Diagnosis of performance of the Noah LSM snow model

Quantifying snow water equivalent (SWE) and runoff timing are vital in hydrologic modeling, especially for snowmelt driven regimes such as the western U.S. and the pan-Arctic. The Noah land-surface model, used in the National Center for Environmental Prediction's operational suite of weather and climate forecast models, has shown a tendency toward ablation of snowpacks earlier than in observations in off-line tests. We report a diagnosis of the sensitivities of various components of the snow scheme in the Noah model that might affect snow ablation. We find that issues with the model's turbulent exchange schemes, and its radiative exchange formulations appear to be the main contributors to the negative SWE bias during the ablation season. Using modified albedo and surface exchange schemes, we test alternative formulations both using point observations at observing sites in mountain maritime locations in the Pacific Northwest, and in the interior of the western U.S. and Canada, as well as in simulations of seasonal variations of snow extent over the entire pan-Arctic drainage. Our results suggest that the a modest change in the snow albedo formulation results in the greatest improvement in model performance, although under some conditions an alternation in the turbulent exchange parameterization produces improved results as well.