38
  A Matlab GUI for use with ISOLA Fortran codes User’s Guide by Efthimios Sokos 1)  and Jiri Zahradnik 2)  1) University of Patras, Seismological Laboratory 2) Charles University in Prague, Faculty of  Mathematics and Physics [email protected] ,  [email protected] Ver. 4.0 February 2009 

A Matlab GUI for Use With ISOLA Fortran Codes

Embed Size (px)

Citation preview

  • AMatlabGUIforusewithISOLAFortrancodes

    UsersGuide

    byEfthimiosSokos1)andJiriZahradnik2)

    1) UniversityofPatras,SeismologicalLaboratory

    2) CharlesUniversityinPrague,FacultyofMathematicsandPhysics

    [email protected],[email protected]

    Ver.4.0

    February2009

  • Contents

    Whatisnewinver.4.0?........................................................3

    GUIPurpose...........................................................................4

    Requirements.........................................................................4

    Installation.............................................................................5

    Method..................................................................................5

    RunningtheGUI...................................................................11

    CrustalModel.................................................................................12

    EventInfo.......................................................................................13

    StationSelection.............................................................................14

    RawDataPreparation.....................................................................15

    TrialSourceDefinition....................................................................18

    GreenFunctionComputation..........................................................21

    Inversion.........................................................................................22

    Plotting...........................................................................................25

    UtilitiesIandII...............................................................................32

    Acknowledgments................................................................34

    AppendixIHowtoprepareapoleandzerofileforISOLA...35

    AppendixII:Filesandfolders................................................38

  • Whatisnewinver.4.0?Thenew versionof ISOLAGUIhas somemajor changes/additions thatmaketheuseofthesoftwarealoteasier.Thefollowingarenew:

    Theinstallationprocessisdifferent,ISOLAGUIrunsfromacertainfolderand ISOLA*.exefileshavetobeadded inthesystempathwhileISOLAGUIfileshavetobeaddedinMatlabpath

    ISOLAGUI isperforminga checkwhen it is called fromwithinafolder and creates the folder structure or copies required filesautomatically ifneeded (e.g.whenwe startprocessingofaneweventandfiles,foldersdontexist)

    Fileswithstationcoordinates,crustalmodelorpolesandzeroescan be stored in the installation folder and ISOLAGUIwill copythemtospecificrunfolder,automatically

    A toolhasbeenadded,theautosacfile importgui,whichtriestoautomate the SAC file conversion to ISOLA format, user selectsonlyonefile,theNScomponentandtheGUIfindsand loadstheothertwo

    Two more file format filters were added PITSA and SEISAN(pitsaimport.m,seisanimport.m)

    Timesearch limitshavebeenextended,theusercanselectupto2500timeshiftsduringtheinversionprocedure

    A new plot of DC versus correlation has been added (seeZahradniketal2007).

    ToolstoconvertplotsinPNGformathavebeenadded. MTsol.txt file has a slightly different format to make it more

    descriptiveandaccurate. Anewfilewithonelinesolution,reportwasadded. Trial source positions are saved in KML format for display in

    GoogleEarth. AnewGUItohelpwithpolezerofilepreparationhasbeenadded. Various bugs have been corrected, and a few news were

    added.!!!pleasesendusyourfeedback plusmanymoresmallchanges..!

    Overallweconsiderthisamajorupdateandurgeuserstoupdate.

  • GUIPurpose

    TheISOLAGUIhasbeencreatedusingtheMatlabGUIDEtool.Itspurposeistohelp

    the user of ISOLAFORTRAN part with data preprocessing, Green function

    preparation,waveform inversionanddisplayoftheresults. Inadditiontothemain

    toolsafewutilitiesarealsopresentthathelptheuserto inspectthequalityofthe

    datausingfilters,converttodisplacement,etc. Inutilitiestherearealsotwoother

    options that could help: Backup files and Create Pole Zero file. The Backup files

    optionsaves inauserspecifiedfolderanumberoffilesthatarecriticalforrunning

    the inversion; in thisway the user can keep track of various attempts during an

    experiment. Finally the import SAC andGCF (GuralpCompressed Format)options

    help convertingSACandGCFbinary files in ISOLAascii format (4 columnsofascii

    dataperstation,i.e.,Time,NS,EWandZcomponent).

    Requirements

    InordertouseISOLAGUI,youneedtohavethefollowingsoftwareinstalled

    Matlab6.5or7.0(thesoftwarehasnotbeentestedusingotherversionsofMatlab)

    GenericMappingTools (GMT,WesselandSmith,1991,1998) (thesoftwarewastestedusingversion4.0and4.1,newerversionsshouldworkalso)

    M_Map,this isafreewareMatlabbasedpackage, it isused inmapcreationwithin Matlab, you can download

    fromhttp://www2.ocgy.ubc.ca/~rich/map.htmlandyouhavetoadditin

    theMatlabpath.

    Ghostscript and GsView for Postscript filesdisplayhttp://www.cs.wisc.edu/~ghost/,youhavetoadditinthesystem

    path.

    ImagemagickconvertsPostscriptfilesinPNGformatfordisplayinwebpages,thisshouldbeinthesystempathtoo.

  • Installation

    ISOLAandISOLAGUI,installationiseasyandneedsjustthreesteps

    Step1:Unzip the zippeddistribution file ina folder likee.g.c:\matlab. Makesure thatyouhave selected theUse foldernamesoption (ifyouuseWinZip)ora

    similaroptioninyourunzipsoftwarethatwillkeepthefoldernamesthatarestored

    in the archive. After extracting you should have a main folder e.g.

    c:\matlab\isola,which is your ISOLAGUI installation folder, and containsvariousMatlabfilesplusthreefolders,\bin,\pzfiles,\fortran.

