use 2 Hz as the upper
frequency limit for our analysis.
We concentrate in this report on coherence measurements at the Adirondack and northernNew York sites (sub-nets I and II in Figure 1), using six vertical-component stations at CTR,
MSNY, BGR, WNY, PNY, and SKT (near GNF). For the most part, the stations are situated onhard-rock with some glacial till cover with low cultural and microseismic noise characteristics
[LDGO field reports]. Inter-station spacing ranges from 24.7 km to 128.8 km, and the median
spacing is 97.7 km. For these preliminary studies, we extracted one minute of PKP signalbeginning with the DF-precursor and tapered the time series with a 10% cosine. We assume thatthe instrument responses are characterized by the nominal poles and zeros given in Lee and Stewart
[19811.
840806.1201
1.0
0.8Xx0N 0.6 -0- CTR*MSNY 129
-o- CTR*BGR 106W ! CTR*WNY 75
0.4 -.- CTR*PNY 129-U- CTR*SKT 25
0t- 0.2-
0.0
0.0 0.5 1.0 1.5 2.0
Frequency (Hz)
Figure 9: Intcr-station cohcrcncc as a function of frequency bctwecn CTR and othcrstations of the Adirondack and northern New York subnets, for one-minute signalduration. Numbers to the right of the legend give inter-station spacing in kilometers.Coherence levels are lower for increased station separation as expected.
Figure 9 shows coherence estimates (.5 Hz spectral window smoothing) between CTR and
the other five stations as a function of frequency. The best coherence is exhibited for CTR*SKT,
which also have the closest inter-station spacing (between 2 and 3 signal wavelengths). Even so,
the coherence drops below .5 above .75 Hz. Not surprisingly, sub-array heterogeneity affects the
53
coherence estimate. Figure 10 displays coherence for two station pairs having nearly identicalinter-station spacing (WNY*PNY and WNY*SKT). Differences in the coherence curves areindicators of differences in station response or station site effects (noise, transmission losses,differences in structure, and so on). The curves have similar fall-offs above about .7 Hz, thoughthe WNY*SKT coherence is lower.
840806.1201
1.0-
IC 0.8x0
0.6S - WNY*PNY
8 0.4 "." WNY*SKT
Io 0.2U
0.00.0 0.5 1.0 1.5 2.0
Frequency (Hz)
Figure 10: Coherence between WNY, PNY, and SKT. WNY is 54 km from bothPNY and SKT. PNY is 109 km from SKT. Differences in coherence indicate that thedistribution of heterogeneities is not isotropic.
840806.1201
1.0-
0.8"=" .1-.3
S 0.6 -o- .4-.6• o- .7-.9
-0- 1-1.20 0.4 -W 1.3-1.5
•0- 1.6-1.8
0.2 -- 1.8-2
0.0--20 40 60 80 100 120 140
Station separation (kin)
Figure 11: Coherence as a function of inter-station spacing for each of several narrowbands. The "hole" between 50 and 60 kin is probably due to improper instrumentresponse.
Figure 1I displays coherence as a function of inter-station spacing for several narrow bands
54
in frequency. The coherence "hole" between 50 and 60 km is presumably due to problems with the
station at WNY (1-second vs. 2-second seismometer, site characteristics), though nothing in the
field installation notes indicates any special siting problems or instrumental defects. Relatively high
coherence is seen at low-frequencies (<.5 Hz) out to distances of about 60 to 80 km (3-5 crustal
wavelengths), somewhat greater than observed at LASA [Aki, 1973] or NORSAR [Bungum et al.,
1971].
While these results are very preliminary (and point to the necessity of obtaining appropriate
instrument response functions), they nevertheless show the potential usefulness of a well-situated
array for teleseismic structural studies and subjacent site structure.
6. Plans for further analysis
Future work will include the reconstruction of individual station calibration responses
(particularly phase responses) from field installation and maintenance notes. We will also develop
more robust coherence estimation techniques, and include more events in our coherence estimates.
