Seite - 129 - in Contributions to GRACE Gravity Field Recovery - Improvements in Dynamic Orbit Integration, Stochastic Modelling of the Antenna Offset Correction, and Co-Estimation of Satellite Orientations
Bild der Seite - 129 -
Text der Seite - 129 -
the weight matrix of the equation system is
PR= (
Σsst+DF A
αΣ A
αˆαˆ (
DFAα )T +DFBαΣ B
αˆαˆ (
DFBα )T)−1
= (
Σsst+Σ A
∆ρ˙AOC+Σ B
∆ρ˙AOC )−1
. (9.2.6)
Here,Σsst is the estimated stationary covariance matrix for the ll-SST observations,
andΣA∆ρ˙AOC andΣ B
∆ρ˙AOC are the AOC covariance matrices from chapter 8. The obser-
vation groups are uncorrelated. The covariance matrix of the derived observations is
thus identical to the complete arc-wise covariance matrix of the improved stochastic
model described in eq. (8.2.1). Also identical to the approach of chapter 8, a complete
set of variance factors, one for each time lag in the stationary covariance function,
one for each short arc, and one for each spacecraft’s AOC covariance per month, are
co-estimated.
The reduced derived observation from eq. (9.1.29) is then
∆λ=− (
ρ˙KBR−f0(x0,α0)−DFAα (αA−α0A)−DFBα (αB−α0B) )
, (9.2.7)
with the Taylor point for the orientation the observations thereof in the first iteration,
and the estimated orientation thereafter. With this information, the normal equa-
tion system can be formed, and∆xˆ and eˆλ are determined in the usual way. Using
eqs. (9.1.20) and (9.1.21), the estimated additions to the satellite orientation are
∆αˆA= Σˆ A
αˆαˆ (
DFAα )T
PReˆλ and ∆αˆB= Σˆ B
αˆαˆ (
DFBα )T
PReˆλ , (9.2.8)
with the final updated estimated orientations
αˆA=αA−∆αˆA and αˆB=αB−∆αˆB . (9.2.9)
Practical Considerations
The algorithm used to co-estimate orientation parameters only affects the covariance
matrix and reduced observations for the ll-SST observables. The hl-SST component
of the equation system remains unaffected. In GROOPS, the co-estimation of the
satellite orientation has been implemented in parallel to the iterative estimation of
the stochastic model with a degree and order 60 gravity field solution. This means
that the updated orientation is not used to re-integrate the dynamic orbits, as this
step is already completed at this point. Further, the updated estimates for the KBR
antenna phase centre vectors are not used in the linearisation of the observation
equations or for computing the updated AOC. Under regular observation conditions
the opening angle β is small for both spacecraft. This is an unfavourable configuration
for a stable estimate of the APC vectors. Until convergence of the stochastic model and
the spacecraft orientation is achieved, estimates of the APC vectors fluctuate wildly.
Allowing these vectors to vary at this point prohibits convergence of the system and
leads to chaotic results.
9.2 Application to GRACE 129
Contributions to GRACE Gravity Field Recovery
Improvements in Dynamic Orbit Integration, Stochastic Modelling of the Antenna Offset Correction, and Co-Estimation of Satellite Orientations
- Titel
- Contributions to GRACE Gravity Field Recovery
- Untertitel
- Improvements in Dynamic Orbit Integration, Stochastic Modelling of the Antenna Offset Correction, and Co-Estimation of Satellite Orientations
- Autor
- Matthias Ellmerr
- Verlag
- Verlag der Technischen Universität Graz
- Ort
- Graz
- Datum
- 2018
- Sprache
- englisch
- Lizenz
- CC BY 4.0
- ISBN
- 978-3-85125-646-8
- Abmessungen
- 21.0 x 29.7 cm
- Seiten
- 185
- Schlagwörter
- Geodäsie, Gravitation, Geodesy, Physics, Physik
- Kategorien
- Naturwissenschaften Physik
- Technik