    Step2:AddtheinstallationfolderinMatlabpath(e.g.withinMatlabgotoFile,Set

    Path,Addfolderandselecttheinstallationfolder,thenpressSaveandClose).

    Step 3: Folder \isola\binholds theexecutablese.g.gr_xyz.exe, isola.exeetc, this folder has to be added in System path (go to Control Panel, System,

    Advanced, Environment variable and add the \isola\bin folder in the Path under

    Systemvariables).ISOLAGUIisnowinstalled.

    Testingtheinstallation

    FromwithinMatlabcommandwindowtypeisolaandpressenter,theISOLAGUIshouldstartinitializinginthecurrentfolder,ifnotchecktheabovesteps.

    Method

    ISOLA isbasedonmultiplepointsourcerepresentationand iterativedeconvolution

    method,similartoKikuchiandKanamori(1991)forteleseismicrecords,butherethe

    fullwavefield is considered, andGreen's functions are calculated by the discrete

    wavenumbermethodofBouchon(1981).Thusthemethodisapplicableforregional

  • and local events. Instrumentally corrected bandpass filtered velocity records are

    used.Thecodetransformsvelocityintodisplacement,invertsthedisplacement,and

    providessyntheticdisplacement.Seismograms(t)isapproximatedbyacombination

    ofelementaryseismogramsei(t), i=1,2,...6, i.e.s(t)=sum(ai .ei(t)).Theelementary

    seismogramscorrespondtobasicfocalmechanismsasfollows:

    istrikediprake

    10900

    22709090

    309090

    4904590

    504590

    6isotropic

    Seismograms(3component),fromasetofstations,representthedata,d,whilethe

    coefficientsaiaretheparameterstobefound,m.Thelinearinverseproblemd=Gm

    issolvedbytheleastsquaremethod:m=(GTG)1GTd,whereGTisGtransposed

    and ()1 is the inverse of the socalled systemmatrix (GT G). Eigenvalues of the

    systemmatrix(GTG)arecalculatedtounderstandhowwellposed(orillposed)the

    inverseproblem is.Ratioofmin/maxeigenvalue is reported.The lower thevalue,

    theworsetheresultsare(although,formally,thequalityofthefitmaybegood,e.g.

    variancereductioncanbelarge).Questionhowtheindividualstationscontributeto

    thewellposedinversionhasnosimpleanswer.Itistobecarefullyanalyzedcaseby

    case.

    Optionally, theweighted inversion isperformed,withweightsprescribedbyuser,

    differentweights for different stations and components. As a rule, upweighting

    records with small amplitudes, and downweighting large amplitudes, is

  • recommended.Anotherstrategyistoupweightstationswhoseseismogramsappear

    tobethemostsensitivewithrespecttosmallmodificationsoftheproblem.

    ThecoefficientsaiarerelatedtomomenttensorMpq(wherep,qrefertogeographic

    coordinatesx(>0,N),y(>0,E),z(>0,up)):

    Mxx=a4+a6

    Myy=a5+a6

    Mzz=a4+a5+a6

    Mxy=Myx=a1

    Mxz=Mzx=a2

    Myz=Mzy=a3

    Moment tensor components in spherical coordinates r, t(=theta), p(=phi), used in

    Harvardcatalogue,orinGMT,aregivenby:

    Mtt=Mxx

    Mpp=Myy

    Mrr=Mzz

    Mtp=Mxy

    Mrt=Mxz

    Mrp=Myz

    Eigenvectorsofthemomenttensorprovidestrike,dipandrake.Eigenvaluesprovide

    scalar moment Mo and decomposition of the moment tensor into three parts:

    doublecouple(DC),compensatedlinearvectordipole(CLVD),andvoluminial(VOL).

    Possiblemomenttensor(MT)inversionmodesareasfollows:

    fullMTinversion:allsixbasicfocalmechanisms,1,2,...6;DC+CLVD+VOL

    deviatoricMTinversion:basicfocalmechanisms1,2,...5;DC+CLVD;VOL%=0;

    DCconstrainedMTinversion:DConly;VOL%=CLVD%=0

    knownandfixedDCmomenttensor(onlypositionandtimeissearched)

  • Opposedtostrike,dip,rakeandMo,theDC%isstronglyunstable.Itcloselyrelates

    withthefactthata largedeviationfrompureDChas,asarule,averysmalleffect

    upontheseismograms.Indeed,itcanbedemonstratedbymanyexamplesinwhich

    syntheticsfor,sayDC%=50andDC%=100,cannotbedistinguishedbyeyefromeach

    other, andquantitativemeasureof their fitwith realdata isnearly identical (e.g.

    variancereduction,introducedbelow,differsby1%).Therefore,wecansaythatthe

    DC%valueprovidedbytheroutinemomenttensorinversionisrarelyofanyphysical

    meaning.ThesameappliesforthevolumepercentageVOL%.

    Optionally,theinversioncanbeconstrainedthroughtheLagrangemultiplierstogive

    anearly100%DCsolution.The(nonlinear)conditionthatthemomenttensormust

    havezerovalueddeterminantissolvediteratively.Itslowsdownthecalculationand

    maycreateproblems.Thereisnospecialadvantageinthisoption,becausethestrike,

    dip and rake of the DCconstrained solution is usually very near to that of the

    (unconstrained)deviatoricMTinversion.

    Thus themost recommended inversionmode is deviatoric inversion (VOL%=0) in

    which,however,theresultingDC%isconsideredwithgreatcaution.

    Itisassumedintheaboveexplanationthatthesourcepositionandtimeareknown.

    This code allows their optimization through a grid search over set of trial source

    positionsandtimeshifts,predefinedbytheuser.Thesearchseeksthetrialsource

    position and time for which the residual error of the least square method is

    minimized. It is equivalent to maximize correlation between the observed and

    syntheticseismograms.Theconceptofoptimizingpositionandtimeisapplicableto

    eachpointelement (thesocalledsubevent)ofthegenerallymorecomplexsource

    model. If thewhole earthquake is represented (in the lowfrequency range) by a

    singlesource,theoptimumsolutionprovidedbythegridsearchcorrespondstothe

    centroid(thecentroidposition,centroidtimeandthecentroidMT).

