420 likes | 551 Views
Quality of Virtual Data. by Pasi Häkli Finnish Geodetic Institute. Use of real-time GNSS applications, especially network RTK, increased rapidly Network RTK: VRS approch – no information about quality due to limitations in data transfer
E N D
Quality of Virtual Data by Pasi Häkli Finnish Geodetic Institute
Use of real-time GNSS applications, especially network RTK, increased rapidly Network RTK: VRS approch – no information about quality due to limitations in data transfer Virtual data backbone (all processing at the user end done with respect to that) Computed zero baselines in order to get the quality and performance of the system without unnecessary biases e.g. site effects Background and motivation
FinnRef + virtual data Data • FinnRef and GNSSnet.fi networks used • zero baselines: • virtual data generated to the locations of the FinnRef stations • computed baseline between FinnRef and virtual data • FinnRef serves the best possible national ETRS89 coordinates (also independent from VRS network) • ETRS89: EUREF-FIN (=ETRF96), epoch: • plate tectonics 1989.0 • intraplate 1997.0
Computation and analysis • Temporal quality • daily solutions – systematic errors and homogeneity • hourly solutions – more details for post-processing applications • Kinematic solutions – for real-time applications • Spatial quality • National • local
Temporal quality – daily solutions • Long-term daily solutions: • good repeatability (small deviation) • some systematics • some site-dependent effects • Influencing factors: • Reference coordinates • Environmental effects • Instrumentation • … Table. Standard deviation and rms of the 10-month time series of daily solutions (gross errors excluded)
What effects on quality? • reference coordinates • Change at DOY 90/2006 METS
What effects on quality? • reference coordinates • Change at DOY 90/2006 OULU
What effects on quality? • Environmental effects • Snow, … SODA
What effects on quality? • Environmental effects • Snow, … ROMU
What effects on quality? • Instrumentation • Antenna/receiver change VIRO
What effects on quality? • Data • Gaps, bad data, … <20h data KEVO
Spatial quality • results show some systematics in up component – are they spatially correlated somehow? • national level: probably land uplift? • local: modelling of biases in the network or inaccuracies in interpolation? Spatial quality studied from the data between 3-5/2006 so far
VIRO excluded Spatial quality – nationwide • land uplift? computed from FinnRef computed from virtual data average of timeseries converted to annual ”error” and relative to METS
Spatial quality – nationwide • land uplift? • at least a good correlation (VIRO excluded)
satellite • 1) Observations at reference stations of GNSSnet.fi at the epoch tc ref. stn Spatial quality – nationwide • land uplift? • EUREF-FIN epochs: • rigid plate motion: 1989.0 • intra-plate: 1997.0 tc
satellite • 1) Observations at reference stations of GNSSnet.fi at the epoch tc 1997.0 ref. stn Spatial quality – nationwide • land uplift? • EUREF-FIN epochs: • rigid plate motion: 1989.0 • intra-plate: 1997.0 • 2) Forcing coordinates to epoch 1997.0 – generating “a land uplift bias” to model tc
satellite • 1) Observations at reference stations of GNSSnet.fi at the epoch tc ref. stn Spatial quality – nationwide • land uplift? • EUREF-FIN epochs: • rigid plate motion: 1989.0 • intra-plate: 1997.0 • 2) Forcing coordinates to epoch 1997.0 – generating “a land uplift bias” to model • 3) Geometrical displacement of reference station data tc 1997.0 vrs stn
satellite • 1) Observations at reference stations of GNSSnet.fi at the epoch tc ref. stn Spatial quality – nationwide • land uplift? • EUREF-FIN epochs: • rigid plate motion: 1989.0 • intra-plate: 1997.0 • 2) Forcing coordinates to epoch 1997.0 – generating “a land uplift bias” to model • 3) Geometrical displacement of reference station data + interpolated biases tc 1997.0 vrs stn
satellite • 1) Observations at reference stations of GNSSnet.fi at the epoch tc ref. stn Spatial quality – nationwide • land uplift? • EUREF-FIN epochs: • rigid plate motion: 1989.0 • intra-plate: 1997.0 • 2) Forcing coordinates to epoch 1997.0 – generating “a land uplift bias” to model • 3) Geometrical displacement of reference station data + interpolated biases = virtual data at 1997.0 tc 1997.0 vrs stn
Spatial quality – nationwide satellite • land uplift? • EUREF-FIN epochs: • rigid plate motion: 1989.0 • intra-plate: 1997.0 • 1) Observations at reference stations of GNSSnet.fi at the epoch tc • 2) Forcing coordinates to epoch 1997.0 – generating “a land uplift bias” to model • 3) Geometrical displacement of reference station data + interpolated biases = virtual data at 1997.0 FinnRef stn tc • 4) FinnRef data at tc 1997.0 vrs stn ref. stn
5) Zero baseline: = land uplift (1997.0- tc) + some additional biases Spatial quality – nationwide satellite • land uplift? • EUREF-FIN epochs: • rigid plate motion: 1989.0 • intra-plate: 1997.0 • 1) Observations at reference stations of GNSSnet.fi at the epoch tc • 2) Forcing coordinates to epoch 1997.0 – generating “a land uplift bias” to model • 3) Geometrical displacement of reference station data + interpolated biases = virtual data at 1997.0 FinnRef stn tc • 4) FinnRef data at tc 1997.0 vrs stn ref. stn
Spatial quality – local • modelling of biases in the network or inaccuracies in interpolation?
Spatial quality – local • modelling of biases in the network or inaccuracies in interpolation? • to find out: • standard deviations • distance to master station Not so obvious but some correlation
Temporal quality – hourly and kinematic solutions 1 week of data – GPS week 1373 (120-126/2006)
Temporal quality – hourly and kinematic solutions KIVE Hourly
Temporal quality – hourly and kinematic solutions KIVE Kinematic
Summary • Virtual data precise (=small standard deviation) but some systematic errors remain • Biases caused by: • Reference coordinates • Environmental effects like snow • Intrumentation • Modelling of errors and/or interpolation errors • … • Land uplift seen in virtual data – how to fix this? Introducing geodynamical model in VRS generation process? • Key issue if cm-level accuracies are wanted in national reference frames!
Thanks! Part of these results were published in: Häkli, P. (2006): Quality of Virtual Data Generated from the GNSS Reference Station Network. Shaping the Change, XXIII FIG Congress, Munich, Germany, October 8-13, 2006. (http://www.fig.net/pub/fig2006/papers/ps05_01/ps05_01_04_hakli_0520.pdf)