A model that accounts for the compaction of snow near the surface of the Greenland and Antarctic ice sheets (aka firn) was implemented into the US Department of Energy's Energy Exascale Earth System Model (E3SM). Simulations of accumulating snow and its transformation into ice via densification demonstrate that, compared to a single-stage compaction scheme more commonly used in seasonal snowpack models, a two-stage compaction model better predicts firn densities at intermediate depths (i.e., 20 to 60 m).
Improving the simulation of firn density in E3SM also results in better estimates of firn air content. Better accounting for firn air content is crucial for improving estimates of surface mass balance, predicting ice shelf hydrofracture, and simulating the land ice and sea-level rise responses to climate change in Earth system models.
We adapt E3SM's snowpack model to accommodate greater depths on ice sheets. Firn depths of up to 60 m are resolved in the E3SM Land Model (ELM) by adding 11 layers to its snowpack discretization. The expanded (16-layer) discretization was used to test three dry snow compaction equations in multi-century ELM simulations. We found that implementing into ELM a two-stage firn densification model resulted in more accurate dry firn densities at intermediate depths of 20 to 60 m. Compared to modeling firn using the equations in the Community Land Model (version 5), switching to the two-stage firn densification model significantly decreased root-mean-square errors in upper 60 m dry firn densities by an average of 41 kg m-3 (31%). Our results suggest that the two-stage model will improve estimates of firn air content in regions where the mean interannual surface temperature is warmer than -30ºC. Our developments in E3SM combine both seasonal snow and firn processes and advance broader efforts to simulate ice sheet evolution and sea-level rise in Earth system models.