C' Shum, C' Zhao, Y' Yi, and P' Luk

1
QUALITY ASSESSMENT OF GFO SENSOR AND DATA PRODUCTS

• C. Shum, C. Zhao, Y. Yi, and P. Luk
• The Ohio State University
• GFO Calibration/Validation Meeting
• NOAA Laboratory for Satellite Altimetry
• Silver Spring, Maryland
• June 12, 2001

2
QUALITY ASSESSMENT OF GFO SENSOR AND DATA PRODUCTS

• Routine quality assessment of GFO data product
• Cycle by cycle GFO sensor and data verifications
• NGDR (GfoM and GfoO) data editing summary
• http//geodesy.ohio-state.edu/gfo
• Restricted web http//geodesy.ohio-state.edu/gfo/
  navy
• Summary of sensor and data assessment
• OODD and GSFC MOE orbit verifications (POE to
  follow)
• Time tag, USO, SWH, ?0, sea state bias
  evaluations
• Radiometer, ionosphere and geophysical
  corrections
• NGDR vs. IGDR comparisons

3
Restricted Web http//geodesy.ohio-state.edu/gfo/
navy
4
Summary of Data Editing Percentage for Recent GFO
NGDR data
Restricted Web http//geodesy.ohio-state.edu/gfo/
navy
5
GFO VERIFICATION WEB-PAGE
Restricted Web http//geodesy.ohio-state.edu/gfo/
navy
6
GFO VERIFICATION WEB-PAGE
Restricted Web http//geodesy.ohio-state.edu/gfo/
navy
7
GFO-1 ORBIT DETERMINATION AND ANALYSIS

• Averaged statistics for OSU orbit, RA time bias,
  range bias, sea state bias

RA bias is relative to T/P MSS and negative
value means range is short

• Selected orbit (6-day arcs) comparison
  (different gravity models)

8
Precise Orbits Nov. 30, 2000 Feb. 5, 2001
data (Operational Data) RMS
in cm
9
Evaluation of Doppler Orbits Using Crossovers
Doppler Orbits Nov. 30, 2000 Feb. 5, 2001
data (Operational Data) Crossover range 32 - 440
cm rms
10
Evaluation of GSFC MOE/SLR Orbit Using Crossovers
MOE/SLR Orbits Nov. 30, 2000 Feb. 5, 2001
data (Operational Data) Crossover range
84-580 mm rms
11
Evaluation of OODD (NGDR) Orbits Using Crossovers
Doppler Orbits April 15 May 18, 2001
data (Operational Data) Crossover range 43 - 185
cm rms
12
Evaluation of MOE (NGDR) Orbit Using Crossovers
MOE/SLR Orbits April 15 May 18, 2001
data (Operational Data) Crossover range 7.6 -
42.5 cm rms (Day 113 picked lower accuracy orbit
than IGDR did)
13
Evaluation of MOE Orbit (IGDR) Using Crossovers
MOE/SLR Orbits April 15 May 18, 2001
data (NOAA IGDR Data) Crossover range 10.4 -
15 cm rms
14
RADIAL ORBIT ERROR ASSESSMENTAFTER
BIAS/TILT/1-CPR ERROR REMOVAL

• Estimated error for MOE/SLR orbits after
  adjustment
• 40,000 km arcs (Ocean-wide crossovers)
• 8.6 cm rms (days 243-259) 9.7 cm rms (days
  260-276)
• 3,000 km arcs (30 Zonal Band between 25S-5N)
• 3.9 cm rms (days 243-259) 4.4 cm rms (days
  260-276)
• 1,000 km arcs (10 Zonal Band between 15S-5S)
• 1.3 cm rms (days 243-259) 2.3 cm rms (days
  260-276)
• Atlantic 1.1 cm (days 243-259) 1.0 cm (days
  260-276)
• Pacific 1.3 cm (days 243-259) 1.2 cm (days
  260-276)
• All crossover data used (No editing)

• Estimated error for OODD (Doppler) orbits after
  adjust.
• 1,000 km arcs (10 Zonal Band between 15S-5S)
• 1.8 cm rms (days 243-259) 6.5 cm rms (days
  260-276)
• Atlantic 1.2 cm (days 243-259) 1.8 cm (days
  260-276)
• Pacific 2.1 cm (days 243-259) 1.3 cm (days
  260-276)

15
SUMMARY ORBIT ERROR ASSESSMENTS

• Near-Real Time (12-24 hours) Orbits
• Opnet Doppler (OODD) orbits estimated accuracy
  30-450 cm rms
• Estimated orbit accuracy for GSFC MOE
  (SLRDoppler) 10-50 cm
• Estimated orbit accuracy after crossover
  adjustment (bias tilt)
• OODD 1.2 - 6.5 cm rms (1,000 km arc)
• MOE/SLR 1.0 - 2.3 cm rms (1,000 km arc)