The one-minute time windows used in our estimation include not only primary signal but some
clearly reverberatory waveforms; future analyses will examine shorter time windows and will use
polarization analysis to detect mode conversions. In addition, LDGO is installing new permanent
instrumentation and acquiring new portable instruments. These will be deployed in various
configurations for both intra-network coherence and PKP studies.
7. References
Aki, K., Scattering of P waves under the Montana LASA, J. Geophys. Res., 78, 1334-1346,
1973.
Aki, K., A. Christoffersson, and E.S. Husebye, Determination of the three-dimensional structure
of the lithosphere, J. Geophys. Res., 82, 277-296, 1977.
Bungum, H., and J. Capon, Coda pattern and multipath propagation of Rayleigh waves at
NORSAR, Phys. Earth Planet. Int., 9, 111-127, 1974.
Bungum, H., E.S. Husebye, and F. Ringdal, The NORSAR array and preliminary results of data
analysis, Geophys. J. R. Astr. Soc., 25, 115-126, 1971.
Capon, J., R.J. Greenfield, and R.J. Kolker, Multidimensional maximum likelihood processing of
a large aperture array, Proc. IEEE, 55, 192-211, 1967.
55
1111 . ...
Chang, A.C., and J.R. Cleary, Precursors to PKKP, Bull. Seism. Soc. Am., 68, 1059-1079,
1978.
Chang, A.C., and J.R. Cleary, Scattered PKKP: further evidence for scattering at a rough
core-mantle boundary, Phys. Earth Planet. Int., 24, 15-29, 1981.
Dahle, A., E.S. Husebye, K.A. Berteussen, and A. Christoffersson, Wave scattering effects and
seismic velocity measurements, in Exploitation of Seismograph Networks, 559-576,
Noordhoff, Leoden, Netherlands, 1975.
Doornbos, D.J., The effect of a rough core-mantle boundary on PKKP, Phys. Earth Planet. Int.,
21, 351-358, 1980.
Doornbos, D.J., and J.C. Mondt, P and S waves diffracted around the core and the velocity
structure at the base of the mantle, Geophys. J. R. Astr. Soc., 57, 381-395, 1979.
Dost, B., Preliminary results from higher-mode surface wave measurements in western Europe
using the NARS array, Tectonophysics, 128, 289-301, 1986.
Dost, B., A. van Wettum, and G. Nolet, The NARS Array, Geol. Mijnbouw, 63, 381-386, 1984.
Engdahl, E.R., Core phases and the earth's core, Ph.D. thesis, St. Louis University, St. Louis,
Mo., 1968.
Gee, L.S., A.L. Lerner-Lam, and T.H. Jordan, Resolving power of higher-mode waveform
inversion for Eurasian upper-mantle structure, Semiannual report, Defense Advanced Research
Projects Agency, 1985.
Haddon, R.A.W., and J.R. Cleary, Evidence for scattering of seismic PKP waves near the
mantle-core boundary, Phys. Earth Planet. Int., 8, 211-234, 1974.
-Humphreys, E., R.W. Clayton, and B.H. Hager, A tomographic image of mantle structure beneath
southern California, Geophys. Res. Lett., 11, 625-627, 1984.
Husebye, E.S., D.W. King, and R.A.W. Haddon, Precursors to PKIKP and seismic wave
scattering near the core-mantle boundary, J. Geophys. Res., 81, 1870-1882, 1976.
Ingate, S.F., E.S. Husebye, and A. Christoffersson, Regional arrays and optimum data processing
schemes, Bull. Seism. Soc. Am., 75, 1155-1177, 1985.
56
"- " ...... .... -'! , ' ' , 0"=11,"11 M11' ' ' " ¢N ', /tI , re gJ
Lee, W.H.K., and S.W. Stewart, Principles and applications of micorearthquake networks, in,Advances in Geophysics, ed. B. Saltzman, Academic Press, 1981.
Paulssen, H., Upper mantle converted waves beneath the NARS array, Geophys. Res. Lett., 12,709-712, 1985.
Nolet, G., and J. Panza, Array analysis of seismic surface waves: limits and possibilities,Pageoph., 114, 775-790, 1976.
