Download - IMCORR

Transcript

MEASURINGGLACIERVELOCITIESONTHEKENAIPENINSULA, ALASKA,USINGMULTISPECTRALSATELLITEIMAGERY WITHFEATURETRACKING by JamesBradleyTurrin Athesissubmittedtothefacultyof The UniversityofUtah in partialfulfillmentof the requirementsforthedegreeof MasterofScience DepartmentofGeography The UniversityofUtah May2010 MEASURING GLACIER VELOCITIES ON THE KENAIPENINSULA, ALASKA,USINGMULTISPECTRAL SATELLITE IMAGERY WITHFEATURETRACKJNG by JamesBradley Turrin A thesis submit tedto the faculty of The University of Utah inpartialfulfillmentof the requirementsforthe degree of Master of Science Department of Geography The University of Utah May2010 CopyrightJamesBradleyTurrin2010 AllRightsReserved Copyright CJames Brad!ey Turrin 20! 0 AURights Reserved T HEU N I V E R S I T YOFU T A HG R A D U A T ES C H O O L SUPERVISORYCOMMITTEEAPPROVAL of a thesissubmitted by James BradleyTurrin This thesis has beenread byeach member of the followingsupervisorycommittee and by majorityvote has been found to besatisfactory. Chair:Richard R.Forster 2-*zH-Jo DorothyK^Hall THEUNIVERSITYOFUTAHGRADUATESCHOOL SUPERVISORY COMMITTEE APPROVAL of a thesis submitted by James Bradley Turrin This thesis has been readby each member of the fol lowing supervisory committee and by majority vote has been foundto be satisfactory. -Z-14.-1(; Chair:Richard R.Forster 2-2'/-10 I'Dorothyall T HEU N I V E R S I T YOFU T A HG R A D U A T ES C H O O L FI NALREADI NGAPPROVAL To the Graduate Councilof the Universityof Utah: IhavereadthethesisofJames BradleyTurrininit sfi n aifo r m andhavefoundthat(1)itsformat,citations,andbibliographicstyleareconsistentand acceptable;(2)itsillustrativematerialsincludingfigures,tables,andchartsareinplace; and(3)thefinalmanuscriptissatisfactoryto thesupervisorycommitteeandisreadyfor submission to The GraduateSchool. DateRichard R.Forster Chair:SupervisoryCommittee Approved for the Major Department Harvey J.Miller Chair/Dean Approved for the GraduateCouncil Charles A.Wight Deanof The GraduateSchool THEUN I VERSITYOFUTAHGRADUATESCHOOL FINALREADINGAPPROVAL To the Graduate Council of the Universi tyof Utah: 1 have readthethesisofJames Bradley Turrininitsfinalform andhavefoundthat(I)itsformat,citations,andbibliographi cstyleareconsistentand acceptable; (2)itsillustrative materials including figures,tables,andcharts areinplace; and(3) the finalmanuscriptissatisfactoryto thesupervisorycommitteeandis readyfor submission to The Graduate School. 5-2 0 -[..6{0 DateRichard RForster Chair: Supervisory Committee Approved for the Major Department Approved for the Graduate Counci l Charles A.Wight Dean orThe Graduate School ABSTRACT PreviousstudieshaveshownthattheglaciersofsouthernAlaskaare recedingand thinning,andcontributingsignificantlytosea-levelrise,andthattemperateglaciersare sensitiveindicatorsofclimatechange.Therefore,a methodisdevelopedtoapplyfeature trackingsoftwareto moderateresolutionmultispectralsatelliteimagerytoderivevelocity fieldsforoutletglaciersoftheHardingIcefield,Alaska,whichwillserveasa basisfor futurestudiesof massbalance,contributiontosea-levelrise,andglacierresponseto climatechange.Thedata producedbyafeaturetracking programaresystematically sortedbylocation,direction,magnitude,correlationstrength,anderrorestimatesto eliminateerrantvectorsusingcustomizedIDL programs.Asuccessfulnewmethodof