Impacts of improved cryospheric surface shortwave radiative transfer scheme SNICAR-AD on fully-coupled Earth system simulations

Many Earth system models predict different snow shortwave radiative properties depending on whether the snow is on land or on sea ice. We developed a hybrid shortwave radiative transfer scheme SNICAR-AD (SNow, ICe, and Aerosol Radiative model - Adding Doubling) to accurately simulate shortwave radiation on all cryospheric surfaces (i.e. snow, sea ice, ponds, and land ice). SNICAR-AD is implemented in the land and sea-ice components of the new Energy Exascale Earth System Model (E3SM). Here we present the impacts of the harmonized and more accurate cryospheric shortwave radiation produced by SNICAR-AD in fully coupled E3SM simulations (1985-2014, 30-year climatology), focusing on snow, sea ice, light-absorbing particles, and their radiative effects.

In polar regions, compared to the default sea-ice module, SNICAR-AD reduces the sea-ice albedo by 0.01, and reduces snow volume on sea ice by 45 km 3 and 89 km 3 in the Northern and Southern hemispheres, respectively. SNICAR-AD changes first-year sea-ice areas in the Northern and Southern hemispheres each by 0.2 million km 2 but in opposite directions, though the total sea-ice area remains approximately the same in each hemisphere. Compared to the default land snow module SNICAR, SNICAR-AD predicts lower diffuse snow albedo. The surface albedo of Greenland and Antarctica is reduced by 0.01-0.03 in summer, which increases shortwave absorption by ~2 W/m 2 in both areas. SNICAR-AD also reduces the seasonal snow albedo in mid-latitudes, altering the spatial patterns of surface albedo, snow coverage, and light-absorbing particles in snow. These net variations of the cryospheric surface warm the global annual mean surface temperature by 0.08 °C.

Comparing to in-situ observations from the JAWS automatic weather station database, the default E3SM underestimates the Greenland Summit and South Pole albedo by 0.03 and 0.1, respectively, in the summer months. Implementing SNICAR-AD in E3SM improves the shortwave radiative transfer algorithm relative to theory and to multi-stream radiation calculations yet amplifies this bias, which implies an unknown bias in the representation of snow physics and/or composition. More evaluation of model-simulated snow cover, snow depth, and light-absorbing particles in snow are in progress.