.

..

&uR. 96-2293

LosAlanmstiATi Ot4AL LABOFIAIOIIY

~C SEISIUXRAWM’ IWIONALDISTANCESEAmmuNm ANDExPLJMIadmuRcEs IN WESTERN

F. N. APQ, r. J. Bos, J. R. Kam

t4mitaring a

ml? MAY1995aim

RECEIVEDJUL191995

OSTJ

(xmpmknsive mast 13an Trmtyr

MASTER

,k,

About This ReportThis official electronic version was created by scanning the best available paper or microfiche copy of the original report at a 300 dpi resolution. Original color illustrations appear as black and white images.

For additional information or comments, contact:

Library Without Walls Project Los Alamos National Laboratory Research LibraryLos Alamos, NM 87544 Phone: (505)667-4448 E-mail: lwwp@lanl.gov

DISCLAIMER

Portions of this document may bc illegiblein electronic image products. Images areproduced from the best available originaldocument.

LA-LK- 96-2293

Synthetic Sciiogrants d Regionnl Distmccs for May t995Earthqua.b and ExplosionSoumcsin WesternChiaa

Fmd N. Appi Randy J. Bo%J~cs KKasIImLos AIamoa National UmrWM’Y

Project No. ST482ASpmsmd by U.S.lkfMIMl ofEnwY

Abstract

Wwefbnns recorded at scved regional seismic 9tations al vatyingazimuths from a#06i~ and cdquske sources in Wcrwru CMnsexhiit madmd variation * station to station. We have pmfixmedtwodimcnsional finite differmm simulations of these evcnq udngmoment tensor fames and simple crustal shudmre models. to generatesynthetic scismogmms at these locations. The syntheticSeiSmOg—sat thm locations exhibit lwhavicx that is qualitatkdy consistent withthe chuL while computstiod results at a fourthstation difl’e f- thedata. We discuss these result% the assumptions of the simulati~and the limitations of this type of mdcling in the contextof qimslseismic propugnthn. Thh report is ● cmdensation of a Loa Alamon

National Laboratory, LA-UR-96- 1600, with the same title,

Keywords:synthetic seismogramscalculatimskimuladmsearthquake ~nd explosion saurces

DtHXAMUt

-. +. .. i .“

C)bjecCive

TMs work is in direet support of U.S. and Comprehensive Test Ban Treaty (CTBT]verification goals. ‘Thcresearch is aimed at providing the teds 10 minimize the fqucncy cdfklse alerts (conventional explosion% mine cdlap~ rock bursts, or earthquakes that aremistakenly inwrpreted to be nuclear tests), yet provide confidcnec that if there were a nudcutest conducted somewhere in the worid, it would be remgnkl u being nuclear, WWeaSthere are numerous seismograms f%xn earthquakes and other conventional aour~ facalibrating many areas of the world, seismograms fkon~ nuclear explosirxw am datively ru%and from a fcw very lwali~ mgitms. ono possihlc way to obtain the nuclear data is throughthe computer generationof synthetic seismograms using realistic desaiptions of the nuclearsmrce and the more important geologic features along the path of the signnl.

This is a report on tho comparison of such synthctioally generated seisnqrams withmeasurements in a region around Lop Nor, the Chinese Nuelcar Test Site Iocatd in WesternChi~ using a rccentlydeveloped LANL finitcdiffmen~ crde named AFD [1], I%c studyincluded dedations for a I5 May 1995 Lop Nor c.xplosion and a 2 May 1995 northwestChina earthquake. The objective was to evaluate the code and identify any improvementsneeded in orderto meet mquiremenls.