systematicallyremovinganomalousvectorswithina velocityfieldviaafeedbackloopis introducedandtested.Pointdataarespatiallyinterpolatedanda vectorfieldisplotted andcolor-codedtodisplayicesurfacevelocitiesona glacier,thenvelocityprofilesare producedtogaugetheinfluenceof topographyonice velocityandtodeterminethe primarymechanismoficeflow.VelocityfieldsforTustumenaandMcCartyGlaciersare derivedthatshowspeedsrangingfrom0000-1-1 -1-1-1-1-1-1-1-1-1-1-1-1-1-1-1 -1-111-1-1-1-1i-1-1I-1-1-1 Figure2.2.Searchandreferencechips,a) Referencechipcenteredinsidesearchchip,b) CorrelationIndexvaluespopulatingthesearchchipwithcorrelationpeakindicatedby coloredpixels. 2.2IMCORR IMCORRisa computerprogramthatcomparestwotemporallysequentialdigital imagesandattemptstofindthedisplacementoffeaturesthathavemovedduringthetime intervalbetweenacquisitionsof the twoimages.Todothis,IMCORRusesasubsetof thefirstimage,calleda "referencearea," or"referencechip," anda largersubsetofthe secondimage,centeredat thesamelocationas the referencechip,calleda "searcharea," or"searchchip" (Figure2.2a).Thereferencechipis movedwithinthesearchchipand foreach positionin whichthe referencechipcompletelyfitswithinthesearchchip,a correlationindexbasedonthe patternof brightnessvaluesbetweenthe twosubsetsis calculatedforthecenterpixelin thereferencechipusingastatisticalalgorithm(Figure 2.2b).Correlationsthatmeetspecifiedparametersaredeemedmatchesandthe displacementof thefeatureis calculated(Scambosetal.,1992)(NSIDC,2009). Thestatisticalalgorithmused byIMCORRisa FastFourierTransform-based 221MCORR and , u""'pts'" fond"'" dioplxern"try upon 2)Y C< I Son by di"",tiCH1 and IrngthinSon by direction, local r.eigbbothoodIongth. $Ubtra Distancefromsouthtonorthalont;!A-A'"( meters) Figure 3. 2. McCany Gl acier longitudinallIelocity profil e along A-A"'. 48 oneithersideoflocationA" ,thereare highlyunrealisticoscillationsinspeedranging from20or30 m/yrupto160 m/yrovera100 meterdistance.Realizingthatglaciersdo notvarytheirspeedinsucha manner,theseapparentabruptchangesin velocity pointto someunderlyingproblemwithintheprocess. ThevelocityfieldforTustumenaGlacier(Figure3.3)isalsospatiallydenseand mostlycomplete,exceptnearthe toe whereablationis highest,andupglaciernearthe snowlineof theAugust2005imagewheretheglacierwasmaskedto preventfalse correlations.The bottom5 kmof thevelocityfieldexhibitsa glaciologicallyplausible trendoficedeceleratingfromspeeds>106m/yr(orangearrows)tolessthan39m/yr (purplearrows)overseveralkilometersapproachingthe toe.Thisis quiteplausible becauseablationratesincreaseatlowerelevationsandthequantityoficeis reducedto sucha degreethatits massis nolongersufficienttomaintainhighspeedsandtheice slowsandmay becomenearlystagnantnearthetoe.In thisarea,IMCORR,withtheaid of theentire parameterspace, produceda velocityfieldthatis plausible.Farther upglacier,thevelocityfieldforTustumenaGlaciersuffersthesame problemseenin McCartyGlacier'svelocityfield.Therearedramaticfluctuationsin velocitywithinvery shortdistancesthatmakethederivedvelocityfieldimpossible.Thelongitudinalvelocity profileforTustumenaGlacier(Figure3.4)showsnumerousareasin whichthespeedof theiceinstantaneously jumpsfrom20or30 m/yrtonearly200m/yr,or viceversa.These rapidpulsesofaccelerationanddecelerationare highlyunlikelywithinthecontinuous massoficeandthereis nophysicalmechanismwithintheglacier(suchassteepinclines