Vous êtes sur la page 1sur 106

SPATIALMODELLINGANDRISKASSESMENT OFSIDOARJOMUDVOLCANICFLOW

RachmanRifai February,2008

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

SPATIALMODELLINGANDRISKASSESMENT OFSIDOARJOMUDVOLCANICFLOW by RachmanRifai Thesis submitted to the Gadjah Mada University (Yogyakarta, Indonesia) and International Institute for Geoinformation Science and Earth Observation (Enschede, The Netherlands) in partial fulfillment of the requirements for the degree of Master of Science in Geoinformation for Spatial Planning and Risk Management, Specialisation: NaturalHazardStudies.

Supervisors: Dr.JununSartohadi(FirstSupervisorUGM) Prof.Dr.V.G.(Victor)Jetten(SecondSupervisorITC) Dr.DinandAlkema(ThirdSupervisorITC)

ThesisAssessmentBoard Prof.Dr.Sutikno(ChairUGM) Dr.JununSartohadi(FirstSupervisorUGM) Prof.Dr.V.G.(Victor)Jetten(SecondSupervisorITC)

Observer: Drs.TomLoran(ProgrammeDirectorITC) Advisor: Dr.Th.W.J.vanAsch(UtrechtUniversity)

GADJAHMADAUNIVERSITY,YOGYAKARTAINDONESIA INTERNATIONALINSTITUTEFORGEOINFORMATIONSCIENCEANDEARTH OBSERVATION,ENSCHEDETHENETHERLANDS

ii

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Disclaimer This document describes work undertaken as part of a programme of study at the International Institute for Geo-information Science and Earth Observation. All views and opinions expressed therein remain the sole responsibility of the author, and do not necessarily represent those of the institute.

I certify that although I may have conferred with others in preparing for this assignment, and drawn upon a range of sources cited in this work, the content of this thesisreportismyoriginalwork. Signed..

iii

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

ABSTRACT
On 29 May 2006 mud volcano occurred in SidoarjoEast Java 23 km south of Surabaya. This is a geological phenomenon due to overpressurized subsurface mud layers.Thismuderuptionappearstohavebeentriggeredbydrillingofoverpressured porousandpermeablelimestoneatadepthof2830mbelowthesurface(Davies,2007). The discharge isveryhighwitha rate of 7,000115,000 m3/day,andhasinundated 4 adjacentvillagesandsofar7000peoplehavedisplaced(UNDAC,July2006). InJune2006thegovernmentbuiltadikearoundthecentreoferuptiontoprotect thesurroundingenvironmentfromfloodandinundation.Todaymorethan12months after the start of the eruption, the mud volcano remains have high flow rates. This conditionisverydangerousbecauseifthedikebreaksorthemudovertops,awiderarea hashighriskofbeingfloodedbythemud. To better understand which areas might be at risk or being flooded due by the mud,apredictiontoolisneededtosimulatethemudflow.Simulationoftheinundation mightbeusedforpredictingwhatwillhappeninthefuturetoknowwhereinundation willtakesplaceincertaintimetocome.Ifonethenknowswhatelementsareatriskof being flooded, and if the vulnerability of the area is known a risk assessment can be done. Quantification of risk is needed as an input for preparation, evacuation, rehabilitationandreconstruction. In this study, a 1D2D hydrodynamic model using SOBEK was implemented to createmudflowmodeling.Inthemodellingapproachfivetimescenariosofmudflow was generated represents five different condition of flowing of the mud. Four time scenarioswererepresentedfourtimesthedikebreakandonetimescenariotorepresents the prediction of inundation without the dike. The scenarios have been implemented also used to test the behavior of mud flow compared with waterflood using SOBEK, sincethissoftwareisdesignedtomodelwaterflood. Themodelresultmostlygive40%agreementontheinundationextent.Thisresult was achieved by comparing the inundation extent from the model with inundation extentfromrecordedeventwhichwascapturedintheIKONOSimage.Themodelalso hasoutputofinundationdepth,buttheresultonlygive10%,thisprobablybecauseof landsubsidencewhichwasoccurredinthestudyarea.Thelandsubsidencefactorwas ignored in this study because the magnitude and direction of land subsidence have spatialandtemporalvariations. The impact assessment also be done in this study by analyzing the result of inundationextentandintegratewithelementatrisk.Theinformationofelementsatrisk wascalculatedfromtopographicmapandIKONOSimagery.

Keyword:Mudvolcano,Mudflowmodeling,SOBEK,Hydrodynamic,DTM,Kriging, Interpolation.

iv

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

ACKNOWLEDGEMENTS
Firstofall,IwanttoexpressmyhappinessbysayingAlhamdulillah,praisestoAllahfor giving me the strength and opportunity to finish my MSc study in the best education environmentatUGMandITC. IwouldliketothanktomyofficeBakosurtanal,especiallytoIbuDra.TitiekSuparwati asmysuperiorandDr.PoentodewoasthechiefofCenterofTopographicalandSpatial PlanningMappingtosupportandgiveopportunitytocontinuemystudy.Iwouldlike also to thank to Bappenas and NESO to provide the scholarship and make this dream cometrue. IwouldliketoacknowledgemylovelywifeParamita,mybelovedmotherandparents in law with the enormous strength and courage bestowed on me to make this study a success.TomychildrenTomyandIndywhobemysourcesofinspirationandstrength. To my late father who had given me the best education from elementary to tertiary education,mayAllahgivehissoulmercyandforgiveness. I deeply appreciate to my supervisors, Dr. Junun Sartohadi for gave me advices, suggestions and encourages. To Prof. Dr. V.G. Jetten who encourage me to conduct researchonthistopicandsupportmewithvaluablecomments. IappreciatewithgratitudetoDr.DinandAlkemaasmythirdsupervisorwhopatiently and tirelessly to guide me with valuable comments and suggestions during this research,especiallyduringthewritingofthethesiswhichisveryshort.Ialsofeelvery lucky because Dr. Alkema took a lot of effort to provide the necessary software and becauseofthatthisresearchfinishsuccessfully. I would like to thank to Dr. van Asch who gave me enlightenment and share his knowledge during the discussion in the Utrecht University and via email. Those discussionshavewidenedmyknowledgeofthistopic. IalsowouldliketothanktoHoebnerkCedricthestudentofUniversityofParistoshare hisvaluabledataduringhisfieldworkinIndonesia,withoutcontributionofhisdatathis researchwasimpossibletosucceed.ToDiaswhohashelpingmeduringthefieldwork andhelpedmetodigitizingthemaps. IamgratefultoallmyfriendsinthedoubledegreeMScprogram,Nugroho,Arif,Budi, Anna,Ebta,Estu,Maya,Mone,Defi,Muktaf,Safrudin,Rudy,Firda,Wulan,Utia,Dodi, PakHosenandBuLily,whohavesharedhappinessduringthisstudy.Andmyprayto lateRhinowhohaspassedawayduringthisstudy,mayAllahgivemercytohissoul.

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

TABLEOFCONTENTS
ABSTRACT................................................................................................................................... iv ACKNOWLEDGEMENTS .......................................................................................................... v TABLEOFCONTENTS.............................................................................................................. vi LISTOFFIGURES ....................................................................................................................... ix LISTOFTABLES ......................................................................................................................... xi CHAPTER1.INTRODUCTION................................................................................................. 1 1.1.Background ....................................................................................................................... 1 1.2.ProblemStatement ........................................................................................................... 1 1.3.ResearchObjectives.......................................................................................................... 2 1.4.ResearchQuestion............................................................................................................ 2 1.5.Limitationofthestudy .................................................................................................... 2 1.6.StructureoftheThesis ..................................................................................................... 3 CHAPTER2.LANDSCAPECHARACTERISTICSOFTHESTUDYAREA ....................... 4 2.1.GeographicalCharacteristics.......................................................................................... 4 2.2.Topography....................................................................................................................... 4 2.3.Geology.............................................................................................................................. 5 2.3.1.Stratigraphy ................................................................................................................ 6 2.3.2.StructuralGeology..................................................................................................... 8 2.4.Landuse ............................................................................................................................ 9 2.5.Landsubsidence ............................................................................................................. 10 CHAPTER3.LITERATUREREVIEW ..................................................................................... 12 3.1.MudvolcanoandDiapirism......................................................................................... 12 3.2.TheoriginandcauseofMudVolcanoinSidoarjo .................................................... 12 3.3.FlowClassification ......................................................................................................... 14 3.4.Rheologicalcharacteristicsofmudflow ..................................................................... 16 3.5.Mudflowmodeling ....................................................................................................... 19 3.6.SOBEKSoftware ............................................................................................................. 20 CHAPTER4.METHODOLOGY .............................................................................................. 21 4.1.Researchphase ................................................................................................................ 21 4.1.1.Preparationanddataacquisition........................................................................... 21 4.1.2.DataProcessingandModeling .............................................................................. 22 4.1.3.Fieldwork .................................................................................................................. 22 4.1.4.PostFieldwork.......................................................................................................... 23 4.1.5.Calibrationandvalidation...................................................................................... 24 4.1.6.RiskAssessment....................................................................................................... 25 4.1.7.Reporting/thesiswriting ......................................................................................... 26 4.2.Materials ........................................................................................................................... 26 4.2.1.Datarequirement ..................................................................................................... 26 4.2.2.Datacollected ........................................................................................................... 28 4.3.Software............................................................................................................................ 28 4.4.Flowchart......................................................................................................................... 29 4.5.ExpectedResult ............................................................................................................... 29

vi

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

CHAPTER5.DTMGENERATION ......................................................................................... 30 5.1.ElevationData.................................................................................................................. 30 a. Landsurveying.......................................................................................................... 31 b.Photogrammetry(aerialphotoandsatelliteimagery) ............................................. 31 c.RadarSatellite(Radarsat).............................................................................................. 32 d.Radarairborne(Insar/Ifsar) ......................................................................................... 32 e.Lidar................................................................................................................................. 32 5.2.DataSource ...................................................................................................................... 33 5.3.DigitalTerrainModelforFlowModeling ................................................................... 34 5.4.DTMInterpolationMethod ........................................................................................... 35 5.4.1.TIN(TriangulatedIrregularNetworkwithLinearInterpolation) ................... 35 5.4.2.MinimumCurvature ............................................................................................... 36 5.4.3.Kriging....................................................................................................................... 37 5.4.4.IDWInverseDistanceWeighted ........................................................................ 39 5.5.TopographyMapAccuracyassessment ...................................................................... 39 5.5.1.HorizontalAccuracy ............................................................................................... 39 5.5.2.VerticalAccuracy ..................................................................................................... 41 5.6.DTMConstruction .......................................................................................................... 43 5.7.DTMQualityAssessment .............................................................................................. 45 5.7.1.RMSE ........................................................................................................................ 46 5.7.2.Accuracyratio ......................................................................................................... 47 5.7.3.MeanandStandardDeviation .............................................................................. 47 5.7.4.AnotherDTMuncertaintymeasurement ............................................................ 47 5.7.5.ResultandDiscussion ............................................................................................ 48 5.8.Manmadeterrainmodeling(dike)............................................................................... 49 5.8.1.DTMDikeintegration ......................................................................................... 49 5.8.2.ResolutionofDTM.................................................................................................. 52 CHAPTER6.MUDFLOWMODELING ................................................................................ 53 6.1.Governingequation ....................................................................................................... 53 6.2.Software........................................................................................................................... 56 6.3.InputParameters ............................................................................................................ 57 6.4.OutputParameters......................................................................................................... 58 6.5.InitialandBoundarycondition .................................................................................... 59 6.6.Surfaceroughness .......................................................................................................... 60 6.7.Modelschematization.................................................................................................... 62 6.8.Scenarioarrangement .................................................................................................... 63 6.9.ModelingResult ............................................................................................................. 64 6.9.1.InundationExtent ................................................................................................... 64 6.9.2.DepthofInundation ............................................................................................... 67 6.9.3.Flowvelocity ........................................................................................................... 70 6.9.4.Computationtime................................................................................................... 72 6.10.ModelCalibrationandValidation ............................................................................. 72 6.11.Sensitivityanalysis....................................................................................................... 75

vii

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

CHAPTER7.IMPACTASSESSMENT .................................................................................... 78 7.1.InundationMap.............................................................................................................. 78 7.2.MaximumDepth ............................................................................................................ 80 7.3.MaximumVelocity......................................................................................................... 81 7.4.MaximumTimeofDuration......................................................................................... 82 7.5.ElementatRisk ............................................................................................................... 84 CHAPTER8.CONCLUSIONANDRECOMMENDATION ............................................... 89 8.1.Conclusion....................................................................................................................... 89 8.1.1.ConclusionfromResearchObjective ................................................................... 89 8.1.2.ConclusionfromResearchQuestion................................................................... 90 8.2.Recommendation ........................................................................................................... 91 REFERENCES ............................................................................................................................. 92

viii

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

LISTOFFIGURES
Fig21.LocationofMudflow.................................................................................................... 4 Fig22.EastJavaProvince ......................................................................................................... 4 Fig23.LocationofMudflowLandsatimagery................................................................... 4 Fig24.InundationArea............................................................................................................. 4 Fig26.GeologicalMapofSidoarjo .......................................................................................... 6 Fig28.ThedistributionofmudvolcanoinCentralandEastJava...................................... 8 Fig29.MudvolcanoinSidoarjo,EastJava............................................................................. 8 Fig210.WatukosekfaultfromLandsatimagery................................................................... 9 Fig211.TopographicalMap1996,Scale1:10.000 .................................................................. 9 Fig212.Ikonosimagery,2002................................................................................................. 10 Fig213.Landusemap(updated2002). ................................................................................. 10 Fig214.Horizontalandverticaldisplacementaffectedontheinfrastructure ................ 10 Fig215.Subsidenceinx(horizontal)direction. .................................................................... 11 Fig216.Subsidenceiny(vertical)direction......................................................................... 11 Fig31.Dashgilvolcano.(http://bulletin.gia.az/).................................................................. 12 Fig32.LokbatanMudVolcanoeruptioninAzerbaijan,2001.(www.wikipedia.org) .... 12 Fig33.TheoriginofSidoarjomudvolcano,afterDavies,etal. ........................................ 14 Fig34.ShearstressasafunctionofShearrateoffluiddeformation................................ 14 Fig35.Classificationofflowmaterialbasedonsedimentconcentration ........................ 16 Fig36.Schematicdiagramofcontinuummechanicsofmaterial ...................................... 17 Fig41.Thepicturesshowsseveralsamplepointlocation .................................................. 23 Fig42.Equipmenttomeasureviscosityvalue(viscotester) ............................................... 24 Fig43.MudinundationMap(Recordedevent) ................................................................... 25 Fig44.UpdatedmapofelementatriskusingIkonosimagery.......................................... 26 Fig45.Schematicdiagramofresearchworkflow................................................................ 29 Fig51a.DigitalSurfaceModelDSM .................................................................................... 30 Fig51b.DigitalTerrainModelDTM.................................................................................... 30 Fig52.Thediagramshowingsatellitesystemderivedelevationdata.............................. 31 Fig53.Lidarsystem,www.satimagingcorp.com ................................................................. 33 Fig54.Thecontourmapofthestudyarea ............................................................................ 33 Fig55.TriangulatedIrregularNetworkTIN ...................................................................... 36 Fig56a.Semivariogram............................................................................................................ 38 Fig56b.RelationshipNugget,SillandRange....................................................................... 38 Fig57.Thetypeofempiricalsemivariogram........................................................................ 38 Fig58a.overvaluedandundervaluedofDEMpixel. .......................................................... 43 Fig58b.DEMafterediting. ...................................................................................................... 43 Fig59.FlowchartofDTMconstruction ................................................................................ 44 Fig510a.DTMcreatedusingTIN ........................................................................................... 45 Fig510b.DTMcreatedusingMinimumCurvature............................................................. 45 Fig510c.DTMcreatedusingIDW.......................................................................................... 45 Fig510d.DTMcreatedusingKriging .................................................................................... 45 Fig511.FlowchartofintegrationDTManddike................................................................ 50

ix

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Fig512.TheDTMwithdikeandIkonosimagery ............................................................... 51 Fig513a.ProfilingfromAB .................................................................................................... 52 Fig513b.CrosssectionfromAB ............................................................................................ 52 Fig61.Thegraphicaluserinterface(GUI)inSOBEK .......................................................... 57 Fig62.InputparameterforBoundarycondition.................................................................. 58 Fig63.Inputparameterforcrosssectiondimension ........................................................... 58 Fig64.Graphicalmenuforinputtingoutputparameters .................................................. 58 Fig65.Modelsimulationtobesetascompletelydrysystem .......................................... 59 Fig66.InputparametersofBoundarycondition.................................................................. 60 Fig67.Landusemapasinputofmanningroughness ........................................................ 61 Fig68.Ikonosimageasinputforlanduseupdate............................................................... 61 Fig69.Landusemapasmanningcoefficientmap .............................................................. 61 Fig610.Schematizationofthemodel................................................................................... 63 Fig611.Flowmodelingresultforfivetimescenarios. ........................................................ 65 Fig612.Inundationextenton06June2006 ........................................................................... 66 Fig613.Inundationextentonseveraltimescenarios .......................................................... 66 Fig614.Comparisonofinundationdepthat29August2006 ............................................ 68 Fig615.Inundationdepthmapat11June2006.................................................................... 68 Fig616.Comparisonofinundationdepthat31October2006 ........................................... 69 Fig617.Inundationdepthmapat27January2006 .............................................................. 69 Fig618.Comparisonofinundationdepthat12December2006........................................ 70 Fig619.Velocitymapon11June2006 ................................................................................... 71 Fig620.Velocitymapsforfourdifferentsimulationscenarios .......................................... 71 Fig621.Buildingfootprintassurfaceroughness ................................................................. 73 Fig622.Theresultofflowmodelforcalibration.................................................................. 74 Fig623.Sensitivitygraphbetweenresolution,discharge,manningandflowvelocity.. 76 Fig624.Sensitivitygraphbetweenresolution,discharge,manninganddepthof inundation ................................................................................................................................... 76 Fig625.SensitivitygraphbetweenresolutionofDTManddepthofinundation ........... 77 Fig626.SensitivitygraphbetweenresolutionofDTMandflowvelocity........................ 77 Fig71.Theinundationsimulationscenarioin10days,20daysand30days. ................. 79 Fig72.TheInundationmap.................................................................................................... 79 Fig73.Therelationshipdepthandflowvelocitywithlevelofhazard............................. 80 Fig74.Housesinundated3050cmdepth .......................................................................... 81 Fig75.Abandonedvillage,afterfifteenmonth .................................................................... 81 Fig76.Floodhazardclassificationbasedondepthandvelocity ....................................... 82 Fig77.Thegraphshowsrelationshipbetweendepth,kineticenergiandtimeof duration........................................................................................................................................ 83 Fig78.Conditionofsettlementareafifteenmonthaftermudflowstarted ..................... 83 Fig79.Thebuildingsweretotallydamage............................................................................ 83 Fig710.Typeelementsatrisk ................................................................................................. 85 Fig711.Inundationareaisdeterminedasareaofelementatrisk..................................... 85 Fig712.ThepictureofElementatrisk................................................................................... 88

