Validation of slant delays acquired by different techniques within GNSS4SWEC Benchmark campaign Current status M. Kačmařík, J. Douša, G. Dick, F. Zus, H. Brenot, E. Pottiaux, G. Möller, P. Václavovic, L. Morel, W. Rohm, P. Hordyniec GNSS4SWEC 3rd Workshop, Reykjavik, March 8 - 10, 2016
Enjoy your stay in Iceland!
Slant Delays - introduction Slant delay represents an influence of neutral atmosphere on GNSS signal in the AZI/ELE of signal STD = SHD + SWD Slant total delay = slant hydrostatic delay + slant wet delay STD = ZHD*mfH + ZWD*mfW + mfG*( GN*cos(A) + GE*sin(A) ) + res – mpt ZHD = zenith hydrostatic delay ZWD = zenith wet delay mfH, mfW, mfG = mapping function for dry part of delay, wet part, horizontal gradients GN, GE = horizontal gradient for north-south, east-west direction A = azimuth angle of observation res = post-fit residual mpt = multipath SIWV (Slant Integrated Water Vapour) = k * SWD
Sources of STDs for validation GNSS solutions: Solution Institution Strategy Software GNSS El. cut-off Mapping function Products ZTD / gradients interval Post-fit residuals CNAM Le CNAM DD Gamit GPS 3 ° VMF1 IGS 1h / 1h NO GFZ GFZ Potsdam PPP EPOS 8 7 ° GMF 15 min / 1h YES and NO GOP_F GO Pecny PPP, forward filter G-Nut/Tefnut IGS final 2.5 min / 2.5 min YES GOP_S PPP, back-ward smoothing ROB_G ROB Bernese 5.2 GPS+GLO CODE final ROB_V TUO 1h / 3h TUW TU Wien Napeos 30 min / 1h
Sources of STDs for validation NWM raytracing: PMF software (GFZ) using fields from global ERA INTERIM and global NCEP GFS model (Florian Zus, GFZ) WUELS software using fields from ??? (P. Hordyniec and W. Rohm, WUELS) BIRA software using fields from regional Aladin-CZ model (H. Brenot), STDs were not delivered yet for the validation Water Vapour Radiometer (WVR) – station POTM/POTS
Selected stations POTS + POTM GOPE LDB2 + LDB0 WTZR + WTZS + WTZZ KIBG SAAL
Validation results – single stations results for whole Benchmark period = 56 days all result statistics in mm noR version = residuals were not used in STD computation GNSS x GNSS = GFZ solution used as a reference solution zenith direction = all STD were mapped back to the zenith using simple 1/sin(elevation)
POTS GNSS vs. GNSS zenith
POTS GNSS vs. NWM zenith
POTS GNSS vs. WVR zenith
POTS, GNSS vs. GNSS, slant
POTS GNSS vs. NWM slant
POTS, GNSS vs. WVR, slant
GOPE LDB2 GNSS vs. GNSS zenith WTZR WTZZ
GNSS vs. NWM ERA zenith GOPE LDB2 WTZR WTZZ
Basic summary GNSS vs. GNSS: GNSS vs. WVR: all solutions (except GOP_F) are in a very good agreement, however, GOP_F was included for specific assessment of real-time capable technique (forward Kalman filter only) bias between GFZ and other solutions is usually below 2-3 mm in zenith direction (except station POTM where GFZ solution has some problem) SDEV between GFZ and other solutions is typically between 4 and 6 mm in zenith direction (except GOP_F) while SDEV between GFZ and GFZ_noR around 3-4 mm fractional SDEV stays almost stable with increasing elevation angle (not showed) GNSS vs. WVR: WVR has a positive bias around 6 mm vs. GNSS in zenith direction SDEV values are a little bit higher than for GNSS x GNSS comparison at all elevations, especially at the lower ones all GNSS solutions perform very similarly in terms of bias and SDEV (except for GOP_F being a little bit worse in SDEV) all GNSS solutions without residuals (noR) perform slightly better than their versions including residuals
Basic summary GNSS vs. NWM: NWM_GFS has a bias of approximately 5 mm versus NWM_ERA in zenith direction in terms of standard deviation the agreement between NWM_ERA and NWM_GFS is similar to agreement between NWM_XXX and all GNSS solutions all GNSS solution perform very similarly (GOP_F is not visibly worse) all GNSS solutions without residuals (noR) perform slightly better than their versions including residuals NWM_WUELS solution has a significant problem (bias around 160 – 200 mm in zenith direction and standard deviation around 20 - 30 mm when compared to GNSS/NWM_ERA/NWM_GFS solutions), WUELS is working on problem elimination, results not shown in the presentation
Validation results – dual stations dual stations are supposed to deliver identical/similar slants results for whole Benchmark period = 56 days all result statistics in mm noR version = residuals were not used in STD computation zenith direction = all STD were mapped back to the zenith using simple 1/sin(elevation)
TUW LDB0 x LDB2 GNSS vs. GNSS zenith GFZ GOP_F GOP_S
Paper in a preparation main topics: M. Kačmařík, J. Douša, G. Dick, F. Zus, H. Brenot, E. Pottiaux, G. Möller, P. Václavovic, L. Morel, W. Rohm, P. Hordyniec, Validation of slant total delays from different techniques, GNSS4SWEC Special Issue main topics: comparison of STDs from different techniques (GNSS, NWM, WVR) for 10 stations and whole Benchmark period evaluation of quality of GNSS STDs evaluation of differences in GNSS STDs obtained by different software/strategies assessment of influence of post-fit residuals on quality of GNSS STDs = using results from dual stations, comparison of GNSS vs. NWM/WVR. Call attention to the problematic information about state of the troposphere stored in residuals.
Extended activities (future) Purpose - deepen the basic study if work on any of proposed topic is done before the submission of the paper (summer 2016), it can be included in it Google document: https://docs.google.com/document/d/1oK- 2JT5E4QSoaeBXd4MxvrXsSp81Y2uB7grhkL_axvg/edit# please join!
Extended activities (future) Detailed study of GNSS post-fit residuals Topics, thoughts: eliminate multipath/azimuth dependent influence from residuals ??? Coordinator: ??? Participant: Michal Kacmarik
Extended activities (future) Detailed study of GNSS post-fit residuals Topics, thoughts: eliminate multipath/azimuth dependent influence from residuals evaluate azimuth-dependency of GNSS post-fit residuals (there might a significant effect of GNSS post-fit residuals in comparison to NWM or WVR data) ??? Coordinator: ??? Participant: Michal Kacmarik
Extended activities (future) STD validation of specific weather events during Benchmark period Topics, thoughts: take a closer look on a selected severe/calm weather events and evaluate behaviour of STD during those specific periods all validation is already done at the level of individual days so it is mainly only a matter of time to work ready results Coordinator: ??? Participant: Michal Kacmarik
Extended activities (future) Improvement of delays and gradients from NWM ray-tracing Topics, thoughts: New methodology for NWM-raytracing of Hugues Brenot (see his presentation in Reykjavik) Coordinator: Hugues Brenot Participant:
Extended activities (future) Evaluation of GNSS processing settings on quality of STDs Topics, thoughts: quantify influence of GNSS processing parameters on ZTD/gradients/post-fit residuals (time interval of ZTD/gradient estimation, time interval of observations, mapping function, cut-off angle, etc.) Coordinator: Participant: Michal Kacmarik