Snowpack and firn densification in the Energy Exascale Earth System Model (E3SM) (version 1.2)
Abstract. Earth's largest island, Greenland, and the Antarctic continent are both covered by massive ice sheets. A large fraction of their surfaces consist of multi-year snow, known as firn, which has undergone a process of densification since falling from the atmosphere. Until now this firn densification has not been fully accounted for in the U.S. Department of Energy's Energy Exascale Earth System Model (E3SM). Here, we expand the E3SM Land Model (ELM) snowpack from 1 m to up to 60 m to enable more accurate simulation of snowpack evolution. We test four densification models in a series of century-scale land surface simulations forced by atmospheric re-analyses, and evaluate these parameterizations against empirical density-versus-depth data. To tailor candidate densification models for use across the ice sheets' dry-snow zones, we optimize parameters using a regularized least squares algorithm applied to two distinct stages of densification. We find that a dynamic implementation of a semi-empirical compaction model, originally calibrated to measurements from the Antarctic peninsula, gives results more consistent with ice core measurements from the cold, dry snow zones of Greenland and Antarctica, compared to when using the original ELM snow compaction physics. In its latest release, the Community Land Model (CLM) (version 5) provides updated snow compaction physics that we test in ELM, resulting in top 10 m firn densities that are in better agreement with observations than densities simulated with the semi-empirical model. Below 10 m, however, the semi-empirical model gives results that more closely match observations, while the current CLM(v5) compaction physics predict firn densities that increase too slowly with depth and are thus unable to simulate pore close off (a phenomenon of particular interest to paleoclimate studies). Because snow and firn density play roles in snowpack albedo, liquid water storage, and ice sheet surface mass balance, these improvements will contribute to broader E3SM efforts to simulate the response of land ice to atmospheric forcing and the resulting impacts on global sea level.
This preprint has been withdrawn.