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

LISTOFTABLES
Table31.Flowclassificationbasedonconcentration,Flo2Dmanualhandbook. ......... 15 Table32.Flowresistanceterm,Naef,et.al,2006.................................................................. 18 Table41.Laboratorytestresultofmudvolcanorheology.................................................. 24 Table42.Typeofresearchdata ............................................................................................... 28 Table51.Optimumpixelsizetovariousmodelapplications(Tennakoon,2004). .......... 34 Table52.Horizontalaccuracyoftopographicalmapinseveralstandards(Maune,2001 inRahman,2006) ........................................................................................................................ 41 Table5.3.ASPRSAccuracyStandardsforlargescalemaps. ............................................... 41 Table5.4.Verticalaccuracyoftopographicalmapbasedoncontourinterval(Maune, 2001inRahman,2006) ............................................................................................................... 42 Table5.5.Flyingheight,minimumpossibleciandheightaccuracy(Petrie,1990inWeng, 2002).............................................................................................................................................. 42 Table56.Algorithmandinterpolationparametersusedinthisstudy. ............................ 44 Table5.7.MeasurementofDTMuncertainty......................................................................... 48 Tabel61.ManningcoefficientderivedfromLandusemap .............................................. 61 Table62.Listofmodelscenarios ............................................................................................ 63 Table63.Comparisonresultbetweeninundationareaandareafromthemodel........... 67 Table64.Thecomputationtimeforeachmodelingscenario ............................................. 72 Table65.Manningcoefficientvalueforcalibrationprocess............................................... 73 Table66.Manningcoefficientforsensitivityanalysis ......................................................... 75 Table71.Elementatriskoutsidethedike(Buildingtype) ................................................. 86 Table72.Elementatriskoutsidethedike(Landusetype) ................................................ 86 Table73.Elementatriskinsidethedike(Buildingtype).................................................... 87 Table74.Elementatriskinsidethedike(Landusetype)................................................... 87 Table75.Elementatriskinsidethedike(Roadtype).......................................................... 87

xi

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

CHAPTER1.INTRODUCTION
1.1.Background An eruption of steam, water and mud occurred in East Java on 29 May 2006 in Sidoarjowhichislocatedat23kmsouthofSurabaya,secondlargestcityinIndonesia. This event, known as a mud volcanic flow is a geological phenomena due to overpressurized subsurface mud layers usually associated with gas (Hensen et.al., 2006).Themudvolcanooccurredwhenmudandgasaccumulatedinseasedimentsthat weretrapped insubductionzoneswhereone tectonicplate slides under another.Mud volcanoeshaveburstoneverycontinent,butareabundantinthesouthCaspianregion (offshore and onshore Azerbaijan) and offshore Indonesia in the East Java basin (Noorden,2006).TheSidoarjomudvolcanoseemstobeahybridbetweenatypicalmud volcanoandhydrothermalventswithmudtemperaturesashighas60to70centigrade and very high concentrations of hydrogen sulphide gas. This indicates that hydrothermalactivityisgoingonatthesametime(Noorden,2006).Thismuderuption appears to have been triggered by drilling of overpressured porous and permeable limestoneatadepthof2830mbelowthesurface(Davies,2007).Thedischargeisvery highwitharateof7,000115,000m3/day),andhasinundated4adjacentvillagesandso far7000peoplehavedisplaced(UNDAC,July2006). Thehotmudflowislocatednearurbanareaswithadensepopulationandithas direct and indirect impacts to the surrounding area. The direct impact is that people havelosttheirhousesorlosttheirjobsbecausethefactorywheretheyusedtoworksank by mud, or were displaced from their own neighborhood. The indirect impact of this hazard is the reduction economic activity in the whole province because Sidoarjo is located along the main transportation line between Surabaya, the capital city of the provinceandthecitiesinsidetheprovince. InJune2006thegovernmentbuiltadikearoundthecentreoferuptiontoprotect thesurroundingenvironmentfromfloodandinundation.Todaymorethan12months after the start of the eruption, the mud volcano remains have high flow rates. This conditionisverydangerousbecauseifthedikebreaksorthemudovertopsthedike,a widerareahashighriskofbeingfloodedbythemud. 1.2.ProblemStatement It is very difficult to predict if and when the mud volcanic flow will stop or whetheritcanbestoppedatall.Manytechniqueshavebeentriedtostoptheeruptionof mud volcano but until this moment nothing was successful. The mud volcano keeps flowing.Tobetterunderstandwhichareasmightbeatriskorbeingfloodedduebythe mud,apredictiontoolisneededtosimulatethemudflow.Simulationoftheinundation mightbeusedforpredictingwhatwillhappeninthefuturetoknowwhereinundation willtakesplaceincertaintimetocome.Ifonethenknowswhatelementsareatriskof being flooded, and if the vulnerability of the area is known a risk assessment can be

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

done. Quantification of risk is needed as an input for preparation, evacuation, rehabilitationandreconstruction. Many scientists have conducted research in the field of mud flow or debris flow modeling. Many modeling approaches were developed and had their advantages and disadvantages.Thedecisiontochoosewhattypeofthemodelshouldbeuseddepends ontheconditionofthearea,thequalityofavailabledata.IncountrylikeIndonesiathe availabilityofdataandthesoftwareisamajorlimitationtoapplymostofthesemodels. 1.3.ResearchObjectives Themainobjectivesofthisresearchare: a. Toreconstructthemudflowinadynamicspatialmodelingenvironment. b. Topredictthepropagationofthemudflows,toestimatetheareathatmightbe affectedinpresentandfuture. c. Tocalibrateandvalidatethemodelbasedoncomparisonwiththeactualevent. d. To asses mud flow hazard which affected elements at risk such as settlements, buildings,roads,factoryetc. 1.4.ResearchQuestion a. What types of data are needed to build a DEM as a basis for the mud flow modeling? b. Howfartheinundationofmudflowsinthenext5,10and20years? c. Howreliableandaccuratethemodel? d. Howmanypropertiesandinfrastructurewillbeaffectedbymudflow? 1.5.Limitationofthestudy Thestudyareaposesseveralrestrictionstotheresearch: a. Theareasurroundingisdangerousduetotoxicgas(H2S),newunpredicted bubbles(explosionofmudvolcano),andhightemperaturesofthemud(70 130 centigrade). This means we cannot move or observe everywhere in the study area freely, there are some place that we cannot enter due to that situation. b. There were some human activities to control the mud flow, therefore the behavior of mud flow was not completely natural. This condition will influencetothecomparisondata. c. Highpotentialinsocialconflict,wehavetotakecaretomakeinterviewwith peoplesurrounding.

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

1.6.StructureoftheThesis Thisthesiscontains8chapters. The first chapter contains the introduction of the research, research background, researchproblems,researchobjectivesandresearchquestions. Chaptertwocontainsadetaileddescriptionofthestudyarea,theclimatecondition,the topography,thereliefandthegeology. Chapterthreecoverstheliteraturereviewofotherresearchrelatedtothisresearch. Chapterfourdescribesthemethodologyoftheresearch,andtheexpectedresult. Chapter five describes how DTM generated, integrate with the dike and the quality assessmentoftheDTM. Chapter six contains a detailed explanation of mud flow modeling using SOBEK software,modelcalibration,modelvalidationandsensitivityanalysisofthemodel. Chaptersevendescribestheimpactassessmentofmudflowtotheelementsatrisk.

Chaptereight,thischaptercontainsconclusionandrecommendations.

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

CHAPTER2.LANDSCAPECHARACTERISTICSOFTHESTUDYAREA
2.1.GeographicalCharacteristics Located23kmfromSurabaya,Sidoarjoliesfrom112.5and112.9eastlongitude to7.3and7.5degreessouthlatitude.TheSidoarjoCityistheCapitaloftheregency, situated in the lowland of the Brantas watershed, located between two river branches, Kali Porong and Kali Surabaya, therefore Sidoarjo is deltaic of Brantas watershed, and the people of this area called the city is Kota Delta or Deltaic City. The regency is bordered on the north by Surabaya municipality and Gresik regency, on the south by Pasuruan regency, on the west by Mojokerto regency and on the east by the straits of Madura.(http://www.petra.ac.id/eastjava/cities/sidoarjo/sidoarjo.htmaccessed20June2007) TheclimateconditionofSidoarjoissimilartoanotherregionsinEastJava,ithasa wet tropical climate, with an average annual rainfall of 1900 mm and a annual temperaturerangeof21to34C.

Fig21.LocationofMudflow Fig22.EastJavaProvince

Central Eruption (Vent)

Fig23.LocationofMudflowLandsatimagery

Fig24.InundationArea

2.2.Topography EastJavaProvincecanbedividedinto3physiographiczone:thesouthernzone is a plateau, the central zone is a volcanic area and northern zone is a folded area. Sidoarjoislocatedinthenorthernarea.Inthismountainousareathereare3volcanoes, whicharecalledAnjasmorogroup,andcompriseofMt.Arjuno(3,239m),Mt.Welirang

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

(3.156meter)andMt.Anjasmoro(2.277meter).Thesethreevolcanoesarestillactiveand somescientistsrelatethemudvolcanotothisvolcanicactivity.Nearbythesemountains thereisasmallhillnamedMt.Watukosekandbyinterpretingthelandsatimagethereis indicationthatthishillwasaresultoffaultandifwemakealinefarawaytothenorth thisfaultwillmeetwithmudvolcanovent. Thisstudyareaislocatedinthelowlandwhichhaselevationrangeof0to5m. This area is deltaic of Brantas watershed, so mostly the relief is flat and nearby to the coastal,thedistancetotheseais15km.

Mt.Arjuno

Mt.Penanggungan Mt.Watukosek

R.Porong BJP#1
Istadi,2006

Fig25.ThreedimensionalviewsofPorongandsurroundings.

2.3.Geology Geologically,SidoarjoislocatedintheEastJavaBasinatthesoutheastmarginof theSundaplatformwhereMesozoicbasicrockandmlangerockarefound(IAGI,2006). The East Java basin is an inverted extensional basin (Matthews and Bransden, 1995, in Daviset al.,2007).Beneaththispart ofJava there isa seriesofhalf graben that havea strikeineastwestdirection,andarefilledwithmarinemudsandcarbonates.TheJava Basin has been geologically active since the Paleogene epoch. The Basin started to become overpressurized during the Oligo_Miocene period (Osborne and Swarbrick, 1997inDavies,2007).Someoftheoverpressurizedsedimentscameuptobecomemud volcanoes, which have been observed at Sangiran Dome and Bleduk kuwu near Purwodadi200kmfromSidoarjo.

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

ThemudthatisreleasednearSidoarjois,accordingtoMatthewsandBransden, 1995(Davis2007)OverpressuredlowerMioceneclayswhichprobablyequivalenttothe Tuban or Tawun Formations (similar age to the Kujung limestone) and the Upper KalibengFormation.AlsoWatanabeandKadar,(1985,inDavies,2007)consideredthis Formationtobethesourceofthemud.

Fig26.GeologicalMapofSidoarjo

2.3.1.Stratigraphy

Kusumastuti,et.al,1999
Fig27.StratigraphyofEastJavabasin

AsdescribedbySuyoto(2006)thestratigraphicalsequenceinSidoarjo,basedon theageofformationfromoldtoyoungcanbeseenbelow:

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

1. NgimbangFormation The lowest formation (the oldest), consist layers of sandstone, debris and siltstones,formerlylocatedinashallowseaenvironmentasdemonstratedbythe dominantcontentofforaminifera.TheageofformationwasearlyOligoceneand thedominantlithologyislimestone. 2. KujungFormation This formation was deposited during the late Oligocene and similar with Ngimbang formation, this formation was deposited in a sea environment with depths between 200 500 m.This is indicated by the abundant benthos. The formationischaracterizedbynapalandclaystonesbutontheupperpartthereis limestone and reef deposit. The biota content consists also of foraminifera and seaweed. 3. PrupuhFormation ThePrupuhformationliesfromPancengheadingtotheeast,throughPaciran PalangTubanthencontinuoustothesouth.Theageofthisformationisthelate Oligocene up to the lower Miocene. The formation is characterized by dirty whitecoloredlimestoneandlightgrayishwhitecoloredbioclasticlimestone. 4. KalibengFormation TheKalibengformationthicknessvariesbetween500700mandisrevealedin theKalibengriver,Jombang.TheformationageisthelateMioceneuptomiddle partofPliocene,whichattheupperpartconsistofinterspersionsoftuffan,tuff, lapilituffandbrecciasandstone. 5. LidahFormation TheLidahformationhastwolayers,theupperpartwassedimentedinashallow seaenvironmentwithadepthof2050mandthebottompartwassedimented intheprotectedmarineenvironmentatthedepthof200300m.Theformation estimatedhasagebetweenPlioceneuptoPleistocene. 6. PucanganFormation ThePucanganformationislocatedincentralwestofKendengzonespreadoutto EasternpartofKendengzone,anditisrevealedinMt.Pucangan,Jombang.The formation consists of tuff napalan and contains mollusca, insertions of breccia, tuffan sandstone and conglomerate of sandstone. The age of this formation is PliocenetolatePliocene. 7. KabuhFormation The Kabuh formation is found in Sumberangin sub district, Jombang. This formation is famous among archaeologists because many fossil of vertebrate werefound,EugeneDuboisfoundtheskullofHomoerectusatSangiranandvon KonigswaldfoundvertebratefossilsatTrinil(Suyoto,2006).Thethicknessofthe formationisvariesfrom100mto700m,andhaveageofMidPleistocene. 8. NotopuroFormation TheformationasitsnamewasrevealedatNotopuroasmallvillageatMadiun, and has age estimated in the late Pleistocene. The thickness of the formation is

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

about 240 m and consists of tuff interspersed with tuffan sandstone, volcanic brecciaandconglomerate. 2.3.2.StructuralGeology There are several Main structures which dominated in the northern part of the EastJavaBasin(Guntoro,2006),Theseare: - AStructuralpatternontheNNESSWthatiswellexpressedinJavaseaand controlledbythefaultingofthebasicrock. - West East patterns that are well expressed on the WestEast fault series (strike slip fault), and folds with WestEast pattern are known as Kendeng fold. Inthesurroundingsofthemudvolcanoin PorongSidoarjo,canbeidentified therearelineamentheadingtoSWNEspreadoutandcontinuetotheMaduraisland whichisseparatethetwofoldedsystems.Thecentralventofmudvolcanointhestudy areaislocated200mSWofBanjarpanji1well(Kadaret.al,2006),borderedinthenorth byKendengzonegentlehillyareaandinthesouthbyaseriesofquaternaryvolcanoes. In the geological map of East Java can be identified a surface structure known as Watukosekfault,thisfaultisnearlylocatedinsamelinewithMudvolcanoinSidoarjo (Porong MV), the similarity of the pattern and direction between the fault and mud volcano were interpreted as there are any relationship between mud volcano and the faultzone(Guntoro,2006.,Kadar,et.al,2006).Thereareanotherseveralmudvolcanoes whichareliesinthesamelinewithPorongMV;PulunganMV,GunungAnyarMVand BujeltasekinMaduraisland(Kadar,et.al,2006).

SofianHadi,2006

MediacenterLusi,2006

Fig28.ThedistributionofmudvolcanoinCentral andEastJava

Fig29.MudvolcanoinSidoarjo,EastJava

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

MudVolcano

Watukosekfault

Fig210.WatukosekfaultfromLandsatimagery

2.4.Landuse The landuse information of study area was extracted from the 1996 Topographical map (RBI) with scale of 1:10.000 that updated using an Ikonos image from 2002 and through field checking. Mainly the land use of Porong subdistrict are characterizedbyhighdensityurbansettlement,industrialestatesandagriculturalland. The types of agricultural lands are paddy field, orchard (sugar cane and tobacco) and mixgardenwhichmostlylocatedinsuburbanarea. Physical infrastructure can also found to support industrial area surrounding suchastollway,bridge,factorybuilding,railwayetc.TheareaofPorongsubdistrictis oneofSidoarjoregencyareawhichintendtobebuiltbecomethelargestindustrialarea inEastJava.

Fig211.TopographicalMap1996,Scale1:10.000

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Fig212.Ikonosimagery,2002.

Fig213.Landusemap(updated2002).

2.5.Landsubsidence Land subsidence is a secondary hazard cause by the mud volcano in Sidoarjo. Thisphenomenacanbeseevisuallyontheinfrastructuresuchasthebridgeinthetoll roadthatbecomefracturedandcrackedbecauseofdisplacementofthefoundation,now thetollroadisabandoned. The cause of land subsidence is mainly because of the weight of mud volcano havebeenspew,untilOctober2006therewas6.15millioncubicsmetervolumeofmud volcano on the ground (inside the dike) and because of tectonic movement of Watukosekfaultinthearea(Mediacenterlusi,October2006,3rdedition).

Timnas,2006

Timnas,2006

Fig214.Horizontalandverticaldisplacementaffectedontheinfrastructure

ThesurveyconductedbyBandungInstituteofTechnology(ITB)concludethere werelanddisplacement in thehorizontal andvertical direction which has spatial and temporal variations in magnitude and direction (Abidin et al, 2006). According to Abidinet.al,theverticaldisplacementprobablycausebygroundrelaxationduetomud flows, mud loading, land settlement and geological structure. The result of Survey on theJuneOctober2006canbeseenbelow:

10

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Abidin,et.al,2006

Abidin,et.al,2006

Fig215.Subsidenceinx(horizontal)direction.

Fig216.Subsidenceiny(vertical)direction.

Fig217.Subsidencemeasurementbargraph(Abidin,et.al,2006)

11

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

CHAPTER3.LITERATUREREVIEW
3.1.MudvolcanoandDiapirism The term mud volcano was coined by Stewart & Davis (2006) and refers to a geologicalstructurewithaconstructionaledificeand somekindof feeder that connect thevolcanotoitssource(Daviset.al,2007).Amudvolcanoeitherontheearthsurfaceor on the bottom of the ocean is probably the result of a pressurized mud diapir called a piercement structure, and is often associated with gas and oil deposits in subduction zones.Mudvolcanoesarealsooftenassociatedwithlavavolcanoes.Theyoftenrelease hydrocarbon gases, such as methane, but sometimes also other gases helium.. (www.wikipedia.orgaccessed19june2007) To explain the triggering mechanism and evolution of mud volcanism and mud diapirismagoodunderstandingofthetectonichistoryisneeded(Sotoetal,2003.They are found in various tectonic settings, but most have been reported in accretional wedges,withcompressionaltectonicforces,suchasinBarbados(Martin&Kestner,1996 inSotoetal.,2003),Makran(Wiedickeetal.,2001,inSotoetal.,2003),EasternIndonesia (Barberetal.,1986,inSotoetal.,2003).

Fig31.Dashgilvolcano.Salsewithintensive bleedingofgas,waterandmud.Spring,2004 (http://bulletin.gia.az/)

Fig32.LokbatanMudVolcanoeruptionin Azerbaijan,2001.(www.wikipedia.org)

3.2.TheoriginandcauseofMudVolcanoinSidoarjo There are many discussion related to the cause of the mud volcano in Sidoarjo. A groupofscientistsandexpertsingeologyandgeophysicshavedifferentopinionabout whatwasthecauseofthemud,butthereare4majorhypothesesasresumebyGuntoro, 2006: - Duetodrillingactivity: This became the dominant opinion in the public. It is true that the first eruption of mud volcano is only 200 m from the drilling activity of a gas exploration,andthatthemudvolcanowaseruptedjustshortlyafterafailure ofdrillingprocess.Inthedrillingreportthedrillingproblemoccurredwhen theyreached9200feetintheKujungformation(Guntoro,2006).

12

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

DuetoYogyakartaearthquake The Yogyakarta earthquake happened 2 days before the mud volcano, at a distance of 200 km from the mud volcano. The existing of fault line surrounding area could be key factor because the energy from earthquake was released along the fault plane rocks and this resulted in a number of deformationprocesses(Guntoro,2006)andmayhavecausedaliquefactionin theKalibengformationatadepthofapproximately2000ft6000ftwhichis overpressured(Guntoro,2006) Duetomudvolcanoactivity The activity of mud volcano is generally indicated by seepage of gas, oil or salt water. This is very common in East Java especially around Sidoarjo. Therewerealsofoundtheexistenceofmethanegaswhichisoneindication of mud volcano activity. Observed data also indicate the existence of methanegasassociatedwithmudflowinthisarea(Guntoro,2006).Guntoro also conclude that the occurrence of mud flows in the study area were probablybecauseofnaturalphenomenonduetomudvolcanoactivity. DuetothepresenceofGeothermalactivity Theexistenceofheatassociatedwithmudvolcanoleadtothishypothesis.So thereisasurfacereservoirwhichiscontinuouslyrechargingthegeothermal source. Geologically in the south of mud volcano vent there is volcanic complex area, and probably this is the source of active magma (Guntoro, 2006).

ThefirsthypothesiswasalsoproposedbyDavies,et.al.,2006.Inthefigure33.a schematicthreedimensionalrepresentationofdevelopmentalstagesofthemudvolcano in Sidoarjo. The sequence of mud volcano eruption process after Davies et.al, are as follows: A. In march to may 2006 drilling activity (Banjarpanji1 well) was drilled to Kujung formation through Kalibeng formation which consisted of overpressuredmud. B. May 2006, Kujung Formation (carbonates) were penetrated and caused a kick (influx of fluid into well bore), this caused hydrofracturing of the overlying strata (probably within the Kalibeng Formation. Mud entered the wellbore and through permeable strata and fracture systems caused of entrainmentofoverpressuredmud. C. May to December 2006, entrainment of Kalibeng Formation caused a subterraneanconduittoform,thewallsofwhichundergocollapse. D. Post 2006, caldera forms around the vent, and gentle saglike subsidence of theregion.

13

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Fig33.TheoriginofSidoarjomudvolcano,after Davies,etal.