oricefalls)thatwouldaccountforsuch movement.Again,the physicallyunlikelyresults 48 on either side of location A", there are highly unrealistic oscillations in speed ranging from 20 or 30 m/yr up to160 m/yr over a100 meter distance.Realizing that glaciers do not vary their speed in such a manner, these apparent abrupt changes in velocity point to some underlying problem within the process. The velocity field for Tustumena Glacier (Figure 3.3) is also spatially dense and mostly complete, except near the toe where ablation is highest, and upglacier near the snowline of the August 2005image where the glacier was masked to prevent false correlations. The bottom 5 km of the velocity fieldexhibits a glaciologic ally plausible trend of ice decelerating fromspeeds> 106 m/yr (orange arrows) toless than 39 m/yr (purple arrows) over several kilometers approaching the toe.This is quite plausible because ablation rates increase at lower elevations and the quantity of ice is reduced to such a degree that its mass is nolonger sufficient tomaintain high speeds and the ice slows and may become nearly stagnant near the toe.In this area, IMCORR, with the aid of the entire parameter space, produced a velocity field that is plausible. Farther upglacier, the velocity fieldfor Tustumena Glacier suffers the same problem seen in McCarty Glacier's velocity field.There are dramatic fluctuations in velocity within very short distances that make the derived velocity field impossible. The longitudinal velocity profile for Tustumena Glacier (Figure 3.4) shows numerous areas in which the speed of the ice instantaneously jumps from20 or 30 m/yr to nearly 200 m/yr, or vice versa.These rapid pulses of acceleration and deceleration are highly unlikely within the continuous mass of ice and there is no physical mechanism within the glacier (such assteep inclines or ice falls) that would account for such movement. Again, the physically unlikely results TUSTUMENAANDHI2GLACIERSVECTORFIELDS7^| Figure3.3. TustumenaandHI2Glaciers'icesurfacevelocitiesderivedusingIMCORR.VelocityprofileA- A' "is shownin Figure3.4.Red,greenand blackstarsindicatelocationsof pointsreferredtoin Figures3.5,3.6,and3.7, respectively,wheredetailedanalysisofIMCORR' sparameterspaceareconducted. Fig... 3.3. TUSlumtna ood1112GIK;"'"ice .."f"", \eloci!i .. 90 m/yr)areadjacenttogreen vectors(45-60m/yr), whichisa reductioninspeedof33 to50%, whichis notrealistic. Overall,COSI-Corr' sfrequentialcorrelatorprovidesthemostconsistentvelocity Figure3.20.McCartyvelocityfieldproducedusingCOSI-Corrfrequential correlatorandASTERband3Nimages. m!)T). Tho ,,..,, \"10.. fie'" proOmI01OtS ..... NCOII.R ,,, edge mapI. dramali. fi"",u"ioo> in. I200: > o 150-> 3006009001200 Distancefrom south to north alongB-B'(meters) Figure3.32.Profilesfor2007/08Tustumenavelocityfield,a) Elevationandlongitudinal velocity profileA-A' .b)TransverseprofilealongB-B' . ,- , t - -! f -- i >

'00 .. '.' M "

