Solar Radiation Tool within ArcGIS For Identifying Areas Prone to Snow Phenomena Santiago Gámiz Tormo Geographical Information Management MSc Student Cranfield University, Cranfield, Bedfordshire, MK43 0AL Supervisor: Tim Brewer - [email protected] www.cranfield.ac.uk/courses/masters/geographical-information-management 1. Background 3. Approach and Methodology 2. Aims and Objectives 4.- Findings and Recommendations 5. Conclusions Photo by Thomas Quentin IDW Kriging Classification and Regression Tree (CART) Roadtravelisnowadaysanessentialpartofoursociety.Roadusersexpecttotravelno mattertheseasonortheweatherconditions. M ainlyduringthewinter,unfavourable weatherconditionsmayhaveseriousconsequencesforsafety,andlessimportanton delays. According to John E. Thornes et.al from the U niversity ofBirmingham,the benefitsofwintermaintenancehavebeenestimatedtobeabouteighttimesthecost. TheincreasingroleofGI Stechnologiesaspartoftheroadwintermaintenanceprograms (developedinpartby Highway Agency inU K)andR oadW eatherInformationSystems ( RWI S )( M etOfficeRoadWeatherForecastinginUK)hasimprovedthesesystemsresults. Between the analysed factorsfor these R W I S,a sky view component has been demonstratedashighlyrelevant( L eeChapm anandJ.E. Thornes). T hus,theSolarR adiationtoolsavailablewithintheArcGI S softwarepackagehavebeen testedtochecktheirsuitabilitywithatrialpredictionmodelforsnowphenomenaforthe Lincolnshireroadnetwork. O verlay ofviewshed with su nmap O verlay ofviewshed with skymap V iewshed m apped onto sky view V iewshed S olarRadiation toolmaptypes:S UNMA P forthe representationofthe su n position overadefined period oftimeand S KYM A P ,whi c h d epi cts the skysectors infl u encing the qu antity ofinsolation (incoming solarrad iation). Theresearchapproachofthisprojectisbased ontheuseofClassificationandRegression Trees(CAR T ).T histechniqueisused topredict variablevaluesbasedonrelationshipsbetween differentdatasets,bythedevelopmentofa classificationbasedonthresholds. Aims Assessthevalueofthesolarradiationtoolsforidentifying“atrisk”areasinbadweather. Objectives Developadatadrivenapproachtoselectareaspronetosnowevents. Assessthesuitabilityofthesolarradiationtoolasacomplementfortheanalysisand identificationoftheseareas Analysepossiblerelationshipsbetweenareaswithalongerdurationofdirectinsolation, higherdirectincomingsolarradiationandspatialsnowmeltingpatterns. Krigingpredictionsofrainfallvaluesweresatisfactory astherewasaconsiderablenumberofsamplingpoints (BADC stations).Blockkrigingwasdemonstrated tobea suitablesupportforthisstudytakingintoaccountthe ESAdatasets,asshownonthecrossvalidationresults. IDW resultswerevery usefultakinginto accountthe samplingsizelimitationsfortheothermeteorological factors. T heCAR T modelperformedw elleventhoughthesizeoftheinputdataset(inrelation to thetaskto bedone).M eteorologicalandgeographicalfactorsidentifiedasimportant bytheliteraturew erealsoidentifiedbytheCAR T. CARToutputsallowedacorrelationanalysisbetweenthespatiallocationofthesnow phenomenaand the SolarR adiation tooloutputs.R esultsfrom thisanalysiswere overlaid with thecurrentgrittingroutesoftheLincolnshireroad networkasthefinal stepfortheconclusionextraction. The resultsofthe approach taken in thisresearch may be furtherimproved in an extended timeframe.Itwould allow theuseofothergeostatisticaltechniquessuchas CoKrigingorM onteCarlosimulations,whichcouldpotentiallyleadintothedevelopmentof ariskmap.Anotherimportantconstraintwasfoundintheavailabilityofspatio-temporal datasetsregardingweatherand snow phenomena,which will improve the CAR T performanceandthustheassessmentoftheSolarRadiationtoolssuitability. Asintheconsulted literature,thesky view factorwasdemonstratedtobeanimportant drivenfactor,whichisapositivefeedbackforafurtheruseoftheSolarRadiationtoolsfor similarpurposes. Therewasaslightrelationshipbetweenareaswithalongerdurationofdirectinsolation, higherdirectincomingsolarradiationandspatialsnowmeltingpatterns Asanoverallconclusionthedepthofresearchheredevelopedisnotenoughtoassess thesuitability ofthesetools.Howeverthismethodology canbeimproved withtheabove suggestionsandatdifferentscales,whichwillallowtheextractionoffurtherconclusions. Thus,asetofmeteorologicalandgeographicalfactors were selected togetherwith the outputsofthe Solar R adiationtools(allofwhichweretreated astheCAR T inputs),toinvestigatetheirrelationshipwithsnow events.SolarR adiationtools from ArcGI S enabledthe quantification(and inclusion withintheCAR T)ofsolar insolationandinsolationhoursparametersoverthe studyarea. M eteorologicaldatawereobtained from theBritishAtmosphericDataCentreweather stations.Inordertoobtainthesamespatialdataforthesepoints(dataavailableonlyfor theweatherstationlocations),KrigingandIDW geostatistictechniqueswereapplied.The krigingtechniqueusedwasO rdinaryKrigingwithablocksupporttobetterfitintotheESA data. Supervisor: Tim Brewer - [email protected] Meterological Factors Geographical Factors Linconlshire W eatherS tations Linconlshire DTM (10m resol.) Representationofthe C A RTinpu ts:meteorologi c aland geographi c alfac torval u es fortheESA snowpoints. C A RToutputs: ESAsnowlocations. C A RTinputs: meteorologi c al and geographi c alfac torval u es fortheESA snow points. Thesnow spatio-temporaldatarequired(outputintheCAR T)was extractedfrom the SM OS L2 SoilM oisture U serDataProduct ( M I R _S M U DP2)forthegridpoints.