3.3.FlowClassification TheNationalResearchCouncilCommitteeonmethodologiesforpredictingmud flows(NRC,1982)wasinitiatedtodelineatebetweenhyperconcentratedsedimentflows andmasswasting(Flo2Dmanual).Thereisabroadrangebetweenflowingwaterand sedimentfromclearwatertolandslidesandthereiscontinuuminthephysicsforthem. The committee proposed four categories: water floods, mud flood, mud flows and landslides. Flowing material is considered to be Newtonian or nonNewtonian based on linear or non linear function when shear stress acting to it. The relationship between eachfunctionisdescribedbyfigure34.asfollows:

Flo2Dmanual

Fig34.ShearstressasafunctionofShearrateoffluid deformation

14

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Hyperconcentratedflowthatoccurworldwideisintherangeof20to55percent concentration by volume (Flo2D manual). Mud Flood events are defined as hyperconcentratedflowswithmorethan20%ofsedimentconcentration,anddefinedas waterfloodiflessthan20%. Mud flood may be difficult to distinguish from water floods but the fluid propertiesofmudfloodsareverydifferentwithmuchhigherdensityandviscositythan water floods (Flo2D manual). Mud flow is non Newtonian, very viscous, non homogeneous and transient flood and the characteristics of its fluid properties are significantly different. Its flow behavior is a function of fluid matrix properties, roughness,slopeandchannelgeometry,thefinesedimentsiltandclayarecontrolsthe flow properties including yield stress, density and viscosity (Flo2D manual). The behaviorofhyperconcentratedsedimentasafunctionofsedimentconcentrationcould beseenatTablebelow:
Table31.Flowclassificationbasedonconcentration,Flo2Dmanualhandbook.
MudflowBehaviorasaFunctionofSedimentConcentration Sediment Concentration FlowCharacteristics by byWeight Volume 0.650.80 0.830.91 Willnotflow;failurebyblocksliding 0.550.65 0.760.83 Blockslidingfailurewithinternaldeformationduringthe slide;slowcreeppriortofailure Flowevident;slowcreepsustainedmudflow;plastic 0.480.55 0.720.76 deformationunderitsownweight;cohesive;willnotspread onlevelsurface 0.450.48 0.690.72 Flowspreadingonlevelsurface;cohesiveflow;somemixing Flowmixeseasily;showsfluidpropertiesindeformation; 0.400.45 0.650.69 spreadsonhorizontalsurfacebutmaintainsaninclinedfluid surface;largeparticle(boulder)setting;wavesappearbut dissipaterapidly Markedsettlingofgravelsandcobbles;spreadingnearly 0.350.40 0.590.65 completeonhorizontalsurface;liquidsurfacewithtwofluid phasesappears;wavestravelonsurface 0.300.35 0.540.59 Separationofwateronsurface;wavestraveleasily;most sandandgravelhassettledoutandmovesasbedload 0.200.30 0.410.54 Distinctwaveaction;fluidsurface;allparticlesrestingonbed inquiescentfluidcondition <0.20 <0.41 Waterfloodwithconventionalsuspendedloadandbedload

Landslide

Mudflow

MudFlood

Water Flood

15

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

The Figure 35. clearly explain about the relationship between sediment concentrationbyvolume,flowclassificationandthefluidtype.

Alkema,et.al.,2007(lecturenote).

Fig35.Classificationofflowmaterialbasedonsedimentconcentration

3.4.Rheologicalcharacteristicsofmudflow Rheologyisthestudyofthedeformationandflowofmatterundertheinfluence ofanappliedstress.ThetermwascoinedbyEugeneBingham,aprofessoratLehigh University,in1920,thetermwasinspiredbyHeraclitussfamousexpressionpanta reioreverythingflows.(www.wikipedia.org) Rheologyconceptisappliedwhentheclassicaltheoryoffluidmechanicorthe concept ofelasticityand Newtonianfluids mechanics of material cannot explain or describe the mechanical behavior of fluid material. The diagram below shows the differentstatesofmaterialandtherelationwithrheologicalcharacteristics:

16

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Elasticity Solidmechanicsor strengthofmaterials Plasticity Continuum mechanics NonNewtonian fluids Fluidmechanics Newtonianfluids Rheology

Fig36.Schematicdiagramofcontinuummechanicsofmaterial,www.wikipedia.org

Differenttypeofrheologicalcharacteristicswillinfluencethetypeofflows whichwillbeapplied(Newtonian,viscous,granularetc).Otherimportantfactor toconsiderintheselectionoftheappropriatemodelsaretheapproximationand numericalmethodemployed(Arattano,et.al).Thisisalsoimportantfactorinthe different type of rheological characteristics of moving material to simulate kinematicmudanddebrisflowovercomplexterrain(Santiagoetal.,2007).This modelisbasedon2Dfinitedifferencesolutionofadepthaveragedformofthe equationofmotionforaflowcontinuum.Anothernumericalmodelalsoapplied but mostly in 1D model (Hungr, 1995; Arattano and Franzi, 2003, Naef et al., 2006.,inSantiagoetal.,2007).
Thereareseveralrheologicalparameterswhicharefrequentlyusedtomodelthe flow material like viscosity, shear stress, yield strength, cohesive strength. Other parameterstobeselecteddependonthetypeofmaterialtomodel. Mostly the discussion of rheological material deal with debris flow, because a real debris flow may show characteristics of a viscous flow, muddy or turbulent type floworgranularflowandthisflowregimescouldbedifferenteventinthesamechannel (Arattano and Franzi, 2004 in Naef, et.al, 2006). Material properties change within the waveandthelimitsbetweentheseflowsregimesaredifficulttodetermineinthefield (Naef,et.al,2006). Differencesofrheologicalcharacteristicswithinflowregimesarereflectedinthe flowresistancerelations(Naef,et.al,2006)andmaybebroadlydividedinto: - onephasemodels,eitherslurryofwaterandfinematerialortheentirefluid solidmixture(Naef,et.al,2006) - twophasemodels,bothfluidandsolidphase(BozhinskyandNazarov,2006; IversonandDenlinger,2001,inNaef,et.al,2006) - hybrid models, different layers with different flow resistance characteristics (Takahashi,2000.,inNaef,et.al,2006)

17

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Table32.Flowresistanceterm,Naef,et.al,2006.

Theconceptualmodelofmudflows(JohnsonandRodine,1984;Costa,1984;in Naef, et.al, 2006) and stony debris flow (Takahashi, 1991; in Naef, et.al, 2006) describe theflowwithproportionoffinematerialwherethecoarserparticlesdominatetheflow behavior(Naef,et.al,2006). Mudflowshavetwodifferentbehaviors,inlaminarflowtheregimeisdescribed as a Bingham fluid, but in nonlaminar flow its behavior explained by the Coulomb viscousrelation(JohnsonandRodine,1984;inNaef,et.al,2006):

Where:
=shearstress

=yieldstress =cohesivestress =soliddensity =frictionangleofthesolidmaterial =Binghamviscosity

=shearrate

18

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

3.5.Mudflowmodeling Mudflowordebrisflowisaformofhyperconcentratedsedimentflowsimilar tofloods,laharsorlandslides.Sucheventsmaybecomeadisasterwhentheyoccurr inthedensehumansettlementssuchasinurbanareas.Scientiststrytounderstand thebehaviorandcharacteristicsofthesephenomenanumericalmodeling. Many methods have been applied to simulate mud flow or debris flow, for example cellularautomatamodelhave beenused to simulate lavaflowbased ona Monte Carlo approach, the model using MAGFLOW was applied to reproduce a lava flow formed during the 2001 Etna eruption (Vicari et al., 2006), similar model were also applied using the latest developments of the cellular Automata model SCIDDICA for simulating debris flow phenomena, this model was applied to simulate of the 1992 Tessina (Italy) earth flow, the 1984 Mt. Ontake (Japan) debris avalancheanddebrisflowthatoccurredint1998atSarno(Italy)(DAmbrossioetal., 2003). AnothermethodtosimulatematerialflowispresentedbyMurtonandBiggs, (2002)whoappliedanumericalmodelofmudvolcanoesandtheirflowsinthegulf of Cadiz. A model of isostatic compensation between the mud column and the sediment overlying the mud source was used to predict the depth to the mud reservoir beneath mud volcanoes. Once erupted, the general behavior of an individual mud flow could be described and predicted using a viscous gravity current model. The model shows that conicalshaped mud volcanoes comprise multiple,superimposedradialflowsinwhichthethickness,eruptionrateandspeed of individual mud flows strongly depends on the viscosity, density and over pressure of the erupted mud. Using these parameters, the model predicts the lowermostflowswillbetheoldest,thickestandhavethegreatestlengthofrunout whiletheuppermostflowswillbetheyoungest,thinnestandshortest.Mikosetal., (2006)appliednumericalsimulationofdebrisflowstriggeredfromthestrugrockfall sourceareainwestSlovenia.Theyusedtwomathematicalmodels,onedimensional modelDEBRIF1DandtwodimensionalmodelFLO2D,usingthismodelMikoset al,werepreparedahazardmapforthevillageforpossiblenewdebrisflow. Martini,et.al.,2004usesa2D1Dhydrodynamicmodeltostudyoffloodwaves and suspended sediment transport in Brenta river, Venetto Italy. The 2D1D hydrodynamicmodelisintegrationoftwodimensionaldepthintegratedmomentum and continuity equation which are modified to take into account the bottom irregularities.Thismodelmerged2Dcomputationaldomainand1Dnetworkwhich isconsistsmallchannelandtheregulationdeviceslikeweirs,sluices,banks,pipes, waterpumpsetc.Thecoupledof2D1Dschemeisusetodescribetheeffectsofsmall channelinthemodelandhasbeenappliedsuccessfullywithanaccurateresolution flowfieldinthestudyarea.

19

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

3.6.SOBEKSoftware There are many software to model the flow material either in 1D or 2D flood modeling. Flood modeling in 1 dimensional software is used to model the flow in the channelsystem,informationonwaterdepthandflowvelocitycanbecalculatedateach cross section of the channel. But the 1D flood modeling has limitation because water flowassumedparalleltothemaindirectionofthechannel.Anexampleof1Dsoftware isHECRASdevelopedbytheUSArmyCorpsofEngineerandMIKE11bytheDanish HydraulicInstitute(DHI). A 2 dimensional flood modeling is used to model overland flow of water. This requiresaninputaboutterraintopographysuchasDTMorDSM.The2Dfloodmodel canprovideinformationaboutinundationextent,theflowvelocityandtheflooddepth. The 2D software for example are FLS, LISFLOOD, Telemac 2D, MIKE_21 and SOBEK (Rahman,2006).The2Dsoftwarearemorepopularbecausethecapabilitytosimulate floodwatersimilartorealcondition. TheSOBEKsoftwarehasbeendevelopedbyWL|DelftHydraulicsinpartnership with the National Dutch Institute of Inland Water Management and Wastewater Treatment (RIZA), and this software is the concrete result of experience from many institutions in Netherlands (SOBEK hlp). This software is combination of 1D and 2D floodmodel,thereforeSOBEKcanintegratebothchannelflowmodelingandoverland modeling.The2DfloodmodelinSOBEKhasbeendesignedtomodeloverlandflowon the initially dry system and in complex topography (Alkema et.al, 2004 in Rahman, 2006). SOBEK as 2D flood model requires DTM as an input of elevation data. This software is designed to interface with another GIS format provided by other software. SOBEKisbasedonhighperformancecomputertechnologythatmeansthesoftwarecan handlewaternetworksofanysize,bigorsmall.ThenetworkisdeterminedbyDTMas network platform and reaches, cross sections and structures. The amount of these parametersinfluencedthecomplexityandthetimeofcomputation. The DTM resolution has to determine first to achieve the optimum result. This decisionisbasedontheapplicationtocreate,forexampleinlargescaleofurbanflood modeling requires detail pixel size and high vertical accuracy, therefore the coverage areahavetoreducedtomaintainthetimecomputationinreasonablevalues. SOBEKwasdesignatedwithadvancedgraphicdisplaycapabilitysoitispossible tosuperimposethemodelresultoveramaporimageryofthearea.Thissoftwarealso havecapabilitytoanimatetheresultoffloodingandthisapplicationwillmakeanalysis offloodingeasier.

20

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

CHAPTER4.METHODOLOGY
This research mainly focuses on mud flows modeling using spatial dynamic environment and risk assessment for surrounding area. Modeling will be done using SOBEKtosimulatetheflowofmudanditsinundation,whileriskassessmentwilluse remote sensing. Remote sensing technique use to obtain the elements at risk of being affected by the mud, such as houses, roads, factories, buildings etc. The conceptual frameworkofthismethodologycanbeseeninFigure45. 4.1.Researchphase To elaborate the main objectives this research are divided into 7 phase, which describeasfollows: 4.1.1.Preparationanddataacquisition a. Literaturestudy This step will be the longest activity in the research process started from beginningtotheend,literaturestudywillbecarriedoutinordertodevelop knowledgeandresearchconcepts.Thispartwillmostlydealwithmudflows andanythingrelatedtomudflowssuchasdebrisfloworsedimentflow,the software,RiskandHazardanddynamicmodeling.Includingintheliterature study was the discussion and consultation with scientists who have experienceandknowledgeinthisstudy,suchconsultationwasdonewithDr. Th.W.J.vanAschfromFacultyofGeoscience,UtrechtUniversitywhohasa longexperienceinstudylandslideandmudflow. b. Determinationofthemodelaccuracy. Modelaccuracydependsontheterrainorelevationdatawillbeused.Inthis researchDigitalTerrainModelwillbecreatedfromthetopographicmapof 1:10.000 scale, which have contour interval of 5 meters. Accuracy of the modelwillbebasedonthisDTM. c. Datacollecting. Data collecting can be divided into two types based on the data type, first typeisdataformodelingandsecondtypeisdataforcalculatingelementat risk. Data for modeling is topographic map with scale of 1:10.000 with contourintervalof5meters,thisdatausedasinputofDTM.Thetopographic mapinthescaleof1:10.000wasthebestscaleavailableinthatarea. Secondtypeofdataissatelliteimagery,thereweremultitemporalsatellite imagery, satellite before disaster happen used to calculate the number of elementatriskwhichaffectedbymudflow.Ikonosimagerywithacquisition yearof2002usetocalculatehouses,roads,buildingetc,whileLandsatETM usedtodelineatelanduseinthesurroundingarea. Ancillary data also collected during this stage, for example characteristic of mudflowsuchasdischarge,densityandviscosity. d. Fieldworkpreparation During this stage preparation of field work will be take place, that will include what type of data will be collected, design route and location of

21

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

points observation, preparation of field equipment such as wireless GPS, PDA, laptop, camera, accommodation in the field and obtaining research permitsfromlocalgovernment. 4.1.2.DataProcessingandModeling In this research modeling of mud flow was based on original terrain or topography, because the condition of this area were not naturally anymore, government have built the dike to protect area surrounding. Potential mud flow inundation was created based on original DEM and this scenario is useful to simulateandpredicthowfartheinundationwillbeincertainperiodoftime. TheDTMcombinedwithdikeorembankmentrepresentedactualchangesof topographyinsurroundingarea,thisnewDTM(actualDTM)usedtocalibrateand validate the reliability of the model, from this result we know how high the reliabilityofthemodel. Thisstageconsistsasetofpracticalworkdescribesbelow: a. GeneratedoriginalDTM. This map represents the terrain before mud volcanic erupted. It is derived from topographic maps. This DTM will be used to simulate the potential inundation of mud flow for 1, 5, 10 or 20 next years. The gridbased DTM was generated from topographic maps at a scale 1:10.000 using krigging interpolation. The grid size was 10 meters to maintain its relief detail and accuracy. b. GeneratedactualDTM This map represents the relief changes on the study area. Man made elevationsuchasdikehasbeenbuiltinthesurroundingareatopreventthe spreadingofinundation. The reconstruction of dike was created based on several sources, Ikonos imagery from CRISP website (www.crisp.nus.edu) which are downloadable for free and have a 6 m resolution. This data was used to assess the dike width after geometrically corrected and digitized, while the report maps whichismadebygovernmentagencyisdataforthedikeheight. c. ConstructedMudFlowModelingsApotentialmudflowinundation. This modeling will simulate the inundation of mud flow without dike or embankment. d. ConstructingofMudFlowModelingtypeB(actualmudflowinundation). This modeling simulate the recent of mud flow, and will use in validation andcalibrationprocess. 4.1.3.Fieldwork Thisstagecomprisesthefieldworkactivitiesandconsistsasetofpracticalwork below: a. Observing the development of mud flow in several observation point samples.

22

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

b. Collectingdataofactualinundationareafromgovernmentauthority(spatial andtabular). c. Observing elements at risk which are affected by inundation of mud flow suchaslanduse,houses,buildings,roads,railway,riveretc. d. Collectingmudsamplesinseveralpointobservation,thenitwillbeanalyzed at laboratory to measure density, viscosity and concentration of mud and water(eitherbyweightorbyvolume).

2 3

2
Fig41.Thepicturesshowsseveralsamplepointlocation

4.1.4.PostFieldwork Thefieldworkwasconductedtwotimes,thefirstwaspreliminaryfieldworkand it was accompanied by two supervisors Drs. Michiel Damen and Dr. Junun Sartohadi.Thisfieldworkhadpurposeasfollows: - observedthestudyareaingeneral, - observedhowfartheinundationarea - checkedelementatriskwhichaffectedbythehazard - checkedthefaultinthefield The second fieldwork was conducted to collect samples of mud volcano, and thenanalyzedthesamplestomeasureviscosity,densityandconcentrationofwater inthemud.Theresultofthelaboratorytestcanbeshownbelow:

23

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Table41.Laboratorytestresultofmudvolcanorheology No. Sample 1 2 3 4 5 Viscosity(Poise) 10 8 12 14 15 Density (g/cm3) 1.3929 1.3661 1.3987 1.4042 1.4176 Watercontent (%weight) 60.9306 62.4943 60.8612 59.3063 59.0002 Temperature (Celsius) 60.6 52.3 51.3 38.1 33.8

Fig42.Equipmenttomeasureviscosityvalue(viscotester)

Duringthisstagethesecondarydatatobeanalyzedandgroupedinto5category : Imagery,picturesandphotographs HardcopyMap:monthlyinundationmapsandheightofinundationmaps DigitalMap:topography,landuse Videorecordedofthehazard Literaturesandpapers 4.1.5.Calibrationandvalidation Calibrationistodeterminethesetofparametersandtheirvaluestogetthebest results from the model and validation is to determine whether the model have a goodaccuracyornot,thisstagecanbedividedintoasetofpracticalwork:

24

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

a. Createseveralscenariosmodelwithdifferentparametersinput.Inthisstudy inputparameterissurfaceroughness.Surfaceroughnessisderivedfromland use map and gives the manning coefficient to every land use types. Several manning coefficient maps is created to be input to the model. The result modelthentobecomparedwithrecordedeventanddeterminedwhichone ofthemanningcoefficientismoresimilartotherecordedevent. b. Digitizingreliablesourcemaps(inundationmaps)fromsecondarydatasuch asreportmaps. c. Calculatingtheaccuracyandreliabilityofthemodel. Calculate the reliability of the model by comparing the condition of the modelwithactualevent.Parametertobecomparedistheareaofinundation andthedepthofthemud. Theareaofinundationandthedepthofthemodelcaneasilymeasureddirect on the model result, this two parameters then to be compared with actual eventbymeasurerecordeddataormap.Recordeddatacanbefoundonthe monitoring map from government authority for inundation events in the past, but recent event cannot be measured by direct measurement due to hightemperatureofsurfacemudandwedonothaveproperlyequipmentsto dothekindofwork. d. Evaluatingthemodel. Evaluationofthemodelwastodeterminetheadvantagesandlimitationsof themodel.

Fig43.MudinundationMap(Recordedevent)

4.1.6.RiskAssessment Thisstageoftheresearchistopredicttheconsequencesofthemudinthefuture. Thefollowingstepshavebeentaken: a. Identifying, and delineating the area of mud inundation into several zones whichhaveacertainlevelofvulnerabilitybasedonflowmodeling.

25

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

b. Interpreting satellite imagery to collect the information about land use, numberofhouses,settlementsandinfrastructurewhichcouldbeaffected. Interpretationofsatelliteimageryusedtoupdatedexistingtopographicmap which produced by Bakosurtanal. There are two type topographic map whichisusedinthisresearch,firstistopographicmapscaleof1:25.000,used tocalculatelanduseaffectedbythishazard,andsecondistopographicmap scale of 1:10.000 used to quantified elements at risk such as number of houses, buildings and roads. Topographic map scale 1:10.000 is large scale map so single building is easy to be identified and to be calculated, this is becomeaninputwhenquantifyingelementsatrisk. c. Toestimatetheconsequencesfortheelementsthatmightbeexposedtothe mud, the potential damage is estimated base on the number of elements at riskwithintheinundationcoverage.