". .) - '00B .... ,,., ., Fi""" JJ2. roo- 2007/01 YM"eIooiry r..kI. _) Ele;...... """ IonsitM,"o! Iodcy pmfiloAA '. b) T.-._ pmfile 8-8'. 97 Tustumena Velocity: TransverseProfileD-D! o Q O > d) 30060090012001500 Distance from south to north along D-D' (meters) Figure3.32Continued.Profilesfor2007/08Tustumenavelocityfield,c)Transverse profilealongC- C.d)TransverseprofilealongD-D' . " "" ~ '", c j , ~ ~ '00 . TranMJrie Profile [).O' "" 0 < 1 '", I ,. > 0 '00 " Figure 3.32 Con'inued. Profiles for 2007108Tu",umena .... Iooi'y f,.Id. d. Thi. ;, .1", the only point in e 2006107 field.,,hid. his many """"Iationo with SSR,. 0.99. bu,the ,,tor fieldforthe 2007108pal, it """finned '0 be """urale Ihfough ,'isual.aloulalionODd""'amli........" . I Figure J 39. 2006/07 Tust"",ena "ector fieldcolor coded.ignal-to-noi .. ratio. 107 Figure3.40.2007/08Tustumenavectorfieldcolorcodedbysignal-to-noiseratio. thoughthereare nocorrelationswithSNR>0.99.The2006/08vectorfieldhasan equitabledistributionofSNR(Figure3.41)amongitscorrelationswithvaluesfrom0.99 (red), totherangeof0.99-0.94(yellow),anddl ( ..ith S:':R Iculalod ba>e '-.Io 0.99 does not mean they should be discarded, but rather that they were produced using a smaller search window size (32x32).The lower SNR and higher velocity difference between the visual and automatic techniques of McCarty Glacier's velocity field suggests higher SNR correlates with better accuracy, but not necessarily better spatial extent, which is why the velocity field produced using a 32x32 search window was chosen for this study. 3.10 Positive Degree Days, Time Span, and Location Comparison While deformation of icesurface featuresdue to large displacements may influence the local success or failure of feature tracking algorithms within a particular area of a glacier, a region's climate and associated melt intensity will determine the maximum time span that can be accommodated by feature tracking. As a relative measure of the amount of ablation on Harding Icefield, the number of positive degree days (PDD) was calculated. The degree day was first introduced by Lt. Gen.Richard Strachey, Chairman of the Meteorological Council and later described by R.H.Scott (Scott,1884, p.354) as "10 F of excess or defect of temperature above or below [freezing]continued for twenty-four hours." Therefore, the PDD may be considered the "accumulated 119 JlII1,11L 200020012002200320042005200620072008 Year Figure3.47.PDDforSeward,AK;NASASE AWSGreenland;Pyramidstation Himalayas. temperature"or"positiveeffectivetemperature"foragivenlocationandtimespan. MeteorologicaldatafromSeward,AK(NOAAwebsite), northeastof HardingIcefield, yieldsPDDfor2000-2008(Figure3.47)rangingfroma highof1713in 2005toa lowof 1140in2008andanaveragePDDvalueof1456. Alapserateof6.5C per1000 mwas appliedtoadjustthetemperatures,recordedat18 metersabovesealevel, toanequivalent temperatureat 400mabovesealevelat the midpointalongTustumena' sablationzone. Thesevaluescan becomparedtoPDDvaluesoflocationswhereothersimilarstudies usingfeaturetrackinghave been performedwithsuccessfulresults. Arecentinvestigation of KhumbuGlacier,Nepal(Scherleretal., 2008)successfullyusedCOSI-Corrto generatevelocityfieldswithsequentialASTERimagesthatspannedupto3.95years (Sept.' 01toAug.' 05). KhumbuGlacierisaslowmovingglacier( ofHording lcofield, yield. PODfor200().2008 (Figut,. POD ,'.II!