cost es1206: contribution of the vienna university of ... · a fast voxel traversal algorithm for...
TRANSCRIPT
-
COST ES1206: Contribution of the Vienna University of TechnologyMöller Gregor, Böhm Johannes, Umnig Elke, Hinterberger Fabian, Weber Robert
Vienna University of TechnologyDepartment of Geodesy and Geoinformation (GEO), Research Group Advanced Geodesy, Vienna, Austria
e-mail: [email protected]
References• Amanatides,J.,Woo,A.,AFastVoxelTraversalAlgorithmforRayTracing,Proc.Eurographics‚87,Amsterdam,TheNetherlands,August1987,pp1-10.
• KarabaticA.,WeberR.,HaidenT.,Nearreal-timeestimationoftroposphericwatervapourcontentfromground basedGNSS data and its potential contribution toweather now-casting in Austria,AdvancesinSpaceResearch,47,2011,pp.1691-1703
• Moeller,G.,Weber,R.,Boehm.J.,GNSSTomographyoftheAtmosphere-ExpectationfromGalileoFOC,4thInternationalGalileoScienceColloquium,Praque,CzechRepublic,4.-6.December2013
• Moeller,G.,Weber,R.,Boehm,J.,Improvedtroposphereblindmodelsbasedonnumericalweatherdata,IONGNSS+2013Proceedings,Nashville,Tennessee,16.-20.September2013,ISSN2331-5911,Pages2489-2495
• Notarpietro,R.,Cucca,M.,Gabella,M.,Venuti,G.,Perona,G.,TomographicreconstructionofwetandtotalrefractivityfieldsfromGNSSreceivernetworks,AdvancesinSpaceResearch47(2011)898-912
IntroductionAtmospheric research isoneof themajoractivitiesat theResearchGroupAd-vanced Geodesy. Hence we are particularly keen to be able to contribute totheEuropeanCOSTAction(EuropeanCooperationinScienceandTechnology)ES1206whichstartedinMay2013withtheaimtoinvestigateadvancedGNSStropospheric products formonitoring severe weather events and climate.WearepresentinganexcerptofourresearchactivitieswhicharerelatedtoWorkingGroup1-GNSSprocessingtechniques.Thisincludesthewidefieldfromtropo-sphericapriorimodelstorealtimeGNSSanalysis.
GNSS analysis for NRT ZTDsTheaimistosetupanautomaticprocessingtoestimatethetropospherictotaldelayinzenithdirection(ZTD)overAustriainnearreal-time(NRT)withanformalerror better than 2mm and a temporal resolution of 15min. The processingschemewasdevelopedwithintheframeworkoftheProjectGNSSMET-AUSTRIAin2010andwasusedagaintoreprocessZTDsoveratimespanofthreemonthin summer2011 inorder to study its impacton thenumericalweathermodelAROME.Nowwe aregoing to apply theprocessing scheme to 90GNSS sitesmostly situated inAustria and some in neighbouring countries; see Figure 4.
TheGNSSsitesareallocatedwithahorizontaldistanceof5-80kilometresandamaximumheightdifferenceofupto2200meters.Todefinethedatum,observationdata of adjacent IGS and EUREF sites are implemented. The ZTDs, based ona double difference approach, are computed bymeans of the BERNESE V5.2package.InTable2theinputdataandmainparameterssetupinBERNESEarelisted.
GNSS Tomography - Expectations from Galileo FOCGNSS observation data allow recovering the structure of the atmosphere. Inparticularthetemporalandspatialdistributionofthehighlyvariablewatervapourin the lower layersof theatmosphere (up to12km) isof interest for regionalweatherforecastingandprecisepositioning.Inordertoobtainthisinformationthe atmosphere is divided into a 3D voxel-model (see Figure 2) and thewetrefractivity ineachvoxel isdeterminedbyapplyingatomographyapproachasdescribedinNotarpietro(2011).InputdataareSlantWetDelays(SWD)derivedbyGNSSobservations.
Figure 2: 3D voxel structure of theatmosphere above Austria with a spatialresolution of ~18.0 km and 16 pressurelevels. A pressure value of 200 hPacorrespondstoaheightofabout11800m.EachbluedotrepresentsaGNSSreferencesite.
IndependentfromthereconstructiontechniqueahighstationdensityandalargenumberofGNSSobservationsarenecessarytoobtainanenhancedresolution(especiallyinverticaldirection)andstabletomographyresults.Upcomingsatellitesystems like Galileo will push on the tomography approch. In order to studyits impact on GNSS tomography a full operational capability (FOC) of Galileosatellitesissimulatedandanalysed.ForGPSandGLONASStheconstellationthatwasoperationalinSeptember2013isassumed.Figure3showsthelines-of-sighttobeobservedatasitenearVienna,Austria.
Figure3:Lines-of-SighttoallGPS(redline),GLONASS(green line) and Galileo (blue line) satellites inviewatGNSS siteBaden (nearVienna) -plottedforasingleepoch.Thedirectionstothesatellitesare calculated from broadcast ephemeris. Both,thesitecoordinatesandthedirectionsareinputforthevoxeltraversalalgorithm;seeAmanatidesandWoo(1987).Thisalgorithmisusedtodetectvoxeltraversalsandtodeterminethenumberofemptyvoxels(voxelswhicharenotpassedbyanyline-of-sight).