Fig44.UpdatedmapofelementatriskusingIkonosimagery

4.1.7.Reporting/thesiswriting In this stage the whole research process start with introduction and end at conclusionandrecommendationwaswrotedown,andbecomethefinalstepofthe thesis.Reportingalsohasfunctionasatooltocommunicatetheresultoftheresearch to other scientists and also to the public. The documentation about the whole research,thedata,picturesandmapshavedocumentedbothinprintedpaperandin digitalformatstoredintheCD. 4.2.Materials 4.2.1.Datarequirement Inordertomakegoodmodelingofmudvolcanicflowitisnecessarytoprovide dataasmainmaterialingoodquality,thedatarequirementtocreatethemodeling areasfollows: Dataofelevationofthestudyarea.

26

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Elevationisprimarydatahavetobeprovided,generallyelevationisfoundon contour map or DEM, but we have to determined first on what level of accuracythecontourmaporDEMisneeded.Tocreateflowmodelingthemore accuracyoftheelevationdata(verticalaccuracy)isthebetter. Anotherfactorhavetotakeaccountiselevationdatahavetohavecapabilityto representallobjectsonthesurfacesuchasbuilding,road,trees,dikeetc.This typeofelevationdatacanbeprovidedbyLIDARorINSAR,bothtypeofdata have capability to represents all object on the surface, but off course very costly,andcountrylikeIndonesiadonothavesuchkindofdata. Dataofmanmadestructure(dike) Dike or dam usually built in area which has high risk due to flood and inundation.Thiskindoffeaturemustbedrawnorconstructedontheelevation data,thisfeaturecanbereconstructedonthemapfromsecondarydatasuchas technical drawing. But in hazard area like in the study area, an emergency workisfirstprioritythanwellplannedwork,sothedikewasbuiltfirstthan planning process, therefore very difficult to find such kind of technical drawing. The reconstruction of dike in this research was done by using data from the regularly report or by interviewing government officer in the field andalsofromIkonosimagery. Dataofhighresolutionsatellite,beforeandafter High resolution satellite was very useful to calculate element at risk in the study area both before and after the hazard, but availability of the data was becomeaproblemduetocostofthedata. Dataofmediumresolutionsatellite,beforeandafter Mediumresolutionsatellitedatawasusedtocalculateandanalyzetheimpact ofthehazardonthelanduseoverlargerarea,andveryusefultopredictland useaffectedinthenext5or10years. Dataofstructuregeology,mediumscale Geological map scale of 1:100.000 used to get overview of geological setting especiallystructuralaspectsuchasfaultorlineament. Recordedeventsuntilrecent(heightanddistributionoftheinundation) Calibrationandvalidationprocessareneeddataofpastevents.Recordeddata of past even provide information such as the spread of mud flow and the heightofinundation. Rheology of mud volcano : viscosity, density, friction, yield stress, concentrationofwaterandmud. This data can be found on the secondary data if the laboratory test has been conducted or the laboratory test conducted during the research. There is not easy to find the laboratory which has capability to test all parameters, for instance in Indonesia we could not find such laboratory during the time of research,onlyparametersofviscosity,densityandconcentrationofwaterand mudcouldbetested.

27

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

- Sample of mud from point observation well distributed from central vent to spillway. To make good quality of the model which simulated the real event similarly, thedataofmudrheologyfromeverypointinthefieldwhichwelldistributed isnecessarybutunfortunatelytheconditioninthefieldisdifferentduetohigh dangeroushazardarea,forinstanceinthenearcentralventthetemperatureis reach 100 degrees centigrade and event more and has high concentration of sulphurintheair.Duetothatconditionpeoplewasprohibitedtoenterinside 500mradiusfromcentralventunlesshavesufficientsafetyequipment. 4.2.2.Datacollected During the fieldwork and post fieldwork the data necessary for this research havebeencollectedfromvarioussources,therearethreetypesource ofdata,from government offices, from private organization and from laboratory work. The type datahavebeencollectedcanbeseeninthetablebelow:
Table42.Typeofresearchdata
No Typeofdataandmap Specification TopographicMap 1 1:10.000 TopographicMap 2 1:25.000 3 IKONOSimage,2002 DigitalFormat DigitalFormat Sources Bakosurtanal Bakosurtanal Status Aim/Purpose

Ready DTM,BaseMap,Elementatrisk Ready Landcover,Manningcoeficient LandCoverorLandUsebefore disaster,Largescaledetail LandCoverorLandUseafter disaster,Mediumscale

DigitalFormat

Bakosurtanal

Ready

4 AsterImage,2006 5 GeologicalMap 6 LandUsemap

DigitalFormat Scanned DigitalFormat

Bakosurtanal DinasESDM Jatim Updated Ikonosimagery Updated Ikonosimagery Laboratorytest DinasESDM Jatim

Ready

Ready Geologicalsetting Ready Elementsatrisk

7 ElementatRiskmap

DigitalFormat

Ready Elementsatrisk

8 RheologyofMud

Tabular

Ready Modelinginputparameter

9 GeologicalMap

Scanned

Ready Geologicalsetting

4.3.Software The software necessary for this research are SOBEK, Ilwis and Arc GIS. SOBEK is to be used to create the model, Ilwis is used to process satellite imagery andsupportSOBEKprogram,ArcGISisusedtoprocessvectordata.

28

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

4.4.Flowchart

Fig45.Schematicdiagramofresearchworkflow.

4.5.ExpectedResult Expectedoutcomesoftheresearchare: a. Adynamicsimulationofthemudflowinthestudyarea b. Evaluationaboutthereliabilityofthemodel c. Inundationmapofmudflow d. ElementatRiskMap e. Anassessmentofthemudvolcanicflowinthestudyarea f. Thesisreport

29

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

CHAPTER5.DTMGENERATION
5.1.ElevationData Elevation data can be represented in various ways such as a contour line, spot heightorDigitalElevationModel.Mostlyelevationdataisrepresentedintheformof DigitalElevationModelthatasstatedbyMaune,(2001inRahman,2006),isthedigital form of the terrain or elevation data to be used, manipulated or analyzed displayed directlybycomputerin3D. TheDEMorDigitalElevationModelisa3Drepresentationofelevationvalues, suchastopographicalelevation,groundwatertable,ageologicallayer,etc.Itisageneral termanditincludesthetermsDTMandDSM. TheDTMorDigitalTerrainModelisa3Drepresentationofthenaturalterrainor groundsurface. TheDSMorDigitalSurfaceModelisa3Drepresentationofthenaturalandnon naturalterrain,itthusincludesmanmadeobjectslikeembankments,dikes,buildings, etc. TheDEMandDTMapplicationareasfollows(www.wikipedia.com): Geometriccorrectionorrectificationofaerialphotographyandsatelliteimagery Modelingwaterflowormassmovement(floodsorlandslide) 3Dvisualization Reliefmapandphysicalmodels Terrainreductionofgravimetrymeasurement Extractingterrainparameters Terrainanalyses(physicalgeography,geology,geomorphology)

www.intermap.com

www.intermap.com

Fig51a.DigitalSurfaceModelDSM

Fig51b.DigitalTerrainModelDTM

The important factor of using DEM or DTM is how accurate elevation at every pixel (absolute accuracy) and how accurately the morphology is represented (relative accuracy).Theabsoluteandrelativeaccuracyaredependingonseveralfactors(Hengl,T et.al,2003): Terrainroughness Pixelsizeorgridspacing Verticalresolution Samplingdensity(elevationdatacollectionmethod) Interpolationalgorithm Terrainanalysisalgorithm

30

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

DigitalElevation/Terrainmodelisavailablethroughseveraltypesofacquisitiondata: a. Landsurveying Landsurveyingorknownassurveyingisthetechniqueandsciencetomeasure the position of an object on the earth surface in three dimensions, and usually used to establish land maps and boundaries for ownership (www.wikipedia.org). This technique has been developed since the beginning of recorded history or about 5000 years ago, the application of surveying is widespread for example construction, transportation,buildingandmapping. Inmapping,surveyingisusedtoacquiretheelevationdataforcreatingaDTM. Elevation data obtained from this technique is the most accurate with accuracy ranges betweencentimeterstomillimeters.Butsincetheoperationofthistechniqueisrelieson ahumanoperator,itstimeconsumingandonlycoveringasmallarea,evenwhenuse theautomaticDGPS.Thusthedisadvantageofthistechniqueisthatitiscostandtime consuming. For the detail application over a small area such as engineering this techniqueisthebestchoice. b.Photogrammetry(aerialphotoandsatelliteimagery) Photogrammetryisscienceofmeasuringtheearthsurfaceandgeometricobject propertieswhicharedeterminedfromphotographicimages.Historicallythistechnique wasdevelopedsincemidnineteenthcentury.Throughthistechniquethreedimensional coordinates of point an object determined by measurements from at least two photographicimagestakenfromdifferentpositions(www.wikipedia.com). Remote sensing techniques, either from airborne (aerial photo) or space borne (satellite imagery), have been used to provide elevation data. DTM generation from remotelysensedimageryhasbeenraisedsincemanysatelliteremotesensinghavebeen launched and operational, there are SPOT5, ASTER, EOS1, CartoSat and ORBVIEW. Theuseofthissatelliteimagerybecauseprovideanexcellentsourcewithlowcostand fairly accurate resolution and coverage a large area, although this technique cannot penetratecloudcoveranddependsonweathercondition.Theaccuracyofthistechnique rangingfromsubmeter(aerialphoto)toseveralmeter(satelliteimagery)(Hutchinson, 1999inTennakoon,2004).TheDTMfromaerialphotographswithsubmeteraccuracyis providegooddataforfloodmodellingbutDTMfromsatelliteimageryhasnotsufficient accuracyfordetailedurbanapplication(Tennakoon,2004).

www.geoeye.com

Fig52.Thediagramshowingsatellitesystemderivedelevationdata

31

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Photogrammetry technique is commonly used in earth science such as topographic mapping, terrain analysis, geomorphology and geology. Nowadays this techniquealsousedinmanyotherfieldssuchasarchitecture,engineering,manufacture, archaeologistsandsoon. c.RadarSatellite(Radarsat) StereoradarorSyntheticApertureRadarisaradargrammetrytechniquesimilar to photogammetry. Photogrammetry is based on optical sensors or passive systems, whileradargrammetryisbasedonaradarsensororactivesystems.Theactivesystem likeradarbasedsensoremitsitsownsignalfromsatelliteandrecordsthesignalreturn reflectedbyobjectonthegroundsurface.Thepassivesystemslikeopticalbasedsensor hastheenergyfromothersourcessuchaslightfromthesun.Thereforetheadvantages ofRadardataisthatitpenetratesthroughcloudcoveranditcanacquiredatabothday and night. Radarsat uses two passes of a radar satellite to generate a digital elevation mapwhichhasresolutionabouttenmeters. The important parameter in Radar is Doppler Centroid which is the basic principleoftheDEMgenerationfromradarimage.TheDopplercentroidgivesthelook angleandtheslantrangewhichhasdistancefromsensortoobjectontheground.These parameterstogetherwithorbitmodelmakepossibletoderivethreedimensionalvector position of object measured. By implementing geodetic equations, this position can be transformedintomapcoordinates(www.radarsat.com). d.Radarairborne(Insar/Ifsar) Terrain mapping from airborne radar is using interferometry technique called Interferometrysyntheticapertureradar.Interferometryasmeasurementtechniquewas developed by Michelson and Morley in 1879 (Tennant, 2000), and since then Interferometry has been used in scientific investigation from space astronomy to differential GPS techniques. For surface terrain observances Interferometry SAR was appliedtotheplanetVenusbutthefirstexperimenttomeasureterrainelevationofthe earthwasreportedbyGrahamin1974,andafterthatittook10moreyearsbeforeitwas usedextensivelyfornonmilitaryapplication.Variousgroupsandresearchagenciesare developed this technique including the US Jet Propulsion Laboratory (JPL), the CanadianCentreforRemoteSensing(CCRS),theEnvironmentalInstituteofMichigan, SandiaLaboratoriesandothers(Tenant,2000). e.Lidar LIDAR stands for Light Detection and Ranging and it is the technology to measure earth surface elevation using pulses of laser light striking the surfaces of the Earth and measuring the time of pulse return. The laser scanner and sensor are mountedinthebottomofanairplane(similarwithaerialcamera)withanIMUinertial measuringunitandGPS(www.satimagingcorp.com).Therecordeddatawillbeusedto measurethetimebetweentheemissionofapulsebythelaseranditsreturnbacktothe sensor to calculate the distance to the terrain surface. This mapping process requires

32

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

groundbaselocationsurveyorGCPmeasurementsimilarwithotherphotogrammetric mapping.Butdifferentwithsurveyswhichuseanyopticalsensor,lidarmappingcanbe donedayornight.

Fig53.Lidarsystem,www.satimagingcorp.com

The Lidar systems also have capability to acquire data of intensity reflectance, thevalueofreflectanceofeachsurfacefeatureswillbedifferentandthiscalledaslidar intensity.Thisintensitycanbefurtherprocessedtoproducedageoreferencedrasterfile andsimilarwithconventionalimageandveryusefultoidentificationoflandcoveron thesurface. 5.2.DataSource The elevation data was derived from topographic map scale 1:10.000 produced by Bakosurtanal, an Indonesian Government Agency for surveying and mapping. The topographic map was produced in 1999 using interferometry synthetic aperture radar dataacquiredin1997,andupdatedusingaerialphotographscale1:30.000forurbanarea and field survey carried out in 1999. This mapping was a part of National spatial planningmappingprojectin6Provinces. The area that were covered in this project were Puncak Bogor and the area surroundingSurabayacityincludingSidoarjo.Thespatialresolutionofthisimagewas 2.5mhorizontaland1.25mvertical.

Fig54.Thecontourmapofthestudyarea

33

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

5.3.DigitalTerrainModelforFlowModeling DTMcanbegeneratedusingvarioussourceofdatasuchasdirectsurvey,optical basedelevationdata,radarorLidar.Theuseofthetypeofelevationdataisdependon the application which intend to create, for flood modeling highly accurate and recent topographic data should be used,but this is notalwaysready due tobudget andtime constraint(Sanders,2007).PrimarilyinfloodmodelingDTMsareusedtoparameterize analgorithmof2Dhydrodynamicsimulationandcomparedwithobservedconditionon thefieldorwithfloodreportandpublishedfloodmaps(Sanders,2007). Highly accurate DTM is required for flood modeling in urban areas because manydetailofinfrastructureandbuildingsuchasroadelevation,curbandobstaclecan be captured. Such structural elements are playing important role in controlling the behavior of water flow (Rahman, 2006). Building as solid blocks will make overestimation on flood extent and depth, but treated building as rough surface could makeunderestimationoffloodextentanddepth(Rahman,2006).Theurbanareawhich have10percentormoreofbuiltupareawillhavesignificanteffectinfloodextent(El Ashmawy2003,inRahman,2006). DTMresolutionisdeterminedbasedontheterrainorlandscape.Inlowreliefthe impacts of the resolution is less but in high relief the impact is significant (Anderson et.alinRahman,2006).ThereforethedeterminationofDTMaccuracyforfloodmodeling isdependsonthereliefofstudyarea.Terraincharacteristicsalsohaveimportantrolein thedeterminationofelevationdatatype,forexampleintherelativelyflatterrainSRTM hasbetterverticalaccuracycomparedwithhighterrain,andinterferometrySARwith1 m vertical accuracy for flood modeling purposes need further processing especially in roughterrainandvegetatedareas(Sanders,2007). Sanders 2007, showed that DEMs which based on airborne LIDAR has the best resultforfloodmodelingapplicationbecauseLIDARhashighverticalaccuracy(0.1m) andtheabilitytoseparatebareearth,builtupareaandvegetation,whileDEMbasedon airborneIfsarneedfurtherprocessingbecauseofsufferfromnoiseandspecklealthough havehighresolution(1m).
Table51.Optimumpixelsizetovariousmodelapplications(Tennakoon,2004).

Application Floodhazardmappinginurbanareas Detailed studies related to velocity measurments,sedimentationanderosion Studies (EIA) relating to individual structuresorreclamationofsmalllands SmallscaleEIAstudiesinregionallevelfor investigating effects on new road/small scalereclamationprojects. DTM Res.5m Adequate DTM Res.7.5m Adequate DTM Res.10m Adequate Adequate

34

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

5.4.DTMInterpolationMethod Generally the elevation data at every location is not readily available. For example, the height of the terrain has to be measured before it can be used, the completenessofthedataoverentireareaisthebestconditionbutitisusuallydifficultor expensive.Therequireddatacanbecollectbymeasurementatsamplelocationandthen various interpolation method can be employed to fill the gaps at unsampled location (Demirham,2003).Thechoiceofinterpolationmethoddependsontheconditionofthe terrainandlandscapeofthesurroundingarea.Surfaceinterpolationoftheterraincreate acontinuousorpredictionsurfacefromsampledpointvalues,thisisthebasicprinciple ofDEMgeneration. Interpolation methods also known as Spatial Interpolation are based on the principle that closer locations will have more similar elevation than those far apart (Demirhan,2003).Thedifferencesinelevationvaluesfrompointtopointarerelatedto their distance, and in the interpolation algorithms the points which have the closest location will have the biggest influence (Rahman, 2006). Although different interpolationmethodsarebasedonthiscommonprinciple,theDEMresultedfromeach methodwillhavedifferentcharacteristicsandaccuracies(Leberl,1973inKidner2003). The DEM interpolation method can be classified into two types : smoothing effect and proximity effect. The smoothing effect is divided into two groups exact and approximate.Alsoproximitycanbedividedintotwogroups;globalandlocal(Shresta, 2003).Anexactinterpolatorpreservesthevalueofthesourcedataintheoutputmap,an approximateinterpolatorestimatesallvaluesoftheoutputmaponthevalueofseveral surrounding points. There are many of the interpolation methods that have been developed, nearest neighbor, moving average, triangular irregular network with linier interpolation,inversedistanceweighting,kriggingandmanymore. The different interpolation approaches should be implemented for different terraincharacteristicsbecausethereisnothebestmethodwhichisclearlysuperiorand appropriate to all surface and landscape condition (Maune, 2001, Shresta, 2003). HoweverthequalityofDEMdependsonhowsmoothnessandtensioncanbedescribed and stream and ridges can be incorporated (Mitas and Mitasova in Shresta 2003). Therefore every terrain characteristics have to be determined with different parameter value.Theflatterrainwillneeddifferentparameterwithhillyormountainousterrain. 5.4.1.TIN(TriangulatedIrregularNetworkwithLinearInterpolation) Triangular Irregular Networks or TIN have been widely used by the GIS community as digital means of representing surface morphology. A TIN is a sets of digitalgeographicdatawhichconstructedbynonoverlappingtrianglescomputedfrom irregularly spaced points that have x, y and z values. TINs are based on triangular element or facets with vertices at the sample points (Moore, et.al, 1991 in Wilson and Gallant,2000),andusuallyusingDelauneytriangulation(WeibelandHeller,2001).The TINs edges form contiguous, nonoverlapping triangular facets and can be used to capture the linier features that have important role such as fault, ridge and stream courses(Moore,et.al1991inWilsonandGallant,2000).

35

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

The DEM generated using TIN interpolation can represent sudden changes in reliefandthatisveryimportantinthemorphotectonicinvestigation,thereforeTINhas advantagesinmorphotectonicanalysis(McCullagh,1988inJordan,2007).TINsarealso very efficient in representing area with uniform slopes therefore very useful for identificationoftectonicallyinducederosionmorphologyofspursandpediment(Riley andMoore,1993inJordan2007). TINsareusuallyusedforhighprecisionmodelingin small areas, such as in engineering application because they allow calculation of planimetric area, surface and volume. But TINs have disadvantages since this method requires a large amount of high accuracy data which are expensive and probably not available(Maune,2001,inRahman,2006)


Fig55.TriangulatedIrregularNetworkTIN

5.4.2.MinimumCurvature Theminimumcurvatureinterpolationiswidelyusedintheearthsciences,ithas been proposed by Briggs (1974) for automatic contouring of geophysical data (Demirhan, 2003). It is not an exact interpolator, minimum curvature generates the smoothestpossibleterrainwhileattemptingtohonourthedataascloseaspossibleand replaces the observation value by the interpolated value in the same location. The minimum curvature interpolation generates a surface similar to a thin linearly elastic platepassingthrougheachoftheinputpointswithaminimumamountofbending.It generates the smooth surface by repeatedly applying an equation over the grid until either the maximum number of iterations is reached or the values are less than the maximumresidual. Demirhan,2003showedtheequationofminimumcurvatureisasfollows, ford=2twodimensionalcubicspline,thecurvatureisobtainedintermofobservation pointvalues f ( xi , y j ) ,i=1I;j=1J whereIandJarethenumberofgridsinthexandydirection hisregulargridspacingofobservation.

