Spectral analysis of harmonic tremor signals at Mt. Semeru Volcano, Indonesia

June 4, 2017 | Autor: V. Schlindwein | Categoria: Multidisciplinary, Geophysical
Share Embed


Descrição do Produto

GEOPHYSICAL RESEARCH LETTERS, VOL. 22, NO. 13, PAGES 1685-1688,JULY 1, 1995

Spectral analysisof harmonic tremor signals at Mt. Semeru volcano, Indonesia Vera Schlindwein•, JoachimWassermann:and Frank Scherbaum Institutfar AllgemeineundAngewandteGeophysik,Ludwig-Maximilians-Universit•it M'tinchen,M•nchen, FRG

Abstract. A detailedspectralanalysisof tremorsignals

In thisstudy,wediscuss theresults of ananalysis of volca-

recorded at Mt. Semeru volcano, Indonesia, in October nic tremor at.Mt. Semeruvolcano,Indonesia,revealingbar-

1992 by the German-IndonesianVolcano Expedition (GIVE'92) revealsclearly harmonicspectracontaining up to 11 integerharmonics.The spectralfeaturescan be explainedby any rhythmicallypulsatingsourceproducing a temporarilystablesourcesignal.As onerealisation of this conceptwe modelthe sourceas a resonating gas volume,generatingseismicsignalssimilarto the sound generatedin a recorder.

monicspectrawith clearintegerharmonksas high as order 11. In contrastto commonlyobservedtremor signals,this observation requiresanexplanation notfor itscomplexity,but ratherfor its extraordinarilysimplespectralstructure.

attentionby seismologists. While mostauthorsagreethat the spectralcharacteristics of volcanictremor- especiallythestable narrowpeakswith centerfrequencies whichare independent of the reco•g site - are mainlydue to a sourceeffect [e.g., Aki et al., 1977; Chouet,1988; Schick, 1992], a great variety of sourcemodelshasbeenproposed.: Early studiesconsidered free oscillations of magmachambersof differentgeometriesas a possiblegeneratingmechanism of volcanic tremor [e.g., Shima, 1958; Shimozuru, 1961]. Thesemodelspredictharmonicspectraconsistingof equallyspacednarrowpeakswhich,however,are generally not observed:tremorspectratypicallyshowa singledominant peak or a non-harmonicsequenceof peaks [Schick, 1992]. Difficultiesalsoarisewhenestimatingthe dimension of the oscillatingsystemnecessaryto generatethe observed tremorperiods.In the caseof the tremorperiodsof Mt. Aso

October 1992, the German-IndonesianVolcano Expedition

magma). Recently, a number of more complex sourcemodels for

Data Analysis

Data Acquisition

Mount Semeruvolcano,the highestmountainon Java,is situatedin theeasternpartof thechainof activevolcanoson Java(Fig. 1). AlthoughMt. Semeruhasbeencontinuously Introduction activefor thepast25 years,it hasonlyrarelybeenthesubject [e.g.:Fadeli andBudi, 1992]. In The origin of volcanic tremor has attractedconsiderable of scientificinvestigations

(GIVE'92) [Hellweget al., 1994]cardedoutseismicexperimentsin orderto investigate thenatureof Mt. Semeru'sseismicity. A 3-component LennartzLE3D 1Hz seismometer was operatedin conjunctionwith a LennartzMARS88 recording systemalonga profilefromthe summitto a distanceof about.11.5 km (Fig. 2). Simultaneously, continuous recordingswere made at a fixed location(base)approximately7.5 km fromthesummitusinga broadband 3-component StreckeisenSTS2 seismometercoupled to a Geotech PDAS-100 data acquisitionsystem. In addition,the crater activity was observedduring the

campaign.The activity was dominated.by frequentsteam explosions anda varietyof ,acoustic events,amongwhicha regularpumpingsoundwasthemoststrikingfeature. The followingdataanalysisis basedmainlyon the shortbroadband volcano(4-7 s [Sassa,1935]), the dimensions of a spherical perioddataset;seismicdatafrom the continuous recording and the observation of the crater activity provide magmachamberare unreasonablylarge (2.5-4.9 km radius, useful supplementary information. assuming1.0 km/s as the acousticwave velocity of the

the seismicdata containabout40 volcanictremorhavebeenproposed, includingtheresonance Amongothersignals, of a fluid-drivencrack [Chouet, 1988], eigenvibrations of signalsidentifiedas harmonictremorsignals.A typical complexdike systems[Dahm, 1991], and unstableflow of a exampleis shownin Fig. 3. The signalsbegingradually two-phasefluid [Martinelli, 1991]. These modelsare able to

generate non-harmonic,peaked. tremor spectra without requiringgeologicallyunacceptable sourcedimensions,but noneof themsatisfactorilyexplainsall of the observedcharacteristics of volcanic tremor. -6'

NowatAWl, Bremerhaven, FRG

andInstitut ftirGeophysik, Uni Stuttgart, Stuttgart, FRG Copyright 1995by theAmerican Geophysical Union.

