530 likes | 546 Views
NGGM ASSESSMENT STUDY Mission Architecture Review ESTEC, Noordwijk, 2 September 2010. NGGM Numerical study. Contribution of SGG measurements to future gravimetric satellite missions. Assumptions. 32-days simulation, 1s sampling period (3/1/96-3/2/96 ) SH degrees 2 to 80
E N D
NGGM ASSESSMENT STUDYMission Architecture ReviewESTEC, Noordwijk, 2 September 2010
NGGM Numerical study Contribution of SGG measurements to future gravimetric satellite missions
Assumptions • 32-days simulation, 1s sampling period (3/1/96-3/2/96) • SH degrees 2 to 80 • Vyy (cross-track) and Vzz (radial) components (Satellite Body Frame) • One gradiometer per satellite (i.e. 2 datasets considered) • “Shaped noise”, (from E2E simulator) • Attitude error negligible • Temporal aliasing assumed to be HIS-0.1AO MT3 models • Frequency-dependent data weighting applied to SGG data inversion
Isotropy of SGG observations Increasing contribution of SGG data
32-days simulations • Single-pair scenarios: • inline-polar (in-90) • pendulum (pend-90) • inline-SSO (in-SSO) • cartwheel (CW-90) • Multi-pair scenarios: • inline-63 + inline-polar (in-63+in-90) • inline-63 + pendulum (in-63+pend-90) • inline-63 + inline-SSO (in-63+in-SSO) • inline-polar + pendulum (in-90+pend-90) • inline-63 + inline-polar + pendulum (in-63+in-90+pend-90)
Overview of SGG contribution Original noise magnitude
Overview of SGG contribution SGG noise 10 times lower
Single-pair scenarios SGG noise 10 times lower
Multi-pair scenarios SGG noise 10 times lower
Overview of SGG contribution SGG noise 100 times lower
Single-pair scenarios SGG noise 100 times lower
Multi-pair scenarios SGG noise 100 times lower
Single-pair scenarios Isotropy improvement
Multi-pair scenarios Isotropy improvement
4 and 7-day simulations • Single-pair: • inline-polar (in-90) • pendulum (pend-90) • Multi-pair: • inline-63 + inline-polar (in-63+in-90) • inline-63 + pendulum (in-63+pend-90)
7-day simulations Nominal SGG noise
4-day simulations Nominal SGG noise
4-day simulations Isotropy improvement
Conclusions • For any scenario and nominal noise amplitude, SGG data is too noisy to make any contribution at all. • For a visible imprevement, SGG noise amplitude (i.e. accelerometer noise) needs to decreased: • 10-fold for single-pair inline scenarios; • 100-fold for dual-pair inline scenarios. • Scenarios (single or multi-pair) considering the Pendulum/Carthweel formation have no gain regarding estimated gravity field error amplitude, only a marginal isotropy improvement at the 100-fold SGG noise down-scaling. • However…
Comments • Short-period estimations important for understanding rapid-changing mass-transport processes, so that de-aliasing models are accurate and temporal aliasing errors minimized. • Minimization of temporal aliasing errors is critical for improvement of the accuracy and resolution of mass transport models at all time scales. • Thanks to the mitigation of spatial aliasing, 4-day estimation periods for single-pair scenarios are only accurate with gradiometric data. • Shorter estimations periods (1-day or less) might only be possible with SGG data, even for multi-pair scenarios (numerical verification needed).
Summary • The added value of SGG data, relative to the original SST scenario, is dependent on: • anisotropy (i.e. inline formations); • temporal aliasing caused by low temporal resolution of the same geographical location (i.e. single-pair scenarios); • spatial aliasing caused coverage gaps (i.e. short estimation periods) • => SGG more significant to inline-polar scenario at 4-days estimation periods (or shorter).
Polar gaps • SST-only • 32-day simulations • Eq. Water H. [m] • -0.5 to 0.5m color scale • Latitude > 65 deg
Polar gaps • SST-only • 32-day simulations • Eq. Water H. [m] • -0.5 to 0.5m color scale • Latitude < -65 deg
SST errors: motivation • The PSD of GRACE’s relative acceleration residuals can be split into two regions: • Temporal aliasing (low-frequency, < 30mHz); • KBR noise (high-frequency, > 30mHz). • First item is a hypothesis under research! Replacing GRACE’s KBR with laser ranging has no (significant) influence on the accuracy of the residuals below 10mHz, given current processing methods.
SST errors: mitigation of temporal aliasing 12 monthly solutions (2008) compared to a long time mean (C) T. Mayer-Gürr • Recent improvements in data processing, taking into account GRACE data-based short-period snapshot (i.e. more accurate handling of mass transport signal), show ~ 3-fold increase in the accuracy of the models (above deg 30).
SST errors: conclusions • Temporal aliasing is a dominant source of error in GRACE. • If temporal aliasing is simulated realistically, the estimated gravity field model errors should be comparable to GRACE-based model errors. • Geoid height error @ deg 80: • Inline-polar SST-only simulations: 0.4mm; • Traditional GRACE-based models: 3mm; • Recent GRACE-based models: 1mm. • Noise ratio between SGG and SST data probably ~3-10 times too large, since SGG data is much less sensitive to temporal aliasing
SST errors: updated nominal SGG/SST error ratio SGG noise 10 times lower
4-day simulations Isotropy improvement
SST errors: updated nominal SGG/SST error ratio SGG noise 10 times lower
7-day simulations Isotropy improvement
SST errors: updated nominal SGG/SST error ratio SGG noise 10 times lower
32-day simulations Isotropy improvement
SST errors: practical added value of SGG data Inline-polar + SGG vs. inline-polar + inline 63o (no SGG): Factor of 2 lower accuracy in inline-polar + SGG; Comparable level of anisotropic error pattern.
Additional issues: attitude of the baseline vector as function of latitude
Attitude of the baseline vector as function of latitude: pendulum high error at low degrees Pendulum-polar
Attitude of the baseline vector as function of latitude: pendulum
Attitude of the baseline vector as function of latitude: inline-polar
Attitude of the baseline vector as function of latitude: inline-SSO
Attitude of the baseline vector as function of latitude: bender
Attitude of the baseline vector as function of latitude: cartwheel
Optimal data weighting Inline-polar
Optimal data weighting Inline-SSO
Optimal data weighting Pendulum-polar
Optimal data weighting Cartwheel
Optimal data weighting Inline-polar + inline 63o