36

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Thetotalsquaredcurvature:

C = i

C(x , y )
j i j

where,

C(xi , yj ) = f (xi , yj+1) + f (xi , yj1) + f (xi1, yj ) + f (xi+1, yj ) 4f (xi , yj ) / h2


f (.) denotestheobservedvalueifitsargument,xandycoincidewithlocationof
observeddata There are several parameters that have to be determined when using minimum curvatureinterpolationmethodasdescribedinSurferhelp: InternalandBoundaryTension This parameter used to control the number of this bowing on the interior, the higherthelessthebowing,forexampleahighervalueoftensionmakesareasin theobservationlooklikefacetsofgemstone.Thevaluesofinternaltensionand boundarytensionarerangingfrom0to1. Relaxationfactor The algorithm of minimum curvature is using a successive overrelaxation algorithm to solve the specified partial differential equation. The interior is updatedusingchessboardstrategyasdiscussedinPress,et.al(1988),thehigher values of relaxation factor the faster algorithms run the lower will cause the algorithmrunslower,thevaluestobedeterminedbytrialanderror. Convergence Convergencefactorissimilarwithmaximumresidual. 5.4.3.Kriging Kriging is originated by Krige in 1951 and developed by Matheron (1971) is an advanced geostatistical procedure that generates an estimated surface from a scattered set of points with zvalues (Demirham, 2003). This interpolation method is a complex procedure that requires knowledge about spatial statistics before using this method. Krigingisbasedontheregionalizedvariabletheorythatassumes,thespatialvariation inthedatarepresentedbyzvaluesisstatisticallyhomogenousintheentiresurfaceand thedistancebetweeneverysamplepointsreflectsaspatialcorrelationandcanbeused todescribethevariationinthesurface(arcgishlp).Inthismethodthemultistepsprocess havetobedone: exploratorystatisticalanalysisofthedata variogrammodeling creatingthesurface The spatial variation is quantified using semivariogram which is estimated by thesamplevariogramthroughcomputationoftheinputpointdataset.Thevariogramis

37

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

a function to analysed and expressed the spatial variability and estimated by the followingequation:

(h) =

1 2n

{Z ( xi ) Z ( xi + h )}
n i =1

Where (h) =semivarianceatdistanceh nisthenumberofobservationseparatedbyadistanceh Z (x) =elevationvalueatpositioni

Z ( xi + h) =elevationvalueathdistancefromi

Fig56a.Semivariogram

Fig56b.RelationshipNugget,SillandRange

The semivariogram depicts the spatial autocorrelation of the measured sample point, measured points which has close to each other will have a smaller difference squaredthanthosefartherapart,thebasicprincipleingeography.Inkriggingwehave todeterminerange,sillandnuggettodescribethemodel.

Linear

Gaussian

Exponential Spherical Fig57.Thetypeofempiricalsemivariogram

38

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

5.4.4.IDWInverseDistanceWeighted The inverse distance weighted interpolation method or also known as Inverse distancetoapowerisweightedaverageinterpolatoranditcanbeanexactorsmoothing interpolator. The basic principle in inverse distance weighted is the points that are locatedclosetooneanotheraremorealikethanthepointsthatlocationfarther.Inthis interpolation dataareweightedduring the processand this willmake the influence of one point to another decline with distance from grid node. The weighting value will control how the weighting factors drop off when distance to a grid node is increase. (ArcGisHelp). Inversedistanceweightedinterpolationusedthefollowingequation:

Zj =

h
i =1 n ij

Zi 1
ij

h
i =1

2 hij = d ij + 2

Where:

hij
) Zj

zi d ij

istheeffectiveseparationdistancebetweengrid nodejandtheneighboringpointi istheinterpolatedvalueforgridnodej aretheneighboringpoints isthedistancebetweenthegridnodejandthe neighboringpointi istheweightingpower isthesmootingparameter

5.5.TopographyMapAccuracyassessment 5.5.1.HorizontalAccuracy Horizontal accuracy standard of USNMAS or US National Map Accuracy Standard which were published in 1947 was the guideline of horizontal accuracy topographic map and because at that time the computer has not been widely used as today,theaccuracystandardwasforpapermap.Itwasstatedthatformapswhichhas scalelargerthan1:20.000hastohad10percentofthepointstestedmorethan1/30inch, andformapsinscaleof1:20.000orsmaller1/50inch,bothinpublicationscale.

39

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

FordigitalmapsthereareanewguidelinehasbeenpublishedcalledNSSDA NationalStandardforSpatialDataAccuracy.Inthisguideline,thehorizontalaccuracy fortopographicmapcanbedescribebelow, RMSEx= [3(x data, i - x check, i ) 2 /n] RMSEy= [3(y data, i - y check, i ) 2 /n] where: xdata,i,ydata,iarecheckpointcoordinatesoftheithinthedataset xcheck,i,ycheck,iarecheckpointcoordinatesoftheithintheindependentsourceofhigher accuracy nisthetotalnumberofcheckpoint iisanintegerrangingfrom1ton HorizontalRMSEis: RMSEr= [3((x
data, i

-x

check, i

) 2 + (y data, i - y check, i ) 2 )/n]

2 2 = RMSE x + RMSE y

WhenRMSEx=RMSEy

If RMSE x = RMSE y
2 RMSEr = (2 RMSE x2 ) = (2 RMSE y )

= 1.4142 RMSE x = 1.4142 RMSE y

Based on assumption that the systematic error is eliminated, that the error is normallydistributedandindependentineachxandydirection.Thefactorof2.4477is used to calculate horizontal accuracy at the level of 95% confidence (Greenwalt and Schultz,1968).Thehorizontalaccuracy(NSSDA)iscomputedusingtheformula: Accuracyr = 2.477 RMSE x = 2.477 RMSE y

= 2.477 RMSEr 1.4142 Accuracyr = 1.7308 RMSEr


When RMSE min/RMSEmax is in the range of 0.6 and 1.0, circular standard error at confidencelevelof39.5%maybeapproximatedas0.5*(RMSEx+RMSEy)(Greenwaltand Schultz,1968).Thehorizontalaccuracyvalueshouldbebasedonformula: Accuracyr ~ 2.4477 0.5 RMSE x + RMSE y

40

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Table52.Horizontalaccuracyoftopographicalmapinseveralstandards(Maune,2001in Rahman,2006) Mapscale (NMAS) 1:1,200 1:2,400 1:4,800 1:6,000 1:12,000 1:24,000 NMASCMAS 90%confidence level(m) 1.006 2.042 4.054 5.090 10.150 12.192 RMSE(NSSDA) (m) 0.671 1.341 2.682 3.353 6.706 8.016 NSSDA Accuracy95% confidencelevel(m) 1.158 2.316 4.633 5.791 11.582 13.899


Table5.3.ASPRSAccuracyStandardsforlargescalemaps.

Mapscale 1:50 1:100 1:200 1:500 1:1,000 1:2,000 1:4,000 1:5,000 1:10,000 1:20,000 Class1Planimetricaccuracy LimitingRMSE(m) 0.0125 0.025 0.050 0.125 0.25 0.5 1.0 1.25 2.5 5.0

5.5.2.VerticalAccuracy NMAS guideline stated that vertical accuracy of topographic map shall be not morethan10percentoftheelevationtestedhasonehalfthecontourinterval,thisisfor papermap.Fordigitaltopographicmap,verticalaccuracyisdescribeas: RMSEz= [3 (z data, i - z check, i ) 2 /n] where: zdata,i,zdata,iarecheckpointofelevationcoordinatesoftheithinthedataset zcheck,i,zcheck,iarecheckpointofelevationcoordinatesoftheithintheindependent sourceofhigheraccuracy nisthetotalnumberofcheckpoint iisanintegerrangingfrom1ton

41

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Based on the assumption that systematic errors have been eliminated and vertical errors are normaly distributed, a factor of 1.9600 is used to calculate vertical accuracyatthelevelof95%confidence(GreenwaltandSchultz,1968).Verticalaccuracy (NSSDA)computedusingformula: Accuracyz=1.9600xRMSEz For topographic maps originated by photogrammetric technique, Weng (2002) described that there are four data sampling pattern to obtain elevation measurements; systematic,random,compositesamplingandcontouring.Systematicsamplingisagrid basedspotheightmeasurementinregularpattern,randomsamplingmeasuresheights atsignificantpointssuchasalongthestream,breaklineorhilltop.Compositesampling combinedbothsampling.Thelastsamplingpatterniscontouringwhichsystematically measures contours in the whole area (Weng, 2002), this mode of measurement yields significantly lower accuracy than the first three method (Petrie, 1990; Shearer, 1990 in Weng, 2002) but this method can be used to calculate the vertical accuracy of topographicmap.TheverticalRMSEcanbecomputedasfollows: 1.645RMSE=contourinterval/2 And, RMSE=0.304*contourinterval
Table5.4.Verticalaccuracyoftopographicalmapbasedoncontourinterval(Maune,2001in Rahman,2006)

NMAS Eq.Contourinterval (m) 0.305 0.610 1.219 1.524 3.048 NMASVMAS 90%confidence level(m) 0.152 0.305 0.610 0.762 1.524 RMSE(NSSDA) (m) 0.091 0.183 0.366 0.457 0.914 NSSDA Accuracy95% confidencelevel(m) 0.183 0.366 0.732 0.914 1.829

Table5.5.Flyingheight,minimumpossibleciandheightaccuracy(Petrie,1990inWeng, 2002) FlyingHeight(m) 1,000 5,000 10,000 15,000 Minimumpossiblecontour interval(m) 0.5~1.0 2.5~5.0 5~10.0 7.5~15.0 RMSE(90%) 0.25~5.0 1.25~2.5 2.5~5.0 3.75~7.5

42

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

5.6.DTMConstruction InthisresearchDTMwasconstructedin4workingstep:datapreparationand editing, interpolation, DTM quality assessment and DTM dike integration. The data sourceoftheDTMisthedigitalcontourmapfromthetopographicmapscale1:10.000 with contour interval 5 m mapsheet : 160813C4,C5,D5,D4,E4, E5; 160814A4,A5; 1608 41C1,D1,E1;160842a1,publishedbyBakosurtanalin1999.Asshowninthefig54.we can see the study area and surrounding is located in relatively flat terrain and from contourlinesthattheheightdifferenceoftheterrainisbelow5m.Wecanalsofindthat thespotheightsinthisareaarewelldistributed.Thedifferencevalueofeachspotheight isintwonumberofdecimalfractionitsmeanthedifferenceheightfromeachspotheight are in order of centimeter. This indicated that elevation data from the study area are adequatetobuildtheDTMasrefertotable51. Inthedatapreparationandeditingtheselectionoftopographicdatawasdone, Severaltopographicmapsasmentionpreviously,havebeenselectedtocoverthestudy area,thecoverageissurroundingofthemudvolcanoventandtheareawhereestimated willhaveinundatedbymudflow.Inthisstepeditingprocesswasdone,thecontourline firsttransformedintoDTMbyimplementedsimpleinterpolationmethodsuchasTINs, and the result DTM then to be checked whether there was an overvalued or undervalued elevation pixel. Fig 58a. below show overvalued and undervalued pixel becauseanerrorinthecontourdatabase.Thecorrectionsonthecontourdatabasehave tobetakentorestorethecorrectvalue.
Undervalued pixel

Overvalued pixel

Fig58a.overvaluedandundervaluedofDEM pixel. Fig58b.DEMafterediting.

Interpolation of elevation data was using four type method which available in the DTM software. There are Triangulated Irregular Network with linier interpolation (TIN), Minimum curvature, Inverse distance weighted and Kriging with linier variogram.Theparameterswhichareusedinthisstepareshownintablebelow:

43

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Table56.Algorithmandinterpolationparametersusedinthisstudy.
Algorithms TINwithliniearinterpolation Minimumcurvature Interpolationparameters Anisotropyratio=1;Anisotropyangle=0 Maxresidual=1.7;Maxiteration=100000 Internaltension=0;Boundarytension=0 Relaxationfactor=1;Anisotropyratio=1 Weightingpower=2;Smoothingfactor=0; Anisotropyratio=1;Anisotropyangle=0 Anisotropyratio=1;Anisotropyangle=0; Variogramslope=0.0298;Nuggeteffect: Errorvariance=3090;Microvariance=0; Drifttype:Nodrift

InverseDistanceWeighted Kriggingwithliniervariogram

In the DTM quality assessment, several statistical equations have been implemented to measure the error or uncertainty of the DTM resulted from the interpolation process. The lowest value of uncertainty error will be the best DTM and usedforfurtherprocessing.Thisstepisdescribedinsection5.7.inthischapter. The government authority has been built a dike in the surrounding area to preventfurtherflooding.ThisdikehastoaddedtontheDTMtomodelthemudflowin therealcondition.ThedescriptionofhowtointegratebetweentheDTManddikeisin section5.8. TheworkflowofDTMconstructioncanbeseenintheflowchartbelow,

Fig59.FlowchartofDTMconstruction

44

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

This step resulted 4 DTMs based on the four interpolation methods that were used.TheDTMscanbeseenbelow:

Fig510a.DTMcreatedusingTIN Fig510b.DTMcreatedusingMinimum Curvature

Fig510c.DTMcreatedusingIDW

Fig510d.DTMcreatedusingKriging

5.7.DTMQualityAssessment ThereareanumberofinterrelatedfactorsthatinfluenceinthequalityoftheDTM;the dataacquisitionmethod,thenatureoftheinputdataandthemethodsusetocreatethe DTM (Shearer, 1990 in Weng, 2002). According to USGS (1995) and Caruso (1987 in Wechsler 2006), the DTM product has three type of errors; blunder error, systematic errorandrandomerror.Blunderserrorareverticalerrorsbecauseofcollectionprocess and usually are identified and removed before the data have been published. SystematicerrorsaretheresultofproceduresorsystemsusedintheDTMgeneration process and follow fixed patterns that can cause bias or artifacts in the final DTM product(Wechsler,2006).Thesetwotypeoferrors,canusuallyberemovedbeforethe DTMispublished,butrandomerrorremaininthedata(Wechsler,2006). Error measurement in DTM is almost impossible because there is rarely determinationofthetruevalueofgeographicfeaturerepresentedindataset(Goodchild et.al.1994;Hunteret.al.1995;inWeng.2002).Wengisusingworduncertaintyinstead oferrorindescribingtheDTMquality.DTMuncertaintyismeasuredbycomparingthe original elevation with elevation in DTM. This comparison will result the height

45

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

differencesbetweentwodataset.ThebetterwayinmeasuringDTMuncertaintyisusing statistical expression such as root mean square error (RMSE) see below, standard deviation and mean (Weng, 2002). In his paper Weng, 2002 was used RMSE as uncertainty originated from the interpolation method when the DTM was created.. Similar way was used by Wood in quantifying uncertainty in the DTM. Wood (1996) was used five parameters to calculate the uncertainty of DTM and 3 of it as described below: 5.7.1.RMSE The RMSE measurement is well known and most popular method to quantify theDTMuncertainty.ItisalsousedbytheUSGS(USGS1987)andtheordnanceSurvey (Ordnance Survey, 1992). In Woods, (1996) the RMSE reflects the differences between interpolated values from the true values or the most probable elevations which is not reflect actual elevation (Wechsler, 2006). This is a dispersion measure because it is the averagedeviationbetweentwodatasets(Woods,1996).Thisuncertaintymeasurementis based on assumption that the distribution of the errors are normal with zero mean (Wechsler,2003inRahman,2006). TheRMSEisvaluableofqualitycontrolstatisticbutdoesnotreflecttheaccurate assessment of the accuracy level of the DTM represents the real elevation. This only represents how the DTM corresponds with the originate data in the field (Wechsler, 2006).ThismeasurementdoesnotrepresentthehorizontalaccuracyofDTM,theusers should calculate the horizontal accuracy on the values that resulted from geometric correctionofsatelliteimageryifthedataderivefromremotesensingmethodorfromthe horizontalaccuracyoftheGPSwhentheDTMderivedfromdifferentialGPS(Rahman, 2006).

RMSE =

(z

zj)

Whereziandzjaretwocorrespondingelevationvalues nisthenumberofelevationpairsmodelled The value of RMSE is the differences between two datasets, the larger of value indicatingthehigherdifferencesbetweentwomeasurements(Woods,1996).TheRMSE isusuallyusetoprovidesingleglobalvalueofdeviation,thereforethereisnoindication ofspatialvariation(Woods,1996).Empiricallythaterrorcanbespatiallyautocorrellated as showed by Carter (1989), Guth (1992), Woods (1993) and Moncton (1994) in Woods 1996.ToovercomethisproblemGoodchild(1986)andHeuvelink(1993inWoods,1996) suggestedtoseparatethevalueofspatialtrendsfromthenonspatialerror.

46

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

5.7.2.Accuracyratio The accuracy ratio is value to represents the result of relative relief effects elimination from deviation measurement, simply by divided RMSE with relative relief measurement. Therefore the accuracy ratio is RMSE value divided by standard deviationofelevationvalue(Woods,1996). Accuracyratio a =

(z z ) (z z )
i j i i

Wherezizjaremeasuredasabove ziistheaverageelevationofi 5.7.3.MeanandStandardDeviation Themeanandstandarddeviationareusedtoestimatetherandomerrorinorder to assess the variability of the forecast about its mean value (Wechsler, 2006). This measurementcannotidentifynonstationaryofthedataset,ifthemeanvalueisnonzero it indicates that overestimates and underestimates of elevation are not equal, and that the overall accuracy of the model can be improved simply by subtracting the mean deviationfromallelevationvalues(Woods,1996).

d zij =

z z
i

s=

[( z

z j ) d zij n

Whereziandzjaremeasuredasabove dzijisthemeandeviationbetweenmodels sisthestandarddeviationbetweenmodels

5.7.4.AnotherDTMuncertaintymeasurement Another uncertainty measurement can be assessed using several statistical estimator as shown in table below, users can be examined all or some of these measurementstatisticalequation(Wechsler,2006). Bias Bias=

(Y Y )
i =1 i i

47

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Relative

) (Yi Yi ) RBias= i =1 Yi N
N

AverageRelativeAbsoluteDifference RelativeRootMeanSquareError

) Yi Yi Y ARAD= i=1 i N
N

) Yi Yi RRMSE= i =1 Yi N
N

LogRootMeanSquareError

) Yi ln Y LRMSE= i =1 i N
N

Where Y referstotheestimatoroftheparameterYiandNisthenumberofsimulation. 5.7.5.ResultandDiscussion The value of DTM uncertainty is depends on three factors, measurement of topographicmap,samplingandmeasurementerrorandinterpolation(Weng,2002).The totalofDTMuncertaintycanbecalculatedusingequationbelow: RMSE = ( RMSE ) 2 + ( RMSE ) 2 + ( RMSE ) 2
total

RMSEm=uncertaintyofsourcemap RMSEs=uncertaintyofsamplingandmeasurementerror RMSEi=uncertaintyofinterpolationprocess


Table5.7.MeasurementofDTMuncertainty Interpolationmethod TIN 10.9s 31.849 123.718 3.885 6.100 127.752 Minimum Curvature 117.1s 250.375 425.866 1 0.269 494.014 Inverse Distance Weighted 694.9s 4.615 18.012 3.902 9.316 18.594 Krigging 00:27:31.9 4.571 18.031 3.945 9.338 18.600

