Assessment of the Greenland ice sheet–atmosphere feedbacks for the next century with a regional atmospheric model coupled to an ice sheet model

In the context of global warming, the projected Greenland sea level rise contribution is mainly controlled by the interactions between the Greenland ice sheet (GrIS) and the atmosphere, in particular through the temperature and surface mass balance – elevation feedback. In order to evaluate the importance of these feedbacks, we used three methods to represent the interactions between the GrIS model GRISLI and the polar regional atmosphere model MAR, under the RCP 8.5 scenario from 2020 to 2150. In the simplest method, there is no coupling, MAR computes varying atmospheric conditions using a constant GrIS geometry (topography and extent) set to observations and GRISLI is forced by these results. The second is a one-way coupling method which represents the interactions by correcting offline the MAR outputs to account for topography changes computed by GRISLI. The third method is a full two-way coupling in which the ice sheet topography and extent seen by the atmospheric model evolve after each ice sheet model time step. Due to the ice sheet elevation feedback, the two-way coupling method amplifies the projected decrease in surface mass balance, the increase in surface temperature and the GrIS surface thinning for the coastal regions, compared to the no coupling method. Compared to both the one-way and the no coupling methods, the two-way coupling allows the changes of fine scale processes to be represented, such as the increase in katabatic winds over the coast. As a consequence, in 2150, the two-way coupling method computes a GrIS melting contribution to sea level rise 9.3 % larger than the no coupling method, and 2.5 % larger than the one-way coupling methods. After 150 years, the GrIS extent seen by MAR in the two-way method is 52 400 km 2 lower than with the no coupling method. Furthermore, in 2150, using a fix ice sheet mask, as in the no coupling method, overestimates by 24 % the SLR contribution from SMB compared to the use of the ice sheet mask as simulated in the two-way method. Beyond the century time-scale, a two-way method becomes necessary in order to avoid an underestimation of the projected ice sheet volume, topography and ice extent reduction. The one-way coupling method however seems to be sufficient to represent the interactions for projections until the end of the 21 st century. The no coupling method always underestimates the projected ice sheet volume loss significantly due to the lack of feedback between the GrIS and the atmosphere.