Deriving Antarctic firn density and post-glacial rebound values from a combination of GRACE and altimetry

Mass balance estimates for the Antarctic suffer from large inaccuracies, in part introduced by a lack of sufficient understanding of postglacial-rebound (PGR) processes. In this work, data from the Gravity Recover and Climate Experiment (GRACE) was combined with altimetry data from Envisat over the period January 2003 - October 2010 to gain an insight into PGR in the Antarctic. The GRACE data, which has known errors in the form of stripes, was smoothed with a 300km Gaussian filter and subjected to an EOF analysis. This reveals months exhibiting significant anomalies / a large amount of noise which can then be removed from the analysis. The Envisat data was interpolated and smoothed to be compatible with the GRACE data. Apart from the errors associated with each satellite, the largest uncertainty in a study of this kind comes from necessary assumptions made about firn density values. Conventionally, a surface density model is employed. When subtracting Envisat from GRACE trends using modelled density values, the assumption is that the difference between the two can be interpreted as a combination of PGR and errors. Consequently, if the modelled firn density values are incorrect, the PGR interpretation will be erroneous. Relating the mass and thickness change of an ice sheet is not trivial, and the effective firn density at which changes appear to occur is a function of the time-scale of variations. Characteristics in the altimetry trend allow attribution of the spatial pattern of surface changes to different densities and causes. A change in the pattern of surface height trends over shorter periods (half the measurement period for instance) points at snow variation rather than variations in ice flow. This was confirmed in areas where altimetry data had already indicated snow-driven changes, such as Ellsworth and Queen Maud Land. Dividing the measurement period into two four-year time series also reveals the domination of errors over shorter time-scales. Furthermore, the difference between gravimetry and altimetry under different firn density assumptions can reveal the time-scales and thus correct density values at which changes occur. Areas like Pine Island Glacier and Wilkes Land are dominated by ice flow and reveal rebound trends. Here, we explore using the difference between GRACE and Envisat to determine appropriate firn density values for the Antarctic continent, which in turn allows us to confidently interpret any remainder as PGR. The errors of both satellites are taken into account and discussed. Even though the size of the error is roughly a quarter of the size of the signal in some cases, areas with a large signal place confidence in our results. Accurate PGR values do not only improve the accuracy of Antarctic mass balance estimates, which we will report on, but also facilitate interpretations of lithospheric density values beneath the Antarctic continent.