Remote sensing and modelling of snow processes

This thesis aims to improve snow processes simulations by land surface models (LSM). The LSM-simulated snow processes define energy and water fluxes over the snow-covered land-surface for the numerical weather prediction models. Therefore, the improvements in snow processes simulations are expected to impact positively the weather and streamflow forecasts. The research described in this thesis contribute to the improvement of LSM simulations by investigating i) advanced methods of model initialization (e.g., assimilation of satellite observation), and ii) modification to the model parameterization.
The thesis quantifies the performance of the existing approaches for (i) retrieval from satellite-based observations and (ii) simulations by LSM of snow properties. The retrieval approaches of snow albedo and fractional snow coverage (FSC) and simulation approaches (also called parameterizations) for snow albedo are found in good agreement with the in situ measurements. However, uncertainties in terms of bias and variance exist because the approaches are based on different assumptions and approximations. These uncertainties adversely affect the simulation – with or without assimilation of satellite-observed snow albedo – of snow properties (depth, albedo, coverage, and duration) and energy and water fluxes (runoff, melt rate, evapotranspiration, upward shortwave radiations).