  • Usercancheck thecorrelationvaluesasa2D functionof the trial sourceposition

    andtime.Thecodehasaninteractivepartinwhichusercanuseautomaticsearchof

    the optimum correlation, or canmake his preferred choice. The preferred choice

    may be the source position or timewith a somewhat lower correlation, close to

    formalmaximum, inwhichhoweversomeotherconditions (constraints)arebetter

    satisfied. In thisway user can apply, for example, his knowledge about the first

    motionpolarities,geologicallydeducedfaultdip,variousassumptionsaboutrupture

    propagation,etc.The2Dcorrelationmayalsohelp in identifyingpoorresolutionof

    someparameters,forexamplethedowndipsourceposition.

    Narrow bandpass inversions around period T can often provide two correlation

    maxima,intimetandt+T/2,characterizedbyrakeandrake+180respectively.This

    is just the casewhereeven a single guaranteed firstmotionpolaritymayhelp to

    decidebetweenpropertime,tort+T/2.

    Thecodeworksiteratively.Apointsourcecontributioniscalledsubevent.Usingdata

    code searches for subevent 1. Once subevent 1 is found, its contribution is

    subtracted from the data to get the socalled residual data. In the next step the

    residualdataareprocessedinthesamewayastheoriginaldataduringthefirststep.

    Subevent2 is found,etc.Originaldataminus the lastresidualseismogramprovide

    theresultingsyntheticseismogram.

    Optionally, the search of spatial and temporal distribution of subevents can be

    performedwithfixed(100%DC)momenttensor.Theuserprescribestheknown(or

    assumed)strike,dipand rake,and thesevaluesarekept fixed forallsubevents. It

    can be recommended when the unconstrained moment tensors vary chaotically

    fromonesubeventtotheother.

    Whenthesolutionisrepresentedbyseveralsubeventsofdifferentfocalmechanism,

    theresultingscalarmomentcannotbeobtainedasthesumofthesubeventscalar

    moments. That is why we also provide an output file where, subsequently, the

  • subevent contributions are summed up as tensors and the scalarmoment of the

    resulting(cumulative)tensorisshown.

    Denotingdataasd,andsyntheticsass,thefitbetweenthemismeasuredbytheso

    calledvariance reductionvarred=1 |ds|2/|d|2,where|.|denotes theL2norm,

    e.g.,|d|2=suma(di2).Thenormiscalculatedasasinglenumbersummingupoverall

    time series, all components, and all stations. The optimum value of varred is 1.

    Wrongsyntheticsmaygiveevenvarred

  • the code calculates varred using d and s=s1+s2, i.e. varred= 1 |d(s1+s2)|2/|d|2.

    Obviously,thevarredvaluereportedafterthesecondsubevent isnotthesameas

    the (squared) correlationbetween r and s2.Moreover, this varred is alsonot the

    sameasthe(squared)correlationbetweendands=s1+s2,becausewedidnotfind

    thesynthetics=s1+s2byasingleleastsquareminimizationof|d(s1+s2)|.

    When nonunit weights are used, and/or some stations are deselected from a

    particular inversion run, the relation between correlation and variance reduction

    getsevenmorecomplicated.

    RunningtheGUI

    Inordertorunthecode,withintheMatlabcommandwindow,changedirectory(i.e.

    usingthecdcommand inMatlabcommandwindow)tothefolderyouwanttorun

    theGUIandtype isola.ThiswillrunthemainGUI fileandaftersome initialization

    messagesyoushouldgetaformlikethefollowing(Fig.1).

    Figure1.MainformoftheISOLAMatlabGUI.

  • When ISOLAGUI startswithina folderchecks if ISOLAGUI specific subfoldersare

    present, ifnot itstartstocreatethemonebyone.Additionally itcopies*.stnand

    *.cru files it finds in ISOLAGUI installation folder to the current run folder.Files

    withextension*.stnarestationcoordinatesfileswhile*.crufilesarecrustalmodel

    files. It is a nice practice to have such files that you use often in ISOLAGUI

    installationfolder.UponstartISOLAGUIwillcopythesefilesandthusiftheuserhas

    afewcrustalmodelsthatusesoftenthesewillbeimmediatelyavailablewithoutthe

    needtobrowseforthefilethroughvariousfolders.Furthermore,theGUIwillsearch

    underinstallationfoldertofindafoldernamedPZFILESifpresentitwillcopyallfiles

    in installation folder\PZFILES folderunder the run folder\PZFILES, ifnot itwillask

    theuserwheretofindthepoleandzerofiles.

    CrustalModel

    Thefirststepintheinversionprocedureisthedefinitionofthecrustalmodel.Thisis

    done by pressing the Define Crustal Model button in themain form,whichwill

    launchtheformshowninFig.2.Thereisafixedoptionforamaximumof15layersin

    thecrustalmodel.Foreachlayer,theusershouldgivethedepthofthelayertop,the

    Vp,Vsvelocitiesinkm/sec,densitying/cm3,Qp,Qsand,optionally,atitle.Inorder

    toallowtheprogramtounderstandthe last layer,theusershouldtype999 inthe

    lastrowasshowninFig.2.

    Therearebuttonsforsavingthedata(Save) inatextformatfilesuitablefor ISOLA

    Fortrancode,aswellas reading itback in the form (Read).The latter isuseful for

    editingapreviouslysavedmodel.Theplottingoption(Plot)generatesgraphsofVp

    and Vs versus depth. Automatic calculation of Vs according to a specified Vp/Vs

    value is also possible, aswell as the calculation of density according to formula

    density (g/cm3)=1.7+0.2*Vp (km/s).PressingSavethecrustalmodel filethatwas

    createdissavedundertheGREENandPOLARITYfolders.

