Errors in Positioning

1
Errors in Positioning
Matt King, Newcastle University, UK
2
Static Issues
3
Propagation of Sub-daily signal

• Propagation of mis/un-modelled periodic signals
  (e.g., ocean tide loading displacements) in 24h
  solutions
• Well described by Penna Stewart (GRL, 2003) and
  Penna et al., JGR, 2007.
• Admittances in float ambiguity PPP solutions up
  to 120 in worst case (S2 north component into
  local up)?
• Depends on coordinate component of mismodelled
  signal frequency geometry
• Output frequencies depend on input frequency
• Annual, semi-annual and fortnightly, amongst many
  others

4
Periodic Signals
mm
Penna et al., JGR, 2007
5
Effect in real data

• King et al, GRL, 2008

6
Local Site Issues
7
Kinematic Issues
8
Day Boundary Jumps

• Day boundary jumps regularly seen in kinematic
  positioning
• Filter edge effects and different parameter
  (trop, real valued ambiguities) estimates either
  side of day boundary -gt different coordinate
  estimates
• Continuous filtering not possible in track
  (yet)?
• Mitigation strategy -gt process 2h overlap, then
  crop back

9
Multipath
lt10m baseline
Ragheb et al., 2007

• (Near-) Sidereally repeating sampling of site
  (baseline) MP (86400-236s)?
• Choi et al. showed mean is closer to 86400-246s,
  but varying in time with satellite constellation
• Mitigation strategy sidereal filtering
• Coordinate domain
• Observation domain
• Ragheb et al suggested little difference
• Caution for glacio applications, unmodified
  approach would remove tidal signal

1cm
2cm
After sidereal filtering
Raw positions
10
Response to tidal forcing how much is real?

• Rutford Ice Stream (W Antarctica) experiences
  tidal modulation of its flow
• How much of this signal is real?

Rutford Ice Stream
Window considered here
11
Response to tidal forcing how much is real?

• Two processing approaches
• Precise point positioning (GIPSY)?
• Relative to a base station (Track), 30km away
• Tidal decomposition of de-trended along-flow
  velocity
• PPP very large response at high frequencies
  from little downstream vertical forcing
• e.g., M2 vertical tide 1.5m 2SK5 probably
  lt0.05m

GIPSY (PPP)?
12
Response to tidal forcing how much is real?

• Relative vs PPP
• LF (fortnightly) terms in good agreement
• Relative processing HF terms not sig.

GIPSY (PPP) and Track (relative)?
13
Response to tidal forcing how much is real?

• Relative vs PPP
• In relative processing, smaller diurnal and
  semi-diurnal vertical tide terms not significant
• Same data
• Why the differences?

GIPSY (PPP) and Track (relative)?
14
Response to tidal forcing how much is real?

• Relative processing is rover minus base (TOLL)?
• How much signal is being differenced by the base?
• Gives tidal error spectrum for SEI1
• HF signal evident at base station on rock
• Common GPS satellite position biases?
• Care needed in interpreting HF velocity signals
  in glaciological GPS time series
• LF velocity signals are reliable in all solutions

15
Solution Precision/Accuracy

• Convincing yourself the solution is robust
• Examine track stats Average RMS should be lt20mm
• Pay attention to track warnings (particularly
  lots of clock edits likely a prior coordinates
  are bad)?
• Vary elevation cut-off angle to examine effects
  of systematics
• Examine residuals (may give idea on where low
  elevation effects increase and hence better
  cutoff angle)?
• Coordinate time series looks sensible!
• Precision of solutions
• Sigmas in track output files are relative to the
  precisions you specify in DATA_NOISE
• Hence are likely unreliable
• Include 2nd static site similar distance away
• Examine linear sections of time series and use
  RMS to scale track sigmas

16
Other Issues

• Incorrect Ambiguity Fixing
• Leads to drift in coordinates of metres
• Dont force fixing unless you are sure!
• Cut-off angle
• No right single choice, but probably some wrong
  ones
• Relative PCVs only down to 10 degrees (including
  converted to Absolute PCV)?
• Track Tropo MF is not as advanced as GAMITs