Upload
everett-shaw
View
218
Download
0
Tags:
Embed Size (px)
Citation preview
SLR in the age of GPS
Frank G. Lemoine, Scott B. Luthcke, Nikita P Zelensky
Brian D. Beckley Code 697, Space Geodesy LaboratoryNASA Goddard Space Flight Center
Greenbelt, Maryland, U.S.A.Geoscience Australia
Canberra, Australia, August 29, 2005QuickTime™ and a
TIFF (Uncompressed) decompressorare needed to see this picture. QuickTime™ and a
TIFF (Uncompressed) decompressorare needed to see this picture.
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.QuickTime™ and a
TIFF (Uncompressed) decompressorare needed to see this picture.
Over the past decade significant advances in the Global Positioning System (GPS), and GPS data processing algorithms and data distribution, have positioned this technology as the primary tracking to support Precise Orbit Determination (POD) in the new era of geodetic satellites.
Jaso
n-1
ICE
Sat
CH
AM
PG
RA
CE
A new generation of geodetic missions all carry aboard a dual frequency codeless GPS receiver (BlackJack) as the primary tool for precise orbit determination (POD).
They also carry an SLR Retroreflector.
Precision Orbit Determination (POD) is a fundamental component in meeting the science goals of geodetic spaceflight missions.
For satellite radar and laser altimetry, POD enables science objectives such as the study of ocean, ice and land topography and surface change
Additionally, in several other applications, such as reference frame and gravity field determination, science is derived directly from the POD. Conversely POD depends on accurate knowledge of the reference frame and good knowledge of the geopotential
These new missions must take advantage of the GPS tracking geometry and near continuous coverage to meet their aggressive orbit accuracy requirements – especially true for the low altitude missions such as GRACE and CHAMP.
Jason-1 has a radial orbit accuracy goal of 1 cm !
The size of a dime!
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
GRACE
TRACKING DATA FOR JASON-1
• GPS (dual frequency codeless BlackJack GPS receiver)
• Satellite Laser Ranging (SLR), to a nine corner cube hemispherical Laser Retroreflector Array (LRA).
• DORIS Doppler.
• Altimetry
Processing multiple data types is the key to high accuracy POD calibration and validation.
Meeting the Challenge – Multiple Tracking Data
Types
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
DORIS beacon at Yarragadee
Reduced Dynamic accelerations are compensating for real physical force modeling error. GPS-based orbits accommodate more force modeling error.
94.5 96.5 98.5 100.5 102.5 104.50
0.002
0.004
0.006
0.008
0.01
0.012
0.014
Millionths
day of year 2002
Amplitude (m/s**2)dynamic slr+doris
red_dyn slr+doris+crossover
red_dyn gps
Can the GPS alone meet the Jason 1-cm goal?First glance at our GPS POD results might tempt you
to say yes.
Solution
DORIS SLR Altimeter Xovers
Type # Pts. RMS
(mm/s) # Pts.
RMS (cm)
# Pts. RMS (cm)
SLR Hi Elev
SLR+DORIS 111304 0.402 3716 1.34 4185 6.04 # Pts. RMS (cm)
GPS 108405 0.401 3709 1.67 4185 5.76 189 1.1
Jason-1 POD Summary Statistics Cycle 9 (10-days)
Radial RMS (cm)
Cross Track RMS (cm)
Along Track RMS (cm)
1.29 2.97 4.59
Orbit Difference – SLR/DORIS vs. GPS
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
So, where does this leave the SLR in the world of POD?
Is it more than just a backup ?It turns out that we need the SLR in order to fully exploit the GPS tracking.
SLR provides an highly accurate, direct and unambiguous observation of the orbit.
-This has proved necessary in fine-tuning the GPS-based orbit solutions.
The GPS tracking is an indirect, ambiguous observation of the orbit.
-This can be problematic, especially when fine-tuning the large GPS POD solution parameter set, and when sorting out systematic errors.
• 30-hr solutions • IGS GPS precise orbits• 33 IGS stations• Double-difference LC ranges to account for clock errors• Ambiguity bias determination per pass ~2700 per 30 hr. arc• Trop. Scale factors estimated every hour per site.• “reduced-dynamic solution”
– Covariance constrained along and cross-track periodic empirical accelerations estimated every: 90 – 20 minutes depending on application. • Need to determine optimal rate, correlation time and sigma
– Drag coefficients estimated every 6 – 1.5 hours depending on application• Need to determine optimal rate.
• Estimate phase center offset – need to determine which components
• … and all the rest of the high-fidelity force and measurement modeling:– e.g. “box-wing” model, antenna orientation, telemetered attitude, phase windup ….
