Quantifying Microphysical Parameterization Uncertainty in Convection-Permitting Forecasts of the 10–12 December 2013 Lake-Effect Snow Event

Lake-effect snow (LeS) presents a substantial forecast challenge for convection-permitting models, due in part to uncertainties in the parameterization of microphysical processes. Stochastic parameterization methods are one approach to better represent physics uncertainties in the development of next-generation, convection-permitting ensembles. Here we focus on understanding microphysical uncertainties for a LeS event that occurred during 10–12 December 2013 during the Ontario Winter Lake-effect Systems (OWLeS) field campaign. Throughout this event, long-lake-axis-parallel snowbands persisted downwind of the eastern shore of Lake Ontario, leading to snowfall accumulations as high as 101.5 cm (liquid precipitation equivalent of 62.5 mm) on the Tug Hill Plateau.

We run nested simulations of the 10–12 December 2013 LeS event at 12-, 4-, and 1.33-km horizontal grid spacing using the Weather Research and Forecasting (WRF) model configured similarly to the operational High-Resolution Rapid Refresh model. All simulations use Thompson aerosol-aware microphysics, with sensitivity experiments conducted by varying fixed parameters within the scheme that alter the distributions of ice categories, such as snow particle size distribution (PSD) equation coefficients, graupel density, and collection efficiencies of cloud water by snow and graupel.

Results from the WRF simulations are compared to detailed observations from OWLeS, including NEXRAD radar data and surface snowfall and crystal habit observations. Additionally, measurements from the University of Wyoming King Air (UWKA) aircraft, including in-situ flight-level thermodynamic and microphysical observations, are used to compare observed and modeled cloud structures. UWKA observations are also used to constrain the range of fixed parameter values in our sensitivity experiments.

For instance, when varying one of the Thompson snow PSD coefficients we simulate more accurate PSDs aloft relative to UWKA PSD observations and a 5-10 mm increase in precipitation on the windward side of the Tug Hill Plateau, reducing a windward underforecast bias of 15-30 mm in a control simulation. In this manner, additional experiments are conducted to investigate the forecast sensitivity of snow and graupel amounts over the Tug Hill Plateau. These fixed parameter experiments are a first step towards development of convection-permitting ensemble forecasts of LeS using stochastic physics.