Research Accomplished

Figure 1 is a gmphic with the ccmtour of Moho depth of the r@n arod Lop NWsupwimposcd on which arc surface vcrlied velocity traces induced by k two seismicsources as reeordcd at several regional stations. Clockwise from the North, these stations arc:TLY (ca, 1600 km NE of Lop Nor), ULN (W 1600 km Eh!E), NIL (ea. i600 km SW). AAK(ea. 1200 km W). and BRVK (ea. 1900 km NW), Depth to Moho is but onc of the duncntito be taken into aunt in pcrfbrrning full wavetbnn modeling in such a ragim Otherpotentially important factors are location, depth and lateral extent of sedimentary basins WI(Isurface topogmphic features, ASshown in 1+’iguro1, there is notable varhion in the recxn-dedbroadband waveforms at these vurious stations. TCIfirst order, an explosion is expected tohave more of its cncrW ●swciakd with the p-wave (or Pg-the earlier arriving part of thesignal) since the sowx is most!y sphcncal, Conversely, tic earthquake energy should beprimarily assoeia[cd with tlw s-waves (or I.g-the Ialcr arrivins signals) sinec tho energy irnreleased through a shcarimg motion Upon inspection of’ the waveforms in Figuw 1, it isappmmi that such a clear part i[ionin~ ot” energy ir IIIM only along ccttain paths, mostnotably the path to Station NIL in the lower Iefi.hand eomw of the plot, At Statiom AAKand IIRVK tho ~}artitioningis less clear but still in evidence, At Sttttions TLY and [JI,N thecxp[osion and earthquake arc practically indistinguishable from cinch other, It is quitsapparant that the signal undcrgow suhti[antial modi fica[lon along its ptith; tlds ohscrvatlnadcrnonstmtes the need for numerous ~tations and pnths for properly characterizing a givensignal as bcinf,j from fin explosion or 4wthqurAo sourc~ Uqmdly importi+nt, from thd

standpoint of CTBT verification needs. wc nd to be able to capture these same observedcliff’ in the mdeling,

A key future cobe included in the modeling is the CNstal structure along each of these paths.The stmcture infbrrnation was obtained from the Corndl Middle East / North Af”csPrqject’s online Profile maker (which atso contains Eurasian data}, using Lop Nor as thesource location. Specifically, values fkx the depth-debasement of sedimentary basins as wellas for the Moho depth wem obtained along each path. Approximate VSIUCSfbr the dasticconstants and attenuation fhctorsfix each mtial (sediment mantl~ crust) were as+-Similarly, the values forthe moment tensor cocfllcients and their time depdcna vmreprescribed. Details on the input parameters used in the calctdathms are described in ~. 2the caladations were run out to simulation times ofca, 500s,

All calculations were mn with the LANL AH) andastic finite diffcmnce cock using momenttensor sources, No detailed source physics, such as strong non-iinear mob: and surfaecspallation, was included in these calculations althoughthat is an option in thcao kinds ofcalculations. AISO,topographic efkts were ignored. The simulations were nut in a i~o-dimensional, plane-strsin (Cartesinn) geometry assuming a flat fti suifacc lt is possible torun the code in either 2-D or 3-D, but the limited capabilities of current compute platformslimit regional-sized calculaticms to 2-D. Among the implications of these assumptions aru:(1) no geornd.rical dispersion effects are incluti (2) 3-D scattering is absent, and (3) nosurface topography cffbcts are present, The main masM these calculations were run inCartesian as opposed to cylindrical gmctry is that we have not implemented the capabilityto account fix the 3-D moment tensor source into the ccx!ing for the cylindrical geontet~equations of motion (ie,, the “2-VZ-D” approach), Due to the nature d’ finite differencesimulations, and the rnemwy and spd limitation of the cmnpiders avsdahh 10 the rc~ht- the results are Iimitcd in frequency content w about onc Hz

Figures 3-S arc graphics containing the crustal sections ofcach path considered together withthe computed surface vertical velocity synthetic seismograms. The crustal structure plots, inwhich the source !ocatinn is at the left, have an expanded vcrticd scale rdativa to dtahorimntal wale, For this series of cdcultitions, the same sito (1~p Nor) was used for tlwIocmion of both the emhqurtke and explosion; in reality, the ca.rthquakc was locatedca, 400 km VJIVWof’Lop Nor, io a region whcm there is h small depression in the Molm Inthe cimulakms, the ewthquaiw was ccnmred at a depth of 33 km. and the expkxdon aX4 km, 1‘fhc veflical scaled’ the vclociti~ vmics betwcwn the calcdations and experiments, so onlyqualitative comparison%should bc made Additionally, atl vcloeitics lmvc been filtered in thapasdwtd 0,2 142.-1,()I IA.

