Detection of GPS station position errors due to instrumentals and environmental effects

1
Detection of GPS station position errors due to
instrumentals and environmental effects

• Thomas Herring, tah_at_mit.edu,
• Department of Earth Atmospheric and Planetary
  Sciences, MIT

2
Overview

• Examine origin and diagnostics of systematic
  deviations in GPS time series due to instrumental
  and environmental effects (not monument
  stability)
• Cases
• Antenna failures
• Environmental effects Snow but also growing
  vegetation.
• Diagnostics
• TEQC MP1 and MP2 Estimates of pseudorange
  multipath.
• Signal to noise ratio
• Phase residual RMS Scatter

3
Diagnostics

• Pseudorange multipath at L1 and L2 MP1 and MP2
• Difference between L1 and L2 pseudorange computed
  from L1 and L2 phase measurements
• Values generated by TEQC using an algorithm that
  accounts for cycle slips and offset between phase
  and range.
• Values shown are an RMS over the duration of the
  rinex file (24-hours)
• Phase RMS scatter
• RMS scatter of the ionospheric free phase
  residuals from post-processing.
• Signal-to-noise ratio
• Commonly available now in rinex files although
  quality is variable
• Here we use an average value for the day
• MP1/2 and SNR can be used before processing data.

4
Multipath diagnostic

• Example of antenna failure
• Site NDAP (Nucleus site)
• Site had strong annual signal in north but not in
  height (rather strange if loading effect)
• Effect of antenna change on position and
  multipath values
• Example a LTUT (no MP1 in this case Should be
  available but is not?)

5
Site NDAP Site had shown annual signal in North
for a long period of time In July 2006, antenna
replaced.
6
(No Transcript)
7
NDAP North/MP1
The MP1 values before antenna switch are
correlated with temperature. Old antenna had
failed patch element.
8
LTUT (another example)
Antenna change
9
SNR analysis

• Many RINEX files now contain S1 and S2 and the
  units are now better defined
• Analysis here
• Fit elevation angle dependent model to SNR at L1
  and L2
• Extract the fitted values at zenith and 20
  degrees elevation
• Example on following slides

10
P708 near Yellowstone
Site notes indicate that so much snow during
visit that they could not get to monument
11
(No Transcript)
12
P708 L1 SNR (no snow)
13
P708 L1 SNR (snow)
Note Low elevations not that effected by snow
14
P708 L2 (no snow)
15
P708 L2 SNR (snow)
As average Use fit value at zenith
16
P708 Height and SNR
17
P708 Height/MP1 overlay
Generated online from interactive
plotshttp//pboweb.unavco.org/plotting/src/publi
c/gps_qc_soh_page.php?stationP708
East effects here
18
P708 East MP1
Jump in east detected in multipath RMS
19
P708 Height/Phase RMS
Phase RMS also catches early event but others
also.
20
Summary for snow events

• Both SNR and MP1/2 work well as indicators of
  snow
• Low elevation angle SNR may indicate horizontal
  position error (not tested)
• Phase noise RMS has problems in that atmospheric
  noise that might not effect position very much
  can increase noise
• SNR and MP1/2 can be computed before data
  analysis as a quality check.

21
Other examples

• Cases generated using on-line tools at
  http//pboweb.unavco.org
• Most of these are probably snow events (cases
  based on site visit notes and height of site)

22
P665
23
P665 North MP1
24
Alaska Volcano (AV04)
25
AV04 time series
26
AV04 Height and MP
MP1/2 indicates that most height signal due to
snow coverSome interesting hysteresis after snow
melt
27
Summary

• Assessment of data quality through use of
  pseudorange multipath measures and SNR analysis
  looks promising. (Available for PBO sites at
  UNAVCO current UNAVCO summary files do not have
  enough significant digits.)
• Detection of antenna problems and snow effects
  possible. Vegetation effects also probably can
  be detected.
• Phase RMS scatter is probably not a good detector
  in that it can be disturbed by effects that dont
  greatly change the station positions.
• Not discussed but needed Detection of sites with
  corrupt data and how these should be treated in
  archives. Important in re-processing so that
  station numbers are not wasted in bad sites
  (i.e., most ACs use fixed number of stations).