06 Fakultät Luft- und Raumfahrttechnik und Geodäsie
Permanent URI for this collectionhttps://elib.uni-stuttgart.de/handle/11682/7
Browse
1 results
Search Results
Item Open Access Applying stochastic constraints on time-variable GRACE data(2009) Lorenz, ChristofSince its launch in the year 2002, the space satellite mission GRACE provides spherical harmonic coefficients, which can be used to observe the time-variable part of the Earth's gravity field. It was initially assumed that the derived gravitational quantities from these coefficients are of high accuracy and would thus deliver reliable large scale mass estimates. However, the provided coefficients of higher harmonic degree and order turned out to be seriously contaminated with noise, yielding an unrealistic signal of mass variations in form of massive north-south stripes. In this work, two methods are investigated, which add stochastic constraints to time variable GRACE coefficients. It is assumed that these techniques are able to reduce the noise level in the monthly datasets by assimilating the GRACE coefficients with more reliable data. Both approaches need prior estimates of a signal and error covariances. Hence, the signal covariance of the time-variable gravity field is assumed to be of isotropic nature and is thus computed as a Kaula-type power law, which is fit into the part where the signal degree variances of the GRACE solutions linearly attenuate. The error covariance is estimated according to the energy balance approach which allows the simulation of a fully populated GRACE covariance matrix. Stochastic constraining in the spectral domain combines both signal and error covariance estimates in a Bayesian type regularization procedure, which constrains the monthly GRACE solutions with the modelled signal covariance. It is shown that Bayesian type regularization can be used to build a spectral filter kernel. Furthermore, the weight between both GRACE coefficients and the regularization term is estimated by a variance component estimation. Tests with a full, block diagonal and diagonal covariance matrix are performed, as it is widely believed that full covariance information can be sufficiently approximated by a block diagonal matrix. Furthermore, the Bayesian type regularization filter is tested with three different monthly GRACE solutions and compared with other widely used filtering techniques. The second approach constrains the time-variable GRACE coefficients with hydrological observations, which are provided as monthly precipitation and run-off values on basin scale. Both the GRACE and hydrological observation group are assimilated in one linear model, which is solved by sequential least squares estimation, yielding an agreement between mass estimates from GRACE and observed hydrology.