• Precise Orbits (NASA GSFC orbits using improved
  models)
• SLR rms 4 cm, crossover rms 8-9 cm
• Estimated radial orbit accuracy 5-6 cm rms

16
SWH AND ?0 CALIBRATION
Geographical SWH comparison 1 14 dm GFO
and ERS-2 (June 1999) Recommended calibrations
(D. Hancock, 2000) Add 0.37 dB to ?0
(1-parameter model) Add 0.24 m to SWH
(1-parameter model) Study based on Tran et
al. 2000, T/P comparisons Buoy validations
(D. Cotton, 2000) Preliminary results
with limited data 2-parameter (bias and
offset) model Ohio State University study (Y.
Yi, 2000) Preliminary results comparison
with TOPEX Applying Hancock biases improves
buoy fit for wind-speed (?0), but not
SWH D. Cotton, pc
Hancock calibration was recommended and
implemented in GFO data processing
17
SWH Buoy Calibration (D. Cotton)
Buoy data fit 12 cm rms (26 cm for TOPEX 32 cm
for ERS-2) Preliminary results (limited
calibration data used)
18
Buoy Wind Speed (?0) Calibration (D. Cotton)
Buoy data fit 1.28 m/s (1.27 m/s for TOPEX
1.23 m/s for ERS-2) Preliminary results (limited
calibration data used)
19
COMPARISONS WITH TOPEX SWH/?0 10-day Averages
within 66S-66N
Preliminary results indicate GFO offsets with
TOPEX SWH and ?0 values, confirming D. Hancocks
calibration results
20
ASSESSMENT OF TIME BIAS AND SSB

• Time Tag Accuracy and Timing Stability
• RA time tag accuracy assessed using crossover
  analysis over three Cal/Val periods and first 4
  operational cycles data sets

Time Tag accuracy 0-3 ms (cal/val), 1.5 ms
(operational) USO Drift Range Correction 15 cm
since launch Lillibridge et al., 2000 Hancock,
personal communication
Sea State Bias One parameter model (dependence
on SWH) estimates based on crossover analysis
varies over different versions of Cal/Val data
(possibly due to changes of ?0/SWH and imperfect
modeling, such as orbits)
Cal/Val I Data (1999) 4.3 SWH Cal/Val II
Data (1999) 3.0 SWH Cal/Val III Data
(2000) 3.5 SWH Cal/Val IV Data (2000)
unavailable Operational Data (2001) 4.7 SWH
21
Precise Orbits Nov. 30, 2000 Feb. 5, 2001
data (Operational Data)
22
GFO Timing Stability Comparisons
Geosat
GFO
23
USO Height Correction Comparisons
Geosat
GFO
24
COMPARISON OF GFO MWR AND NCEP WET DELAY
GFO MWR - ERS2 MWR -11 ? 31 mm (Dec 6-22, 1999
data)
GFO MWR NCEP Mean (mm) RMS
(mm) Cycle 2 (Jan 3-19, 2001) 1.6
25.7 Cycle 6 (Mar 12-18, 2001)
3.5 24.2 Cycle 8 (April 15-May
01) 3.6 25.3 Cycle
9 (May 02-18) 2.1
24.8
25
COMPARISON OF GFO MWR AND ERS-2 MWR (ATSR) WET
TROPOSPHERE DELAY
26
VALIDATION OF IONOPSHERE CORRECTION

• Comparison of JPL and CODE GIM (Global Ionosphere
  Map) and IRI95, high solar activities (Sept. 00
  April 01)
• CODE and JPL GIM, bias 2.9 mm, rms 1.5 cm
• IRI95 and CODE GIM, bias -10 mm, rms 5.4 cm
• IRI95 and JPL GIM, bias -7.1 mm, rms 5.1
  cm
• Sea level drift error could be 0-3 mm/yr directly
  due to ionosphere correction error during this
  period
• Assessment of IRI-95, GIM (CODE) versus TOPEX
  truth GIM is better than IRI-95 during
  high solar activities on both RMS and drift and
  comparable to IRI95 during low solar activities
  
• IRI95-T/P 5.6 cm rms, GIM T/P
  2.3 cm rms
• (TOPEX data used Sept 00
  - Feb 01)
• IRI95-T/P drift 0.35 mm/yr (Data
  1993-1999)

27
COMPARISON OF GFO IONOSPHERE DELAY FROM IRI95 AND
JPL, CODE GIM (12/1999-2/2001)
28
Apparent sea level rise 3 cm/60 days
29

GFO Range (SSH) Data Noise Estimation

• SSH (Sea Surface Height) NGDR data for
  July-August, 1999 (Cal/Val I) and for Aug-Oct.
  2000 (days 243-259, Cal/Val IV)
• Track segments 400 sec. long (12 tracks used)
• Single repeating cycles used
• Method
• Estimate SSH signal by cubic splines
• SSH Noise SSH - SSH Signal
• Selected data in deep oceans with least
  variability (1 Hz data)
• 11-19 mm rms in Atlantic and 18-22 mm rms in
  Pacific (1-3 m wave heights)
• Comparison with other altimeters
• ERS-2 SSH Noise 19/27 mm rms (Sep, 1997)
• TOPEX SSH Noise 11/15 mm rms (Sep, 1999)