57
DISTRIBUTION LIST
Dr. Monem Abdel-Cawad Woodward-Clyde Consultants
Rockwell International Science Center Attn: Dr. Lawrence J. Burdick
1049 Camino Dos Rios Dr. Jeff Barker
Thousand Oaks, CA 91360 P.O. Sox 93245
Pasadena, CA 91109-3245 (2 copies)
Professor Keiiti AkiCenter for Earth Sciences
University of Southern California Dr. Roy Burger
University Park 1221 Serry Rd.
Los Angeles, CA 90089-0741 Schenectady, NY 12309
Professor Shelton S. Alexander Professor Robert W. Clayton
Ceosciences Department Seismological Laboratory
403 Deike Building Division of Geological and Planetary
The Pennsylvania State University Sciences
University Park, PA 16802 California Institute of Technology
Pasadena, CA 91125Professor Charles B. Archambeau
Cooperative Institute for Research in Dr. Vernon F. CormierEnvironmental Sciences Earth Resources Laboratory
University of Colorado Department of Earth, Atmospheric and
Boulder, CO 80309 Planetary Sciences
Massachusetts Institute of Technology
42 Carleton StreetDr. Thomas C. Bache Jr. Cambridge, MA 02142
Science Applications Int'l Corp.
10210 Campus Point Drive Professor Anton M. Dainty
San Diego, CA 92121 Earth Resources Laboratory
Department of Earth, Atmospheric and
Dr. James Bulau Planetary Sciences
Rockwell International Science Center Massachusetts Institute of Technology
1049 Camino Dos Rios 42 Carleton Street
P.O. Box 1085 Cambridge, MA 02142
Thousand Oaks,'CA 91360Dr. Zoltan A. Der
Dr. Douglas R. Baumgardt Teledyne GeotechSignal Analysis and Systems Division 314 Montgomery StreetENSCO, Inc. Alexandria, VA.223i45400.Port Royal Road
Springfield, VA 22151-2388 Prof. Adam Dziewonski
Hoffman LaboratoryDr. S. Bratt Harvard University
Science Applications Int'l Corp. 20 Oxford St.
10210 Campus Point Drive Cambridge, MA 02138
San Diego, CA 92121
Professor John Ebel Professor John Ferguson
-Department of Geology & Geophysics Center for Lithospheric Studies
Boston College The University of Texas at Dallas
Chestnut Hill, MA 02167 P.O. Box 830688
Richardson, TX 75083-0688
Dr. Jeffrey W. Given Professor Charles A. LangstonSierra Geophysics Geosciences Department11255 Kirkland Way 403 Deike Building
- Kirkland, WA 98033 The Pennsylvania State UniversityUniversity Park, PA 16802
Prof. Roy Greenfield Professor Thorne LayGeosciences Department Department of Geological Sciences403 Deike Building 1006 C.C. Little BuildingThe Pennsylvania State University University of MichiganUniversity Park, PA 16802 Ann Arbor, MI 48109-1063
Dr. George R. MellmanProfessor David G. Harkrider Sierra GeophysicsSeismological Laboratory 11255 Kirkland Way-Division of Geological and Planetary Kirkland, WA 98033
SciencesCalifornia Institute of TechnologyPasadena, CA 91125 Professor Brian J. Mitchell
Department of Earth and AtmosphericProfessor Donald V. Helmberger SciencesSeismological Laboratory Saint Louis UniversityDivision of Geological and Planetary Saint Louis, MO 63156
SciencesCalifornia Institute of Technology Professor Thomas V. McEvillyPasadena, CA 91125 Seismographic Station
University of CaliforniaProfessor Eugene Herrin Berkeley, CA 94720Institute for the Study of Earth & ManGeophysical LaboratorySouthern Methodist University Dr. Keith L. McLaughlinDallas, TX 75275 Teledyne Geotech