• Jason: ~500.000 GPS DDLC obs. in 10 days vs. ~4000 SLR obs.
Nominal GPS POD Method
If SLR were not available we would have to determine the optimal GPS solution parameter set from the GPS tracking data residual and orbit overlap performance.
As we move to a more “reduced dynamic” solution both the orbit overlap and GPS DDLC range residuals become more meaningless as an orbit precision and accuracy metric.Estimate more parameters get a better fit.
Not necessarily a better orbit.
Increasing the frequency of the empirical accelerations –
Have less independent data for each set of parameters during overlap
Simply follows the data
“Waters down” orbit overlap metric.
If you only had the GPS tracking could you guess which is the better orbit ?
Which one is CHAMP, which one is Jason-1?
Satellite GPS DDLC Fit RMS (cm)
Radial Orbit Overlap
RMS (cm)
3D Orbit Overlap RMS (cm)
1 0.93 0.9 2.6 2 0.84 1.1 2.9
POD Summary Statistics Over Several 30-hr. arcs.
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
Can you figure it out now ?
Satellite GPS DDLC Fit
RMS (cm)
Radial Orbit Overlap
RMS (cm)
3D Orbit Overlap
RMS (cm)
Indep. SLR Fit
RMS (cm) 1 0.93 0.9 2.6 1.8 2 0.84 1.1 2.9 4.5
1 is Jason-1
2 is CHAMP
We need the SLR to discriminate between solutions and to ultimately fine-tune the GPS solutions to fully exploit the GPS tracking technology.
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
Can the GPS data alone “weed-out” gross systematic modeling errors?
First we can perform a little experiment:
- Use Jason-1 cycle 9 GPS solutions (ten 30-hr. solutions)
- Input a gross systematic modeling error
- Slightly rotate all of the GPS orbits – frame error
-Re-determine the Jason-1 orbits
-Using the GPS data alone, can you see the systematic error?
94.5 94.7 94.9 95.1 95.3 95.5DOY 2002
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
Meters
Introduced GPS Orbit Error
Rad. RMS = 0.0 cmCrs. RMS = 19.7 cmAlg. RMS = 19.4 cm
95.5 95.7 95.9 96.1 96.3 96.5DOY 2002
-0.10
-0.05
0.00
0.05
0.10
Meters
Resultant Jason Orbit Error
Rad. RMS = 0.2 cmCrs. RMS = 6.1 cmAlg. RMS = 4.2 cm
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
94 96 98 100 1020.85
0.90
0.95
1.00
Centimeters
Jason GPS RMS Fit per 30-hr Arc (DD1W LC Phase)
w/ error RMS = 0.931 cmw/o error RMS = 0.929 cm
94 96 98 100 102
DOY 2002
1
2
3
4
5
Jason Total Position Overlap RMS (6-hr. overlap)
w/ error RMS = 3.369 cmw/o error RMS = 3.317 cm
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
94 96 98 100 102DOY 2001
0.5
1.5
2.5
3.5
4.5w/ error RMS = 2.81 cmw/o error RMS = 1.83 cm
JASON Independent SLR RMS Fit per 30-hr. GPS Solution
GPS RMS Overlap RMS Indep. SLR RMS
0
15
30
% Change
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
GSFC
JPL
FIGURE 3 GPS antenna phase center maps (azimuth clockwise; 0º to 90º elevation from outside to center) The APC map estimated post -launch with GPS tracking data, is very important to POD. The 5ºx5º azimuth/elevation GSFC map was estimated in a formal solution usi ng eleven 30-
hr. arcs sampled over a â’
cycle, and for which low elevation GPS double-difference phase data was edited. The JPL map (Haines, pers. comm.. 2003) was computed averaging one year of un -differenced GPS phase residuals into 5ºx2º azimuth/elevation bins.
Jason-1 GPS Measurement Modeling Improvement
GPS Antenna Phase Center Correction Map developed post-launch.
Significant improvement in independent SLR fits: Cycles 8-24:
-No APC Map, 2.3 cm;
-With APC Maps, 1.7 cm.
0.5
1.5
2.5
3.5
4.5
5.5
80 100 120 140 160 180 200 220 240 260
DOY 2002
Independent SLR RMS per 30-hr. arc (cm)
No APC MapGSFC APC Map
Fixed Yaw Fixed Yaw Fixed Yaw
The 1 cm orbit: Jason-1 Precision Orbit Determination Using GPS, SLR, DORIS and Altimeter Data, Luthcke, S. B. et al., Marine Geodesy, 26(3-4), July-Dec, 2003.
