Challenges in simulation of snow microstructure and implications for remote sensing of snow mass

One of the greatest challenges for global measurement of snow mass is quantification of the snow microstructure. Radiative transfer models are more sensitive to the snow structure metrics used than to snow depth, so microstructure must be well quantified in order to retrieve snow mass from satellite observations. Principles of physics have been used to simulate microstructure in many years of avalanche and climate research, although these have different accuracy requirements to remote sensing applications.

Growth of snow crystals is dependent primarily on the snow temperature gradient, but also the temperature and density of the snow. Forced with the same meteorological data, different models simulate different snow temperatures. Even with the same grain growth assumptions, this leads to different rates of microstructure evolution. This must be taken into consideration if snow models are to be used to give the necessary parameters for retrieval of snow water equivalent.

The JULES Investigation Model snow model (JIM) has a highly configurable structure that allows different layering assumptions to be used. It incorporates all major components from existing snow models, which enables the simulation of an ensemble of 1701 members. JIM was used to quantify the impact of different model parameterizations such as snow compaction and thermal conductivity on simulated microstructure (previously referred to as ‘grain size’) for each of four different grain size parameterizations from the Crocus, MOSES, SNICAR and SNTHERM models. The Helsinki University of Technology snow microwave emission model was then used to demonstrate the impact of different snow model assumptions on the simulation of microwave brightness temperature. This paper discusses potential snow mass retrieval errors due to uncertainties in snow parameters from snow evolution models, and how these may be mitigated through techniques such as data assimilation.