314 Montgomery StreetProfessor Robert B. Herrmann Alexandria, VA 22314Department of Eakth and Atmospheric__ SciencesSaint Louis University Professor Otto W. NuttliSaint Louis, MO 63156 Department of Earth and Atmospheric
SciencesProfessor Lane R. Johnson Saint Louis University'Seismographic Station Saint Louis, MO 63156University of CaliforniaBerkeley, CA 94720 Professor Paul G. Richards
Lamont-Doherty Geological Observatoryof Columbia University
Professor Thomas H. Jordan Palisades, NY 10964Department of Earth, Atmospheric and
Planetary SciencesMassachusetts Institute of Technology Dr. Norton Rimer -Cambridge, MA 02139 S-Cubed
A Division of Maxwell LaboratoryDr. Alan Kafka P.O. 1620Department of Geology & Geophysics La Jolla, CA 92038-1620Boston CollegeChestnut Hill, MA 02167
I
Professor Larry J. Ruff Prof. John H. Woodhouse
Department of Geological Sciences Hoffman Laboratory1006 C.C. Little Building Harvard University
University of Michigan 20 Oxford St.Ann Arbor, MI 48109-1063 Cambridge, MA 02138
Professor Charles G. Sammis Dr. G. Blake
Center for Earth Sciences US Deptof Energy/DP 331University of Southern California Forrestal BuildingUniversity Park 1000 Independence Ave.Los Angeles, CA 90089-0741 Washington, D.C. 20585
Dr. David G. Simpson Dr. Michel Bouchon
Lamont-Doherty Geological Observatory Universite Scientifique etof Columbia University Medicale de Grenoble
Palisades, NY 10964 Laboratoire de Geophysique
Interne et TectonophysiqueI.R.I.G.M.-B.P. 68
Dr. Jeffrey L. Stevens 38402 St. Martin D'Heres
S-CUBED, Cedex FRANCEA Division of Maxwell Laboratory
P.O. Box 1620 Dr. Hilmar BungumLa Jolla, CA 92038-1620 NTNF/NORSAR
P.O. Box 51
Norwegian Council of Science,Professor Brian Stump Industry and Research, NORSARInstitute for the Study of Earth N-2007 Kjeller, NORWAY
and ManGeophysical Laboratory Dr. Alan DouglasSouthern Methodist University Ministry of DefenseDallas, TX 75275 Blacknest, Brimpton, Reading RG7-4RS
UNITED KINGDOMProfessor Ta-lhang TengCenter for Earth Sciences
University of Southern California Professor Peter HarjesUniversity Park Institute for Geophysik
Los Angeles, CA 90089-0741 Rhur University
BochumDr. R. B. Tittmann P.O. Box 102148
Rockwell International Science Center 4630 Bochum I1049 Camino Dos Rios FEDERAL REPUBLIC OF GERMANYP.O. Box 1085Thousand Oaks, CA 91360 Dr. James Hannon
Lawrence Livermore National LaboratoryProfessor M. Nafi Toksoz P.O. Box 808
Earth Resources Laboratory Livermore, CA 94550Department of Earth, Atmospheric and
Planetary SciencesMassachusetts Institute of Technology Dr. E. Husebye
42 Carleton Street NTNF/NORSARCambridge, MA 02142 P.O. Box 51
N-2007 Kjeller, NORWAYProfessor Terry C. WallaceDepartment of Geosciences
Building #11 Dr. Arthur Lerner-LamUniversity of Arizona Lamont-Doherty Geological ObservatoryTucson, AZ 85721 of Columbia University
Palisades, NY 10964
Mr. Peter Marshall Mr. Jack Raclin
Procurement Executive USGS - Geology, Rm 3C136
Ministry of Defense Mail Stop 928 National Center
Blacknest, Brimpton, Reading RG7-4RS Reston, VA 22092UNITED KINGDOM
Dr. B. Massinon Dr. Frode Ringdal