-9'

104'

Papernumber95GL01433 0094-8534/95/95GL-01433

INDIAN OGEAN

11•8'

11'2'

11'6'

Figure 1: Locationof Mt. SemeruvolcanooneasternJava.

$03.00 1685

1686

SCHLINDWEIN ET AL.: SPECTRAL ANALYSIS OF HARMONIC TREMORSIGN/X•S PC

$umtr•l hmghl 3•76m a s I

Crat•J• 3100

•xRVKA _•ED -:'i 26OO

OR

, I I i

i

2100

RP

Slake165 , Stnke' 30'

..•

!

S

1

2

3

4

5

6 D•stance[km]

7

8

9

10

11

N

Figure2: Sketchof a NS cross-section of Mt. Semeru (vertical exaggeration 2:1).Recording sitesonthe profilearemarked bysolidtriangles, recordings sitesofftheprofilewithopentriangles. ratherthan abruptlyand last between40 s and about1000 s.

3) In threeof thecases(e.g.,Fig. 5), systematically lower Theenvelope of thesignalscanoftenbedescribed asspindle spectralamplitudesof the evenharmonicsare observed. shaped,sometimes with a superimposed beatingpattern.The 4) Anotherdistinctobservation is thatthe frequency conenlargedsection(Fig. 4) of a tremorsignalclearlydisplaysa tentof thetremorsignalsmayvaryin time.Thepeaksshiftin repeatingwaveformwhichremainsessentially unchangedunisontowardslower or higherfrequencies.The harmonic

duringseveral Cycles ofthedominant tremor period. structureis maintained(Fig. 7). Thee•xtraordinary character ofMt.Semeru's tremor signals 5) In contrastto thesimply-structured frequency contentof

is revealedin a detailedspectralanalysis.FastFourierTrans- the tremorsignals,the spectralamplitudes of the peaks forms were calculated for successive 20 s time windows

behavein a complicated way.They vary independently of whichoverlap by 10s.Thespectral amplitudes areplotted as eachotherandirregularlyin spaceandtime. a functionof frequencyand time in Figs.5 to 7. Typical Theseresultsareconfirmed by the reference recording at

tremor spectracan be characterisedas follows:

thebasestation.The tremorspectraof the broadband record-

1) The energyof the tremorsignalsis confinedto a frequency•interval between0.5 Hz and8.0 Hz andthe spectra consistof strictlyharmonicsequences of narrowpeaks.The fundamentalfrequencyin the rangeof 0.5 Hz to 1.7 Hz is

ings exhibit the samecharacteristics.The fact that the peaks appearat the same frequenciesand that shifts of the peaks occur simultaneouslyat both recordingsites, stronglysuggeststhatthe structureof the harmonictremorspectrais gen-

accompanied byasmany as11integer harmonics (Figs. 5,6).

erated at the source.

On average, 4 to 6 pronounced peakscanbe identified. The peaksappearfor all seismometer components at thesamefre- Interpretation quencies, ascanbe seenfromthespectrum in Fig. 3. 2) For close peak spacing(< 0.75 Hz) the fundamental

modeis generallymissing.Examplesare shownin Figs.5 and 6o

Following the initial observationsof harmonic tremor

spectraat Sakurajimavolcano[Kamoet al., 1977] andLangila volcano[Mori et al., 1989], the extraordinarily clearhar-

monic spectraobservedat Mt. Semerufinally require'an a)

449.0

-399.0 449.0

-399.0.

449.o ', ', ', ', ', ', ', ', ', : ', : ', : ', I ', ', I I I I I I I I I I I I

--•99., 0

I I I I I I I I I I I I I I ! I I I I I I I I I I I I I I I

0.0

0.5

1.0

1.5 TIME

2.0

2.5

b) 0.20261 0

0.51fi67

. _,• .... '

'

•_•J•,...

'

'

,.o-,.

