We have a new open-access study out in the current volume of The Cryosphere that looks at geothermal heat flow beneath the Greenland Ice Sheet. Geothermal heat flow is an important boundary condition for ice flow models because it influences the temperature of the ice-bed interface, which in turn influences how easily ice can deform and flow. Right now, there are about seven widely used estimates of geothermal heat flow beneath the Greenland Ice Sheet (Figure 1). These different heat flow maps come from different research groups, using different methods. It hasn’t been entirely clear what the influence of choice of heat flow map has on modeled ice flow. For example, the ensemble of Greenland ice flow projections within the Ice-Sheet Model Inter-comparison Project for CMIP6 (ISMIP6) used differing heat flow maps. Our new study simply spins up a Greenland ice flow model with all seven heat flow maps and tries to understand the resulting differences in ice thickness and velocity.
Although the average geothermal heat flow only varies by about ±10 mW/m2 across the seven heat flow maps, there are pronounced variations in the spatial distributions of this heat flow. This means that there can be local heat flow differences of up to ±100 mW/m2 between individual heat flow maps. When the ice sheet model is spun up in a fully transient mode with these different heat flow maps, there are lots of areas where the difference in ice thickness from ensemble mean exceeds ±150 meters (Figure 2). This is due to spatial differences in basal ice temperatures, which strongly influence the viscosity and deformation of ice. Across the seven transient spin ups, the discharge of icebergs into the ocean varies by about ±10 gigatonnes per year, which is equivalent to about ±2.5% of the total ice-sheet iceberg discharge.
We also spun up the ice flow model with the seven different heat flow maps under a so-called “nudged” spin up that was a key part of ISMIP6. Unlike a fully transient spin up, a “nudged” spin up constrains modeled ice thicknesses with observed ice thicknesses; it generally ensures a very realistic geometry for the simulated ice sheet. Although the ice geometry is more-or-less constant across these “nudged” simulations, there are still pronounced differences in the magnitude and spatial distribution of ice velocity. This largely results from large differences in the extents of the frozen and thawed ice-bed areas. Depending on choice of heat flow map, between 22 and 54% of the ice-bed area is simulated as thawed (Figure 3). This has strong implications for the proportion of the ice sheet beneath which water-dependent processes, like basal sliding, can occur.
So, what is the way forward after showing that the choice of heat flow map has a non-trivial impact on Greenland ice sheet simulations? Well, deciding which heat flow map is “the best” is one approach. We highlight a small, but growing, database of in situ temperature measurements against which simulated ice temperatures can be compared. There are also qualitative hints as to which heat flow map might be most appropriate. For example, does it preserve the widespread frozen basal conditions that we find in North Greenland? In terms of community ice-sheet projections, we recommend that it may be prudent to limit the direct inter-comparison of ice-sheet simulations to those using a common heat-flow map. In terms of ISMIP specifically, we suggest that future ensemble should perhaps use a range of basal geothermal forcing scenarios, similar to how it employs a range climate forcing scenarios.
Zhang, T., Colgan, W., Wansing, A., Løkkegaard, A., Leguy, G., Lipscomb, W. H., and Xiao, C.: Evaluating different geothermal heat-flow maps as basal boundary conditions during spin-up of the Greenland ice sheet, The Cryosphere, 18, 387–402, https://doi.org/10.5194/tc-18-387-2024, 2024.