Living Planet Symposium Bergen June 2010 1
Gradiometer In-FlightCalibration
Alternative In-Flight Calibration of the GOCE
Gradiometer: ESA-L Method
Daniel LamarreMichael Kern
ESA
Living Planet Symposium Bergen June 2010 2
Gradiometer In-FlightCalibration
Topics
Differences between TAS-I & ESA-L methods
Comparison between TAS-I & ESA-L results
Improvement of scale factor retrievalwith star tracker combination
Evolution of gradiometer parameters
Living Planet Symposium Bergen June 2010 3
Gradiometer In-FlightCalibration
Two Main Methods for ICM Determination (Note also the ESA-K/Gradnet method: See poster session by C. Siemes)
TAS-I ESA-LImplemented in: Ground segmentOff-line
Retrieval per: OAG Whole grad’r
Computes: ICMs Grad’r parameters
Equations: 9 12
Scale factors (SF) found 6 1 by comparing with STR:
STR vs Grad’r Misalignment: Assumed null Retrieved
Baselines (Lx Ly Lz): Assumed known Assumed known
Convergence criteria: Per parameter Simultaneous forall parameters
Linear/angular coupling Assumed null Some info could factors: be retrieved
Living Planet Symposium Bergen June 2010 4
Gradiometer In-FlightCalibration
The 12 Equations Used by ESA-L Method
Gradients cannot be expressed as linear combination of linear and angular accelerations acting on the spacecraft:
Vxx=0 Vyy=0 Vzz=0 Bandwidth
Vxy=0 Vxz=0 Vyz=0 (50 to 100mHz)
Estimates of linear accelerations from different OAGs are the same (Michael Kern’s equations):
ax14 = ax25 = ax36 Bandwidth
ay14 = ay25 = ay36 (50 to 100mHz)
az14 = az25 = az36
These and the assumed knowledge of the 3 baselines, ensure coherence between all 18 accelerometer gain estimations.
Living Planet Symposium Bergen June 2010 5
Gradiometer In-FlightCalibration
Comparison with Star Tracker Angular Rates
Star Tracker Gradiometer
Absolute Gain: Perfect Wrong
Gains along 3 axes: Same Same
Reference frame: Perfect Orthogonal but rotatedabout 3 axes
By best fit are retrieved: Gradiometer single scale factor
Fixed rotations of grad’r about x, y and z
Best fit performed in bandwidth: ~ 0.7 to 2.0mHz
Living Planet Symposium Bergen June 2010 6
Gradiometer In-FlightCalibration
Living Planet Symposium Bergen June 2010 7
Gradiometer In-FlightCalibration
Star Tracker Systematic Errors- FOV dependent errors appear as orbital harmonicson a short time scale
- Impacts retrieval of gradiometer absolute scale factor
- Can be reduced by:
1) Removing orbital harmonics in comparison betweengradiometer & star tracker angular rates
2) Combining readings from 2 (or 3) star trackers
Living Planet Symposium Bergen June 2010 8
Gradiometer In-FlightCalibration
Living Planet Symposium Bergen June 2010 9
Gradiometer In-FlightCalibration
Calibrations Performed in Latest Configuration
Shaking Date Available Star Trackers
#3 Oct/2009 STR1, STR2
#4 Jan/2010 STR1, STR3
#5 Mar/2010 STR1, STR2 #6 May/2010 STR1, STR2
Merging of the 2 available star trackers with a least square algorithm from C. Siemes Yields a ‘virtual star tracker’ STRV
Living Planet Symposium Bergen June 2010 10
Gradiometer In-FlightCalibration
Comparison of ad14x (Vxx) ICM rows: Absolute Values
ESA-L Values:SHK3: 0.0175226 0.0000121 -0.0000082 1.0237767 -0.0000237 0.0000577 SHK4: 0.0176962 0.0000123 -0.0000068 1.0239178 -0.0000294 0.0000638 SHK5: 0.0177480 0.0000120 -0.0000066 1.0236419 -0.0000240 0.0000558 SHK6: 0.0178763 0.0000116 -0.0000051 1.0235056 -0.0000286 0.0000640 TAS-I Values:SHK3: 0.0172522 0.0000126 -0.0000110 1.0075948 0.0000000 0.0000366 SHK4: 0.0180007 0.0000129 -0.0000099 1.0416350 0.0000000 0.0000366 SHK5: 0.0177637 0.0000125 -0.0000093 1.0246993 0.0000000 0.0000359 SHK6: 0.0181930 0.0000126 -0.0000083 1.0417186 0.0000000 0.0000368
ESA-L Variations (ppm):SHK4vs3: 174 0 1 141 -6 6 SHK5vs4: 52 0 0 -276 5 -8 SHK6vs5: 128 0 1 -136 -5 8 TAS-I Variations (ppm):SHK4vs3: 749 0 1 34040 0 0 SHK5vs4: -237 0 1 -16936 0 -1 SHK6vs5: 429 0 1 17019 0 1
