Towards dedicated snow modelling in the Arctic that allows quantification of the impact of light absorbing impurities in snow

Abstract

Snow is an important factor in the Earth System as it influences the global energy balance due to its high albedo. Light-absorbing impurities (LAI) in snow reduce its albedo, leading to enhanced absorption of shortwave radiation that warms the snowpack and stimulates feedback mechanisms that are partly responsible for faster warming of the Arctic than in any other part of the world. However, to date there is a lack of sufficient measurements to quantify the concentrations of LAI in Arctic snow, their seasonal evolution and trends. Therefore, it is difficult to investigate, quantify and understand the effects of LAI in snow on the evolution of snow properties, including the snow albedo, associated feedback mechanisms and the radiative forcing, especially in remote areas. To address this gap, models simulating the evolution of snow properties taken into account the effects of LAI on the radiative energy balance can be applied. However, reliable measurements to force these simulations, i.e. reliable deposition rates, are as well limited in time and space. This dissertation has the ambitious goal to simulate reliably the impact of LAI on the radiative energy balance in snow. Several milestones had to be reached to achieve this goal. The first milestone was to find reliable forcing data for remote regions of the Arctic. The second milestone was to develop, for a snow model designed for application in the European Alps, fit-for-the-Arctic parameterisations that describe sufficiently well the evolution of snow properties. After reaching these milestones, the effects of LAI in snow, using exemplarily black carbon (BC), on the evolution of snow properties could be investigated. The analyses were performed at a site in northeast Greenland, using atmospheric in-situ and snow depth data from Villum Research Station (VRS) (2014 to 2018) together with additional snow measurements carried out during the Polar Airborne Measurements and Arctic Regional Climate Model Simulation Project (PAMARCMiP) campaign 2018, the modern global atmospheric reanalysis ERA5 and the regional atmospheric reanalysis CARRA-West, and the detailed snow model Crocus. My results show that reanalyses are in principle able to represent prevailing atmospheric conditions, but suffer when resolving precipitation (amount and timing) and high wind speeds. Thereby, CARRA-West resolves small scale events and orography better than ERA5. When comparing CARRA-West and ERA5 forced simulations the performance vary from year to year. But overall, CARRA-West driven simulations perform slightly better than ERA5 driven simulations. Surprisingly, though, CARRA-West and ERA5 forced simulations agree better with snow measurements than simulations forced by atmospheric in-situ measurments. In general, Crocus performs well with regard to snow depth evolution, while there are weaknesses in simulating vertical profiles of snow density and snow specific surface area (SSA). But, there are still shortcomings with respect to the simulation of snow depth. Snow depth decreases too rapidly in spring, possibly due to a combination of overestimated compaction and snow metamorphism formulations that are insufficient for Arctic conditions resulting in a too rapid decrease in SSA. This has implications for the simulation of snow albedo and hence the simulation of the effects of BC in snow. The study shows that the presence of BC reduces the surface snow albedo, which leads to a shortening of the snow season. Further effects are a decrease in snow depth and warmer ground temperatures. Efforts to improve model performance by introducing a new snow density parameterisation adapted to Arctic atmospheric conditions and modifications in a parameterisation influencing the upper snow density during strong wind events result in higher agreements between measured and simulated snow properties. In conclusion, this dissertation substantially advances the understanding of the performance and sensitivity of the snow model Crocus in the Arctic, including knowledge gain on processes that are important in the Arctic. The results of this dissertation provide an improved understanding of the reasons for deviations between measured and simulated snow properties caused by the forcing data and model parameterisations. Thus, it provides a good basis for the design of future research studies and field campaigns to evolve further snow modelling in the Arctic. Reliable snow simulations in the Arctic are needed to analyse feedback mechanisms and processes leading to enhanced Arctic warming. Finally, this dissertation presents the prevailing BC concentrations in Greenlandic snow and demonstrates the effects different BC deposition rates have on the evolution of snow properties.