  • Figure2.TheCrustalmodeldefinitionform.

    ThenativeformatofthecrustalmodelfileinISOLAFortrancodeisthefollowing:

    Crustal model Tomography model number of layers 8 Parameters of the layers depth of layer top(km) Vp(km/s) Vs(km/s) Rho(g/cm**3) Qp Qs 0.0 5.47 2.700 2.560 300 300 2.0 5.50 2.860 2.800 300 300 5.0 6.00 3.230 2.940 300 300 10.0 6.20 3.240 2.940 300 300 15.0 6.48 3.400 2.980 300 300 20.0 6.70 3.800 2.980 300 300 30.0 6.75 3.810 2.980 300 300 40.0 8.00 4.660 3.360 1000 1000 ************************************************************************* Afterfinishingwiththecrustalmodelitistimefortheusertogivesomeinformation

    relatedtotheeventhewantstoanalyze;this isdonebypressingEvent Info inthe

    mainform(Fig.3)

    EventInfo In this form theuserprovides informationthatwillbeused inall thesubsequent

    stepsof the inversion, like theevent Latitude, Longitude,Depth,Magnitude,Date

    (the lasttwoarentused inanycalculationandare included justforclarity),Origin

    Timeof theevent,StartTimeof thedata file (seismograms)and, finally, theTime

    Lengthofthedatathatwillbeusedfortheinversion.Forusersconvenience,there

  • are5predefined time length intervals thatshouldcovermostapplicationsat local

    andregionaldistances.AfterpressingtheupdatebuttontheGUIwritesorupdates

    theappropriate filese.g.event.isl, rawinfo.isl,duration.isl; thoseare ISOLA (*.isl)

    related text files thathold theabove informationandare stored in ISOLA running

    folder.

    Figure3.TheEventInfoform.

    Nextstepistheselectionofthestationsthatwillbeusedfortheinversion(Fig.4).

    StationSelection

  • Figure4.TheStationSelectionform.

    Optionsinthisformaresimple;theusershouldpressMakeMapSelectstationsin

    order to select his (previously prepared) text file containing the geographical

    coordinatesof thenetwork. Thename and locationof the file isdisplayed at the

    SelectedStationFilesection.TheGUIwillselectthefileautomatically, ifafilewith

    *.stnextensionexists in ISOLA installation folder. Iftheuserwantstochange ithe

    can browse to select a different file. The file should be an ascii file with three

    columns, Station Name, Latitude, Longitude, separated by spaces without any

    header.TheStationNameshouldhave3characters.Ifitislongerawarningisissued

    andthestationnameisreducedtothefirstthreecharacters,cautionisneededsince

    stationnameslikeforexample,ABCD,ABCFwillresultinthesamestationnamefor

    both,ABC, causingproblems.PressingMakeMap Select stations (and assuming

    thatM_MAP isproperly installed)amapof thestationdistribution iscreated,and

    theusercanselectthestationshewantstoincludeintheinversion.Selectionofthe

    stationsisdoneusingtheleftmousebutton;thelaststationshouldbepickedusing

    the rightmousebutton.Afterpressing the rightmousebuttonawarningwindow

    comesupwiththetotalnumberoftheselectedstations.Laterduringthe inversion

    stage,ifnecessary,theusercanagaindeselectsomestations,i.e.toremovethem

    from the inversion process, but once the station was selected the corresponding

    seismogramsmustbeprovided.

    * Sometimeswhen stations are too close clicking on one of themmay result in

    selectionofboth, toavoid this,aparameterhasbeenadded (defaultvalue0.1) in

    Stationselectformadjustingthis(makingsmaller)couldremovetheaboveproblem.

    After finishing the selectionof the stations theuser should justpress Exiton the

    Station Selection form. Doing this the GUI will create files like stations.isl,

    station.dat,source.dat(inGREENfolder)andallstat.datinINVERTfolder.

    RawDataPreparation

  • NextstepistheRawDataPreparation(Fig.5),maybethemostimportantoptionof

    theGUI,sinceitprocessestheseismograms(doingtheinstrumentcorrection,origin

    time alignment, resampling) and produces data files ready for the inversion. The

    data,i.e.theuncorrectedseismograms,mustbeavailablein4columntextfiles(with

    the strictly required column order, i.e., Time in seconds , NS, EW, Z in counts,

    separated with spaces), one file per station. The filenames are not completely

    arbitrary;theyhavetostartwiththestationname,forexample,fortheLTKstation

    thefilenamemuststartwiththe lettersLTK,whereLTK isthestationnameused in

    theStationSelection.Theremainingpartofthefilename(includingtheextension)is

    arbitrary,the locationofthefile isalsoarbitraryanduserwillbeguidedtobrowse

    forfiles.

    *** ISOLAGUI has some default folder for storing the uncorrected files it is the

    \datafolderandthefilesifcreatedusingtheISOLAGUItools(e.g.sacimport)havethefollowingfilenamee.g.LTKunc.dat.

    Theusershouldalsomakeavailablethefilescontainingtheparametersthatwillbe

    usedfortheinstrumentcorrection.Anexampleofsuchapoleandzerofilefollows.

    A few pole and zero files for several instruments are includedwith the software

    distributionandthenewGUIversionhasatoolforpreparingthesefiles(makepz).