Computationtime Mean StandardDeviation Coef.Ofvariation Coef.Ofskewness RMSE

48

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Table 57. showes the results of measurement of the DTM uncertainty in this study, four interpolation method has been implemented. The TIN interpolator has the lowest computation time, and the value of RMSE is the second lowest after Minimum curvature. Besides the result from this interpolator show triangular facets in the flat terrainindicatingthatdataistoofewinthisarea.Minimumcurvatureinterpolatorhas the biggest RMSE, the mean is the lowest and it has the biggest uncertainties. Kriging and IDW have almost similar values of RMSE but krigging takes the longest in the computation time. The DEM resulted from IDW has terraces features on slope terrain and bull eye features in flat terrain, therefore overall krigging interpolator has the less uncertaintyandhighestqualityofDEMinthisstudy. ThetotalvalueoftheDTMbasedon1:10.000topographicmapinthisstudyis: Sourcemap: RMSEm=0.304x5=1.52 Interpolationprocess: RMSi=18.600 TotalRMSE: RMSEtotal= (1.52) 2 + (18.6) 2 = 2.3104 + 345.96 = 348.2704 =18.66 5.8.Manmadeterrainmodeling(dike) 5.8.1.DTMDikeintegration A Dike forms a protection of an area to flooding by acting as an obstacle to waterflow.Otherobjectsonthesurfacewhichhavesimilarfunctionlikeobstaclesuchas building, elevated road, curb and another man made features. All these features will guide the flow water, and will play an important role in the flow velocity and the maximumextentoftheflood. Thestudyareaislocatedinurbanareaandmanmadefeaturesareeverywhere. Toincludetheelevationofmanmadefeatureisverydifficultbecauseoftimeandcost constraints,LIDARandIfsararetwotypesofremotesensingthatcancaptureelevation ofmanmadefeatures,butsincethecostaremajorconstrainttheavailabilityofsuchdata isverylimitedorevendoesnotexist. Data about location and boundary of dike in this study are derived from IKONOS imagery which can downloaded for free from CRISP website (www.crisp.nus.edu), from this website we can download the IKONOS imagery with lowerresolution(6m).Theheightofthedikeisderivedfrommonthlyreportmapfrom governmentauthority.TheflowchartofintegrationofDTManddikecanbeseenbelow :

49

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Fig511.FlowchartofintegrationDTManddike

Five DTMs were constructed in this study to represent 5 time scenarios of the inundation.ThedatesoftheDTMsarethefollowing: 06June2006thedikehasnotbeenbuilt 29August2006thedikewasbroke 31October2006thedikewasbroke 06December2006thedikewasbroke 05January2007thedikewasbroke TheDTMandimagerywhichrepresentsthetimescenarioabovecanbeseen below:

50

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow


Imageunnecessary

Dike11/06/2006

Dike29/08/2006

Ikonos29/08/2006

Dike10/31/2006

Ikonos10/31/2006

Dike12/06/2006

Ikonos12/06/2006

Dike05/01/2007

Ikonos05/01/2007 Fig512.TheDTMwithdikeandIkonosimagery

51

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

5.8.2.ResolutionofDTM ResolutionofDTMisimportantfactorinflowmodelingbecauseDTMresolution indicatedhowdetailedtheDTMrepresentsthesurfacerelief.Theresolutionalsoplays an important role for the computation time of the simulations. Many computers with highspeedcomputationcanbefoundtoday,butsincemostmodelingsoftwareworkon raster data, the more data the slower the computation process, and of course more advanced computer technology will make the computation faster but the cost will be higher. In this study the resolution selection that resulted in acceptable computation times, was achieved by trial and error. The final resolution was also selected by comparison between the DTMs with various resolutions. The optimal resolution was determined by the crosssection or traverse method which was implemented in some parts on the DTM. Fig 513a shows the profiling from AB and Fig513b is the height differenceforeachtypeofresolution.DTMforthisstudywasoriginatedfromcontour topographicmapwith5mcontourinterval.Basedonthecrosssectionandcomputation time needed by SOBEK software the DTM with 20 m resolution is reliable enough to representsterrainreliefforthisapplication. B A

Fig513a.ProfilingfromAB Fig513b.CrosssectionfromAB

52

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

CHAPTER6.MUDFLOWMODELING
6.1.Governingequation The behavior of fluid material can be described either by Newtonian or Non Newtonian Law of physics. The Newtonian fluid considered when shear stress is a linearfunctionoftheshearrateforexampleinflowofwater,floodmodelingusingthis principletodescribetheflowofwater.ThenonNewtonianfluidconsideredwhenthe shearstressisnotalinearfunctionoftheshearrateandwhenviscousbehaviorisnon linear. Mud flow behavior can be described as non Newtonian fluid if the viscosity is highenoughdependsontheconcentrationoffineparticleinthemixture. The flow behavioroftwo type of fluidas explained above can be explainedby mass and momentum conservation principle. The mass and momentum conservation equationsare:

h uh vh + + =0 t x y

z f ox uh u 2 h vuh + + = g h sin x cos x + h x t x y x


z f oy vh uvh v 2 h = g h sin y cos y + h y + + y t x y
where, h

=theflowdepth

(uh, vh) =depthaverageofflowmomentuminthexandy z f = z b + h =heightorverticalcoordinateofthefreesurface(m) z b ( x , y )


=bedsurface =bedgradientinthexandy =bulkunitweight(KPa.m1) =bottomshearstress

Theequationsabovearevalidfornewtonian,viscousandgranulartypeofflows (Santiago and van Asch, 2007). The use of equation above to the type of fluid flows is dependsonhowtheformulationofflowresistancetermandgenerallynonNewtonian fluid will use the definition of the HerschelBulkley model (Santiago and van Asch, 2007)andcanbedescribedbythefollowingequation:

= y + t
m

53

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

where, =shearstress(KPa) y =yieldstrength

=strainrate
=coefficientofdynamicviscosity(kPa.s)

misanempiricalparameterwhichhastobedeterminedinHerschelBulkleymodeland equalzeroforBinghammodelandthismodelcanbeusedtodescribedmudflow. The Newtonian fluid such as water described using shallow water equation basedonSaintVenantequationlikeusedinSOBEK.Theshallowwaterequationcanbe describedusingfollowingrelation:

A f t

Q = qlat x

where, A f =wettedarea

qlat
Q t x

=lateraldischargeperunitlength(m2/s) =discharge(m3/s) =time(s) =distance(m)

SOBEK using integration of 1D and 2D hydrodynamic modelling which using thefollowingequation: Momentumequation1Dflow:

A f t

Q x

Q2 Af

qQ Q h + g . Af . + 2 Wf wi = 0 w x C RAf

where, A f =wettedarea(m2)

qlat
Q t x g h

=lateraldischargeperunitlength(m2/s) =discharge(m3/s) =time(s) =distance(m) =gravityacceleration(=9.81m/s2) =waterlevel(m)

54

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

C R Wf wi

=chezycoefficient(m1/2/s) =hydraulicradius(m) =flowwidth(m) =windshearstress(N/m2) =waterdensity(kg/m3)

Continuityequation2Dflow:

( uh ) ( vh ) + + =0 t dy x
where: u v V h d =velocityinxdirection(m/s) =velocityinydirection(m/s) =velocity: V = u 2 + v 2 =waterlevelaboveplaneofreference(m) =totalwaterdepth: +d(m) =depthbelowplaneofreference(m)

Momentumequation2Dflow:

uV u u u d +g + g 2 + au u = 0 +v +u y x t x C h
vV v v v d +u +v +g + g 2 + av v = 0 t x y C h x
where: u v V h d a C =velocityinxdirection(m/s) =velocityinydirection(m/s) =velocity: V = u 2 + v 2 =waterlevelaboveplaneofreference(m) =totalwaterdepth: +d(m) =depthbelowplaneofreference(m) =wallfrictioncoefficient(1/m) =chezycoefficient(m1/2/s)

55

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

6.2.Software In the flow modeling the selection of the software is primarily based on the characteristics of flow material. There are three type of flow material based on flow category,streamflow,slurryandgranularflow(Alkema,et.al,2006).Thesetypeofflow categories are classified based on concentration of water and fine material. If the concentration of water exceeds 70% it has category as streamflow like water and mud flows, and if water content decreases until 80 % it has category as slurry flow for example debris and solifluction. For every flow category there is the law of physics which works either Newtonian or NonNewtonian. Newtonian law of physics used to describe mostly water with low viscosity and its flow behavior are determined by friction, while the flow of water with very high particle content and high viscosity are determinedbyBinghamflow(Alkema,et.al,2006). Bothtype ofmaterialflowas explained abovewereimplemented into different software. SOBEK is software for hydrodynamic modeling developed by Delft Hydraulicsisusedtomodelwaterflowmaterial.Waterwithveryhighparticlecontent andhighviscositycanbemodeledusingPCraster,anopensourcesoftwaredeveloped by University of Utrecht the Netherlands. As an open source software PC raster has capabilitytoincorporatethedesignofflowmaterialbasedonflowpropertieswhichis definedbyusers,butSOBEKasbuiltinsoftwareisablackboxsystemwhichrequired aninitialknowledgeofthesoftware. Inthisstudymudvolcanoasflowmaterialhasspecificpropertieswhichisvery uniquedependsonitslocationanditsgeologicalcharacteristics.Forexampleinseveral literaturemudvolcanosarehavehighviscosityanddensitybutinthiscasethedensity isalmostthesamewithwater(1.4gr/cm3)anditviscosityisrelativelylow(1.4N.s).So this phenomenon is relatively new because there was limited research previously conducted. PC raster which allows users to make their own designed model using programming script is very useful and ideal tool to model the flow behavior of mud volcano.ForthisreasonPCrasterwaspreviouslyuseinthisstudy,buttherearesome limitation causing this work was not successful. The first constraint is there were no scripthas beendeveloped for mud volcano, whiletodevelop thenew scriptis takesa longtimewhichisbeyondthetimelimitofM.Scstudy.Thesecondreasonisinsidethe developer of PC raster itself, such flow phenomenon which called diffusive flow is underconstruction(Jetten,lecturenotes2006),thereforethisresearchinthefuturecould be one of the study case. In order to make a good model of mud volcano it required deep study of mud flow properties such as viscosity, the yield strength and its temperature,andthiswillbefulfilledifthereisenoughtime. SOBEK was selected to replace PC raster software based on the properties of mud volcano which has density similar with water and has low viscosity. The similar researchwasdonebyMartini,et.al,2003,whichconductedresearchonthefloodflows and suspended sediment transport in the Brenta River, Italy. Martini, et.al was used ID2DhydrodynamicmodelingbasedontheSaintVenantequation.

56

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

6.3.InputParameters SOBEK has powerful module to define an input data called NETTER. In this module users can defineinput data graphically using advance graphical user interface orGUI.NETTERalsohasatooltoattachvectorfileorimageasabackgroundsouser candefineinputdataeasier.SOBEKrequiredseveralinputdatasuchas: - Elevationdata(DTM) - Simulationperiodandtimesteps - Dimensionofthechannel - Waterlevel - Dischargesofwater - Coefficientofroughness

Fig61.Thegraphicaluserinterface(GUI)inSOBEK

TheprocessofinputdatainSOBEKareintwosteps,firstisdefinedinputfor1D featuresuchasriverorchannelandthesecondisinputfor2Dfeaturesuchasoverland flow.The1Dfeaturelikeriverorchannelisdefinedusing1Dnodesanditcalledflow model,andthisnodesconnectedasnetworkusingstraightlineconnectorcalledreaches. The reach segmenthasfunction as flow modelor overland flow modelconnector. The whole input which contain node and reaches then work together as network and simulatestheflowmodeling. In this study the data input are the DTM, discharge of mud flow and the cross section of the channel.As a matter of fact there wasno channel in the field except the channel of gas exploration or the well bore. In SOBEK users has to define the channel dimensionin1Dmodule,forthisreasonthechanneldimensiondefined.

57

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Fig62.InputparameterforBoundarycondition

Fig63.Inputparameterforcrosssectiondimension

6.4.OutputParameters Output parameters in SOBEK can be classified into three type, there are water depth, water level and velocity. This output parameter then can be displayed or analyzedasGISmap,andcontainsthefollowingvariables(SOBEKhlp):

maximumwaterdepth(d_max) timeofmaximumwaterdepth(t_dmax) maximumwaterlevel(h_max) timeofmaximumwaterlevel(t_hmax) maximumvelocity(v_max) timeofmaximumvelocity(t_vmax) timeofwetting(t_wet) timeofwettingperclass(steadystate)(t_wet/class) timeofdrying(t_dry) timeofdryingperclass(steadystate)(t_dry/class) rateofchangeofwaterdepth(dd/dt)


Fig64.Graphicalmenuforinputtingoutputparameters

58

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

SOBEK also provide an output result as time series output files in ascii format (*.asc). SOBEK will create the series output files only when users choose water depth, water level and velocity checkbox is switched on. This will create many files representingsomeparametersasstatedbefore.Thenumberoffileswhichwillcreateare depend on the number of time steps of the simulation, for example if users make 24 hourssimulationwithoneminutetimestepsotheoutputmapscreatesbySOBEKwill be24x5x60equals7200outputfiles. 6.5.InitialandBoundarycondition InSOBEKinitialconditionistheinitialinputdefinedbyusers,therearetwotype ofinitialcondition,thedepthsorwaterlevelsandthedischargesatthebeginningofthe simulation.Oneoftheinitialconditionmustbedefinedfirst,butusersareallowstouse both, for example at certain model there are specific discharges at the beginning of simulation but it has also water inundation at certain level. In such cases the program willfirstfilleduptheareawithwateratcertainlevelandthenthesimulationwillstart further from that level. The simulation also can be built from the result of another simulation,theinitialsimulationwhichissaveinrestartfilewillbethebasisofthenext simulation. Inthisstudytheinitialconditionwasthedischargesfromthepointoferuption, and the water level starts from zero, it means this is completely dry system. Mud volcanointhisstudywascomeuptothesurfacefrombelowbecauseofafailureofgas explorationactivity,thereforethesimulationwasstartfromcompletelydryconditionto inundated.


Fig65.Modelsimulationtobesetascompletelydrysystem

Boundaryconditionisrepresentstheexternalflowortheamountsofwatercome upinthesystem.Boundaryconditioncanbespecifiedaswaterlevel,discharge,depth or as rating curve. There is also an option to specify the boundary condition as time seriesinput.Forthismodeladischargeofmudflowisdefinedasboundarycondition. Thedischargeisdefinedinthe1Dboundarynodewhichrepresentsaneruptionventof

59

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

mud volcano. Several discharges were defined to model several dike breaks in five differentscenariosasexplainedinscenarioarrangementsubtopic.

Fig66.InputparametersofBoundarycondition.

6.6.Surfaceroughness Surface roughness has function as a resistance force in the flow equation and usedtorepresentthebottomfriction.Thereareseveraltypeofsurfaceroughnessusein SOBEK, chezy, strickler, Colebrook, manning, and commonly used to represent the surfaceroughnessoffloodplainandchannel(Tennakoon,2004). In this study mannings roughness was used as an input of surface roughness. The manning coefficient was derived from land use map of 1:25.000 produced in 1996 andupdatedusingIKONOSimageryof2002. Themanningsequationisexpressedas:

(R V=

2/3

S 1/ 2 N

where, V =Flowvelocity(m/sec) R =Hydraulicradius(m) S =Channelgradient N =Manningsroughnesscoefficient Thelandusemapwithanattributecontainsmanningcoefficientthenconverted intorasterfileformatusingILWIS,thisrastermapthenexportedintoASCIIfileformat anduseasaninputinSOBEK.

60

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Fig67.Landusemapasinputofmanning roughness

Fig68.Ikonosimageasinputforlanduse update

Fig69.Landusemapasmanningcoefficientmap

Tabel61.ManningcoefficientderivedfromLandusemap MANNING COEFFICIENT 0.0200 0.1500 0.0200 0.0300 0.0350 0.0300 0.0130 0.0350 0.0300 0.1000 0.0600 0.0600

NO. 1 2 3 4 5 6 7 8 9 10 11 12

LANDUSETYPE BarrenLand/OpenSpace Builtuparea Fishpond Grassland MixGarden Paddyfield Pavement,AsphaltorConcrete Plantation(Sugarcane) River Settlement Shrub Swamp

13 Waterbody/Sea 0.0300 Source:Pieterse,et.al,1988;BrushCreekWatershedStudy,2004.

61

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

6.7.Modelschematization ThefirststeptocreatesimulationinSOBEKisModelSchematization,inthisstep users designed the network and the attribute of each nodes, reach segments, cross sectiondataandboundarycondition.Riverandchannelmustbedefinedin1Dnetwork. The 1D network is constructed from boundary nodes, connection segments, cross sections and calculation points. The 2D network is constructed from several elements such as nodes and reaches segments, and in the 2D network the boundary conditions suchaswaterlevelanddischargearedetermined. The elevation data such as DTM is imported from ASCII file format. The DTM will read in SOBEK as a set of grid, each grid can be determined as single value of roughnessoronevalueofroughnessforthewholegrids.Thevalueofroughnesscould bedefinedasvariablewithreferencetothelandusemap. In this study the schematization was built by determine flow boundary node which connected to another flow module node by reach segment. The Cross section node will be defined in the flow segment and acting like a small channel to flow a certain numbers of discharges. This is the basic procedure in 1D module in SOBEK, without this the simulation will not work. In real condition there is no channel in the field,becauseinthisstudymudflowwascomeupfromwellborebelowthesurface. Several history stations were defined in the surrounding area to calculate the velocityandthedepthofthemudflow.Thesestationsarebuiltinsideandoutsidethe dike. The stations inside the dike are to calculate flow parameter when mud flow fulfilledthedike,andwhenthedikebrokeorovertoppedthestationsoutsidethedike willdocalculation. At a certain location the elevation of dike was built half to one meter lower to simulate the broken dike. The mud flow from this point will inundate the area surroundingandSOBEKwillsimulatethefloodextent.

62

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Fig610.Schematizationofthemodel

6.8.Scenarioarrangement Mud flow in this study has been occurred for almost two years, and the dike which is built area surrounding to prevent from flooding has been broke many times. Thisstudyisintendedtomodelthemudflowbysimulatingtheareainundatedwhen thedikewasbroke.Fivescenariosofinundatedeventsintheareawerechosenbasedon theavailabilityofthedatatoverify.Theavailabledatawereusedtomodeledmudflow areIkonosimageryandthepublishedmaps. Thescenariosofthemodelaredescribedinthetablebelow:
Table62.Listofmodelscenarios No. 1 2 3 4 5 6 Dateofevent 29May06 11Jun06 29Aug06 31Oct06 6Dec06 5Jan07 Discharge (m3/day) 50,000 50,000 120,000 120,000126,000 120,000126,000 120,000 Discharge (m3/s) 0.578 0.578 1.388 1.3881.458 1.3881.458 1.388 Simulation time(days) 0 15 60 60 40 30

63

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

There are some basic assumption were implemented in this model due to very complicated condition in the field. Some human activity to control the mud flow has been done and therefore the behavior of mud flow is not totally natural. The first assumptionwasthe dischargeofmud flow to be determined asconstantvalueduring themodellingprocessbecausethereisnorecordofdischargechangesfromtimetotime. TheonlyavailablerecordisfromMediaCenterLusi,oneofofficialNewsmediafrom governmenttoprovidethenewsofthishazardevent. Thesecondassumptionwasthewaterleveldefinedaszeroheightorthesystem assumed as completely dry. It means for every model scenario the initial condition is start from zero water levels and water will come up in certain level following the discharge. The water then spilled out or overtopped after reach a certain elevation of dikeandthissimulateasmudflow. The value and magnitude of land subsidence in this model were ignored, this because of the pattern and magnitude of the subsidence were random in this area (Abidin,et.al,2006). Thisstudywasfocusedonthebehaviorofflowingofmudvolcano,thereforethe majorconcernwasthecharacteristicsofmudflowsuchastheextentofinundatedarea, thedepthandthevelocityofflow. 6.9.ModelingResult In SOBEK the output of flow modeling is inundation extent, depth and flow velocity. All output is in time series file and can be converted into GIS file format, thereforemodelinganalysisusingthissoftwareiseasierandcanintegratewithanother GISsoftwareforfurtheranalysis. 6.9.1.InundationExtent There are five output of inundation extent based on scenario arrangement and thentheresulttobecomparedwithreportedmapwhichrepresentsactualevents.Fig6 11. below showed five output inundation extent maps for five time scenarios. From theseresultsareaofinundationextentaretobecalculatedandtobecomparedwitharea ofinundationextentfromreportedmap.

64

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

06June2006

29August2006

31October2006 Fig611.Flowmodelingresultforfivetime scenarios.

06December2006

05January2007

Table 63. consist of comparison of the area of inundation between the actual eventandthemodel.Theoverlappedareaalsocalculatedtoseehowlargeareabetween theactualeventswiththemodel.Thepercentagesofareaoverlapshowhowgoodthe flow model represents the actual events. This simple analysis was tried to find out whether SOBEK can simulate mud flow or not, and whether the mud flow can be assumedhavecharacteristicssimilarwithwaterornot. Fig 612 and Fig 613 showed comparison between mud flows recorded from IKONOSimagerywithflowmodelingresult.Thefiguresshowedthatoverlapareaisthe areaofmudflowwhichrepresentsbothactualeventandmodelingevent.

