Application of GRACE to the assessment of model-based estimates of monthly Greenland Ice Sheet mass balance (2003–2012)
- 1University of California, Los Angeles, Los Angeles, CA, USA
- 2Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California, USA
- 3Geological Survey of Denmark and Greenland (GEUS), Øster Voldgade 10, 1350 København, Denmark
- 4University of Liège, Department of Geography, 4000 Liège, Belgium
- 5Institute for Marine and Atmospheric Research Utrecht (IMAU), Utrecht University, Utrecht, the Netherlands
Abstract. Quantifying the Greenland Ice Sheet's future contribution to sea level rise is a challenging task that requires accurate estimates of ice sheet sensitivity to climate change. Forward ice sheet models are promising tools for estimating future ice sheet behavior, yet confidence is low because evaluation of historical simulations is challenging due to the scarcity of continental-wide data for model evaluation. Recent advancements in processing of Gravity Recovery and Climate Experiment (GRACE) data using Bayesian-constrained mass concentration ("mascon") functions have led to improvements in spatial resolution and noise reduction of monthly global gravity fields. Specifically, the Jet Propulsion Laboratory's JPL RL05M GRACE mascon solution (GRACE_JPL) offers an opportunity for the assessment of model-based estimates of ice sheet mass balance (MB) at ∼ 300 km spatial scales. Here, we quantify the differences between Greenland monthly observed MB (GRACE_JPL) and that estimated by state-of-the-art, high-resolution models, with respect to GRACE_JPL and model uncertainties. To simulate the years 2003–2012, we force the Ice Sheet System Model (ISSM) with anomalies from three different surface mass balance (SMB) products derived from regional climate models. Resulting MB is compared against GRACE_JPL within individual mascons. Overall, we find agreement in the northeast and southwest where MB is assumed to be primarily controlled by SMB. In the interior, we find a discrepancy in trend, which we presume to be related to millennial-scale dynamic thickening not considered by our model. In the northwest, seasonal amplitudes agree, but modeled mass trends are muted relative to GRACE_JPL. Here, discrepancies are likely controlled by temporal variability in ice discharge and other related processes not represented by our model simulations, i.e., hydrological processes and ice–ocean interaction. In the southeast, GRACE_JPL exhibits larger seasonal amplitude than predicted by the models while simultaneously having more pronounced trends; thus, discrepancies are likely controlled by a combination of missing processes and errors in both the SMB products and ISSM. At the margins, we find evidence of consistent intra-annual variations in regional MB that deviate distinctively from the SMB annual cycle. Ultimately, these monthly-scale variations, likely associated with hydrology or ice–ocean interaction, contribute to steeper negative mass trends observed by GRACE_JPL. Thus, models should consider such processes at relatively high (monthly-to-seasonal) temporal resolutions to achieve accurate estimates of Greenland MB.