< of Alap$1000 La 500 CO SE Greenland Svalbard Antarctic y vHimalayas Alaska 500100015002000 Positive Degree Days New Zealand 25003000 Figure3.48. Comparisonof glaciervelocityand positivedegreedays(PDD)forselected glaciatedregions. studieshaveusedimageryspanninga singleablationseasoninareasoutsidethepolar regions: NewZealand(Kaab,2002;Quincey&Glasser,2009), Norway(Jacksonetal., 2005),the KarakoramRange(Coplandetal., 2009),andKyrgyzstan(Mayeretal., 2008). Tofurtherillustratethe unique positionofsouthernAlaskanglacierswithregards toablationandlocation,a plotof PDDvs. glaciervelocityforfiveselectedglaciated regionsis provided(Figure3.48).Althoughthisrepresentsa smallsamplingof worldwideglaciatedareas,southernAlaskaclearlystandsout withrelativelyslow velocitiesand highablation,whichreaffirmsthe potentialdifficultiesofsuccessfully applyingopticalfeaturetrackingtechniquesthere. Basedon the resultsof thisstudyand thechoicesmade bynumerousother '" Now Zealand (KUb. 2002, Quingon\S ..ol"';I,.ly !'orrOSI"""", data. ;COSI.(".., 1;.\ VIm.. ho,,, ,I or 4bands: .Bandl.. """'-" ... offoou I' d,,,,,""')" ,Ill _""0 ,'" Illc n&hl .Bar.l2 i. oonh ....,h olf.... (y d,ra.;"')" ,Ill pOSi'''. Y 10,be lOp ('",,"''''' 10 .1\lCORR', .. oo"h.....)....... ) "II>oh."'".) ,F""'JU""! ..I """"I."". .rnt.. 4 bands, .ut,,,.i """"n Jbands 134 readu,,cosi_corr_array;readenvifileintoarray params=widget_auto_base(title-Enterparametersforconversion',/xbig) list=['stepsize:','windowsize:','spatialresolution(m):'] vals=[4,64,15] output=widget_edit(params,uvalue='parameters',/auto,prompt='Enternew value',dt=2,list=list,vals=vals) params_result=auto_wid_mng(params) ? ? grid_spacing= pamms_rGSu\t.parameters[] ;scalefactorforchangingsizeofdisplayedimage,usevalues1.0orlarger initial_window_size=params_rQSu\t.parameters[1 ] ;scalefactorforchangingsizeofdisplayedarrows resolution=pa.mms_resu\t.parameters[2] ;spatialresolution,usedtoconvertoffsets(pixels)tovelocity(m/yr) outfile=widget_auto_base(title-Selectlocationof UNSIFTEDfolderforoutput',/xbig) output=widget_outf(outfile,uvalue='output_fileVauto,/directory,xsize=5(,prompt-Enter nameofoutputfile(.txt)') outfile_result=auto_wid_mng(outfile) openw,:,outfile_result.ow/p^/z/e,/append ;thisis theIMCORR-styletextfilewhichwillbe processedwith;otherIDLprograms. inform=['ConvertingtoIMCORRformat....'] envi_report_init,inform,base=percent_complete,title='ConvertingCosi-CORR data',/interrupt;begin%completewidget fori=0], columns-1dobegin ;Forloopstocreatethe9datacolumnswhichmimicanIMCORRdatafile for j =, rows-dobegin x_coor=i*grid_spacing+(*initial_window_size) ;x coordinate,takingintoaccountgridspacing y_coor=j*grid_spacing+(0.5*initial_window_size) ;y coordinate,takingintoaccountgridspacing '" _ "HI,,,_a ut bOJ!r,1o"koch ,IIbt proilow_lizo) .x..... ,.... on,o _001"II p,.t """'"1 Looor-j ' grid_ op>oal ,ij.porl_.t.l.pero.... u'omplOl i. col""'.... l ;up.bt .. perourompkt< a Hlget",th ca.chmplOloJ finioh ;tntO'(Of .-",h .... Z)......crmed InEX\-) ,n ... hoch.1It Iloe, .. has ,-.I.,.. of Iand ;lW1'OIIndUlI lCfmn ..... '"""'" oro. '" lPEG200.) rom.., .J) r,1e l1O ",.ku'''........ ''''..Ul.I_,emp .......... ;open '..",plal. '" ..... OK"'0; """"un: nhIfi.ld J , mo,. I... or p;oIh......... '..,lOrnlco .... ,of............. and ,fltoI limy m 138 glacier_mask_path=envi_pickfile(title=,Findglaciermask(jpeg2000 format)',filter=**.jp2') glacier_mask=read_image(glacier_mask_path) ;readglaciermask,mustbe jpeg2000format land_mask_path=envi_pickfiIe(title='Findlandmask(jpeg2000format)',firter='*.jp2') land_mask=read_image(land_mask_path) ;readglaciermask,mustbe jpeg2000format reversed_mask=reverse(glacier_mask,1) ;reversemasksoycoordinatesaretransposedandoutputis orientedcorrectly.Thisis ;necessarybecauseIDLandvelviewhavedifferentcoordinatesystemsin whichthey ;directionisreversed. glacier_image_path=envi_pickfile(title-Findglacierimage(jpeg2000 format)',filter='*.jp2') glacier_image=read_image(glacier_image_path) ;read jpeg2000imageofglacier how_big=size(glacier_image) columns=how_big[] rows=how_big[] flag_array=intarr(columns,rows);createarraytoreceiveflagvalues flag_array[*,*]=0 filled_flag_array=intarr(columns,rows) ;createarraytoreceiveenlarged,color-codedflagvalues filled_flag_array [ *,*]=() glacier_directory=widget_auto_base(title='SelectfoldercontainingSIFTEDand ACCURACYsubfolders',/xbig) output=widget_outf(glacier_directory,uvalue='output_directoryVauto,/directory,prompt= 'Selectfolder',xsize=75) directory_result=auto_wid_mng(glacier_directory) get_name=widget_auto_base(title='Enternameofglacier') output=widget_string(get_name,uvalue='glacier_name',/auto_manage) name_result=auto_wid_mng(get_name) Ii"";"_mW;..J>aLb_,'I-pkkrll,i.l.-'Flnd Ii""'''' mask (jpoJlOOD fonnoI).(olt ....&I",i_mask-"" ..... ,j, J" K1IT""'Y infonn-('SiR,"lI Vector rw.....1 ,.,-1_pon J .I,'; nfonn.boo."""'...._compl ....lillo-'Propam ....ru .... .rlnlqmoo.upkt. ",Jgcl for j"" I . m 1 do boll. If ;!I'IroI'T _"",y(lJl ;rr.,.,..}P.i]101.1- -X_cUon[Q'f! .... MtIoI If i..-. ...... l'I . j] CT 0 ..... lq,. ,'"du,pi..."".M........... -..1"' prJoU, 11. lbmw-1m. 21'10.' . I'. 41'10.4)" im:tiool[Q_ ., .......... ""i"."......_.... 0imoon"_"""yJJ LT Q . ... bttll '" 142 ;is xnegativeandynegative? printf,I I,format='(2I5,2F10.4,14,4F10.4)',imcorr_array[*,j] printf,33, format='(2I5,2F10.4,14,4F10.4,A115)', imcorr_array[*,j],runlist(i) vectors_in_glacier=vectors_in_glacier+i endif endifelseif directionEQ'sw' thenbegin if imcorr_array[,j]GT0 ANDimcorr_array["\j]LT0 thenbegin ;is xnegativeandypositive? printf,i I,format='(2I5,2F10.4,14,4F10.4)', imcorr_array[*,j]printf,33,format='(2I5, 2F10.4,14,4F10.4,A115)', imcorr_array[*,j],runlist(i) vectors_in_glacier=vectors_in_glacier+ ? endif endifelseif directionEQ'ne' thenbegin if imcorr_array[j ]LTANDimcorr_array[j ]GTthenbegin ;is xpositiveandynegative? printf,, format='(2I5,2F10.4,14,4F10.4)',imcorr_array[*,j] printf,33, format='(2I5,2F10.4,14,4F10.4,A115)', imcorr_array[*,j],runlist(i) vect or si ngl aci er=vect or si ngl aci er+ endif endifelseif directionEQ'se' thenbegin if imcorr_array[ %j]GT0 ANDimcorr_array[5,j]GT0 thenbegin ;is x positiveandypositive? printf,11, format='(2I5,2F10.4,14,4F10.4)',imcorr_array[*,j] printf,33, format='(2I5,2F10.4,14,4F10.4,A115)', imcorr_array[*,j],runlist(i) vectors_in_glacier=vectors_in_glacier+ endif endif endif endif endif endif 11 . f""",, -'(1IS.lfllH. 14. 4F10.4),. i"""""_arra}i JI "rI.,f. " . _ -,(liS. lfIO.4. 14, 4FW.4. All S). i"""""_array( JI. runlisl(i) ''1O1O(imcorr_lmIrl IOase-'IdItt,. o.,o)ldod '.,;-11.......... '... .... If 1IIL....y(iJJ[Q < ..... 6Ikdjl.....arny(i- 1:i l .j-lti l t-" 1t 1llL.....y(iJJ[Q .dono 6Ikd)1.......... y(;" I:i. l .j- l :,i' 1)al If ''''''Y(iJl [Q 1 ..... II ,....y(;JJEQI.. fiU"'UlaLomIy(,. 1:;.l J. 1,., 1-1'1 coI.....,.. 1 ... 1tqIot ;.-11'0< Ioopo I"'" .......",11. :A "m,l.. "'"" .. """ .'''''''"'"8 ,,, "CIIl!Il>""En1cr_ ,-.lilt',dl- l ,liOl- liSl, ".1 .....1.) ..lI.._"""'h.... ! o_ "",l"tion _J>OlV'IO_reoul'.po>"'''''''!'.)