65

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Model Mudflow Overlap

Fig612.Inundationextenton06June2006

DriedMud flow

Model Inundation

DriedMud flow

Model Inundation

Overlap Overlap MudFlow MudFlow insidethe dike 29August2006 06 December MudFlow 2006 MudFlow MudFlow insidethe dike 31October2006 MudFlowModel Model Inundation Inundation

Overlap

Overlap

MudFlow insidethe dike

MudFlow insidethe dike

DriedMud DriedMud flow flow

05January2007 Fig613.Inundationextentonseveraltimescenarios

66

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Table63.Comparisonresultbetweeninundationareaandareafromthemodel Dateof event 11Jun06 29Aug06 31Oct06 6Dec06 5Jan07 Data comparison source ReportedMap Ikonosimage Ikonosimage Ikonosimage Ikonosimage AreaActual (m2) 471986.71 48648.178 141681.661 851547.709 1365194.356 Area Modeling (m2) 3770941.201 347644.597 1023800 1383112.445 906313.991 Area Overlap (m2) 99646.48 35515.601 70526.49 367760.455 563413.092 Difference (%) 21.11214 73.005 49.77814 43.1873 41.2698082

No. 1 2 3 4 5

Based on the table above we can see that the result of overlap area has percentagedifferencerangesfrom21%to73%butmostlywith40%overlap,fromthis pointtheflowmodelingusingSOBEKrepresents40%ofareainundatedbymudflow. This result probably because of no input of building height and elevation data of man madestructuresuchasroad,irrigationchannel,curbetc.Suchmanmadefeaturesplays important role as obstacle and friction force and influence to inundation extend. The building footprint was treated as surface roughness and the result indicated wider inundation area comparing with the result which use original value of manning coefficientasexplainedinthemodelcalibration. The value of mud flow velocity also plays role primarily on the flow velocity, viscosity defined as fluids internal resistance to flow and influence to lower flow velocity.Thereforeviscosityalsoinfluencestheinundationextent. 6.9.2.DepthofInundation The depth of inundation is also one of output parameter in SOBEK. By comparing the result from flow model and actual event we can also find out whether SOBEK can simulate the mud flow or not. In the figures below there are five maps of inundationdepthforfivetimescenario,onlythreeofthemtobecomparedwithactual depthofinundationbecausetherewereonlythreerecordeddatawhichmatchwiththe timescenarioofthemodel.ThedataofinundationdepthwascomefromPT.Lapindo Brantas,thegasexplorationcompany. Fig614toFig616showedthatonlyFig614showthedepthofinundationon 29 august 2006 has a good match with recorded data except in the central eruption where the recorded data has deeper elevation. There are probably because of several causes: - landsubsidence,inthefieldthereweresubsidenceoccurredwhichhasrandom magnitudeanddirection,inthemodelthisfactorwasignored - the DTM resolution, in order to maintain the optimum result of computation time,theDTMresolutionwasresampleintocertainvalue(20m)andthiswill causedtheterrainreliefreduced - the DTM quality, the DTM input of the model was generated from 1:10.000 topographicmapswhichhas5mcontourinterval,inflatterrainthespotheight

67

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

wasaddedtorepresentsthedifferenceofelevation,probablythisspotheightwas notadequateandadditionalmeasurementofspotheightisnecessary

August29,2006

Fig614.Comparisonofinundationdepthat29August2006

Fig615.Inundationdepthmapat11June2006

68

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

October29,2006

Fig616.Comparisonofinundationdepthat31October2006

Fig617.Inundationdepthmapat27January2006

69

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

December12,2006

Fig618.Comparisonofinundationdepthat12December2006

6.9.3.Flowvelocity Similartotheinundationdepththeflowvelocityofthemodelalsocomputesat every timestep of simulation and converted into GIS file format. In Figure 619 and Figure 620 can be seen the result of flow velocity variation for five time simulation scenarios.Theresultofflowvelocityshowsthatfivesimulationscenarioshavethesame value and indicated that flow velocity influenced mainly by slope of the terrain. Unfortunatelytherewereobserveddatafromthefieldtoverifythisresult. The flow velocity is important aspect to calculate kinetic energy resulted from floodevent,andthisismainparametertodeterminethepotentialdamageinthearea.

70

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Fig619.Velocitymapon11June2006

29August2006

31October2006

6December2006

5January2007 Fig620.Velocitymapsforfourdifferentsimulationscenarios

71

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

6.9.4.Computationtime Thecomputationtimeisoneofmajorlimitationinflowmodeling.Mostlyflow modelingconsumealargenumberoftimedependofthepixelresolutionandthearea coverageofmodelingextent.InSOBEKthecomputationtimealsodependonsimulation periods and time steps of computation. Therefore users has to be determined the resolution, coverage area and the computation time steps in order to get the optimum timeofcomputationandtheresultexpected. In this study several initial run of modeling has been done in order to get the optimum parameters of computation before the real modeling was run. The result of computationofthisstudycanbeseeninthetablebelow:
Table64.Thecomputationtimeforeachmodelingscenario

No. 1 2 3 4 5 6 DTM 11Jun06 29Aug06 31Oct06 6Dec06 5Jan07 DTMfor Risk Assessment Res (m) 25 20 20 20 20 30 Number ofpixel 263x260 412x264 200x215 175x193 178x191 411x250 Simulation periods (days) 15 60 60 60 60 60 Time steps output 1hrs 1hrs 1hrs 4hrs 5min 1hrs Number oftime steps 1057 2977 2833 4321 17857 1417 Timeof computation (hrs:min:sec) 0:34:32 7:53:51 9:08:33 5:32:23 10:03:07 2:45:22

6.10.ModelCalibrationandValidation Model calibration was conducted in this study by implemented three scenarios with three different values of manning roughness coefficient. The first scenario was using the value roughness of manning coefficient which was compiled from several literatures, the second scenario was using manning coefficient which has value 25% higher from the first scenario. The third scenario was using the value of manning coefficient of the first scenario but also including the manning coefficient of building footprint.Dataofbuildingfootprintwasgeneratedfromtopographicmapscale1:10.000 whichupdatedusingIKONOSimageryof2002. Three scenarios of different manning coefficient were implemented to find out thecorrectvalueofroughnesscoefficientinthestudyarea.Whilebuildingfootprintwas added to get the real impact of buildings and houses in the modelling. The building footprint was acted as roughness surface rather than three dimensional objects in the fieldbecausethethreedimensionaldataofbuildingsandhouseswerenotavailable.

72

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Fig621.Buildingfootprintassurfaceroughness

Table 65. showed three different value of manning coefficient which implementedinthecalibrationprocess.
Table65.Manningcoefficientvalueforcalibrationprocess

NO. LANDUSETYPE
1 2 3 4 5 6 7 8 9 10 11 12 13 14 BarrenLand/OpenSpace Builtuparea Fishpond Grassland MixGarden Paddyfield Pavement,Asphaltor Concrete(Road) Plantation(Sugarcane) River Settlement Shrub Swamp Waterbody/Sea Buildingfootprint(single building)

InitialRun
0.020 0.150 0.020 0.030 0.035 0.030 0.013 0.035 0.030 0.100 0.060 0.060 0.030

Calibration1
0.025 0.1875 0.025 0.0375 0.0438 0.0375 0.0163 0.0438 0.0375 0.125 0.075 0.075 0.025

Calibration2
0.03 0.225 0.03 0.045 0.0525 0.045 0.0195 0.0525 0.045 0.15 0.09 0.09 0.045 1

Inthecalibrationprocesstheparametersusedinthemodellingwere: - DTM resolution is 20 m, in this resolution modeling result was showed the highestsensitivityasexplainedinsensitivityanalysissubtopic. - Theinundatedscenarioisusingthesixthscenario(5January2007)asdescribed in the scenario arrangement because this scenario has the most complicated terrainofDTM. - Thedischargeis1.4m3/s

73

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Initialrun

Calibration1

Fig622.Theresultofflowmodelforcalibration

Calibration2

Theresultofcalibrationthenanalyzedbycomparingtheinundationextendand the flow velocity. As we see in the Fig 622. the first scenario showed the lower inundationextendsandhigherflowvelocitycomparedtoanotherscenarios.Thesecond and the third result showed the bigger inundation extend indicated the surface roughnessactasobstacleonthesurfaceandmadeinundationspreadwider.Thevalue of building footprint as the highest of surface roughness gives significant effect on the flowmodel,fromtheresultwecanseetheinundationextendwaswiderthanthesecond scenario. In the model calibration we found that roughness coefficient has an important roleinflowbehavior,byimplementingtherightvaluewewillgettheappropriateresult which can construct the real event in the field. But in this study we still get the result which has different inundation extent compared to the real events, this probably becauseofthedifferentcharacteristicsofflowmaterial.Therealeventisthemudflow which has different viscosity with water. The value of viscosity acts as friction force whichmakesthevelocityofflowingslower. Another reason regarding to different result is the detail terrain of study area were not captured well enough in the topographic maps. The topographic maps is in scale of 1:10.000 which has 5 m contour interval, it means topographic maps can only represents the relief which has 5 m different elevation. In steep elevation this maps is

74

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

adequate as input data of flow modeling, but in flat terrain like in the flood plain it requiresadditionalmeasurementofspotheight. Inthisstudythetopographicmapshasmanyspotheighttorepresentthedetail of relief and the different of each spotheight is in order of centimeters unit, but the distance between one spotheight to another is in the range of 100 250 m, this is probably the reason why the spotheight could not represent the detailed of the relief. From this point additional measurement of elevation data are necessary, but due to conditioninthefieldanewmeasurementisnotpossible. 6.11.Sensitivityanalysis Sensitivity analysis was done using all parameters used in the modelling. The parametershavebeenusedinthisstudywereDTMresolution,dischargeoftheeruption ofmudvolcanoandsurfaceroughness.Sensitivityanalysisalsoimplementedtoeachof parameters and one of them will explain in the following sub topic and the rest parameterscanbeseenontheappendices. Sensitivity of the model was analyzed by inputting varying value of each parameterand observedtheresult of themodel.Everyparameter wasdesigned into 5 categories,eachcategoryhasvaluedifferent25%,forexamplethedischargevalueused is0.5,0.75,1,1.25and1.5m3/s.TheDTMusedisDTMwithresolutionof10,15,20,25 and30m.Themanningcoefficientalsodesignedinto5categories,thefirstvalue was derived from literatures and another was designed into value of 25% lower and 25% higher.Table66.belowshowedthemanningcoefficientsusedforsensitivityanalysis.
Table66.Manningcoefficientforsensitivityanalysis NO. 1 2 3 4 5 6 7 8 9 10 11 12 13 LANDUSETYPE BarrenLand/OpenSpace Builtuparea Fishpond Grassland MixGarden Paddyfield Pavement,Asphaltor Concrete(Road) Plantation(Sugarcane) River Settlement Shrub Swamp Waterbody/Sea 25% Lower 0.015 0.1125 0.015 0.0225 0.0263 0.0225 0.0098 0.0263 0.0225 0.075 0.045 0.045 0.0225 50% Normal 25% 50% Lower value Higher Higher 0.01 0.075 0.01 0.015 0.0175 0.015 0.0065 0.0175 0.015 0.05 0.03 0.03 0.015 0.02 0.15 0.02 0.03 0.035 0.03 0.013 0.035 0.03 0.1 0.06 0.06 0.03 0.025 0.1875 0.025 0.0375 0.0438 0.0375 0.0163 0.0438 0.0375 0.125 0.075 0.075 0.0375 0.03 0.225 0.03 0.045 0.0525 0.045 0.0195 0.0525 0.045 0.15 0.09 0.09 0.045

75

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Theresultofflowmodelsthentobeplottedintoliniergraphiccharttofindout whichparameterswasthemostsensitiveinthemodel.Fig623.belowshowedthatfor flow model in this study the parameter of manning coefficient is the most sensitive to thevelocityofflowanddepthofinundation.
SENSITIVITY ANALYSIS RES, Q, MANNING VS VELOCITY
0.01200

0.01000

0.00800 VELOCITY (m/s)

0.00600

0.00400

0.00200

0.00000 50% Lower 25% Lower Normal PARAMETERS INPUT DTM RESOLUTION (m) DISCHARGE (m3/s) MANNING COEFFICIENT 25% Higher 50% Higher

Fig623.Sensitivitygraphbetweenresolution,discharge,manningandflow velocity
SENSITIVITY ANALYSIS RES, Q, MANNING VS DEPTH
0.60000

0.50000

0.40000 DEPTH (m)

0.30000

0.20000

0.10000

0.00000 50% Lower 25% Lower Normal PARAMETERS INPUT DTM RESOLUTION (m) DISCHARGE (m3/S) MANNING COEFFICIENT 25% Higher 50% Higher

Fig624.Sensitivitygraphbetweenresolution,discharge,manninganddepthof inundation

Sensitivityanalysisalsohavebeenimplementedintoeachgroupofparameters, for example for parameter of DTM resolution, there were five different value of DTM resolution have been used. Through this analysis we try to find out the value of DTM resolutionwhichhasthemostsensitiveinthemodel.

76

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Fig625andFig626.belowshowedtheliniergraphicchartofvariousvalueof DTMresolutionrespecttotheflowvelocityanddepthofinundation.Forbothvelocity anddepththe20mresolutionDTMwashavethemostsensitiveinthemodel.Another resultofsensitivityanalysiscanbefoundintheappendices.

1.60000

1.40000

1.20000

1.00000 DEPTH (m)

0.80000

0.60000

0.40000

0.20000

0.00000 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82 85 88 91 94 TIME (hrs) DTM RES 10 M DTM RES 15 M DTM RES 20 M DTM RES 25 M DTM RES 30 M

Fig625.SensitivitygraphbetweenresolutionofDTManddepthofinundation

SENSITIVITY ANALYSIS DTM RESOLUTION VS VELOCITY


0.06000

0.05000

0.04000 VELOCITY (m/s)

0.03000

0.02000

0.01000

0.00000 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82 85 88 91 94 TIME (hrs) DTM RES 10 M DTM RES 15 M DTM RES 20 M DTM RES 25 M DTM RES 30 M

Fig626.SensitivitygraphbetweenresolutionofDTMandflowvelocity

77

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

CHAPTER7.IMPACTASSESSMENT
7.1.InundationMap Impact assessment in this chapter are describing the characteristics of mud volcanicflowinthestudyarea,theflowbehaviorlikethedepth,velocityandduration ofinundationandtheinfluenceofthishazardtotheelementatriskinthesurrounding area. As explained in the chapter 3 mud volcanic flow is a geological phenomenon whichcausedoferuptionofgas,waterandclayoutofsurface.Themixtureofwaterand clay causing flood which has longer time duration compared with stream flood or another type of flood. In this case geologist and geophysicist based on seismic and gravity data have calculated the potential reservoir of mud volcano in the subsurface were1,155billioncubicfeetandiftheaveragedischargeofmudflowwere100,000m3 perday,thenthemudflowwillbestoppedorfinishedin31years(MediacenterLusi, 2006).Thisisanenormousvalueofinundationtime,andthestudyofsuchfloodinghas neverbeenconductedbefore.Thereforetheinundationmapinthisstudywasdesigned intheperiodof30days,insuchperiodpredictedthatgovernmentandpeoplebeableto evacuatethemselvestosaferlocation. Inundation map was generated from flow modeling and designed using 30 m resolution of DTM and cover an area wider than the area used to model flow characteristics. For this reason resolution of DTM was reduced into 30 m, this also to maintain computation time in range of reasonable value. In the chapter 6 is explained howthecoverageareaandtheresolutioninfluencethetimeofcomputation.Therefore inthiscaseitwasimpossibletosetthetimestepsofsimulationinlongerperiodssuchas oneortwoyears.Inthisstudythetimestepsforinundationmapwassetinto30daysof inundationsimulation. TheinundationmapwasusedtheDTMwithnodikesurrounding,thisscenario haspurposetosimulatethedepthandmagnitudeofinundationwhentherearenodike hasbeenbuiltandthereforeweabletopredictwhichareawillbeinundatedforthefirst 30days.Fromthispointpeoplewholocatedinsideinundationareacouldbewarnedto evacuatetheirfamiliesandbelongingsotheycanminimizepropertieslossanddamage. Figure71showedtheresultofinundationsimulationforfirst10days,20days and 30 days. The designed of inundation simulation into 3 periods was based on the priorityofwarningtimesopeoplewillknowwhentheyshouldtoevacuate.The10days simulation was determined to be the first zone, 20 days inundated simulation was the secondzoneand30dayssimulationwasdefinedasthethirdzone.

78

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

10daysinundated

20daysinundated


Fig71.Theinundationsimulationscenarioin10 days,20daysand30days.

30daysinundated

Fig72.TheInundationmap

79

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

The Fig 72. above is inundation maps in three different time scenarios which then to be defined as different zoning area as explained before. This zonation is very usefulto helpgovernment tomanage the hazard in the first 30 days. These zones are representsthelevelofdangerfromeveryarealocatedineachzoneandcanbeusedas inputdataforlocalgovernmenttodeterminedthebufferareaandprotecttheareafrom utilizationbypeopleinthefuture. 7.2.MaximumDepth Generallypeopleusethedepthofinundationasprimaryindicationthelevelof flood hazard. Visually we can see the severity of flood hazard from the depth of inundation.Thedepthandvelocityusuallyworktogetherassubstituteparameter,one becomeoppositetotheother,forinstancefloodingwithhigherdepthwillhavevelocity lowerandviceversa. In the study area there are two conditions due to inundation depth, one is the depth inside the dike and second the inundation depth because of dike broke. The inundationdepthinsidethedikehaveconstantvalue,itrangefrom3mintheedgeof the dike and 8 m near of the central eruption, this information can be found on the reportedmaps. Thedepthofinundationresultedfromthedikebreakwasfoundfromtheresult of flow modeling because there were no observed information. We can found that the depthoutsidethedikeisrangefrom0to50cm.Instreamfloodsuchdepthcategorized as low hazard as showed in the Fig 73. below, but in this study the real condition is totallydifferent. Mud flow hazard has severity higher than any other hazard caused by water flood. In this case the hazard not only generated by water it self but also from concentration mud inside the mixture. Once the flood has stopped than the clay will remain and become dried and this will become serious problem for properties, health andlife.Thereforethedepthinundationevenat25cmwillbecomeseriousproblemand createhighhazard.

Source:USDept.ofLandandSoilConservation,inTennakoon,2004

Fig73.Therelationshipdepthandflowvelocitywithlevelofhazard

80

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Fig 74. below showed the impact of mud flow in the settlement. From this picturewecanseethedepthofinundationwasnottoohigh,itwasabout30to50cm but it start create the disaster. People can not work or live in such condition, transportation can not through it and probably the mud contain the toxic material becausethismudcomesupfromoilandgasreservoirbeneaththesurface. Fig75.alsoshowedthevillagewhichhasabandonedbypeoplebecausethereis impossibletoliveandstayinthatarea.

photobyAmienW/TimnasPSLS
Fig74.Housesinundated3050cmdepth

Fig75.Abandonedvillage,afterfifteenmonth

7.3.MaximumVelocity Flood impact assessment mainly using only parameters of depth and time of duration (Chen, 2007) this because the availability of data are only depth and time durationwhilevelocityisusuallyunavailable.Flowvelocityisimportantparameterto determine the value of maximum impulse and kinetic energy in flood (Alkema, 2003). Physicaldamagecanbedeterminedthroughimpulseandkineticenergyandcanbealso determinedtheleveloffloodhazard.

81

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Fig 76. below showed the relationship the depth and flow velocity, this figure alsodescribethevalueofdepthandflowvelocityrespecttotheresultontheelementat risk.Inthisstudythemaximumflow velocity was only 0.5 m/s, thisvalue is very low comparing to other flood events which have high level of flood hazard. Mud flow hazard has different characteristics, by low flow velocity and depth its still have destructiveforceespeciallybecauseofclayconcentrationinsidethemixture.

Source:USDept.ofLandandSoilConservation,inTennakoon,2004

Fig76.Floodhazardclassificationbasedondepthandvelocity

7.4.MaximumTimeofDuration Timeofdurationisoneoftheimportantparameterinimpactassessmentofflood hazard, the longer time of hazard duration will cause the higher hazard and potential damage and losses. In this study the maximum time of duration is unknown because until this thesis written the hazard still occurred. From the mud flow started (29 May 2006)ithasbeenalmosttwoyearsandthereisnoindicationthathazardwillstop. Asexplainedintheprevioussectionexpertspredictsthatmudflowhazardwill stopinthenext30years,thisisamazingphenomenonbutalmostineveryplaceinthe worldmudvolcanicflowwillremainuntilyears.Inthiscasetheinundationconditionin thatareahasbeencontrolledbydikebuiltbygovernment,butinseveraltimesthedike hasbeenbrokeandinundatedareasurrounding. Duetoverylongofdurationtimethishazardconsideredtobehighhazardrefers tographicinFig77.Inthisstudyalthoughthedepthandvelocityissmallbutthetime duration is very long and based on this situation the mud flow hazard in this area is high. Fig78.showedthepictureofsettlementareaatthetimeof8August2006which ismorethanoneyearfromthehazardstarted.Thehousesareinundatedinmoreofone year and people were moved and most of them were live in the shelter. From this conditionwecanfoundthattheinundationareaofmudflowhazardcannotberestore orrecover,becausetherecoverycostinthatareaisveryhigh.

