EGU 2009,Session G4,21 April,2009 1 EPOS-RT: Software for Real-time GNSS Data Processing Maorong Ge, Junping Chen, Gerd Gendt Department of Geodesy and Remote Sensing, Deutsches GeoForschungsZentrum Contact: junping.chen@gfz-potsdam.de EGU 2009,Session G4 Vienna, Austria 19-24 April, 2009
EGU 2009,Session G4,21 April,2009 2 G-SEIS Project GPS-SurfacE Deformations WithIn Seconds Landslide, Volcano, GPS Shield,... GITEWS Project German-Indonesian Tsunami Early Warning System Motivation GPS Shiled, GPS Buoy and Reflector Real-Time IGS(IGS-RTPP) Real-time orbits and clock products
EGU 2009,Session G4,21 April,2009 3 Special Consideration Base of the future GFZ software Multi-technology GNSS (GPS, GALILEO, GLONASS, COMPASS, ) SLR VLBI Multi-Function Real-time/Post-mission Static/Kinematic/Dynamic High spatial&time resolution Integrated solution
EGU 2009,Session G4,21 April,2009 4 New Concepts Software Structure Architecture Content Software evaluation Real-time PPP Real-time network solution Investigation of IGS-RTPP Conclusions
New Concepts The station-satellite-concept does not fit to the requirement. Think about a LEO with GPS receiver, SLR reflector, KBR and even a GPS transmitter (Pseudo satellite). What is it, a station or a satellite? Platform(PF)/Instrument(INST)/observation(OBS) Parameters are defined within these elements Data and parameters for the modelling of an INST1(receiver) PF1(marker) observation are available via pointer or simple index. OBS INST2(transmitter) PF2(satellite) No difference to handle data from different systems and different techniques EPOS-RT(SRIF) is developed based on these new concepts EGU 2009,Session G4,21 April,2009 5
Main Features of EPOS-RT(SRIF) Blackbox design For users, only one ctrl file is needed Multi-GNSS processing GPS,GLONASS,GALILEO Real-time and Post-processing supported Epoch-wise processing Network solution and PPP supported Un-difference modeling Same software(srif) for orbit determination/clock estimation/ppp/network solution EGU 2009,Session G4,21 April,2009 6
EGU 2009,Session G4,21 April,2009 7 Architecture - PPP based Real-Time GFU Orbits Positioning Service Internet connection IGS Ultra IGS hourly data GFU orbits OBS Streaming RT Clock NTRIP Products NTRIP Products EPOS Hourly data IGS Ultra Rapid GFU prod SPX3ORB ORB/CLK BNC BNC2SRIF SRIF(Clk) NTRIPcoder Orb/Clk/ERP (UPD) NTRIPdecoder SRIF(PPP) Positions Control Unit / User Interface RT PPP GPS receiver
EGU 2009,Session G4,21 April,2009 8 Architecture - Real-Time Network Deformation Monitoring GFU Orbits Internet connection IGS Ultra IGS hourly data GFU orbits OBS Streaming RT Network NTRIP OBS EPOS Hourly data IGS Ultra Rapid GFU prod SPX3ORB ORB/CLK BNC BNC2SRIF SRIF(NET) Positions Local Control Unit / User Interface NTRIP OBS communication NTRIP OBS GPS receiver
Clock Estimation&PPP with IGS final GFZ IGR Network Orbits&ERP fixed to IGS final products Station coordinates fixed to IGS combined solution Estimated satellite&receiver clocks and ZTD at each stations Compared to IGS final clocks, STD=0.04ns (12mm) PPP: Kinematic position DUBR, 2 cm each component orbits Clock difference(ns) Clock RMS (ns) 0.5 0.4 0.3 0.2 0.1 0-0.1-0.2-0.3 PRN_G02 PRN_G03 PRN_G04 PRN_G05 PRN_G06 PRN_G07 PRN_G08 PRN_G09 PRN_G10 PRN_G11 PRN_G12 PRN_G13 PRN_G14 PRN_G15 PRN_G16 PRN_G17 PRN_G18 PRN_G19 PRN_G20 PRN_G21 PRN_G22 PRN_G23 PRN_G25 PRN_G26 PRN_G27 PRN_G28 PRN_G30 PRN_G31-0.4 0 24 48 72 96 120 144 168 192 216 240 264 288 0,20 Epoch number (sample rate 300 s) Position Difference 0,15 0,10 0,05 0,00 PRN_G02 0,20 0,15 0,10 0,05 0,00-0,05-0,10-0,15 PRN_G04 PRN_G06 PRN_G08 Epoch Number(Sampling rate 30s) EGU 2009,Session G4,21 April,2009 9 PRN_G10 PRN_G12 X Y Z PRN_G14 PRN_G16 PRN_G18 PRN_G20 PRN_G22 PRN_G24 PRN_G26 PRN_G28 PRN_G31-0,20 60 300 540 780 1020 1260 1500 1740 1980 2220 2460 2700