....,1"10.,.., .....t to ....wen offSCl$ (p",ol., 10,.Iocoty (m,),) ye .... - P"fOI"$_reoullJ"' ....""" , ] .11 ....Sj""""ol ,,,,.,.,. ,n )....... to J _displx.",enltp;.",O.i). pi .., I.ill GT ""'an_to'-d;", toIdiOp. Wi.-- WJlEII(l::otp_",,&ImI)' "fcOd_x _ disp Ia=ncn, - n' Ii"""i..._ ilDaJn_array(x,yJ ,.... ,,..],d_. _displ""crn",ul' ,yJ - . _displ"".......,'( y) fIO'ld..y_ disploc"",en'(. ,y I y _disp'"""",en' !',r J [ ,sOFOR U - n, liilocicr j'''''II"_ oolWll1tf4locier_image _ """"J ,=alo """YO ;10 IooldIItIdrd,-II"""""""''' 10he """" toplol ..,..,...,' 157 V=fltarr(glacier_image_columns,glacier_image_rows) good_total_displacement=fltarr(glacier_image_columns,glacier_image_rows) good_corr_str_array=fltarr(glacier_image_columns,glacier_image_rows) FORi=0L,glacier_image_columns-lDOBEGIN FORj =0L,glacier_image_rows-lDOBEGIN IFgood_arctan_array[i,j]NE'NaN'THENBEGIN ;if theindiceholdsalegitimatevalueandnotNaN,proceed U[i,j]=good_x_displacement[i,j] ;transferxdisplacementsintoarrayTJ' V[ij]=good_y_displacement[i,j] ;transfery displacementsintoarray' V good_total_displacement[i,j]=sqrt(U[i,j]A+V[i, j]A) ;transfertotaldisplacementintoseparatearray ENDIF ENDFOR ENDFOR FOR1 =, glacier_image_columns-lDOBEGIN ;forloopstotransfercorr.str.andparametersintonewarrays ;atlocationsof goodvectors FORj =0L,glacier_image_rows-lDOBEGIN IFABS(good_total_displacement[i,j])GT0.0THENBEGIN good_corr_str_array[i,j]=corr_str_array[i,j] ENDIF ENDFOR ENDFOR rotated_U=reverse(U,2);rotateso tiffswilldisplay properlywhen;viewedinENVI r ot at edV=reverse(V,2) U=congrid(U,glacier_image_columns*scale,glacier_image_rows* scale) V=congrid(V,glacier_image_columns* scale,glacier_image_rows*scale) v.. lIIarl'(JlKier. ..... coh .......' ....... ....... ......., JO>Od_..... .. ....rrUJaci(.) yO .MIoI(y) yl u (y) _.....IIeMI.l-.0yr.o( 0( 1)) , ( -_"" 10 n..InlcF "'"'" y _oaop-lINI(yl-)OyllMl(o(. ....... mool,ondo,.:It b) II mumar... ([_ .. .. .... (oboI. VJI>OdIy_"""')J) .... . ........'I4lmumq) .... - _Imjph (vllQOd/maxmq) r . ' anjI. ll.< !dtor .. - r '.,.(......) .. .r ': 1J< HotS .1 .. J.(ct d(i) _0( I )) ---pIob,jxO .. S .1 .. l-{.. .w.d.... d)1yd)".dJ.S 5 11 .11-(""d)1yd-" ' 4xIld)' )':I J. S ""Iao-coIorJ&vid(i) mod 1( 1)) d.)111""""1m 22.5 dcp Icnll'h ofhnd

large_num_pts=n_elements(good) loadct,13;rainbow Ip.thick= 1 if dat aori gi nEQV thenbegin ; commandstocolorcodevectorsfromIMCORRprocess.... fori=01, large_num_pts-1dobegin if good_corr_str_array[good[i]]LE2.0thenbegin !p.co/or=25;purple xO = x(good(i)mods(l)) dx=sina(i) xl=x0+dx yO =y(good(i)/s()) dy=cosa(i) yl=yO +dy xd=x_step yd=y_step plots, [x0,x 1 ,x 1 -(ct*dx/xd-st*dy/yd)*xd,$ xl,xl-(ct*dx/xd+st*dy/yd)*xd],$ [y0,y 1 ,y 1 -(ct*dy/yd+st*dx/xd)*yd,$ yl,yl-(ct*dy/yd-st*dx/xd)*yd],$ color=color,/device endifelseif good_corr_str_array[good[i]]LE4.0thenbegin \p.color=50;blue xO = x(good(i)mod s ( l ) ) dx=sina(i) xl=x0+dx y0=y(good(i)/s()) dy =cosa(i) yl=yO +dy xd=x_step yd=y_step plots,[xO,xl,xl-(ct*dx/xd-st*dy/yd)*xd,$ xl,xl-(ct*dx/xd+st*dy/yd)*xd],$ -.. ,1 .... _

Ir ",,",_oripn.. botI ............. '" 0010< o5(i) _0( 1)) cb - Ii..... ) d-.O'" ,0.Nood(il 1 0( 1)) dy - cooa(i) yl ,o dy