Societe Radlomana NTNF/NORSAR27, Rue Claude Bernard P.O. Bot 5175005, Paris, FRANCE N-2007 Kjeller, NORWAY
Dr. George H. RotheDr. Pierre Mechler Chief, Research DivisionSociete Radiomana Geophysics Directorate27, Rue Claude Bernard Headquarters Air Force Technical75005, Paris, FRANCE Applications Center
Patrick AFB, Florida 32925-6001Mr. Jack MurphyS-CUBED Dr. Alan S. Ryall, Jr.Reston Geophysics Office Center for Seismic Studies11800 Sunrise Valley Drive 1300 North 17th StreetSuite 1212 Suite 1450Reston, VA 22091 Arlington, VA 22209-2308
Dr. Svein Mykkeltveit Dr. Jeffrey L. Stevens
NTNF/NORSAR S-CUBED,P.O. Box 51 A Division of Maxwell Laboratory
N-2007 Kjeller, NORWAY P.O. Box 1620La Jolla, CA 92038-1620
Dr. Carl Newton
Los Alamos National Laboratory Dr. Lawrence TurnbullP.O. Box 1663 OSWR/NEDMail Stop C 335, Group ESS3 Central Intelligence AgencyLos Alamos, NM 87545 CIA, Room 5G48
Washington, DC 20505
Dr. Peter Basham Professor Steven Grand
Earth Physics Branch Department of Geology, Department of Etergy and Mines 245"Natural History Bldg
* 1 Observatory Crescent 1301 West Green StreetOttawa, Ontario Urbana, IL 61801CANADA KlA OY3
Professor J. A. Orcutt
Geological Sciences DivisionUniv. of California at San Diego
La Jolla, CA 92093
Dr. Frank F. Pilotte
Director of GeophysicsHeadquarters Air Force Technical
Applications CenterPatrick AFB, Florida 32925-6001
Professor Keith Priestley
University of NevadaMackay School of Mines
Reno, Nevada 89557
IM
DARPA/PM U.S. Geological Survey1400 Wilson Boulevard ATTN: Dr. T. HanksArlington, VA 22209 National Earthquake Research Center
345 Middlefield RoadMenlo Park, CA 94025
Defense Technical Information Center SRI InternationalCameron Station 333 Ravensworth AvenueAlexandria, VA 22314 (12 copies) Menib Park,-CA 94025
Defense Intelligence Agency Center for Seismic StudiesDirectorate for Scientific and ATTN: Dr. C. Romney
Technical Intelligence 1300 North 17th StreetWashington, D.C. 20301 Suite 1450
Arlington, VA 22209 (3 copies)
Defense Nuclear Agency Dr. Robert Blandford
Shock Physics Directorate/SS DARPA/GSDWashington, D.C. 20305 1400 Wilson Boulevard
Arlington, VA 22209-2308
Ms. Ann KerrDefense Nuclear Agency/SPSS DARPA/GSDATTN: Dr. Michael Shore 1400 Wilson Boulevard6801 Telegraph Road Arlington, VA 22209-2308Alexandria, VA 22310
Dr. Ralph Alewine IIIAFOSR/NPG DARPA/GSDATTN: Director 1400 Wilson BoulevardBldg 410, Room C222 Arlington, VA 22209-2308Bolling AFB, Washington, D.C. 20332
Mr. Edward GillerAFTAC/CA (STINFO) Pacific Sierra Research Corp.Patrick AFB, FL 32925-6001 1401 Wilson Boulevard
'Arlington, VA 22209
Science Hlorizons,. Inc.AFWL/NTESC Attn: Dr. Bernard MinsterKirtland AFB, NM 87171 Dr. Theodore Cherry
710 Encinitas Blvd., Suite 101Encinitas, CA 92024 (2 copies)
U.S. Arms Control & Disarmament Agency Dr. Jack Evernden
ATTN: Mrs. M. Hoinkes USGS - Earthquake StudiesDiv. of Multilateral Affairs, Rm 5499 345 Middlefield RoadWashington, D.C. 20451 Menlo Park, CA 94025
Dr. Lawrence BralleDepartment of GeosciencesPurdue UniversityWest Lafayette, IN 47907
Dr. L.A. Bollinger Dr. Muawia BarazangiDepartment of Geological Sciences Geological SciencesVirginia Polytechnical Institute Cornell University21044 Derring Hall Ithaca, NY 14853Blacksburg, VA 24061