0.0 _ ._• 0.0 1.0

2.0

'

'

3.0

[mln]

explanationfor this phenomenon. In theory,the basicfeaturesof the harmonicspectracanbe modelledby a seriesof equallyspacedspikesconvolvedwith an arbitrary sourcesignal, such as the time seriesshown in Fig. 8a with its spectrum: 1) The resultingspectrumconsistsof a fundamentalfrequency,definedby the time intervalbetweenthe spikes,and its integerharmonics.The envelopeof the spectrumis determinedby thesourcespectrum(Fig. 8b). 1011.0

•-•.,., ......

................

3.0

FREQUEHEY

6.0

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

-1053.0 1011.0

':" 0 ............... 4.0 5.0

'

L•_.....

'

7.0

, _ 8.0

, .... ,_ 9.0

[Hz]

-1053. 1011.0

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

'

-1053.0 0.0

5.0

10.0

15.0

20.0

TIME [sec] Figure3: a) Harmonic tremorsignalrecorded atAR (cf.Fig. 2). Theamplitudes aregivenin counts(1 countequals80urn/ Figure 4: Harmonictremorsignalrecordedat PC (cf. Fig.

sinthepassband oftheseismometer). b)Amplitude spectrum2). Theamplitudes aregivenin counts(1 countequals80 urn/ (FFF) of themarked30 s timewindow.Thespectral ampli- s in the passbandof the seismometer).Note the almost tudes are normalized to the maximum value of the three com-

uncha_nged waveformon the north componentbetweensec-

ponents.

onds 5 and 20.

SCHIJ2qDWEIN ET AL.' SPECTRAL ANALYSIS OF HARMONIC TREMORSIGNALS

1687

1008 972 936 9oo 864 828 792

756 72o 684 648

•70

576'

504

LU

468

•--

432 396

•'

'•

540



::::

•40

120

100

36O

324 288 252 216 180 144 108 72 36 0

i

160 150

612

::i

7O

40 30 ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: :::::::::::::::::::::: ::::::::: ::: ,

0

ß

0

I

2

3

4

5

6

7

======================================================================================= ::-:''-:'-:':"-':-:: '-- :v .-.... -:.-..,

::::

8



lb

,

FREQUENCY [HZ]

FREQUENCY[Hz]

Figure 5: Time-dependent spectrum of thelongestharmonic tremorsignalrecordedat AR (cf. Fig. 2). Due to thelongsignal duration,spectrawerecalculatedfrom 40 s time windows overlapping by 4 s. The beginningof eachtime windowis

Figure 7: Time-dependentspectrumof a tremor signal recordedat AR (cf. Fig. 2). The beginningof each 20 s time window is marked on the vertical axis. The frequencyvariationsof the spectralpeaksare obvious.The peaksshiftin unimarkedon the vertical axis. The signalshows11 harmonics. son,the harmonicstructureis maintainedthroughout.. Evenharmonicshaveclearlyreducedspectralamplitudes.

posede.g. by Shima [1958] or Shimozuru[1961]. Oscilla2) Changesin the time intervalsbetweenthe spikesresult tions of a cylindrical gas volume closed at both ends in shiftsof the spectralpeaks. representa sourcesignalwhich is intrinsicallyperiodic.Its 3) Spectralamplitudesof the evenharmonicsare systematspectrumconsistsof a fundamentalfrequency,which is ically reducedif a secondspikeserieswith lower amplitudes definedby the lengthof the cylinderand the acousticwave is shiftedby haft an interval and addedto the original spike velocity a of the gas, and its integerharmonics[Bergmann series. Consequently, any seismicsourceradiatingan arbitrarybut temporarilystablesourcesignalcanexplainthebasiccharacteAsticsof Mr. Semeru'sharmonictremor spectra.In addition, from numericaltests,deviationsof severalpercentfrom strictperiodicitycan be allowed to obtainharmonicspectra up to order 11. Furthermore,the observedtemporalchanges of the spectralamplitudescan be interpretedas time variationsof the sourcesignal. One simplephysicalrealismionof thisgenericmodelcould be the free eigenvibrationsof fluid or gas volumes as pro-

and Scl•fer, 1990].

The generationof harmonictremorsignalsat Mt. Semeru volcanocouldbe modelledby oscillationsof a gas-filledvertical dike boundedby a lava plug (which actuallyexistsin the centerof the crater) at the upperend and a magmacolumnat the lower end. The maximum length of the dike (about 500 m) is calculated from the smallest fundamentalfrequency (0.5 Itz) and an order of magnitudeestimateof the acoustic wave velocity in a hot gas (air at 400øC:a = 500 m/s [Bergmann and Schltfer,1990]). The observedshifting of spectral peakswould requirerisingor falling of the magmacolumnat a maximumvelocity of roughly 1.15 m/s.

