Sensitivity of inverse glacial isostatic adjustment estimates over Antarctica

• Main Authors: Willen, Matthias O., Horwath, Martin, Schröder, Ludwig, Groh, Andreas, Ligtenberg, Stefan R. M., Kuipers Munneke, Peter, Broeke, Michiel R.
• Format: Text
• Language: English
• Published: 2019
• Subjects: East Antarctica; Antarc*; Antarctica
• Online Access: https://doi.org/10.5194/tc-2019-95; https://www.the-cryosphere-discuss.net/tc-2019-95/

Glacial isostatic adjustment (GIA) is a major source of uncertainty in estimated ice and ocean mass balance that are based on satellite gravimetry. In particular over Antarctica the gravimetric effect of cryospheric mass change and GIA are of the same order of magnitude. Inverse estimates from geodetic observations are promising for separating the two superimposed mass signals. Here, we investigate the combination of satellite gravimetry and altimetry and how the choice of input data sets and processing details affect the inverse GIA estimates. This includes the combination for almost full GRACE lifespan (2002-04/2016-08). Further we show results from combining data sets on time-series level. Specifically on trend level, we assess the spread of GIA solutions that arises from (1) the choice of different degree-1 and C 20 products, (2) different surface elevation change products derived from different altimetry missions and associated to different time intervals, and (3) the uncertainty of firn-process models. The decomposition of the total-mass signal into the ice-mass signal and the apparent GIA-mass signal depends strongly on correcting for apparent biases in initial solutions by forcing the mean GIA and GRACE trend over the low precipitation zone of East Antarctica to be zero. Prior to bias correction, the overall spread of total-mass change and apparent GIA-mass change using differing degree-1 and C 20 products is 68 and 72 Gt a −1 , respectively, for the same time period (2003-03/2009-10). The bias correction suppresses this spread to 6 and 5 Gt a −1 , respectively. We characterise the firn-process model uncertainty empirically by analysing differences between two alternative surface-mass-balance products. The differences propagate to a 21 Gt a −1 spread in apparent GIA-mass-change estimates. The choice of the altimetry product poses the largest uncertainty on debiased mass-change estimates. The overall spread of debiased GIA-mass change amounts to 18 and 49 Gt a −1 for a fixed time period (2003-03/2009-10) and various time periods, respectively. Our findings point out limitations associated with data processing, correction for apparent biases, and time dependency.