Secular gravity variation at Svalbard (Norway) from ground observations and GRACE satellite data

The Svalbard archipelago, Norway, is affected by both the present-day ice melting (PDIM) and Glacial Isostatic Adjustment (GIA) subsequent to the Last Pleistocene deglaciation. The induced deformation of the Earth is observed by using different techniques. At the Geodetic Observatory in Ny-Ålesund, precise positioning measurements have been collected since 1991, a superconducting gravimeter (SG) has been installed in 1999, and six campaigns of absolute gravity (AG) measurements were performed between 1998 and 2007. Moreover, the Gravity Recovery and Climate Experiment (GRACE) satellite mission provides the time variation of the Earth gravity field since 2002. The goal of this paper is to estimate the present rate of ice melting by combining geodetic observations of the gravity variation and uplift rate with geophysical modelling of both the GIA and Earth's response to the PDIM. We estimate the secular gravity variation by superimposing the SG series with the six AG measurements. We collect published estimates of the vertical velocity based on GPS and VLBI data. We analyse the GRACE solutions provided by three groups (CSR, GFZ, GRGS). The crux of the problem lies in the separation of the contributions from the GIA and PDIM to the Earth's deformation. To account for the GIA, we compute the response of viscoelastic Earth models having different radial structures of mantle viscosity to the deglaciation histories included in the models ICE-3G or ICE-5G. To account for the effect of PDIM, we compute the deformation of an elastic Earth model for six models of ice-melting extension and rates. Errors in the gravity variation and vertical velocity are estimated by taking into account the measurement uncertainties and the variability of the GRACE solutions and GIA and PDIM models. The ground observations agree with models that involve a current ice loss of 25 km3 water equivalent yr−1 over Svalbard, whereas the space observations give a value in the interval [5, 18] km3 water equivalent yr−1. A better modelling of the PDIM, which would include the precise topography of the glaciers and altitude-dependency of ice melting, is necessary to decrease the discrepancy between the two estimates.