Improved De-Aliasing for Gravity Field Modelling with GRACE
Final Report Abstract
The overall goal of the project was to develop improved methods and models for the representation of high frequency mass variations in the analysis of gravity field mission data. This is required in order to avoid aliasing effects in the monthly time variable gravity field time series determined from GRACE data. An improved estimation of high frequency mass variations leads to more correct geophysical mass variability estimates and finally to better conclusions, which can be deduced from them, especially for Earth science applications. Two major items were addressed by the collaborative project. These were (1) the identification of representative error parameters for atmospheric and ocean models used for determination of high frequency mass variations and (2) the development and implementation of a method to use these error parameters in a 3D approach for the computation of gravitational corrections to be applied in GRACE data processing. An attempt was made to estimate representative error measures for the ECMWF operational analysis model parameters, which serve as main information source for modelling the high frequency atmospheric mass variations, but also used for forcing the ocean circulation model applied in this study. For this purpose not only uncertainties delivered with the model, but also inter-comparisons with other models and in-situ observations, like results from GPS radio occultation observations from CHAMP, were applied. Intensive statistical analyses of uncertainties and model differences have been performed, which led to the following main conclusions. From the ECMWF model error estimates it was identified that surface pressure data over land have less uncertainty than the pressure data over water and that the given sigma data show a direction-dependent coupling. From model comparisons between ECMWF and NCEP it was observed that the bias over the oceans varies in the range of -3 to +3 hPa, while largest deviations occur over Antarctica and the Himalaja region. In addition it was identified that the bias is larger over the winter hemisphere compared with the summer. From comparisons of CHAMP radio occultation with ECMWF model profiles it could be concluded that assimilating such kind of information into the ECMWF model had significant impact on the model parameters to be applied for this study. Implicitly one could assume that ECMWF parameter uncertainties could be reduced from the time when this kind of information has been applied. Over the oceans, bottom pressure variability from GRACE, sterically corrected altimetry, and from the GECCO ocean model was analyzed in order to derive approximate error estimates for the ocean bottom pressure (OBP) maps. The estimated error maps are consistent with the misfits of individual fields against OBP sensor data, with the caveat, that a general underestimation of the signal-strength as a common, correlated, error in all products, cannot be recovered by the method. In addition detailed investigations on the application of monthly GRACE time variable gravity field for ocean studies were done. These include processing and filtering procedures of GRACE data, analysis of OBP seasonal cycle in amplitude and phase, monthly and short term ocean variability and its validation against in-situ observations. The error maps determined from atmospheric models analyses as well as those determined over the oceans are further used for the subsequent error propagation to the de-aliasing coefficients. The Software for propagating error estimates of atmosphere and ocean parameters into uncertainties of gravity field potential harmonics was developed from scratch. The mathematical procedure requires a 3D error integration of the atmosphere. For this purpose the complete procedure was reformulated as a least squares problem in order to enable varlance-covariance propagation of the input error fields to the resulting harmonic coefficients. The computational demands increased significantly compared to the conventional approach applying numerical integration and used in the standard GRACE processing. The Software extensively was used for analysis of the impact of error parameters on the GRACE performance. For this a-priori and a-posteriori observation residuals and also monthly gravity field models were analyzed in detail. From these investigations several conclusions could be derived. Atmospheric surface pressure errors are the main error source for GRACE de-aliasing and the de-aliasing coefficients strongly depend on their characteristics. The impact of atmospheric and oceanic model uncertainties is hardly above the current performance of GRACE, but, with improved de-aliasing, one could expect a reduced GRACE error level as well (iterative process of improvement). A-priori GRACE range rate residuals could be reduced for some analysis periods, which implies that the model world fits better to the observations made in the real world and consequently which means, that the newly computed high frequency de-aliasing coefficients perform better than the conventional solutions. The latter conclusion still needs to be confirmed based on a systematic analysis of longer time spans. Finally, an investigation about the impact of the chosen mean field for the de-aliasing process led to the conclusion that the mean field doesn't play an important role for the quality of the resulting de-aliasing harmonic coefficients. The developments made in this project are used for further investigations in the context of defining an optimal scenario for de-aliasing for future gravity field missions.
Publications
-
(2008): A long-term model for non-tidal atmospheric and oceanic mass variations and its implications on LAGEOS-derived solutions of Earth's oblateness, GFZ Scientific Technical Report STR08/12, GFZ Potsdam
Flechtner F., Thomas M., König R.
-
(2008): Comparison of ECMWF analyses with GPS radio occultations from CHAMP; Annales Geophysicae, Vol. 26, Nr. 11, pp 3225-3234, European Geosciences Union
Schmidt, T.; Wickert, J.; Heise, S.; Flechtner, F.; Fagiolini, E.; Schwarz, G.; Zenner, L.; Gruber, T.
-
2008: Decadal Sea Level Changes in the 50-Year GECCO Ocean Synthesis. J. Clim., 21, 1876-1890
Kohl, A., and D. Stammer
-
(2009): The role of the atmosphere for satellite gravity field missions; in: Sideris, M.G. (eds.) Observing our Changing Earth, lAG Symposia, Vol. 133, pp 105-112, Springer, ISBN (Print) 978-3-540-85425-8, ISBN (Online) 978-3-540-85426-5, ISSN 0939-9585
Gruber, T., Peters, T., Zenner, L.
-
(2010), Comparison of GRACE and Model-based Estimates of bottom Pressure Variations against in situ Bottom Pressure Measurements, In: F. Flechtner et al. (eds.). System Earth via Geodetic-Geophysical Space Techniques, 297 Advanced Technologies in Earth Sciences, Springer-Verlag Berlin Heidelberg 2010
Stammer, D. A. Kohl, V. Romanova, and F. Siegismund
-
(2010): Improved Non-tidal Atmospheric and Oceanic De-aliasing for GRACE and SLR Satellites, Flechtner et al. (eds.). System Earth via Geodetic-Geophysical Space Techniques, Adv. Technologies in Earth Sciences, Springer-Verlag Berlin Heidelberg 2010, 131-142
Flechtner F., Thomas M., Dobslaw H.
-
(2010): Propagation of atmospheric model errors to gravity potential harmonics - impact on GRACE De-Aliasing. Geophysical Journal International, Vol. 182, Nr. 2, pp 797-807, Wiley, 2010
Zenner, L., Gruber, T., Jäggi, A., Beutler, G.
-
(2010): The Release 04 CHAMP and GRACE EIGEN Gravity Models, Flechtner et al. (eds.). System earth via Geodetic-Geophysical Space Techniques, Adv. Technologies in Earth Sciences, Springer-Verlag Berlin Heidelberg 2010
Flechtner F., Dahle, Ch., Neumayer, K.H., König R., Forste, Ch.
-
(2011), Ocean Bottom Pressure Variations Estimated from gravity, non-steric sea surface height and hydrodynamic model simulations, J. Geophys. Res., 116, C07021
Siegismund F., A. Köhl and D. Stammer
-
(2011): Simulation of the time-variable gravity field by means of coupled geophysical models; Earth System Science Data, Vol. 3, Nr. 1, pp 19-35, Copernicus Publications, ISSN 1866-3591
Gruber, T.; Bamber, J. L.; Bierkens, M. F. P.; Dobslaw, H.; Murböck, M.; Thomas, M.; van Beek, L. P. H.; van Dam, T.; Vermeersen, L. L. A.; Visser, P. N. A. M.