    A0 Comment133310 A0valuecount>m/sec Comment3.33e10 conversioncountstom/seczeroes(ReandIminrad/sec) Comment3 Numberofzeroes0.00.0 Zero0.00.0 Zero51.50.0 Zeropoles(ReandIminrad/sec) Comment5 NumberofPoles272218 Pole272218 Pole56.50 Pole0.11110.1111 Pole0.11110.1111 Pole

  • Itisstrictlyrequiredthatthefiledescribestheinstrumenttransferfunctionforinput

    velocity.ThefilenamemustbeacombinationoftheStationNameandtheBH*.pz

    string; forexample, for thestationnamedLTKand theEWcomponent, itmustbe

    LTKBHE.pz,thusthreepoleandzerofilesperstation,areneeded.Thelocationofthe

    pzfiles isnotarbitrary, theusermustplace them in thePZFILES folder.Caution

    mustbegiventotheappropriateunits.ISOLAneedsthepolesandzerosinrad/sec,

    andtheconstantofthetransferfunctionA0(guaranteeingthattheplateauofthe

    amplituderesponseequalstounity),shouldbeconsistentwiththat.Incasethatthe

    user has the zeros and poles in Hz, they all must be multiplied by 2, and,

    correspondingly,A0mustbemultipliedby(2)NPNZ,whereNPandNZisthenumber

    ofthepolesandzeros,respectively.

    Figure5.TheDataPreparationform.

    PressingLoadAscii file,astandardopen filedialogappearsand theuserselectsa

    datafile.Whenthefileisread,thesamplingfrequencyisdisplayed(inred)atthetop

    oftheform(Fig.5),sidebysidewiththeresamplingfrequencyofthedata inblue.

  • The resampling ismade automatically,basedon specificmethodical requirements

    andtheeventinfoinputfromtheprevioussteps.

    AfterdataloadingtheCutandInstrumentcorrectionbuttonsareenabled.Ifnotall

    of the data is neededusing the Cut button the last part of it can be deleted.By

    pressingtheInst.Correctionbuttontheprogramsearches inPZFILESfolderforthe

    appropriatepoleandzerofileandthenperformsthe instrumentcorrection.During

    thisoperationafiltercanbeapplied(Butterworth,2ndorder,zerophase)alongwith

    the DC and trend removal, all using the builtin Matlab functions. The baseline

    optionisanalternativeoftheDCremoval,inwhichthebaselineisdeterminedfrom

    thebeginningoftheseismogramonly.

    Next step is the time alignmentof the seismogram, i.e., irrespectivelyof its Start

    Time,sincenowitwillbeginattheearthquakeOriginTime;thisisdonebypressing

    theOriginAlignbuttonwhichbecomesactiveaftertheinstrumentcorrection.

    Finally the user should press the Save Data button in order to save the files in

    INVERTfolder.Thefilesaresavedastextfiles(4columns,Timeinsec,NS,EW,Zin

    m/sec),onepera station,always containing just8192pointsof theappropriately

    resampleddata.TheGUIautomaticallycreatestheproperfilenameforthecorrected

    resampleddata,whichforstationLTKwouldbeLTKraw.dat.Thisfilenamemustnot

    bechangedbytheusersinceitwillbeusedintheinversionstep.

    TrialSourceDefinition

    The finaldata that theuser shouldprepare is the trial source location and this is

    donebypressingtheSeismicSourceDefinitionbuttonintheISOLAmainform.The

    userhastoselectsinglesourceormultiplesourcedefinition.

    Inthecaseofsinglesource(Sourcesbelowepicenter),hecanvarythedepthonly,

    thustheoptionsintheformareStartingdepth,DepthstepandNoofSources,the

  • explanationofwhich isgiven inFig.6. Inthiscase,allthetrialpositionsarebelow

    theepicenter.BypressingExit theprogramwrites theappropriate files inGREEN

    andGMTFILES folders. These files are src*.dat files inGREEN and sources.gmt in

    GMTFILES.Thesources.gmtfileisformattedforGMT(psxy)plotting.

    It isassumedthatthesinglesourceoptionwillbealwaysusedtoretrieve (ateach

    trial depth) just a single subevent, that one which predominates in the low

    frequency inversion. The grid search over the depth then serves to provide the

    optimumsourcepositioncharacterizedbythebest fitbetweenthewaveformdata

    and synthetics. Thus we get a first approximation of the centroid. By the first

    approximationwemeanthat(opposedtoassumptionofthisoption)thecentroidis

    notnecessarilybelowtheepicenter.

    Figure6.TheSingleSourcedefinitionform.

    AspecialcaseofsinglesourceinversionistheOneSourceatHypocenterwherejust

    onesourceisusedfortheinversion,theonethatislocatedatthehypocenter.

    IftheuserwishestousetheMultipleSourceinversionanewformappears(Fig.7).

    Thisinversionhasthreeoptions:sourcesalongthestrikeofthefault,sourcesalong

    thedip,or sourcesonaplane.Theuserhas to select the fault strikeanddip, the

    number of the trial sources he wants to try, their spacing, and the first source

    Starting Depth

    Depth step

  • position.AfterpressingtheCalculatebuttontheprogramwritesthesource files in

    GREENfolderandproducesamapwiththetrialsourcedistribution.

    TheMultipleSourceoptionhastwodifferentapplications.Thefirstoneistoprepare

    trialpositionsforthegridsearchofthecentroid, inspectingusuallypositions inthe

    vicinity of the optimum depth found with the single source inversion. In this

    applicationwedodesignamultiplesourcestencil,butmake inversion insuch low

    frequencies that just a single subevent represents thewhole source. The second

    possible application of the multiple source option is to design the trial source

    positionsforinversionoftheearthquakeintoacomplexsourcemodel,consistingof

    more than a single subevent. Obviously, as a rule, the latter is achieved by the

    inversion includinghigher frequenciesandusingadensergridofthetrialpositions

    thanofthefirstapplication.