170 100 0.0

1.0

2.0

3.0

140' :::::::::::::::::::::::::::::::::::::::::$J?:•:::::":::::::" .............. !:::i ....... i:i:::::! ......... ::::::::::::::::::::::•:i:i:•:i:i:i:!'i:!:i:i:i:i TIME

?.

LM

100

0.0

2.0

4.0

6.0

8.0

10.0

FREQUENCY

........ :..

4.0

[sec]

12.0

14.0

16.0

10.0

[Hz]

..................................................................... :: ......... :..... :**i:'..,:,,..... b)•.......... ,••.....

2

0.0

5.0

10.0 TIME

15.0

20.0

[sec]

0

0

I

2

3

4

5

6

7

8

9

10

FREQUENCY [Hz]

0.0

2.0

4.0

6.0

8.0

10.0

FREQUENCY

12.0

14.0

16.0

18.0

[Hz]

Time-dependent spectrumof a tremorsignal Figure 8: Rhythmicallypulsatingsource.a) Seismicsource signaland its amplitudespectrum. b) Periodicsequence of window is marked on the vertical axis. 9 harmonics can thesesourcesignals(cf. Fig. 4) and the resultingharmonic amplitudespectrum(cf. Fig. 3b). clearlybe identified.The fundamental modeis missing.

Figure 6:

recordedat AR (cf. Fig. 2). The beginningof each20 s time

1688

SCHLINDWEIN

ET AL.: SPECTRAL

ANALYSIS

OF HARMONIC

TREMORSIGNALS

Gas oscillationsas a possiblephysical realisationof a rhythmicallypulsatingsourceat Mt. Semeruaresupported by

Geowissenschaften undRohstoffe,DeutscheForschungsgemeinschaft(Germany);LembagaIlmu Pengetahuan Indonesia,Badan severalarguments: PenkajianDan PenerapanTeknologi,Volcanological Surveyof Primarily,the naturalhomogeneityof gaswouldallow an Indonesia,andGadja Mada UniversityYogyakarta(Indonesia). undisturbeddevelopmentof perfectlyharmonicoscillations. Members of GIVE'92: Kirbani Sri Brotopuspito, Fadeli In addition,the gas volume itseft would reactpassivelyto Abdulrahman,Imam Suyanto, Budi Eka Nurcahya, Arnold Brodscholl, Adi Susilo, Dieter Seidl, Wolfgang Brtistle, magmamovement.Thus,the wavefieldin the gaswouldnot IohannesSchweitzer,HorstRademacher, MargaretHellweg.

be severelydisturbedby thevolumechangeandits harmonic

spectrumwould be maintainedduringthe frequencyshifts. Gasoscillationscouldbe excitedby degassing processes of the magma itseft. Varying amplitudesof the tremor signals and a maximumsignal durationof 1008 s suggestthat the oscillationsare producedby a continuousexcitationmechanism. In analogyto the generationof soundin a recorder,we proposethe following feedbackmechanism:The pressure oscillations,onceinitiatedby a pressurepulsecausedby gas releasefrom the magma,control in turn the degassingprocess.Missing fundamentalmodesat low frequenciesin the harmonictremor spectra,equivalentto overblowingof low tonesin a recorder,are a typical phenomenonrelatedto this kind of feedbackmechanism[BergmannandSch•fer,1990]. At the upperend of the gascolumn,smallamountsof gas could be pushedperiodicallythroughthe lava plug by the pressureoscillationsand producethe pumpingsoundwhich is heardin connectionwith harmonictremorsignals. However, this simplistic model leaves severalquestions unanswered:

We are •rateful to Mamaret Hellwee. Christor>her Ster>hensand

an anonymousreviewer for their helpful commentson this manuscript.

References

Aki, K., M. Fehler and S. Das, Sourcemechanismof volcanic

tremor:Fluid-drivencrackmodelsandtheirapplication to the 1963Kilaueaeruption,I. Volcanol.Geother.Res.,2, 259-287, 1977.

Bergmann, L. andC. Sch•ifer, Lehrbuch derExperimentalphysik, BandI: Mechanik- Akustik- W•irme,10thed., 902 pp., Walter de Gruyter,Berlin, 1990.

Chouet, B., Resonance of a fluid-driven crack:Radiation propertiesandimplications for thesourceof long-period-events and harmonictremor,J. Geophys.Res.,93, 4373-4400,1988. Dahm,T, Eigenvibrations of magma-filled dike systems with complexgeometry, in R. SchickandR. Mugiono(eds.),Volcanictremorandmagmaflow, 97-114,KFA Jtilich,1991. Fadeli, A. und E.N. Budi, Eruptive seismiceventsof Mr.

