Understanding how uncertainty in the forcing irradiances impacts simulations of snow

Thesis (Master's)--University of Washington, 2014Snowmelt in mountains is an important part of the water and energy cycles and provides water for 1/6th of the world's population. The downwelling irradiances are the primary drivers of this melt, however, they are rarely observed. The use of estimated irradiances, few observations, lack of evaluation of alternative sources of data, and the unique climate of mountain environments all lead to substantial uncertainties in the radiative fluxes used to force simulations of snow. The net irradiance of snow is determined by external forcing irradiances, the downwelling irradiances, and by the upwelling irradiances, which are functions of the internal model feedbacks. Errors in the forcing irradiances can be masked by errors in the internal processes that control the outgoing irradiances. The impact of uncertainties in the forcing irradiances for simulations of snow is evaluated in a series of idealized modeling experiment that split into two parts: 1) understanding errors in the forcing irradiances alone and 2) understanding the feedback and compensation between errors in the forcing irradiances and the internal processes that control the outgoing irradiances. In the forcing irradiances, it is shown that longwave biases of magnitude greater than 20 Wm-2 and shortwave biases of magnitude greater than 40 Wm-2, typical of methods for estimating irradiances in complex terrain, have substantial impacts on simulated snow water equivalent (SWE) and the simulated energy balance across a range of mountain climates. Random noise in the forcing irradiances has a negligible effect on modeled snowmelt and energy balance. The exception is warmer sites, which were found to be sensitive to nearly all errors in the forcing irradiances. The internal processes that control the outgoing fluxes can significantly impact the net irradiance of the snow. Two processes are explored: 1) albedo parameterization that controls the reflected shortwave irradiance and 2) the turbulence parameterization that controls the outgoing longwave irradiance through the surface temperature. Tuning of albedo parameters, an approach typically taken in modeling set-ups, can completely compensate for biases in the forcing irradiances when evaluating model performance using SWE. Varying turbulent flux parameters was found to have a much smaller impact on simulated snowmelt than albedo parameters - calling the role of the stability feedback into question. However, the surface temperature does depend strongly on the turbulence scheme selected. Finally, the application of these results is shown for a variety of mountain environments and methods. In general, the uncertainty in the albedo terms is larger than the uncertainty in the forcing irradiance terms. The structure of errors in the forcing irradiances is either uniform offsets that do not vary substantially throughout the year or shorter punctuated periods where the irradiance values are substantially different