Resolving sea level contributions by identifying fingerprints in time-variable gravity and altimetry

We have studied the ability of the GRACE gravimetry mission and Jason-1 altimetry to resolve ice and glacier induced contributions to sea level rise, by means of a fingerprint method. Here, the signals from ice sheet and land glacier changes, steric changes, glacial isostatic adjustment and terrestrial hydrology are assumed to have fixed spatial patterns. In a joint inversion using GRACE and Jason-1 data the unknown temporal components can then be estimated by least-squares. In total, we estimate temporal components for up to not, vert, similar 80 individual patterns. From a propagation of the full error-covariance from GRACE and a diagonal error-covariance from Jason-1 altimetry we find that: (1) GRACE almost entirely explains the mass related parameters in the joint inversion, (2) an inversion using only Jason-1 data has a marginal ability to estimate the mass related parameters, while the steric parameters have much better formal accuracy. In terms of mean sea level rise the steric patterns have a maximum formal accuracy of 0.01 mm for an 11 week running mean. In general, strong negative error correlations (ρ < - 0.9) exists between the high and low elevation parts of the ice sheet drainage basins, when those are estimated independently. The largest formal errors found are in the order of 40 Gton for small high elevation subbasins in the southern Greenland ice sheet, which are difficult to separate. In a simplified joint inversion, merging high and low elevation basins, we have investigated the ability of the GRACE and Jason-1 data to separate the geocenter motion into a present-day contribution and a contribution from glacial isostatic adjustment (GIA). We find that the GIA related signal is larger than the present-day component with a maximum of -0.71 mm/year in the Z direction. Total geocenter motion rates are found to be -0.28, 0.43, -1.08 mm/year for the X, Y and Z components, respectively. The inversion results have been propagated to the Jason-1 along-track measurements. Over the time period considered, we see that a large part of the variability in the Pacific, Atlantic and Indian ocean can be explained by our inversion results. The applied inversion method therefore seems a feasible way to separate steric from mass induced sea level changes. At the same time, the joint inversion would benefit from more advanced parameterizations, which may aid in fitting remaining signal from altimetry.