Figure 2 shows the results along the path to AJUC which hirs a relatively smooth, nearly levelMoho and virtual absence of scdimentrq basks. Tlw.rc is qualitative agnxrncnt bctwccn thedata and simulation for the explmion insohr as tlie Lg is somewhat cnhanccd relative to thePg. !jimilariy, although somewhat less mxuratcly, the eadtquake traces exhibit the canonical“earthquake” property ofaigmificant.lyenhanced Lg relative to Pg IrIboth simulations thereis a strongly attenuated Lg coda (rela[ivc to the data), s chamctcrktic that is evident in allsimulations shown in this repcxt.

Figure 3 presents the results along the path f~ NIL which exhibits IIstrongly dipping Mohostrucom along the Tibetan suture zow as well as a pronounced sedimentary basin. AgaiRthere is approximate qualitative ap~cnt between the data and simulations. The cxpltionsynthetic shows enharrccd Pg mlativc to L& allhwgh nd as strongly pronounced as in thodata. The earthquake synthetic shows an Q enhancement relative to Pg that is morepronounced than in the data,

Figure 4 shows the results akmg the path to TLY, which has a mildly varying Moho andmincw sedimentary structure. Approximate qualitative agrccmcnt between the data andsimulations again obtains, [n this W, both eqdosion and earthquakeahibi: the carmrdcal“cmthquake” property of much cnhimccd Lg relative to Pg. In both simulations theswis tina strongly attenuatedLg @da.

Figuro 5 Sives the rcsuhs ulong [he path to ULN, which diiks only slightly in ●zimuth fromthe path to TLY (see Fig, 1) and exhibits a similar #tincture consisting of a mildly vaqingMohri md nc@igitdc scdimcntnry basin. In this CU,SO,however, there is ncdable ~@ ofappro~imate qualitative agwcment betwceu the data and simulatiorw The waveform data forboth cxphxiorrand cartbquakc exhibit an enhanced 1’4stmcltm (which is also evident in thobroudbmd data shown in Fig. l); this fcatuw is ~ reproduced in the synthetics, vA; .,exhibit characteristics similar to those shown in the TLY simulations.

Conclusions ●nd Ilecommtndntions

The u)mpu[ational results dewribcd above largely exhibit rwsonoble qualitative agnmncntwith the data. This is a aatist~ing outcome, irtdhr as only basic path material, and sourceinfmrntition were used M input, Uwther improvement in thi6 input, usin~ e.g., more rdlnedmwtal structure (stratigraphic) models and improved material pqwty irtfcmnaticrq wouldvery likely incrcasc [hc tiddity of the results 10some dqpoc Similwl y, a more rellned mesh(i.e., using a smaller r~mputational zone size) would allow for higlw frqucncy resolution inthe calculated results,

The computiitional rernults, however, also edlihil cmtain Iinlitakns of khig method I:Wwtamplc, the weak l-g coda in the simulations ia likely due to the lack of ucattaring in thesimuhttiomt Vresunutbly lhis clttrrwtcristic would be impr~weulby path mitlg 141, in~tcmlof2-D, simulations: tho cmlwncedscattering in thu 3-Dw probubl~ contlibulcs to tho Q

.

coda. A more serious dcficie+wyis shown in the comparison of the data with computationsfor the LILN path (see Hg, 5): the calculathms do not exhiiit at all the Pg structuru seen inthe data. There arc several factors that may plausibly contribute 10 this discrepancy, amongwhich are: (1] the eflkct ofobiique 3-D subsurface stmcums thatcannot be accountedfbr inthe 2-D cmsml %Iicc” simulated, (2) the crustal structure information may be inadequate -incorr~ (3) the mutcrial properties used may be inappropriate (4) near-rcceiv= godogyatior struchnm etTecrs that wwe completely absmt from the simulation.

Synthetics generated with Ihc current code. evemlimiting the oalculaths [o 2-D, can provideimproved undmstanding of processes important in shaping wavcfums rdoq specific paths.An intrinsic strength of the finite difference technique is thal it allows fill inspection of thephysical processes occurring thrcwghout the entire shws field during the aimulaxbna. Withproper qwdiflcation. the waveforms mq be useabte as crude substitute wavefms in areasfor which there is no nucleur explosion data. However, in order to capturu all of the sdicntfealures of a regional wavcfnrm, there is a very real need 10 install such improvements astopography into the cnd~ optimize *O @c tominirnizamemory requirementsand executiontim~ mbfi impkrncnt the code onto mow powerful compute platfbrmsl As time andresources permig wc intend to dcvdop the capability for using these simulations as anadmittedly imperfect substitute fbr dala along ftaths for which seismic data is eitherunavailable or unattainable (or bosh), Given such a ~pability, the outcome of simulationswould be prinmrily )imikd by infom~ation on [he geology,

References

1. J, R. Kamm, R. J, Ilos & II. M. Jones, U.wr ‘J Guidl’ I;) AFD v. /.0, Los Alamos Nationa!LabOr&UOrymQOfiLA-UK-96-853, March, I‘~6.