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
Jason-1 Tracking Data Residual Summary
Table 5 Independent and Dependent Data Residual Summary for Cycles 8-24 Solution Type GPS DDLC
RMS (cm) DORIS RMS
(mm/s) SLR RMS
(cm) Xover RMS
(cm) Xover mean
(cm) Below with JGM3
SLR+DORIS Dyn. 0.421 1.710 5.926 0.229 SLR+DORIS RD 0.418 1.665 5.867 0.219 SLR+DORIS+Xover RD 0.418 1.914 5.780 0.048 GPS RD 0.75 0.419 1.698 5.766 -0.026 GPS+SLR RD 0.77 0.419 1.341 5.750 -0.029
Below with GGM01S SLR+DORIS Dyn. 0.419 1.524 5.859 0.129 GPS RD 0.74 0.419 1.596 5.754 0.024 GPS+SLR RD 0.76 0.419 1.249 5.735 0.012
Jason-1 Orbit Difference Statistical Summary
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
Table 6 Orbit Difference Statistics Computed per Cycle and Summarized over Cycles 8-24 Solutions Differenced Avg.
Radial RMS (cm)
Avg. 3D RMS (cm)
Avg. / Stdev of RSS[mean XY] per cycle (cm)
Avg. / Stdev of mean Z
per cycle (cm) Below with JGM3
GPS RD – SLR+DORIS Dyn
1.365 6.053 0.479 / 0.586 0.071 / 0.362
GPS RD – SLR+DORIS RD
1.141 4.809 0.376 / 0.437 0.069 / 0.380
GPS RD – SLR+DORIS+Xover RD
1.063 5.375 0.412 / 0.425 0.109 / 0.500
GPS RD – GPS+SLR RD
0.405 1.226 0.067 / 0.125 0.075 / 0.119
SLR+DORIS+Xover RD – SLR+DORIS Dyn.
0.946 5.396 0.591 / 0.271 -0.044 / 0.299
Below with GGM01S GPS RD – SLR+DORIS Dyn
1.178 5.015 0.144 / 0.570 0.422 / 0.327
GPS RD – GPS+SLR RD
0.370 1.122 0.089 / 0.113 0.131 / 0.112
YARA7090GREE7105
GRAS7835RIYA7832
MNPE7110HART7501
GRAZ7839MCDO7080
AJAC7848POTS7836
RGO_7840STRO7849
AREL8403ZIMM7810
0.0
0.5
1.0
1.5
2.0
RMS (cm)
(a) GPS RD Solution High Elevation Independent SLR Fit
RMS = 1.25 cm
0 2 4 6 8 10 12 14mm
0
5
10
15
20
number of differences
(b) GPS RD Solution Radial Orbit Overlap Performance
mean = 4.4 mmmedian = 4.1 mm
Jason-1 GPS-based 1-cm POD
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
AVERAGE ALTIMETER CROSSOVER RESIDUALS.Crossover residuals averaged over 5ºx5º bins for Jason cycles 8-24.
Jason Xover (SLR+DORIS Dyn. JGM3)
RMS = 17.0 mm
Jason Xover (GPS+SLR RD JGM3)
RMS = 14.7 mm
Jason Xover (SLR+DORIS RD JGM3)
RMS = 16.4 mm
Jason Xover (GPS+SLR RD GGM01S)
RMS = 13.7 mm
Jason-1 – TOPEX/Poseidon Mean Sea Surface Height: Verification Phase
During the first 21 Jason repeat cycles, Jason and T/P were flying in formation sampling the same ground track separated by 72 seconds. This provided an unprecedented opportunity to isolate orbit and instrument differences between the two altimeters since the atmospheric and oceanic conditions at the geo-referenced locations virtually cancel enabling a more accurate measure of the global inter-mission instrument bias.
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.QuickTime™ and a
TIFF (Uncompressed) decompressorare needed to see this picture.
Jason-1 – TOPEX/Poseidon Mean Sea Surface Height: Verification Phase
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.QuickTime™ and a
TIFF (Uncompressed) decompressorare needed to see this picture.
Two lessons emerge from the joint analysis of SLR, DORIS, GPS and altimeter data on Jason-1: (1) The GPS+ SLR Reduced dynamic orbits perform the best; (2) The DORIS+SLR Reduced dynamic orbits perform next best. We can demonstrate that these replacement orbits reduces the geographical signature of the orbit error. Moreover, we have validated an approach that will allow us to reprocess the entire Topex/Poseidon time series.