ESA-L vs TAS-I (ppm):SHK3: 270 0 3 16182 -24 21 SHK4: -305 -1 3 -17717 -29 27 SHK5: -16 0 3 -1057 -24 20 SHK6: -317 -1 3 -18213 -29 27
Living Planet Symposium Bergen June 2010 11
Gradiometer In-FlightCalibration
Comparison of ad14x (Vxx) ICM rows: Relative values (ie each row divided by CSF)
ESA-L Values:SHK3: 0.0171156 0.0000118 -0.0000080 1.0000000 -0.0000232 0.0000563 SHK4: 0.0172828 0.0000120 -0.0000067 1.0000000 -0.0000287 0.0000623 SHK5: 0.0173381 0.0000117 -0.0000064 1.0000000 -0.0000234 0.0000545 SHK6: 0.0174658 0.0000113 -0.0000050 1.0000000 -0.0000279 0.0000625 TAS-I Values:SHK3: 0.0171221 0.0000125 -0.0000109 1.0000000 0.0000000 0.0000364 SHK4: 0.0172812 0.0000124 -0.0000095 1.0000000 0.0000000 0.0000352 SHK5: 0.0173355 0.0000122 -0.0000091 1.0000000 0.0000000 0.0000350 SHK6: 0.0174644 0.0000121 -0.0000079 1.0000000 0.0000000 0.0000354
ESA-L Variations (ppm):SHK4vs3: 167 0 1 0 -6 6 SHK5vs4: 55 0 0 0 5 -8 SHK6vs5: 128 0 1 0 -4 8 TAS-I Variations (ppm):SHK4vs3: 159 0 1 0 0 -1 SHK5vs4: 54 0 0 0 0 0 SHK6vs5: 129 0 1 0 0 0
ESA-L vs TAS-I (ppm):SHK3: -6 -1 3 0 -23 20 SHK4: 2 0 3 0 -29 27 SHK5: 3 0 3 0 -23 20 SHK6: 1 -1 3 0 -28 27
Living Planet Symposium Bergen June 2010 12
Gradiometer In-FlightCalibration
Comparison of Results ESA-L vs TAS-I
- Excellent agreement for differential parameters
- Excellent agreement for common misalignments
- ESA-L retrieved common scale factors much more stable
Living Planet Symposium Bergen June 2010 13
Gradiometer In-FlightCalibration
Why should we use the ESA-L retrieved scale factors ?
-In principle, ESA-L method is more robust because only 1 scale factor is retrieved, and grad’r vs star tracker misalignment is retrieved as well.
-ESA-L gives more stable results, property more often associated with more accurate method than with less accurate method.
-ESA-L gives results more in-line with expected stability.
-ESA-L results are more consistent with the variation of differential parameters.
-ESA-L results are ‘validated’ by external calibration investigations.
Living Planet Symposium Bergen June 2010 14
Gradiometer In-FlightCalibration
Living Planet Symposium Bergen June 2010 15
Gradiometer In-FlightCalibration
Living Planet Symposium Bergen June 2010 16
Gradiometer In-FlightCalibration
Living Planet Symposium Bergen June 2010 17
Gradiometer In-FlightCalibration
Living Planet Symposium Bergen June 2010 18
Gradiometer In-FlightCalibration
Conclusion wrt Comparison with Star Tracker
- Fusion of data from 2 star trackers improves significantly scale factor & misalignment retrieval
- Filtering of orbital harmonics helps a lot if data from only 1 star tracker is available
Living Planet Symposium Bergen June 2010 19
Gradiometer In-FlightCalibration
ICM Comparison: ESA-L 6th vs 3rd Shakings, STRV. Difference (ppm)
OAG14 271 5 -6 -354 1 -3 -4 851 0 0 -224 3 6 0 259 3 -2 -249
Vxx -354 1 -3 271 5 -6 0 -224 3 -4 851 0 3 -2 -249 6 0 259
OAG25 521 -9 1 141 -2 -1 8 474 -1 1 190 1 0 1 925 3 -1 81 141 -2 -1 521 -9 1
Vyy 1 190 1 8 474 -1 3 -1 81 0 1 925
OAG36 653 -1 -3 15 1 1 0 1181 1 0 -17 1 2 -1 624 0 -1 10 15 1 1 653 -1 -3 0 -17 1 0 1181 1
Vzz 0 -1 10 2 -1 624
Living Planet Symposium Bergen June 2010 20
Gradiometer In-FlightCalibration
Evolution of In-Line Differential Scale Factors OAG14: Vxx OAG25: Vyy OAG36:Vzz
Living Planet Symposium Bergen June 2010 21
Gradiometer In-FlightCalibration
Conclusion Concerning Grad’r Evolution
- Alignment is very stable
- Common scale factor variation ~< 100 ppm/month
- Differential scale factor variation seems continuous:
Vxx < 50 ppm/month
Vyy < 30 ppm/month
Vzz < 2 ppm/month
Interpolation between shakings should be investigated:- Eg external calibration, or ESA-K (Gradnet) method- Can take advantage of stable alignment