2. F. N, App, R. J. 130s& J, R. Kamrn. .Synthctic .Wstnogrums d rcgtond )Xwwces.fur Mqv1995 Ih’lhqtulhw WUI l’kpl(AvioJl ,fhm.’es itl Wvskvw (YIhMi, I AK AlRrnos NationalLaboratory report LA-UR+6” 1600.

Ew 110’

N

w w

Lop N(X AAK

i Free Surface 1 ., 1. I I 1. I

0 200 400 600 800 10001200Range (km;

.Fig.IrcL Crustd sWucWt betwwm Lap Nor ~d s&fio~ AN4 idKwo l-ted G+-lzm km =W of Lap Nor: eafi?quake and cxploslon .~w~vefams a? this station [right). There is zapproximate qualitative agremel’u betw?en >

h ihta ad simulation for !xh sources- Thesimulations cx.fdMta tack of Lg coda, !ikc[ydue to a tack of scattw-ing in the 2-iJUhMions.

Waveformsfiltered @ 1 Hz

Explosion.z>z t’ 1 I I.Q o 100 200 360 400 500 Datar- ‘“~ Earthqualw

f’ I 1’ 1 i 10 100 200 300 400 500, I I 1 +-

+-+’’’+$4Explosion1 r

1 I 1 i

0 100 200 300 400 500 Shulatkm

l-’+-+ ‘atihquake+ I

I i I f-

6 100 200 300 400 500Time (s)

Lop ?dO~ NIL

I I* Free Surface +

0 400 800 1200 1600Range @T,j

Figure3. Gustal stmcitxre between lAJph-orand StAorl 3!! (abme), located Cid.16(MkmSW of Lop New; eartnqwk and cxplosicmwavtfomns at W station (right). ‘h pathexh]%itsa s&o@)- dipphig Who ~ well ~%aproncmnced 5edimatir? tmsin. There isapproximate qualitative agmme.nt betweenthe data ad simulation for both sources.

dEsiEE3Exp’”s~ O @l 290 300 49500 Simulation

‘-E””’’”””0 100200300460500

Time (s)

Lop ~Of TLY

)+ Free Surface 1

() “_.._~ - . .!. . 1i:-.. .’”: : , . ...

gQ -60

0 400 000 1200 1600

1., 1, , $. .1 1, .,,1,.

+,1 1“=1’” [“0 100200300400500

0 100260360 tio 500

Waveforms

filtered @ 1 Hz

Explosion

Data

Earthquake

Range [km)

Figure -!. Custa! structure klween hp Norand station TLY (above), !octted cat.1400km $= ’’’’”s’O”

KE of bp ?ior; earthquake and explosion ~ O ih 200 300 400 500 simulation

waveform al lhk station (tight). The pathexhibits a mikliy varying Moho and m-norsedimentary basin. There is approximatequz!itatiw agrmmnt between the &ala and ‘E3El!3Ea”hq”ake0 i60 200300400500-simulation for both sourc~ with the exception Time (s)of the weak Lg ctia in the simulations.

*

J

LOP Nor llLN

o

~ -20sz -40zi!? -60

-80

1 Free Surface i

-0 100200300400500

=0 100260300 400500

Waveformsfiltered @ 1 Hz

Explosion

Data

Earthquake

o 400 800 1200 1600Range (km)

Figure 5. (MMal suuchwe betwwn Lop Norand SQtion ULN (abow), locamd ca.1600kmENE of f.+ Nor; earfbquake and etplm-inawaveform at this station (right). The path issimilar to that to station TI.Y (lig. 4), Thereis nolable disagreement between W data andsirnulaf$on for time sourcts the data [or boththe earthquake and c~plosion show Q c!is-tinctive Pg thitistacking In the comespmding

0 100200300400500& ““” ‘“ i I I ,

Explosion

Simulation

Earthquake

0 100200300400500Ilme (s)

simulations.