Galileoprovidesabout50%additionalobservationsinnewazimuthandelevationangles(comparedtoGPS+GLONASS).Thusthedistributionofslantpathdelays,especiallythehorizontalcoverage,becomesmorehomogeneousandthenumberofemptyvoxelscanbesignificantly reduced.Weanalysed theobservationofallthreesatellitesystemsandfoundthatafter15minutesinaverage50%ofthevoxelsremainemptyifonlyGPSobservationsareused.IfGLONASSandGalileoobservationsareadded, thenumberofemptyvoxels canbe reducedby35%anditstemporalvariation(duetothechangingnumberofsatellitesinviewovertime)iswidelycompensated.
Infurtherconsequencethetomographyapproachwillbenefitfromtheseadditionaldataanditssolutionwillbecomemorestable.InourcooperativeprojectGNSS-ATomwecontinueour studies inorder tofind the „best“ approach toobtainSWD(differenced/undifferenced)fromGNSSobservations,toreconstructNwetfromSWDandtoassimilateitintohigh-resolutionweathermodels.
Real Time GNSS processingRecentlytheworkontheprojectPPPServehasbeenfinished.Withintheframeworkoftheprojectwedevelopedarealtimecapablesoftwarefortheestimationofso-calledsatellite-phase-biaseswhichallowfor integerambiguity resolution inPPP.TheprocessingschemeofthesoftwarealsoincludestheestimationoftheZTDatevery stationofa chosennetwork.Currentlyourmodeluses theNiellMF, thatconsidersdifferentobliquity factors for thewetanddrycomponents.Thewettroposphericdelayisestimatedontopofanapriorimodelasarandomwalk process within a Kalman filter together with the other parameters. FirstcomparisonsbetweenourZTDs,estimatedwithin thefixedPPPsolution,withthe ZTDs of a network solution indicated a very good agreement at the fewmillimeter-level.
Tropospheric error correction models GPT2(w)In2013twoupdatestotheGlobalMappingFunction(GMF)andtheGlobalPressureandTemperaturemodel (GPT)havebeen releasedwhichare calledGPT2andGPT2w.Inputparametersarejustthedateandthestationcoordinates(longitude,latitude and ellipsoidal height). GPT2 provides pressure, temperature, watervapourpressure,temperaturelapserateandmappingfunctioncoefficientson5degreegridsasaveragevalues,annualandsemi-annualvariations.Therewithitispossibletocalculatethezenithhydrostatic(ZHD)andwetdelay(ZWD)andthemappingfunctionforanyreceiveronearthatanytime.
Figure 1: Tropospheric delay in Zenithdirection (ZTD) at GNSS site Malindi (Kenya,Africa) delivered by CDDIS / IGS (blue), thetropospheric delay model RTCA MOPS (red),GPT2 (black) and GPT2w (green). The ZTDcourse is dominatedby semi-annual periods.GPT2wbenefitsfromthenewparametersandtheimprovedwetdelaymodel.
Theextension‚w‘inGPT2wisrelatedtotheadditionalparameterswatervapourdecreasefactorandmeantemperaturewhichareprovidedon1degreegridsinordertocalculatethezenithtroposphericdelay(ZWD)morereliably.ThepositiveimpactofthenewparametersandtheimprovedwetdelaymodelisseeninthecomparisonwithZTDsprovidedbyIGSin2012(seeFigure1).ThebiasandRMScouldbereducedsignificantly-especiallyincomparisontocommontroposphe-remodelsforSatellite-basedAugmentationSystems(SBAS)likeRTCAMOPS.
Model Bias (cm) RMS (cm)GPT2w -0.03 4.13GPT2 -0.37 5.05
RTCA MOPS -2.60 6.39
COST ES1206 Workshop - GNSS4SWEC | 26. - 28.02.2014 | MUNICH
Observation data GPS/GLONASS, L1 & L2, 30 sec sampling rate, 7 degree cut-off
Orbits and ERP IGS Ultra Rapid Products
Datum definition Constraints on IGS & EUREF sites (0.1 mm, ITRF2008, 2005.0)
A priori model & MF GPT and wet GMF
Parameter Spacing 15 min (ZTD), 12 hours (Gradients)
Rel. a priori sigmas 2 mm (ZTD), 0.2 mm (Gradients)
Ambiguity Solution Sigma-Strategie (L5/L3), L1/L2
Table 2: Processing parameters
Table 1: Bias and standard deviation of the differen-ces between ZTDs delivered by CDDIS (IGS) and the three blind models RTCA MOPS, GPT2 and GPT2w - calculated for 320 GNSS sites over the period 2012.
AcknowledgmentsTheprojectsPPPServe,GNSSMET-AUSTRIAandGNSS-ATomarefundedbytheAustrianResearchPromotionAgency(FFG).GPT2whasbeendevelopedintheframeworkoftheprojectTROPSYwhichisfundedbytheEuropeanSpaceAgencyin theframeof theTechnologyResearchProgram.Wewould like to thankourprojectpartnersfromTUGraz,TCAandZAMGandinparticularAntonioMartellucciandRaulOrusPerezatESA/ESTECfortheirvaluablecontribution.
Re
al T
ime
N
ear
Real
Tim
e
Pos
t Pro
cess
ing
/ A p
riori
mod
el
Figure4:Distributionofthe90sitesprovidingGNSSobservationinnearreal-time