On the Determination of Sea Level Changes by Combining Altimetric, Tide Gauge, Satellite Gravity and Atmospheric Observations

The determination, monitoring and understanding of sea level change at various spatial and temporal scales has been the focus of many studies during the past decades. The advent of satellite altimetry and the multitude of unprecedented in accuracy and resolution observations that it offers allowed, in combination with tide gauge data, precise determinations of sea level variations. The realization of the GRACE mission and the forthcoming GOCE data offer new opportunities for the estimation of sea level trends at regional and global scales and the identification of seasonal signals. In such studies, even though the data combination and processing strategies have been carried out carefully with proper control, a point that has been given little attention is error propagation and variance component estimation of the data variance-covariance matrices. The latter two are of significant importance in heterogeneous data combination studies, since on one hand error propagation can provide reliable estimates of the output signal error while variance component estimation allows for a rigorous control of the data covariance matrices and subsequent sound decisions on statistical testing of hypotheses involving least-squares residuals and the estimated deterministic parameters. This work presents some new ideas towards the estimation of sea level change through the combination of altimetric, tide gauge, atmospheric, and GRACE- and GOCE-type observables. The combination scheme is based on a hybrid deterministic and stochastic treatment of the data errors and an estimation of sea level changes through least-squares collocation with considerations for glacial isostatic adjustment and continental water outpour effects. The deterministic model parameters treat datum and geophysical correction model inconsistencies in the data used, while the stochastic part allows for a simultaneous determination of stochastic parameters included in the data in terms of residual signals. Within this mixed adjustment scheme with stochastic parameters, variance component estimation is carried out using the iterative almost unbiased estimator method. The analytical equations for the prediction of the adjusted input and output signals are presented along with possible modifications of the observation equation for the determination of solely steric and atmospheric-driven sea level changes.