Semeru: Howto investigate, 17thAnnualScientific Meeting How can seismicwavesbe generatedby the gas oscillaof theHAGI andInternational Workshop on Physical Volcations?Effective energytransmissionbetweenthe compresnology,October12-14 1992,Yogyakarta, Indonesia. sionalwavesin the gasinto the surroundingsolidis restricted Hellweg,P.,D. Seidl,K. Sri Brotopuspito andW. Brtistle,Team to the endsof the gascylinder.The lava plug andthe magma investigates activityat Mr. Semeruvolcano,lava, EOS,Transbeneaththe gasvolumemay be involvedin the generationof actions, AGU, Vol.75, No. 28,313-317,July12, 1994. theharmonicseismicwaves,whichconsiderably complicates Kamo, K., T Furuzawa andJ. Akamatsu,Somenaturesof volcaa quantitativeunderstanding of the process. nic micro-tremors at theSakurajima volcano,Bull. Volcanol. Soc.lapan,22, 41-58, 1977. How realistic is a resonatinggas volume of =500 m in Mechanismen ftirdieEntstehung dimension?How consistantis the presenceof moltenlava in Martinelli,B., Fluidinduzierte yonvulkanischen Tremor-Signalen, Dissertation, ETH Ztirich, the craterarea,as is suggested by visualobservations [Hell1991. weg et al., 1994] with an oscillatinggasbody? Mori, J., H. Pafia,C. McKee, I. Itikarai, P. Lowenstein, P. de

SaintOursandB. Talai,Seismicity associated witheruptive

Conclusions

activityat Langilavolcano,PapuaNew Guinea,J. Volcanol.

Using methodsof spectralanalysis,harmonictremor signals recorded at Mt. Semeru were charactersed.

The tremor

Geother.Res., 38,243-255, 1989.

Sassa,K., Volcanicmicrotremors and eruption-earthquakes,

Mem. Coil. Sci. Univ.Kyoto,Ser.A, 18, 255-293,1935. spectraconsistof an unusuallyclear seriesof up to 12 evenly Schick,R., Volcanictremor:seismicsignalsof (almost) spacedpeaks.Harmonic spectraof similar quality are rarely unknown origin,in P.Gasparini, R. Scarpa andK. Aki (eds.), observed.The basic structureof the tremor spectracould be IAVCEI Proceedings in Volcanology, vol.3, VolcanicSeismology,157-167,Springer-Verlag, New York, 1992. explainedtheoreticallyby a sourcewhich periodicallyradivolcanic micro-tremor atvolcano Aso, ates an arbitrarybut temporarilystablesourcesignal. As a Shima,M., Onthesecond Bull.DisasterPrevent.Inst.KyotoUniv.,22, 1-6, 1958. physicalrealisationof this conceptwe proposedthat the sigD.,Volcanic micro-seisms - discussion ontheorigin, nals are generatedby oscillationsof a gas volume, which is Shimozuru, Bull. Volcanol.Soc.Japan,5,154-162, 1961. boundedby magma at its lower end and by a lava plug at its top. While for example the excitation of gas oscillations could be explainedin terms of this model, the generationof V.Schlindwein,Alfred-Wegener-Institute for polar and marine research,Columbusstr.,27568 Bremerhaven,FRG, (e-mail: seismicwaves by gas oscillationsrepresentsa [email protected]) lem. An intensivesurveyof the crateractivity(recordinge.g. Geosound in a wide frequencyrange) combined with seismic J. Wassermann,Institutfar Allgemeineund Angewandte physik, Theresienstr. 41/IV, 80333 Mtinchen, FRG, (e-mail: array measurements(providing more detailedinformationon jowa•flaucher.geophysik.uni-muenchen.de) the radiated wave field) could give essentialhints on the F. Scherbaum,Institut far Allgemeineund AngewandteGeoapplicabilityof the proposedmodel to Mt. Semeru'sharphysik, Theresienstr.41/IV, 80333 Mtinchen,FRG, (e-mail: monictremorsignals. frank• bavaria.geophysik.uni-muenchen.de)

Acknowledgements. We thankthefollowing organisationsforReceived:August 17, 1994 their support:Kernforschungsanlage Jtilich, Bundesanstalt ftir

Accepted:March 22, 1995

Revised:January27, 1995

Lihat lebih banyak...

Comentários

Copyright © 2017 DADOSPDF Inc.