Rondout AssociatesDr. L. Sykes Attn: Dr. George SuttonLamont Doherty Geological Observatory Dr. Jerry CarterColumbia University D'r. Paul PomeroyPalisades, NY 10964 P.O. Box 224
Stone Ridge, NY 12484 (3 copies)
Dr. S.W. Smith Dr. M. Sorrells
Geophysics Program Geotech/Teledyne, University of Washington P.O. Box 28277Seattle, WA 98195 Dallas, TX 75228
Dr. Bob Smith
Dr. L. Timothy Long Department of GeophysicsSchool of Geophysical Sciences University of UtahGeorgia Institute of Technology 1400 East 2nd SouthAtlanta, GA 30332 Salt Lake City, UT 84112
Dr. Anthony Gangi
Dr. N. Biswas Texas A&M UniversityGeophysical Institute Department of GeophysicsUniversity of Alaska College Station, TX 77843Fairbanks, AK 99701
Dr. Freeman Gilbert Dr. Gregory B. Young
Institute of Geophysics & ENSCO, Inc.Planetary Physics 5400 Port Royal Road
Univ. of California at San Diego Springfield, CA 22151P.O. Box 109La Jolla, CA 92037
Dr. Pradeep Talwani Dr. Ben MenaheimDepartment of Geological Sciences Weizman Institute of ScienceUniversity of Sduth Carolina Rehovot, ISRAEL 951729Columbia, SC 29208
Weidlinger Associates
University of Hawaii Attn: Dr. Gregory WojcikInstitute of Geophysics 620 Hansen Way, Suite 100Attn: Dr. Daniel Walker Palo Alto, CA 94304Honolulu, HI 96822
Dr. Leon KnopoffUniversity of California
Dr. Donald Forsyth Institute of Geophysics & PlanetaryDepartment of Geological Sciences PhysicsBrown University Los Angeles, CA 90024Providence, RI 02912
Dr. Kenneth H. OlsenLos Alamos Scientific Laboratory
Dr. Jack Oliver Post Office Box 1663Department of Geology Los Alamos, NM 87545Cornell UniversityIthaca, NY 14850
Prof. Jon F. Claerbout AR;L/XO
Prof. Amos Nur Hanscom AFB, MA 01731-5000
Dept. of GeophysicsStanford UniversityStanford, CA 94305 (2 copies)
AFGL/LW
Dr. Robert Burridge Hanscom AFB, MA 01731-5000Schlumberger-Doll Research Ctr.Old Quarry RoadRidgefield, CT 06877
AFIL/SULLResearch Library
Dr. Eduard Berg Hanscom AFB, MA 01731-5000 (2 copies)Institute of GeophysicsUniversity of HawaiiHonolulu, HI 96822 Secretary of the Air Force (SAFRD)
Washington, DC 20330
Dr. Robert Phinney
Dr. F.A. Dahlen Office of the Secretary Defense
Dept. of Geological & Geophysical Sci. DDR & E
Princeton University Washington, DC 20330Princeton, NJ 08540 (2 copies)
Dr. Kin-Yip Chun
Geophysics Division HQ DNAPhysics Department Attn: Technical Library
University of Toronto Washington, DC 20305Ontario, CANADA M5S IA7
New England Research, Inc. Director, Technical Information
Attn: Dr. Randolph Martin III DARPAP.O. Box 857 1400 Wilson Blvd.
Norw-ich, VT 05055 Arlington, VA 22209
Sandia National Laboratory Los Alamos Scientific- Laboratory
Attn: Dr. H.B. Durham Attn: Report LibraryAlbuquerque, NM' 87185 Post Office Box 1663
Los Alamos, NM 87544Dr. Gary McCartor
Mission Research Corp.735 State Street Dr. Thomas WeaverP. 0. Drawer 719 Los Alamos Scientific LaboratorySanta Barbara, CA 93102 Los Alamos, NM 87544
Dr. Al Florence
Dr. W. H. K. Lee SRI InternationalUSGS 333 Ravenswood AvenueOffice of Earthquakes, Volcanoes, Menlo Park, CA 94025-3493
& Engineering* Branch of Seismology
345 Middlefield RdMenlo Park, CA 94025
97I