    Wehavetonoteherethatthismanualdescribesthecodeusageandcodeoptions.It

    cannotdescribeallpossiblestrategieshow touse thecode.Forexample, theuser

    mayusethemultiplesourceoptioninthebeginningofhisstudytofindthecentroid

    (the first application), which might need trial positions relatively far from the

    hypocenter,andwitharelativelylargegridspacing.Thenhemaywishtorepeatthe

    sameprocedurewith finergrid spacing to find theexactpositionof the centroid.

    Finally, he can try the application of the second type, i.e. to resolve several

    subevents. The lattermay often fail, due to lacking data, or simply because the

    earthquakehasnoclearcomplexity, thenhis result remains tobe just theprecise

    centroidsolution(withnoadditionaldetails).Thisdiscussionexplainswhyitisnearly

    impossible to give an alwaysworking hint ofwhere and how densely the trial

    sourcepositionsshouldbedesigned.

  • 1

    2

    3

    5

    4

    6

    S tr ike }Spac ing

    Ne w Hypoc ente r a t so urc e 1

    Sh ift to E a st

    Shift

    to N

    orth

    H ypo ce nte r

    Figure7.TheMultipleSourcedefinitionform.

    GreenFunctionComputation

    After finishingwith the trial sourcedefinition the inversion can startand the first

    step is to prepare theGreen functions. This is done by pressingGreen Function

    Computation in the ISOLAmain form.Thegreenpre.m is running (Fig.8)and, for

    simplicity,theuserhasonlyoneoptiontofill in,this isthemaximumfrequencyof

    theGreen functions tobe computed (fmax). (Obviously,nohigher frequency than

    thisonecanberequestedlaterintheinversion.)PressingRunthenecessaryfilesare

    created inGREEN folderandtheexecutableFortrancodesarecalled fromMatlab:

    gr_xyz.exe and elemse.exe, for the Green function and elementary seismograms

    computation, respectively. The execution is monitored through the systems

    CommandWindow.Theuser shouldwatch theexecutionof theFortranprograms

    andpressOK in the Copy files in invert folderdialog,onlywhen theprocedure

    ends in theCommandWindow, a relevantmessagewill appear there.After that,

    copying startsandamessage is issuedwhen the filesarecopied in INVERT folder

    andtheusercanproceedtotheinversion.

  • Forsimplicity,thisversionofthecodedoesnotspecifyanyparticularslipratetime

    function. Itmeans that the assumed slip rate is that givenby theGreen function

    calculation. By definition, the Green function is response of the medium to the

    impulsive(Diracdeltafunction)source,butnotethatweworkinalimitedfrequency

    band.Therefore,workingwith thebandlimiteddelta function, theactual sliprate

    function applied in the code has duration approximately equal to dur=1/fmax,

    wherefmaxisthemaximumfrequencyconsideredintheinversion.Theonlycorrect

    applicationofthiscodeisforsubeventswhosetruedurationsareshorterthandur.

    Figure8.TheGreenfunctioncomputationform.

    Inversion

    The laststepstartsbypressing Inversion in the ISOLAmain form; it runs invert.m

    thatproducesthefollowingform(Fig.9).

  • Figure9.TheInversionform.

    Beforerunningtheinversion,theuserhastoselecthisoptionsregardingtheFilter,

    Type of Inversion,Number of Subevents to be retrieved, and parameters of the

    TimeSearch(temporalgridsearchofthesubevents).Thesamefilterisautomatically

    usedforthedataandsynthetics.Propervaluesoftheaboveparametersdependon

    the specific problem e.g. on the event magnitude, on signal to noise ratio, on

    locationuncertaintyetc.Thus it isupon theuser to select thebest values forhis

    problem.

    The usermay press the Compute weights buttonwhich automatically calculates

    weights to be applied in the inversion. (Not pressing this buttonmeans that unit

    weightswill be used.) The automaticweights are inversely proportional to peak

    valuesofgrounddisplacementofthe individualcomponents intheusedfrequency

    band.InfactwhatarecomputedbytheComputeweightsfunctionarejustthepeak

    valuesforthetracesandlatterisola.exetakesthe1/peakvalueastheweight.Finally,

    pressingRun,themainexecutableFortrancodeisola.exe isrunninginthesystems

  • Commandwindow(Fig.10).Whenfinishingwiththefistsubeventretrieval,thecode

    makesapauseandaskstheusertoaccepttheautomaticselectionofthetrialsource

    position(thatofmaximumcorrelation),ortoselectsomeotheruserspreferredtrial

    source position. To simplify the decision, the user can plot a correlation diagram

    usingthePlotCorrelationoption(Fig.11);GMTandGsviewareneededforthisplot.

    Inthis2Ddiagramthecorrelationvalueandfocalmechanismareplottedagainstthe

    trialsourcepositionandtimeshiftforthesubeventunderstudy.

    PressingtheDeselectStationsbutton,theallstat.datfileopensusingnotepad,and

    theusercan removeastation from the inversionbyediting thesecondcolumnof

    the file (1=use the station, 0=not use it). The columns 3, 4, 5 contain the above

    mentionedpeakdisplacementvaluesfortheNS,EW,Zcomponent,respectively.The

    user canmanually edit them, thusmaking his preferred specification ofweights.

    Notethatthelargerthenumberinallstat.dat,theloweristheweight.

    SeealsotheMethodchaptertolearnmoreabouttheweightsandcorrelation.

    Allstat.datfileexampleincludingweights.

    GUR16.057019e0052.130505e0043.082369e004

    LTK16.179501e0052.366762e0042.285469e004

    DID11.139388e0041.451932e0041.765126e004

    PYL11.145303e0041.020548e0041.030644e004

  • Figure10.Exampleoftheinversionrun.

    Figure11.Exampleofthecorrelationplot.

    Plotting

  • The inversion code isola.exe createsanumberofoutput files that canbeplotted

    after the inversion run. This isdonebypressingPlot Resultsbutton on themain

    ISOLA form. The plotres.m is called and produces the form of Fig. 12. There are

    options for plotting Synthetic versus Real data, the MT results, as well as for

    comparingthesolutionwithfirstmotionpolarities.

