Potential of using satellite derived snow products for estimating snow aerodynamic roughness length and evaposublimation across spatio-temporal scales

Evaposublimation of snow plays an important role in the energy balance of snow, particularly in low- and mid-latitude mountain regions where this process can contribute substantially to overall snow mass partitioning. The evaposublimated snow, driven by the exchange of turbulent latent heat fluxes between the snow surface and the atmosphere, have significant implications for water resources management, as they reduce the meltwater released to the soil and rivers.  

A key parameter in quantifying turbulent heat fluxes is the aerodynamic roughness length, which represents the height above the surface where the horizontal wind speed drops to zero. This parameter is intrinsically linked to the surface roughness of snow, which is highly dynamic and evolves with the snowpack's physical state. As the snow transforms, its surface characteristics, and consequently its aerodynamic roughness length, can vary substantially, influencing the magnitude of turbulent flux exchanges. Modeling turbulent latent heat fluxes however often suffers from limited knowledge of spatio-temporal evolution of aerodynamic roughness length, leading to significant uncertainty in evaposublimation rate estimates.

Remote sensing offers a valuable tool to monitor snow properties across spatio-temporal scales. In this study, we investigate the potential of satellite derived products related to the current state of snow such as snow cover fraction, albedo, snow grain size, and land surface temperature in combination with in-situ meteorological measurements, to predict aerodynamic roughness lengths of snow, and consequently turbulent latent heat fluxes in the European Alps on a spatio-temporal scale using machine learning regression models. Validation is conducted using roughness lengths and turbulent latent heat flux data obtained from three FLUXNET eddy-covariance stations. This approach assesses the feasibility of generalizing predictions of evaposublimation from the ground across different locations and temporal scales contributing to a better understanding of its implications for snowpack dynamics and water resource management.