Real-Time Clock Estimation & PPP IGS real-time network from BKG NTRIP, about 85 stations,13. February 2009 Orbits&ERP fixed to GFZ ultral-rapid products (3h update) Station coordinates fixed to our PPP results Differenced Observations between epoch Estimated satellite&receiver clocks and ZTD EGU 2009,Session G4,21 April,2009 10
EGU 2009,Session G4,21 April,2009 11 0.5 0.25 0-0.25 Real-Time Clock Estimation & PPP -0.5 50400.00 54000.00 57600.00 61200.00 64800.00 68400.00 72000.00 75600.00 79200.00 82800.00 STD of clock(ns) 0.2 0.16 0.12 0.08 0.04 0 G03 G04 G05 3 hour initialization G06 G07 G08 G09 G10 G11 G12 G13 G14 G15 G16 G17 G18 G20 G21 G22 G23 G24 G25 G26 G27 G28 G29 G30 G31 G32 Comparison with GFZ Rapid Clock, session by session Few short gaps Average STD =0.1ns PPP precision after convergency(bor1): (2.0,1.3,2.2) cm
EGU 2009,Session G4,21 April,2009 12 Network solution in Post-processing 16.04.2006,sampling 30sec GFZ ultra-rapid orbits LEIJ & BOR1 fixed, LEIJ WTZJ 248km Ambiguity fixing with Lambda method(teunissen et al, 1995) Kinematic coordinates compared to IGS weekly solution (WTZJ) (1 cm in Horizontal, 2 cm in height) initialization
EGU 2009,Session G4,21 April,2009 13 Network solution with single frequency receiver 16.04.2006,sampling 30sec WTZS WTZJ 700m Ambiguity fixing with Lambda method (L1 only) Kinematic coordinates compared to IGS weekly solution (WTZJ) (5 mm in Horizontal, 1 cm in height) Too few observed satellites
EGU 2009,Session G4,21 April,2009 14 Real-Time Network solution Data from BKG NTRIP Streaming:1 hz,8 stations(6 fixed,2 kinematic)16-17,april,2009 Ambiguity fixing with Lambda method GFZ ultra-rapid orbit (no update of orbits) Kinematic coordinates compared to IGS weekly solution (BZRG)
EGU 2009,Session G4,21 April,2009 15 Investigation of IGS-RTPP Status of IGS-RTPP, satellite clock: 0.3ns compared to IGR clock (but we have 0.1 ns) Network reason?? Orbit update strategy??
Network effect Orbits&ERP fixed to GFZ ultral-rapid products (predicted part, session from 02h-08h) Different networks Comparison with IGS Final Clock Num. 85 70 60 50 40 30 20* bias(ns) -0.25-0.25-0.25-0.25-0.24-0.24-0.34 STD(ns) 0.10 0.10 0.10 0.10 0.11 0.11 0.17 No difference when station number more than 30 (globally well distributed) * Satellite G15 lose tracking in the first few hours EGU 2009,Session G4,21 April,2009 16
Network effect Orbits&ERP fixed to GFZ ultral-rapid products (predicted part, session from 02h-08h) Different networks Comparison with IGS Final Clock Num. 85 70 60 50 40 30 20* bias(ns) -0.25-0.25-0.25-0.25-0.24-0.24-0.34 STD(ns) 0.10 0.10 0.10 0.10 0.11 0.11 0.17 No difference when station number more than 30 (globally well distributed) * Satellite G15 lose tracking in the first few hours EGU 2009,Session G4,21 April,2009 17
EGU 2009,Session G4,21 April,2009 18 Orbit update Orbits&ERP fixed to GFZ ultral-rapid products (predicted part) Network with 30 globally distributed real-time station Comparison with IGS Final Clock Clock estimation using different orbits Orbit igu igu igu igu igr 00-03h 01-04h 02-08h 00-24h 00-24 bias(ns) 1.23 0.54-0.24 1.41 1.20 STD(ns) 0.10 0.10 0.11 0.20 0.12 Significant improvement using updated orbits Similar precision: < 6h predicted orbits and estimated orbits No big difference between orbit updating every 3h and every 6h
EGU 2009,Session G4,21 April,2009 19 Conclusions New concepts are implemented No difference between handling of different systems and techniques Performance of EPOS-RT Satellite clock estimation precision : ~0.03ns (post-processing.gfz igr network) ~0.10ns (real-time. igs real-time network) PPP kinematic coordinate precision: Horizontal 1~2 cm, 3D <4 cm Kinematic coordinate in network solution: Horizontal ~1 cm, Height 2-3 cm Investigation of the status of IGS RTPP No difference in clock estimation using all or a part of existing globally distributed network No difference with different orbit update interval (6 hour is sufficient) Follow-on Co-operate with partners and implement in Tectonic active area.
Thank you! EGU 2009,Session G4,21 April,2009 20