    TheMT(momenttensor)plottingisdoneusingGMT,andtheuserhastheoptionfor

    a single source summaryplot (Fig.13)aplotofCorrelationversusSourcenumber

    andDC%(Fig.14)ormultiplesourceresultsplotinmapview,pressingPlotSources.

    In all the plots (except the one in the single source summary) the fault plane

    solutions are displayed by beach balls without showing the nonDC component,

    because,asexplainedabove,thenonDCpartisusuallyunstable.

    Notethatthesinglesourceplottingoption isavailablealso forthe inversionwhich

    resulted in more than one subevent. The code prompts the user to specify the

    subevent he wishes to plot in Fig. 13 and Fig. 14, and only this is plotted. The

    variance reduction given in the Single Source plot in Fig. 13 then, refers to the

    situationafteraddingthechosensubevent(forexample,whenplottingsubevent2,

    thevarredvalueapplies for the fitbetweendataand the synthetics composedof

    subevent 1 and subevent 2). The fit is calculated only from stations used in the

    inversion(thedeselectedstationsarenotconsidered).

  • Figure12.ThePlottingform

    PressingthePlotSourcesbuttontheplsource.mopens.ThiscreatestheformofFig.

    15.Theuserhasalotofoptionsinordertocreateamapwiththemultiplesubevent

    results (Fig.16).Mostof theoptionsare selfexplanatory,but someknowledgeof

    GMT isneeded.Themostcriticalare theBeachballshiftandBeachballXshift to

    preventoverlappingofthebeachballs.Theplotformally includesallthecalculated

    subeventsbutlotofcareisneededtofindoutwhichofthemarephysicallyjustified.

    Thiscomplicatedphysicalproblemisnotsolvedhereinthemanual.See,forexample,

    Zahradniketal.(2005).

  • Figure13.TheSinglesourcesummaryplot

    Figure14.TheCorrelationvsSourcenumberplot.

  • Figure15.Themultiplesourceplottingform.

    Figure16.Plotofthemultiplesourceinversionresults.

  • Fullnumerical resultsof the inversion are available in the files inv1.dat, inv2.dat,

    inv2c.dat, inv3.dat, inv4.dat, located in the INVERT folder. For a detailed

    explanation, the readme filesof the ISOLAFortranguide shouldbeconsulted (see

    below).Mostoften,simpleroutput(thatprovidedinthePlottingform)issufficient.

    FinallytheuserhastheoptiontoplottheMT inversionresultsalongwithselected

    polarities(Fig.17) inordertocheckvalidityofthe inversion.Afewfilesneedtobe

    updated for thisoption.Namely the station.dat filehas tobemanuallyedited to

    assign the observed polarities per stations. This can be done within the GUI by

    pressing Input Polarities afterwhich the station.dat file opens in an editor. The

    other two files thatneed tobe createdareonemech.datandmoremech.dat that

    theuserhastocreateinthePOLARITYfolder.Theonemech.datcontainsdatatobe

    shown in red (e.g. the solution from ISOLA),whilemoremech.dat can include, for

    comparison,severalothersolutions(e.g.thosefrominternationalagencies).Thetwo

    filescanbeconstructedautomaticallybytheISOLAGUIbypressingEdit*mechfiles,

    theGUIwillchecktheINVERTfolderforresults(theINV2.DATfile),preparethetwo

    filesbasedonthe inversionresultsandcopythemtoPOLARITYfolder.Then itwill

    open the files in an editor and theuser canmake changes ifhewishes, e.g. add

    anothersolutioninmoremech.dat.

    Figure17.Plotofpolaritieswithinversionresults.

  • Moremech.datexample(spaceseparatedtextfile;onlycolumns4,5,6=strike,dip,

    rakeareimportanthere)

    4 6.66 .185761E+18 151. 76. 77. 13. 19. 131. 251. 61. 44. 32. 88.9 .7718E+00 1 3.96 .560486E+17 201. 50. 144. 316. 63. 46. 76. 83. 175. 39. 68.4 .8426E+00

    In the new version of ISOLAGUI, there is an option for plotting extra station

    polarities, e.g. from short period stations thatwerent used in the inversion but

    provide polarity information that can help in constraining the moment tensor

    solution. To do this the user should prepare a text file called extrapol.pol in the

    POLARITY folder theGUIwillcheck for thepresenceof this fileandadd theextra

    polaritiesinthefinalplot.

  • UtilitiesIandIIBesidestheinversionrelatedtools,anumberofoptionsisprovidedforassistingthe

    user in converting his binary data to ISOLA ascii format and inspecting the data

    quality.Toolsareavailablefor importingdata intwoformats:SACandGCF(Guralp

    Compressed Format). The tools are quite easy to handle; just read the files,

    componentbycomponent,andthensavethefile(Fig.18).Ifthethreecomponents

    at a station dont share the same starting time, or dont have the same length,

    optionsfortheirtrimmingareprovided.

    Figure18.TheGCFfileimportform.

    Thetoolsfor importingdataarebasedonfreeMatlabroutinesfromtheWeb(see

    acknowledgmentpartofthemanual).Theusercanusethemasanexampletoeasily

    addhispreferredformatconversiontothepresentGUI.

    AfterconvertingthedatatoISOLAformattheuserhastheoptiontoinspectthedata

    qualityeitherusingtheInspectDatatoolortheTryFilterstool.Datacanbefiltered

    invariousfrequencybandsinordertodefinetheproperbandforinversion(i.e.the

    band with a good signaltonoise ratio). Data can be also inspected as the

    displacement. It is extremely important to inspect the displacement without

  • filtrationandinstrumentcorrectionsinceitallowsdetectionofvariousdisturbances

    thatareoftenhiddeninthevelocityrecordsand/orinthedisplacementbandpassed

    records(ZahradnikandPlesinger,2005).