Jason – T/P Mean SSH
Effect of precision orbits on mean sea level determination
GSFC replacement orbits
GDR based orbits
Mean = 152.1 mm
Sdev = 4.5 mm
Improved agreement seen in lower left image results from improved TOPEX (SLR+DORIS RD) and Jason-1 (GPS+SLR RD) orbits. The improvement was achieved by using the more current ITRF2000 reference frame and associated station complement in the orbit computations, the GGM02C gravity field, and other improved force and measurement models.
GSFC Replacement Orbit - GDR, Mean Orbit Differences
Generating Radar Altimeter Climate Data Records ForGlobal Mean Sea Level Estimates
Consistency Issues
Precise Orbit Ephemerides
• terrestrial reference frame (ITRF).
• gravity field (GRACE).
• optimal tracking strategy (SLR, DORIS, GPS).
Geophysical Models.
Sea State Bias & Ionosphere.
Ground Retracking Philosophy
Continuous Long Term Monitoring
Tide Gauge Validation
• land motion adjustment
Radiometer
• GPS and VLBI
• SSM/I
Significant Wave Height
• NOAA deep ocean buoys
SLR provides a high accuracy, direct, unambiguous observation of the orbit.
SLR is absolutely necessary in realizing the full potential of the GPS tracking data
- Fine tuning solution parameterization- Calibrating measurement system parameters-“weeding out” systematic errors
SLR and GPS in combination are essential to achieving the best possible orbits for applications such as Mean Sea Level Determination.
Backup slides
GSFC Topex/Poseidon Reduced Dynamic Replacement POE
ITRF2000 vs.CSR95L02
Descending TracksGSFC Reduced Dynamic (ITRF2000) - GDR (cy.
344-364)
Ascending TracksGSFC Reduced Dynamic (ITRF2000) - GDR (cy.
344-364)
RMS= 4.7 mm
Jason SLR passes January-August 2005
0
100
200
300
400
500
600
SIME1873 MAID1864BORO7811 GRAS7835 MCDO7080
BEIJ7249SHO_7838 CHAC7237 GRF17105 POT37841 HART7501 SANF7824
RIGA1884 AJAC7848MNPE7110 MATE7941
RIYA7832 ZIMM7810 STL37825RGO_7840 GRAZ7839 WETZ8834 YARA7090
number of passes
POD Methodology
• NASA GSFC’s GEODYN precision orbit determination and geophysical parameter estimation software.– Simultaneously process a myriad of data types including
radar and laser altimetry
• Observations– GPS: double differenced ionosphere free phase (DDLC)– SLR, DORIS, altimeter crossovers
• Stations:– ITRF2000 GPS, SLR and DORIS – 33 GPS station network (selected based on best
performing and geographical distribution)– Phase center offsets modeled along with surface
deformation– For GPS, tropospheric scale factor estimated every 60
min.
5 10 15 20 25
5.4
5.6
5.8
6
6.2
Jason cycle
crossover residuals rms (cm) slr+doris dynamic (5.93 cm)
slr+doris red_dyn (5.84 cm)
gps red_dyn (5.77 cm)
gps+slr red_dyn (5.75 cm)
Jason-1 GPS-based reduced dynamic orbits demonstrate significant improvement in crossover performance.
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
Jason GPS+SLR Reduced Dynamic Orbits
Tracking Data Fits (Cycle 9 Only)Models GPS RMS
(mm)SLR RMS(cm)
BASELINE+JGM3
7.61 0.918
BASELINE + GGM02C 7.56 0.832
+GOT00 Tides+ATM GRAV
7.47 0.822
+IERS2000 Tides (Stations)+ ATM LOAD
7.47 0.820
Monitoring the Jason-1 Microwave Radiometer with GPS and VLBI Wet Zenith Path Delays
D.S. MacMillan, B.D. Beckley, P. Fang
VLBI and GPS Station Sites Jason-1 JMR is monitored using coincident wet zenith delay estimates from VLBI and GPS geodetic sites near altimeter ground tracks. The solid circles are older GPS sites, and the open circles are for sites that began observing after 1997. Two offsets have been observed centered about cycles 30 and 69.
JMR - ECMWF JMR - GPS __________________________________________
cycle 30 -3.6 +- 0.2 mm -5.4 +- 0.7 mm
cycle 69 -8.7 +- 0.4 mm -7.7 +- 0.8 mm
Errors based on the RMS of the cycle values before and after offsets.
Wet path delay differences between JMR and TMR are largest in tropics and coastal areas where most tide gauge and calibration sites are located.
Sensitivity of Tide Gauge Drift Analysis