Jason1 Orbit Improvement by Combining GPS with SLRDORIS

1
Jason-1 Orbit Improvement by Combining GPS with
SLR/DORIS

• Key-Rok Choi, John Ries and Byron Tapley

This research is supported by NASA/JPL Contract
961429.
2
GPS Data Pre-processing

• TEXGAP software (the university of TEXas Gps
  Analysis Program)

Sample the raw RINEX and convert the data to
binary files Correct the receiver time
tags Generate the DDHLP observables from the L1
and L2 undifferenced observables Generate the
ionospheric-free DDHL observables Edit data and
fix cycle slips Correct the DD ambiguity term
approximately Merge and Sort the DDHL data
- - -
- - - -
3
Double Differenced High-Low Observable
Single-differenced phase (SDP)
Computed phase difference between q j at
Phase of the p-th satellite received by j-th at
phase integer ambiguity between j p.
Phase generated by the j-th station receiver at
Double-differenced high-low phase (DDHLP)
4
Dynamic Models
5
Measurement Models
6
Parameterizations for Empirical forces
Two Orbit Comparison from two different
parameterizations

• Heavy parameterization 1.8738 hour for Drag
  and 5.6214 hour for Empirical Forces.
• Lighter parameterization 0.1725 day for Drag
  and 0.6898 day for Empirical Forces.

• One of the benefits of the GPS data over the
  SLR/DORIS data for the POD purpose is its dense
  and homogeneous tracking. With the dense
  observation set, heavier parameterizations are
  possible to accommodate the force model errors.
  How far we can push the parameterization is still
  a question.
• The sub-arc lengths for this investigation were
  chosen by considering several factors such as the
  orbit period, the integration arc length as well
  as the yaw events.
• Based on the crossover rms and SLR residuals,
  1.8738 hr for Drag and 5.6214 hr for Empirical
  1-cpr forces was a level of parameterization
  which generally performed best.

7
Optimal Weights for each Data Type? (1)
SLR data NOT used for POD

• Table above clearly shows that combining GPS
  with SLR/DORIS can improve the orbit quality.
• To find optimal weighting for GPS, 3 cm, 10
  cm, 15 cm,20 cm, 25 cm, 30 cm and 50 cm for GPS
  were weighted for each case, with the SLR weight
  fixed to 10 cm, and with the DORIS weight fixed
  to 2 mm/sec.

8
Optimal Weights for each Data Type? (2)

• The optimal weight for GPS resides somewhere
  between 10 cm to 30 cm.
• The experiments with the various DORIS weight
  from 1 mm/sec to 4 mm/sec show that the orbit
  quality is not sensitive to the DORIS weight.
• For the nominal orbits, 25 cm was chosen for
  the optimal GPS weight and 2mm/sec for DORIS.

9
Effect of Ground Station Numbers on Orbit (1)
Set 1 ( 5 stations)
Set 2 ( 11 stations)
Set 3 ( 20 stations)
Nominal Set (43 Stations)
Set 4 ( 24 stations)
10
Effect of Ground Station Numbers on Orbit (2)

• The crossover mean comparison shows that the
  orbits of cycle 8 from Set 1 (5 stations) and Set
  2 (11 stations) are possibly miscentered.
• The crossover RMS comparison shows that the
  orbit from Set 3 (15 stations) performs close to
  the nominal orbit which was obtained from 43
  stations.

11
Effect of Ground Station Distribution (1)
Optimal Set for Cycle 8 (37 stations)
12
Effect of Ground Station Distribution (2)

• An Optimal Set was selected considering 1)
  geographical distribution, 2) tracking
  performance, and 3) coordinates accuracy out of
  208 ITRF2000 stations. The optimal geographical
  distribution set for cycle 8 consists of only 37
  stations. Adding more stations actually hurts the
  distribution.
• Orbits determined with the Optimal Set perform
  no better than orbits with the Nominal set.
• GPS-only orbits with heavily biased distribution
  set such as only European or American sites show
  a significant miscentering. By adding the
  SLR/DORIS data, the miscentering was improved.
  This implies that SLR/DORIS can help the GPS
  orbit with the orbit centering.
• The Z-bias is affected only at the few mm level
  by the non-uniform hemispherical distribution of
  stations.

13
CoM Offset X and Z components

• For the Nominal orbits, DX and DZ were
  simultaneously estimated for each day.
• The Z component was fairly stable at -3.4 cm,
  but the X component varied between 1 to -4 cm
  with a mean of -1.3 cm.
• Addition of the SLR/DORIS data to the GPS data
  didnt change them much.

14
External Comparison 1) Crossover RMS

• Crossover RMS comparison shows that JPL_gps and
  CSR_mix orbits perform best.

15
External Comparison 2) SLR Residuals

• The orbit solution with combining three data
  types may be approaching the 1cm RMS radial
  accuracy.

16
Summary

• Nominal Orbits combining GPS and SLR/DORIS
• JGM3 model
• All 43 stations were fixed (ITRF2000 reference
  stations)
• - Well distributed and well performing 20
  stations may be enough for the POD purpose.
• Estimates every 1.8738 hr for Drag and every
  5.6214 hr for empirical forces
• 25 cm for the GPS weight, 10 cm for SLR and
  2mm/sec for DORIS
• Both X and Z offsets estimated ( Z -3.4 cm
  stable, X -1.3 cm unstable)
• The orbit from the mixed data types clearly
  showed improvement over the orbit from a single
  measurement type. The SLR residuals with high
  elevation data shows that the orbit solution by
  combining three data types may be approaching the
  1cm RMS radial accuracy.