    FinallythedatacanbeshiftedusingtheShiftdataoption.Thisoptionmaybeused

    (withgreat caution)when, forexample,wehavean indication that theemployed

    crustalmodelisnotappropriateatsomestationasregardsthearrivaltime.

    AnewmemberoftheUtilitiescodesismakepz.ThisisaGUIthatwillassisttheuser

    toprepareapoleandzero file for ISOLA. Itsuse isverysimple, theuserneeds to

    presstheNewZeroandNewPolebuttonstoaddthepropernumberofpolesand

    zerosforhissystem.Thenhehastoaddallthepropernumbersinthetextboxesand

    finally save them in a file suitable for ISOLA, by pressing Save file. The GUIwill

    proposeaproper filenamebasedontheStationnamegivenandtheusercanalso

    plottheresponsecurvebypressingPlot.

    Figure19.ThemakepzGUI.

  • AcknowledgmentsSpecialthanksgotoPetraAdamovaforextensivetestingofISOLAGUI.Thefollowing

    mfilesarepartofotherprogramsthatwefoundusefulandaddedtotheGUI.

    rsac.m,lh.mbyMichaelThorne, readgcffile.mbyM.McGowan,GuralpSystemsLtd., Bandpass.mCoraldistribution.

    In the Fortran part, several subroutineswere used,written by other authors:M.

    Bouchon,O.Coutant,J.Sileny,J.JanskyandO.Novotny,J.A.Snoke,P.Reasenberg

    andD.Oppenheimer.Codes from FortranNumericalRecipeswerealsoused.This

    workwas supported by the EC project 0040433HAZCorinth and by the Charles

    UniversityinPraguegrant,GAUK279/2006/BGEO/MFF.

  • Appendix I - How to prepare a pole and zero file for ISOLA. Preparingaproperpoleandzero filehasalwaysbeenaproblem.Thuswe try

    heretogiveanexampleonhowtoprepareacorrectpoleandzerofileforISOLA.

    The format of the file is not discussed here since it is described in the ISOLA

    manual.Inthisdocumentthecorrectproceduretocomputethepropervaluesis

    escribed.d

    ThusinordertoprepareapoleandzerofileforusewithISOLAweneedtoknow

    hefollowing:t

    ISOLAneedsthepoleandzerovaluesinrad/s. The pole and zero should describe the VELOCITY response of the

    instrument

    nsetooTheA0valueshouldbegivenforVELOCITYrespo Besides theA0weneed toprovide thevalueofm/secthatequalsto1

    count.

    Before proceeding with instrument correction we can use option PlotResponse(fromRawdataPreparationGUI)thiswillplotthe instrument

    response of the instrument, (Fig.1). At this plot we should check if the

    frequencyrangeofflatresponseisthecorrectoneandiftheamplitudeof

    theresponsefunctionisone,thismeansthatourA0valueiscorrect.

    WepresenttheaboveusingtheexampleofaTrilliumseismometerandaTrident

    (Nanometrics, http://www.nanometrics.ca/) digitizer. From the manual we

    knowthefollowing:

  • AndTridentissetatthe1cou

    huswehavethefollowing:

    ntpermicrovoltsensitivity.

    T

    A

    0=133310

    Using the Trillium sensitivity (1500 V/m/sec) and digitizers sensitivity

    1cnt/microV)wegetthevalue6.66667e10asm/secper1count.(

    inallyweputtheabovevaluesinatextfilewiththefollowingformat.F

    A0

    133310

    count>m/sec

    6.6667e10

    zeroes

    3

    0.00.0

    0.00.0

    51.50.0

    poles

    5

    272218

  • 272218

    56.50

    0.11110.1111

    0.11110.1111

    Example of Pole and Zero file for ISOLA (Trillium40 seismometer and Trident

    digitizer)

    igure1.FrequencyResponseforTrillium40seismometer.F

    What do ifyouhaveaRESPfile.

    If the user has his response info in SEED RESP format, creating an ISOLA pole

    zero file is easy. First search for the B053F07 A0 normalization factor: entry, this is

    your A0 value. Then copy the pole-zero part from RESP file to ISOLA pole zero file,

    the values are exactly the same but BE CAREFUL to have the RESP file for input

    velocity since this is needed for ISOLA. Finally search for the B058F04 Sensitivity:

    to

  • entry, this value is the inverse of the value needed by ISOLA to specify sensitivity,

    thus the inverse of this number should be put in ISOLA pole zero file in the

    ount>m/secoption.c

    Appendix II: Files and folders

    In this sectionexamplesof filesareprovidedaccording to the folders theyare

    ocated.l

    \

    herootrunningfolder,shouldcontainfileslike*.isl,*.stn,*.cru.T

    \data

    hisfoldershouldcontainuncorrecteddatainISOLAformat(4columnasciifiles)T

    \gmtfiles

    FilesforplottingwithGMTaresavedhere,bothdatafilesandbatchfilesusedfor

    the calling of GMT commands e.g. event.gmt, sources.gmt, psources.bat,

    lotselsta.batetc.Moreover*.kmlfilesforusewithGoogleEartharesavedhere.p

    \green

    ilesusedinGreenfunctioncomputation.F

    \invert

    ilesusedforinversionandinversionresults.F

    \output

    raphicsfiles,*.pngaresavedhere.G

    What is new in ver. 4.0 ?GUI PurposeRequirementsInstallationMethodRunning the GUICrustal ModelEvent InfoStation SelectionRaw Data PreparationTrial Source DefinitionGreen Function Computation InversionPlottingUtilities I and II

    AcknowledgmentsAppendix I - How to prepare a pole and zero file for ISOLA.Appendix II: Files and folders