30
GFO Noise Estimate July - August, 1999 (Cal/Val
I Data)
Blue 1 Hz Red 3-sec. average
31
GFO Noise Estimate (Cal/Val IV Data, 2000,
September)
32
Precise Orbits Nov. 30, 2000 Feb. 5, 2001
data (Operational Data)
33
GFO RANGE BIAS
Bias Rel. to T/P (Cal/Val I, II, III)
-3 5 cm
Lake Michigan GPS- Buoy Campaign (March 24,
1999) 31 42 cm
Note Precise GDR not Available for March
24, 1999 GFO Pass
34
COMPARISON OF NGDR AND IGDR

• NGDR and IGDR MOE agree well on time tags, SWH,
  ?0, AGC, solid Earth tide and wet troposphere
  delay
• Ionosphere delay, ocean tide and dry troposphere
  delay have differences as different models were
  used
• NGDR MOE and OOD agree perfectly on time tag,
  SWH, ?0, AGC and all other corrections except
  ionosphere delay NGDR MOE and IGDR MOE orbits are
  not always same
• Differences exist in Uncorrected Altimeter Range
  between NGDR MOE, IGDR MOE and NGDR OOD products
• Comparisons conducted using Cycle 8 and 9 data
  products
• Mismatch orbit (1 Hz vs 10 Hz processing) was
  suspected and confirmed by other studies (J.
  Lillibridge, S. Klingenberger, R.
  Vaughan et al.)

35
(No Transcript)
36
(No Transcript)
37
(No Transcript)
38
(No Transcript)
39
(No Transcript)
40
GFO Sensor and Data Assessment

• Estimated Doppler orbit accuracy 30-440 cm GSFC
  MOE orbit accuracy 10-50 cm GSFC precise orbit
  accuracy 5-6 cm rms
• Orbit accuracy due primarily to availability of
  SLR tracking
• Time bias insignificant (0-1.5 ms) USO Drift is
  15 cm/mission
• Noise of GFO SSH (1Hz) 19 mm TOPEX noise 13
  mm rms, ERS-2 noise 23 mm rms, over same
  regions
• Sea state bias 4.7 of SWH. GFO absolute range
  bias -3 cm
• Recommend to adopt David Hancocks suggested
  calibrations for SWH and ?0 add 0.24 m to SWH
  and add 0.37 db to ?0
• Wet troposphere correction 0.2 cm bias, 2.6 cm
  rms (w/NCEP Model) 1 cm bias, 3 cm rms (w/
  ERS-2)
• CODE GIM and IRI95 provide ionosphere corrections
  accurate to 1-5 cm rms during medium to high
  solar activities GIM performs better during high
  solar activities

41
FUTURE POTENTIAL IMPROVEMENTS

• Estimation of 4- or 5-parameter sea state bias
• Improved calibration of SWH and ?0 needed
• Crossover analysis using best processed data
  set
• Barotropic ocean tide models
• NAO99.2b Matsumoto et al., 2001, GOT99.b Ray,
  2000
• Ionosphere models
• GIM (CODE or JPL?), IRI2001 Bilitza, 2001
• Dry troposphere and IB
• NCEP operational, FNOC (?), ECMWF (for GDR?)
• Need Wallops internal calibration
• Concerns USO drift large, inland lake/sea data
  quality, lack of waveform data, land-sea flag,
  improved center of gravity (SLR retroreflector
  and RA), further gravity field improvement?

42

GFO Great Lakes data are being edited due
primarily to large SSB(?)
43
ASSESSMENT OF TIDE ERROR USING MODEL COMPARISONS

44
Assessment Of Tide Error Using Model Comparisons
(RSS of 8 constituents)

6 recent models used NAO99.2b, GOT99.b,
CSR4.0, DW98, Delft, YATM4D Shum et al. 2001
45
TIDE MODEL EVALUATIONS USING ALTIMETER SEA
LEVEL DATA Shum et al. 2001Global coastal
ocean (depthlt1000m)

• GEOSAT (Residual
  rms, cm)
• Tide Models GM ERM ERS-1
  T/P
• CSR3.0 52.59 111.04 26.72
  21.53
• YATM4d 58.43 110.96 34.60
  33.56
• CSR4.0 51.96 53.99 25.40
  19.09
• GOT99.2 46.96 48.53 25.15
  17.75
• NAO99 51.90 53.67 24.70
  17.04
• Smith 47.97 47.51 32.37
  33.28

Sea Level Altimeter SSH - OSUMSS95
Latitude weights applied Yi and Rapp, 1995
Edit criteria 1000 cm