82

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Source:USDept.ofLandandSoilConservation,inTennakoon,2004

Fig77.Thegraphshowsrelationshipbetweendepth,kineticenergiandtimeofduration

Fig78.Conditionofsettlementareafifteenmonthaftermudflowstarted

PhotobyAmienW/TimnasPSLS
Fig79.Thebuildingsweretotallydamage

83

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

7.5.ElementatRisk Element at risk is defined as the properties, population, economic activities, services and another human activities, at risk in a certain area (Westen, Lecture note 2007).Basedonthisdefinitionthecalculationofelementatriskisimportantinputdata in impact assessment. This parameter is not only necessary for assessing the value of potential loss and damage but also needed to designed evacuation plan, rehabilitation andreconstruction,aswellasthepreparednessplanbeforethehazardoccur. In this study the element at risk was calculated based on topographic map of 1:25.000forderivelanduseinformationandthenumberofbuildingwascalculatedfrom topographicmap1:10.000andalsousingIKONOSimagerytoupdateinformationabout buildingandhouses.Thisstudyofelementatriskwasonlyforphysicalinfrastructure which can derive from topographic maps and remote sensing imagery, and another elementatrisksuchasthepopulation,economicactivitieswasexcludingfromanalysis. Thestudyaboutelementatrisksuchaspopulationandeconomicactivitiesrequiredeep explorationonstatisticalparameterandfieldworklikeinterviewsusingquestionnaire. Duetolackofstudytimebothresearchactivitiescannotbedone,andtheconditionin thefieldwhichwasverydelicateregardingdisputebetweenpeopleandgasexploration company,interviewactivitywasimpossible. Thefirststeptoassesselementatriskistoidentifythetypeelementatriskitself, inthestudyareathetypeofelementatriskcanbeidentifiedbelow: Buildingandhouses Publicfacility Transportationfacility Factory Landuse

84

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Factory

Buildingandhouses

Tollroad

Railwaytrack

Fig710.Typeelementsatrisk

Therearetwo typeelement atriskin the study area, the first is element at risk which is located inside the dike and near surrounding the dike, these element at risk weresankundermudflow.Thesecondiselementatriskwhichlocatedoutsidethedike and can be identified through flow modeling, these element at risk was threatened by thebreakdike.Fig710,areclearlydescribedtheelementatrisk.

Inside thedike

Outside the dike

Fig711.Inundationareaisdeterminedasareaofelementatrisk

85

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Elementatriskwasdeterminedbydelineatingtheareainsideinundatedextent bothinsidethedikeandoutsidethedike,thenusingGISsoftwareitcanbecalculated basedonthetypeofbuildingsorlandusefromtopographicmaps.Allelementsatrisk were determined as totally damage, this condition is based on characteristics of mud flowitself. Elementatriskwascalculatedusingflowmodelingwhichsimulatedinundated areain30days.Meanwhiletheelementatriskinsidethedikewasusingtheinundated eventat05January2007.Theresultofcalculationofelementatriskcanbeseenbelow, TheresultofcalculationofElementatriskoutsidethedike(resultfrommodeling)as follows:

Table71.Elementatriskoutsidethedike(Buildingtype)

No. 1 2 3 4 5 6 7

ElementatRisk(Buildingtype) Buildingandhouses Governmentoffice(village) Mosque Moslemgraveyard School Hospital Factory Total

Number 3063 6 9 16 19 2 2 3117


Table72.Elementatriskoutsidethedike(Landusetype) No. 1 2 3 4 5 6 7 ELEMENTATRISK(TYPEOF LANDUSE) BarrenLand Fishpond Grassland MixGarden PaddyfieldIrrigated Plantation Shrub Total AREA (Ha) 5.565 652.167 12.944 40.555 187.732 320.365 60.254 1279.581

86

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

TheresultofcalculationofElementatriskinsidethedike(resultinundatedeventat 5January2007)asfollows:
Table73.Elementatriskinsidethedike(Buildingtype) No. 1 2 3 4 5 ElementatRisk(Buildingtype) Buildingandhouses Governmentoffice(village) Mosque Moslemgraveyard School JumlahTotal Number 1628 7 6 18 14 1673

Table74.Elementatriskinsidethedike(Landusetype) No. 1 2 3 4 5 6 7 ELEMENTATRISK(TYPEOF LANDUSE) BarrenLand Fishpond Grassland Mixgarden PaddyfieldIrrigated Plantation Shrub Total AREA (Ha) 134.414 3.951 24.048 0.100 29.739 190.773 23.509 406.534

Table75.Elementatriskinsidethedike(Roadtype) Length No. ElementatRisk(Roadtype) (m) 1 2 3 4 5 TollRoad LocalRoad OtherRoad Pathway Railwaytracksingle Total 4769.382 4477.128 38513.648 7142.929 461.098 55364.185

87

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

Settlement

Tollroad

Mudflow

Agriculturalarea

PhotobyAmienW/TimnasPSLS Fig712.ThepictureofElementatrisk

88

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

CHAPTER8.CONCLUSIONANDRECOMMENDATION
8.1.Conclusion 8.1.1.ConclusionfromResearchObjective a. To generate and construct the mud volcanic flow model in dynamic spatial modelingenvironment. Thisresearchwasdonesuccessfullyandtheresultisfiveflowsmodelingwhich usedtoanalyzetheflowmodelgeneratedfrom1D2Dhydrodynamicmodellingwith the behavior of mud flow occurred in the study area. The inundation extent of the flowmodelfits40%withtheinundationextentofthemudflowhazard,butthedepth of the inundation indicated only less 20% overlap between flow model and the mudflowhazardthisbecausethemodelwasnotincludedthelandsubsidencewhich occurredinthatarea. b. To know the propagation of the mud volcanic flows, how large the area will be affectedinpresentandfuture. ThemodelresultfromSOBEKhasoutputparameterssuchasinundationextent, thedepthandflowvelocity,thiscanbeusedtofindouttheareaofinundationina certain of time. The simulation within period of time is constraint by a number of pixel of DTM, resolution of pixel and the timesteps, therefore there must be adjustment to set the combination of that parameters. In this study the highest computationtimeswas10hourswith60dayssimulationperiodand178x191number ofpixelin20mresolution.Itmeansthecoverageareais4kmx4km,foralongtime simulationsayit2yearsitwillrequirehighercoverageareaanditwillconsumealot of computation time. Therefore it is impossible to create simulation in longer time unlessthereispowerfulsupercomputerdedicatedtothiswork. c. Tocalibrateandvalidatethemodelitself,howisthereliabilityandaccuracyof theresultbasedoncomparisonwithactualevent. Therewereseveralparameterswithdifferentcombinationvaluesusedtocalibrate and validate the model, and through this step we can conclude that the surface roughness play important role to the result of the model. The surface roughness whichactasresistingforceintheflowofthematerialneedtobedefinedasaccurate aspossiblewiththerealconditioninthefield.Theparameterlikesinglebuildingor building footprint was also give significant influence in the model. The building footprint as surface roughnessmadethe area of inundationmore larger andsimilar withthemudflow,andifthebuildingfootprintactaselevatedbuildingitwillgive betterresult.Theaccuracyofthemodelalsodeterminedbyviscosityofthematerial andthisparameterwasignoredinthismodel.

89

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

d. To asses mud volcanic flow hazard which affected the surroundings such as settlements,buildings,roads,factoryetc. Theinundationextentresultedfromflowmodelintegratewithanotherdatasuch as Land use maps and Ikonos imagery, can be used to calculate element at risk for exampleroads,buildings,housesandlanduse.Anotherelementatriskforinstance thepopulationandeconomicactivityrequirefurtherresearchandanalysis. 8.1.2.ConclusionfromResearchQuestion a. WhattypesofdataneededtobuildDEMasaplatformtogeneratethemudflow modeling? The DEM as a representation of relief change has to be able to capture the detail of relief change on the surface, this basic principle will require advance techniqueandcost.Thesetwoaspectsarethemajorconstraintforthedevelopment countries like Indonesia. Generally such data is not available. In this study topographic map scale 1:10.000 is adequate to represent the elevation data in steep slopeforflowmodeling.Inflatterrainlikeinthefloodplainspotheightasadditional elevationdatatorepresentthereliefarenecessary.Inthiscasethetopographicmap in scale of 1:10.000 have well distributed spot height, but there is not adequate in density,thereforeadditionalspotheighttodensifytheelevationisnecessary. b. Howfartheinundationofmudflowsinthenext5,10and20years? The computation time is major constraint to simulate flow model in longer periodoftime.Thesoftwarewillrequirethepowerfulsupercomputertoprocessthis model. c. Howreliableandaccuratethemodel? The reliability and accuracy the model is depend on varying parameters. The characteristicsorthepropertyofthematerialisoneofparameterstobedetermined first.Thebehaviorofflowmaterialwillbedeterminedfromthischaracteristic.The otherparameterisroughnesscoefficientwhichisactasfrictionforceintheflowof material, these parameters must be determined good enough in order to get optimumresult. d. Howmanypropertiesandinfrastructurewillbeaffectedbymudflow? The element at risk can be identified and calculated using data of inundation extentfromflowmodeling.InundationextentintegratewithanotherGISdatacanbe determine the element at risk affected by mud flow. The impact assessment in chapter7isdescribedthistopic,andtheresultofelementatriskcalculationcanbe foundinthischapter.

90

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

8.2.Recommendation Thestudyofmudflowisveryimportantsincethisphenomenoniswidelyfound inseverallocationsintheworld.Thishazardisuniqueandhaspowerfulenergywhich canputhumanliveingreatdanger.Mudflowhazardifoccurredintheurbanareawill causethemaximumlossinthepropertiesandlife.Inthishazardthousandpeoplewere displaced,thirteenpeopleweredeadandthousandpropertiesweredamageorloss.For this reason the proper assessment of the hazard will safe lives and properties, by simulatingtheflowofthemudpeopleandgovernmentcanpredictwherehazardwill goandabletomanagethehazardinagoodway. This study of mud flow hazard probably is the beginning research since the research conducted on this field was limited especially in Indonesia. Therefore similar researcheshavetobeencouragedinordertosolvethehazardproblemwhichcannotbe imaginedbefore.Theresearchinthesocialaspectofthemudflowhazardalsohaveto be encourage so people and government can use the result to face the problem of the hazardinthefuture. Thisstudyasalimitedresearchofcoursehasmanylimitationandconstrainsand this has to be solve in the future research. The use of proper software as main tool to simulatethehazardhastowelldetermineaswellastheunderstandingoftheproperties ofmudflow.PCrasterasopensourcesoftwarelettheuserstoexplorethecapabilitiesof thesoftwareinthemaximumlimit, but itrequireadequateknowledgeofthesoftware itself.Thereforeinthefuturetheresearcherhastoprovidegoodknowledgebeforeuse PCrasterasatooltosimulatethemudflowhazard.

91

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

REFERENCES
1. Arattano,M.,Franzi,L.,Marchi,L.,2006.Influenceofrheologyondebrisflow simulation.,Nat.HazardsEarthSyst.Sci.,6,519528. 2. Arattano, M., Franzi, L., 2003. On the evaluation of debris flows dynamics by means of mathematical models. Natural Hazards and Earth System Sciences. 3, 539544. 3. Atkinson, P.M. Lloyd, C.D. 2007., Nonstationary variogram models for geostatisticalsamplingoptimization:Anempiricalinvestigationusingelevation data.,Computers&Geosciences33.,12851300. 4. Blomgren, Sten., 1998., A digital elevation model for estimating flooding scenarios at the Falsterbo Peninsula., Environmental Modelling & Software 14., 579587. 5. Bowles, Frederick A., Faas, Richard W., Vogt, Peter R., Sawyer, William B., Stephens, Kevin., 2003., Sediment properties, flow characteristics, and depositional environment of submarine mudflows, Bear Island Fan., Marine Geology,197pp6374. 6. Balka,F.,Alka,A.,Kurucub,Y.,Alk,Z.,ValidationofRadargrammetricDEM Generation from Radarsat Images in High Relief Areas in Edremit Region of Turkey. 7. Chen,Pengyu.,2007.FloodImpactAssessmentusingHydrodynamicModelling in Bangkok, Thailand.MSc. Thesis ., International Institute for Geoinformation ScienceandEarthObservation,Enschede,TheNetherlands. 8. Davies,RJ.,Swarbrick,RE.,Evans,RJ.,Huuse,M.,2007.Birthofamudvolcano: EastJava,29May2006.GSAToday:vol.17,no.2. 9. Demirhan, M., Ozpinar, A., and Ozdamar, L., 2003. Performance evaluation of spatial interpolation methods in presence of noise. International Journal of RemoteSensing24:12371258. 10. DAmbrosio,D.,DiGregorio,S.,Iovine,G.,Lupiano,V.,Rongo,R.,Spataro,W., 2003. First simulations of the Sarno debris flows through Cellular Automata modeling.Geomorphology.54,91117. 11. GeospatialPositioningAccuracyStandardsPart3.,NationalStandardforSpatial Data Accuracy Subcommittee for Base Cartographic Data., Federal Geographic DataCommittee.1998. 12. Hengl, T. Gruber, S. Shrestha, D.P., 2003., Digital Terrain Analysis in ILWIS., Lecturenotesanduserguide.InternationalInstituteforGeoinformationScience andEarthObservation,Enschede,TheNetherlands. 13. Hensen,C.,Nuzzo,M.,Hornibrook,E.,Pinheiro,LM.,Bock,B.,Magalhaes,VH., Bruckmann, W., 2006. Sources of mud volcano fluids in the Gulf of Cadiz indications for hydrothermal imprint. Geochimica et Cosmochimica Acta, 71, 12321248. 14. Habib, E. M. Kim, M. Morgan, I. Couloigner. Dem Generation from High ResolutionSatelliteImageryUsingParallelProjectionModel.

92

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

15. Jacob, Mathias. Hungr, Oldrich., 2006., Debris flow hazards and related phenomena.,SpringerPublishedinassociationeithpraxispublishing,chichester, UK. 16. Jordana, Gyozo., 2007., Adaptive smoothing of valleys in DEMs using TIN interpolation.fromridgelineelevations:Anapplicationtomorphotectonicaspect analysis.,Computers&Geosciences33,pp573585. 17. Kusumastuti,Arse.,Darmoyo,AgungB.,Suwarlan,Wahyudin.,Sosromihardjo, S.P.C.,1999.TheWunutField:PleistoceneVolcaniclasticGasSandsinEastJava., PROCEEDINGS,INDONESIANPETROLEUM ASSOCIATIONTwentySeventh AnnualConvention&Exhibition,October1999 18. Kidner, D.B. 2003. Higherorder interpolation of regular grid digital elevation models.InternationalJournalofRemoteSensing24:29812987. 19. Li, Zhe., Zhang, Juntao., 2001. Calculation of Field Mannings Roughness Coefficient.,AgriculturalWaterManagement,49:153161. 20. Martini, P., Carniello, L., and Avanzi, C., 2004. Two dimensional modelling of flood flows and suspended sediment transport: the case of the Brenta River, Veneto(Italy).NaturalHazardsandEarthSystemSciences,4:165181 21. Murton, BJ., Biggs, J., 2003. Numerical modeling of mud volcanoes and their flowsusingconstraintfromtheGulfofCadiz.MarineGeology.195,223236. 22. Mikos,M.,Fazarinc,R.,Majes,B.,Rajar,R.,Zagar,D.,Krzyk,M.,Hojnik,T.,C etina, M. 2006. Numerical simulation of debris flows triggered from the Strug rockfallsourcearea,WSlovenia.Nat.HazardsEarthSyst.Sci.6,261270. 23. Malet, JeanPhilippe. Rematre, Alexandre. Maquaire, Olivier. 2004., Runout modelling and extension of the threatened area associated with muddy debris flows.,Gomorphologie:relief,processus,environnement.3,p.195210. 24. Naef,D.,Rickenmann,D.,Rutschmann,P.,McArdell,B.W.,2006.Comparison of flow resistance relations for debris flows using a onedimensional finite elementsimulationmodel.,Nat.HazardsEarthSyst.Sci.,6,155165. 25. Otieno., Jennifer. A., 2004. Scenario study for Flood Hazard Assessment in the Lower Bicol Floodplain The Philippine using A 2D flood model Based on the 1988 flood event caused by typhoon Yonning., MSc. Thesis ., International Institute for Geoinformation Science and Earth Observation, Enschede, The Netherlands. 26. Portugues, S.B., Van Asch, T.W.J., A Numerical simulation of the propagation anddepositionofmudanddebrisflowsovercomplexterrain.,DraftPaper. 27. Pieterse, N. M., Venterink, H. Olde., Schot, P.P., Verkroost, A.W.M. 1998., Streamflow,aGISbasedEnvironmentalAssessmentToolforLowlandStreams., The Netherlands Centre for Geoecological Research (ICG), Functioning of LandscapeEcosystemsResearchGroup.,DepartmentofEnvironmentalScience, UtrechtUniversity,Utrecht. 28. PlatteCounty,Missouri.,2004.,BrushCreekWatershedStudy.

93

SpatialModellingandRiskAssessmentofSidoarjoMudVolcanicFlow

29. Rezel, Rohana., 2007., A 2D multiresolution urban flood propagation model usingsimplifiedNavierStokesequations.MSc.Thesis.,InternationalInstitutefor GeoinformationScienceandEarthObservation,Enschede,TheNetherlands. 30. Rahman,M.Z.B.A.(2006)DigitalSurfaceModel(DSM)ConstructionandFlood Hazard Simulation for Development Plans in Naga City, Philippines. MSc. Thesis ., International Institute for Geoinformation Science and Earth Observation,Enschede,TheNetherlands. 31. Sanders, Brett f. 2007., Evaluation of online DEMs for flood inundation modeling.Advancesinwaterresources30.,18311843. 32. Soto,J.I,Comas,M.C.,Talukder,A.R.,2003.EvolutionoftheMudDiapirismin theAlboranSea(WesternMediterranean).,AAPGInternationalConference. 33. Tennakoon, K.B.M., 2004. Parameterisation of 2D Hydrodinamic Models and Flood Hazard Mapping for Naga city, Philippines., MSc. Thesis ., International Institute for Geoinformation Science and Earth Observation, Enschede, The Netherlands. 34. Tomislav,H.,Stephan,G.,Dhruba,P.S.,2003.DigitalTerrainAnalysisinILWIS. International Institute for Geoinformation Science and Earth Observation, Enschede. 35. Vicari,A.,Alexis,H.,DelNegro,C.,Coltelli,M.,Marsella,M.,Proeitti,C.,2006. Modeling of the 2001 lava flow at Etna volcano by a Cellular Automata approach.EnvironmentalModellingandSoftware.22,14651471. 36. Van Noorden, R., 2006.Mud volcano floods Java, disaster plagued Indonesian islandfacesnewthreat.,http://www.nature.com/news.,accessed19June2007. 37. Weng,Q.2002.Quantifyinguncertaintyofdigitalelevationmodelsderivedfrom topographicmaps.In:AdvancesinSpatialDataHandling,SpringerVerlag,New York,pp.403418. 38. Werner,M.G.F.,2001.ImpactofGridSizeinGISBasedFloodExtentMapping Usinga1DFlowModel.,Phys.Chem.Earth(B),26,No.78,pp.517522,2001 39. Wood, J. 1996. The Geomorphological Characterisation of Digital Elevation Models.PhD,LeicesterUniversity,Leicester. 40. Wechsler,SuzanneP.,2003.,PerceptionsofDigitalElevationModelUncertainty byDEMUsers.,URISAJournalVol.15,No.2. 41. Wechsler, Suzanne P., Kroll, Charles N., 2006., Quantifying DEM Uncertainty and its Effect on Topographic Parameters., Photogrammetric Engineering & RemoteSensingVol.72,No.9.,pp.10811090. 42. Wilson, John P. Gallant, John C., 2000., Terrain Analysis: Principles and Application.JohnWiley&Sons,Inc. 43. Wise,S.M.,2007.,EffectofdifferingDEMcreationmethodsontheresultsfroma hydrologicalmodel.Computers&Geosciences33.,13511365.

94