GRACE-derived linear and non-linear secular mass variations over Greenland

Abstract

In the past, GRACE (Gravity Recovery and Climate Experiment) monthly gravity field solutions have mainly been exploited to derive secular and seasonal mass changes on the Earth’s surface. After seven years in operational mode, the satellite mission makes accelerated and decelerated mass variations detectable. Here we investigate the temporal characteristics of secular trends by fitting both linear and higher-order polynomials to the mass-change time-series. Our findings have been derived from GRACE gravity field time-series provided by CSR, GFZ and JPL. As a case study, we look at recent ice-mass variations over Greenland. Based on various model selection criteria (Akaike and Bayesian information criterion, cross-validation, hypotheses testing), our investigations show that linear regression is unable to describe recent deglaciation. Instead, the secular trend is best represented by a secondorder polynomial, confirming accelerated deglaciation of the Greenland ice sheets, which increased by 250% between April 2002 and March 2009.