Comparison of ILRS Station Positions A Series only

1
Comparison of ILRS Station Positions (A Series
only)

• Van Husson

2
SINEX Issues

• Apriori positions
• AUSLIG old positions for
• Zimmerwald (x 0.176, y 0.024, z 0.184)
• San Fernando (x 0.244, y -0.023, z 0.172)
• Greenbelt (x -0.004 y 0.001, z 0.042) from
  991205
  
• Computer round-off/truncation
• JCET (occasionally 0.5 mm)
• CSR (occasionally 0.5 mm)
• AUSLIG (occasionally 0.5 mm)
• NERC (occasionally 0.5mm)
• Bias Parameters

3
Effect of Modeling Errors

• Any modeling errors (e.g. GM, satellite CoM,
  troposphere, biases, data weighting, data
  treatment) will be absorbed into the station
  coordinates, mostly height.
• What can be done to mitigate these errors?

Radial Height
Baselines Scale
Modeling Error
Range Bias
4
Site Position Estimation

• Case I (perfect conditions)
• No biases (range, time, scale) in any sites
• Modeling is perfect (i.e. perfect orbit, bias
  estimates 0)
  
• Site positions perfect

Case I conditions do not exist.
5
Site Position Estimation (cont)

• Case II (near perfect conditions)
• No biases, except a fixed RB in one site XYZ0
• A) Modeling is perfect, can accurately detect
  the RB in site XYZ0
  
• Site positions are perfect,
  even for site XYZ0
  
• B) Modeling is almost perfect, but slight
  error in RB estimate for site XYZ0
  
• Site XYZ0 will have a height
  error (same direction of RB estimate error),
  
• other site heights will have a
  slight height error in the opposite direction
  
• C) Modeling is almost perfect, assumed 0
  bias for all sites including XYZ0
  
• Site XYZ0 height is lower
  than truth by RB (opposite
  
• direction of true RB),
  other site heights slightly higher
  

Case II conditions do not exist.
6
Site Position Estimation (cont)

• Case III (TRUTH)
• Few mm RBs exist in best sites, TBs should not
  exist in any site.
  
• A few sites occasionally have significant TBs.
• A few sites occasionally have scale biases.
• Modeling is not perfect. Errors in
• Bias Estimation
• LAGEOS/Etalon Center of Mass
• Troposphere
• Earth Center of Mass
• Earth Rotation
• GM
• Etc.

7
Issues

• Can range bias be separated from height and to
  what level?
  
• What are the limiting factors?
• Some solutions violate REALITY. Is this a
  problem?
  
• What causing the variations in these solutions?

8
1999 NP Data Volume
9
1999 Bias Problems
10
ILRS Network
11
Bias Corrections
12
Sites in Solution
13
Sites missing in 032899 Solution
14
Data Treatmentby site/28 days
1. DGFI estimated pass-by-pass biases, when the
suspected range or time bias exceeded 30mm
or 30 microseconds, respectively.
2. IAAK assumed no biases, except a constant RB
for Maidanak and Komsomolsk 3. For the core sites
, assumed no biases, for other sites RB estimated
every 7 days per satellite
There are significant differences in data
treatment.
15
Mean Range Bias of CORE Sites
16
Mean Heights of CORE Sites
17
ASI LAGEOS RB Analysisof the CORE Sites
18
1999 Range Bias Comparison
Longitude
Estimates are in excellent agreement. Are these
biases real?
19
Greenbelt (7105) Time Bias Analysis
DGFI estimates diverge on day 122 from reality.
Day 122 marks the beginning of their next 7-day
Arc
20
DGFI TB Estimation
LAGEOS-1
LAGEOS-2
DGFI estimated time biases for perfectly
good data from 6 other sites, near the
beginning of its new LAGEOS-2 7-day Arc
21
Influence of Modeling Error
DGFIs Greenbelt Time Series
jump in height baseline lengths
22
Dangers of Bias Estimation

• The DGFI LAGEOS-2 orbit was skewed (i.e. had an
  along-track error) due to the timing problems in
  Greenbelt.
  
• This along-track error was absorbed in the
  latitude and longitude (not height) of any sites
  tracking LAGEOS-2 on May 2 or May 3, 1999.
  
• The Greenbelt frequency bias went into
  Greenbeltss height.
  
• Coordinate transformations solutions may MASK
  this error.

23
Conservative Reality Checks(for the BEST
Performers)

• Range Bias Estimation
• variations between LAGEOS-1 2 5mm
• Baseline Length
• variations 20mm
• Height
• variations 20mm

24
ASI Analysis
Monument Peak
McDonald
Greenbelt
OMC(mm)
Herstmonceux
Grasse
Graz
OMC(mm)
Yarragadee
Mt. Stromlo
OMC(mm)
25
CRL Analysis
Monument Peak
McDonald
Greenbelt
OMC(mm)
Herstmonceux
Grasse
Graz
OMC(mm)
Yarragadee
Mt. Stromlo
OMC(mm)
26
CSR Analysis
Monument Peak
McDonald
Greenbelt
OMC(mm)
Herstmonceux
Grasse
Graz
OMC(mm)
Yarragadee
Mt. Stromlo
OMC(mm)
27
CSR European Heights (2.5 year time series)
28
DGFI Analysis
Monument Peak
McDonald
Greenbelt
OMC(mm)
Herstmonceux
Grasse
Graz
OMC(mm)
Yarragadee
Mt. Stromlo
OMC(mm)
29
IAAK Analysis
Monument Peak
McDonald
Greenbelt
OMC(mm)
Herstmonceux
Grasse
Graz
OMC(mm)
Yarragadee
Mt. Stromlo
OMC(mm)
30
NERC Analysis
Monument Peak
McDonald
Greenbelt
OMC(mm)
Herstmonceux
Grasse
Graz
OMC(mm)
Yarragadee
Mt. Stromlo
OMC(mm)
31
(No Transcript)
32
(No Transcript)
33
(No Transcript)
34
(No Transcript)
35
(No Transcript)
36
(No Transcript)
37
(No Transcript)
38
(No Transcript)
39
(No Transcript)
40
JCET vs CSR EUROLAS Heights
41
(No Transcript)
42
(No Transcript)
43
(No Transcript)
44
JCET vs DGFI EUROLAS Heights
45
Estimated Parameters
46
Conclusions/Questions

• Range bias analysis, by itself, tells us a lot
  about the quality of the solutions.
  
• Data treament is a major source of variation
  between analysis center solutions.
  
• To what level can range bias be separated from
  height variations?
  
• Should solutions pass REALITY checks prior to
  be combined?

47
Recommendations

• Standardize bias treatment
• For sites with stable bias either estimate a
  common LAGEOS-1, -2 range bias, per site, per
  28-day or estimate NO BIAS
  
• For sites with instability and/or frequency
  errors and/or low data quantity, edit the data
  
• Develop and implement the BEST models
• LAGEOS CoM
• Tropospheric Models
• GM
• Implement data integrity checks prior to
  combinations
  
• Eccentricities
• Range bias
• Baseline lengths
• Site heights
• Implement bias corrections
• Each center should QC their own solution

48

• Improved Modeling
• Satellite CoM
• Troposphere
• GM
• Pressure Loading
• Earth CoG

• Improved Data Treatment
• One RB/site/28-days
• Edit Problem Data
• Fix Known Bias Problems

• Data Integrity Checks
• Eccentricities
• Range Biases
• Baselines
• Heights

Less Variation between Solutions and
More Accurate Site Positions