)'II-y_..... .. I .. H .. WlcoHl dyIyd) .oS.5 . 1.. ,-{.. dxIxd ... dylyd) .d). $ [yO,y 1 ,y 1 -(ct*dy/yd+st*dx/xd)*yd,$ yl,yl-(ct*dy/yd-st*dx/xd)*yd],$ color=color,/device endifelseif good_corr_str_array[good[i]]LE6.0thenbegin !p.co/or=100;aqua xO = x(good(i)mods(l)) dx=sina(i) xl=xO+dx yO =y(good(i)/s()) dy =cosa(i) yl=yO +dy xd=x_step yd=y_step plots,[x0,x 1 ,x 1 -(ct*dx/xd-st*dy/yd)*xd,$ xl,xl-(ct*dx/xd+st*dy/yd)*xd],$ [y0,y 1 ,y 1 -(ct*dy/yd+st*dx/xd)*yd,$ yl,yl-(ct*dy/yd-st*dx/xd)*yd],$ color=color,/device endifelseif good_corr_str_array[good[i]]LE8.0thenbegin \p.color=\50;green xO = x(good(i)mods( 1)) dx=sina(i) xl=xO+dx yO =y(good(i)/s()) dy=cosa(i) yl=yO +dy xd=x_step yd=y_step plots, [x0,x 1 ,x 1 -(ct*dx/xd-st*dy/yd)*xd,$ xl,xl-(ct*dx/xd+st*dy/yd)*xd],$ [yO,yl,yl-(ct*dy/yd+st*dx/xd)*yd,$ yl,yl-(ct*dy/yd-st*dx/xd)*yd],$ color=color,/device endifelseif good_corr_str_array[good[i]]LE10.0thenbegin !p.co/or=200;yellow xO = x(good(i)mods( I)) dx=sina(i) xl=xO+dx [)O.ylJHOI dy/)'d ... lW);d)"yd. S 11.yliOl d)'lyd ... $ """'_A\t )'I,),.1!q> plo...,(.O"I .. I-{OI d)' yd1. S __Jndonl dol ""_ of"'" OIls!< of.1I ' ..:',... " "h," Si"',r,w",nolo""," A ,,,,,,or,, "h an ""jII" "h,," ",.oro. "'" >IOndanl deI,.hm Ihn:shokl"_plotll A""", .....p"",,,,1 .,,,1 ... , ...........W.,...".,Ionc 'n ,,,....1.."'" "..... ",10 A o;m,lor ""'"II< -. ..-.J,nllO ""!Of icngoh....ell. ,Th,. f""II'M' ,0/;"' .. '"P"' ,he fik o _'oon " r"""" ..'0Orr 1) ;popu1>1.toL dul'. ,ord )"d,"1'.ak"J ... -( AT I\.s(y _d;loa '''"\>0 :lol!>ole.,lion '"" ,-ee,,,, at I of'SOO o/xe .... the$1, 00"If." ,II "'SPlay f'IO]>VI roul ' oIl0l,In! ..... :>uk!....,1,,n d" ,do (I Y I I) 1I'W .........([ ... >(.I>s("llOOdI. _.. ....I "... 1fOC>III1_ ..cpl)J) ..... - ....,.,.'1enJth (........... od(l) mod 0(I d.-.i ..... i) .1 - .{l+d, yO _ 11( 1)) dy - 4. S . 1.d.(O\ d.,.d.. S l)O.yl.yl-{CI dyI)d .. .j.O"'l .. l-/a ,w,d.. d,o'yd)".d. S .1 .. 1-{...... S ()O.yl,yl-(ctdylyd .. .t...d)"jd. S )'1.yl-(.... < J!-('....,Ian..,."1hrooho1d'. 'toto.I bokb) .........,hrnholdo_rcsu[ 'OId;>JI _IhrJIiom!,.I '"'orpola""" " 9>.9 \; ......1.H,. J, "r".""" .. bu....lI} tho{lOO formaIr.fi l' ...... &IoOtCtn,. oompl .... i. ooun,, 1 .updates po>rI,Of , ...., 0'(",, ) :an>< fOr)"I .... IF IiMOD1(0) F.Q0 I prl . """,,,I,edJ'_d"l'l..,e,,,.,,,(x,r] . - val"... pi"" ... ;n onIacl b)' 1>;ond0. .!.co.hen: '" i""'iJi.,-..' hfrt(inf,rutd _indioes[ I ..W " .. .. .. .. md comrrunok toomooth thI... .. .. .. ...... rot>ted_" _ displ""coie.rov.',' ocoJ.) ;cnloooa!"""Id,,,,,I,,,,.......,,oonjjn .... coo'Glla!,OU];oI"",os'..,..lo. """,sc.le) .....1.1 1 .r:unbow rolor lable d ..'I. d(Fod(1) _ 0( 1)) OSi'ion/O....;plM:< q ' _icc ;ploth"", defio,,,, =>!ill ."'''"''' .yosiliOfl/.)-.; 1 aNI Y"-'" P>O"""'" ""0 am) Y-",,"ilion[ nJ.oyl . ... 1;bp, JaoooaI_ ; ..J>OIIilioni UJt; wi. -(.Iope P- . pi" y)ntrn:"I" I )' 1 ,G z(.pi:< I"'!OW _di ... pi:. ..,J>OOi!io( P}+i )1IiJr-(oIopt 2' xpix..,._ iIIlaupo2)'r ., ".pi>.J-tdfrr 17......oct . pi>..)pi>: I .._rOO" di ...."d.bo( ..,JIOOition( r I' JIOSi'ioni ! I) .1op di."tay """,,,'Iy .. ben openedon"\' 1 ,,-rl.._ IIff, ditt l itl-{'n.. (P ...lIH...".."j,'...... [pi.eI.q '''IP{ I,''] ""'put .. _ NlI,(.,....."l,."" u.-'porarnoIo'-oljirruo&e.....I. 1DI / ......./ odae.....,) '''PI''1 fi", fil .... "Ii",h ., ""'"....,""'..) fill..- -",reato kemel to bo:oppl'ro ocrooo """" I... bands kernel, ;"",, I. '' I.),< 10 F' an Hluco In .....Y muf2......(f2Jnan) f2- f2 ' {Z"6Irna: .. "",..... ho< .... _".,u ..1.inheri'-inhhcd___ b.W'( title-1'. "" amCompla&< , al .....oompk1c',prompt-'S"""...folly dosInp