Snow Level Forecasts Improved for Operations in Canada

Snow level (SL) forecasts using operational numerical models in Western Canada were improved by explicitly including diabatic cooling. The existing Canadian Meteorological Centre (CMC) Regional Deterministic Prediction System (RDPS-10 km) model with our post processing SL forecasts were too high, and better snow level guidance was requested. The RDPS-10 km model dry bulb freezing levels compared well to observations (using the top of the radar bright band and the upper freezing level), however the new most advanced CMC High Resolution Deterministic Prediction System (HRDPS-2.5 km) model was often too warm in major precipitation events, and the discrepancy in precipitation guidance during major weather events was confusingly large due to hydrometeor drift “spillover”.  With further post processing within SCRIBE (Canadian weather forecast text generator), the snow levels were also problematic. This combination of issues led to missed forecasts for heavy snowfall events and false alarms for heavy rainfall, especially for big storms crossing mountain passes.  To create better SL forecasts for this study, diabatic cooling is simulated using an Enhanced Wet Bulb Freezing (EWBZ) algorithm with CMC numerical models used in operations. The diabatic cooling is computed from the total latent heat exchange during precipitation as snow melts into rain when the atmosphere is saturated.  

The distance between the upper freezing level (FZLVL) and SL is called the Melting Snow Layer Thickness and is computed from Precipitation Intensity (QPS mm/3hr) and the atmospheric heat content (dQ/dz) available for melting below the freezing level. SL is defined as where snow accumulates on the ground, and this is compared to the computed EWBZ, WBZ, and FZLVL.  The base of the isothermal Wet Bulb Freezing Layer is the SL.  The heaviest precipitation rates are observed near the SL when the FZLVL was near/below the mountain summits. Observed lapse rates are steep and very unstable just below the SL during heavy snowfall, increasing convection and associated precipitation rates.  Interior and coastal British Columbia sites are examined using model calculations compared to vertical profiler, radar, radiosonde, surface weather and snowfall accumulation observations.  This new EWBZ algorithm more accurately predicts snow levels, when the model precipitation rates agree with observations.   

A new diagram was created that plots Melting Snow layer Thickness versus Precipitation Intensity to evaluate the SL forecasts during different atmospheric situations and to better validate results.  Major Atmospheric River (AR) events are typically associated with large Pacific Cyclones which deliver the highest precipitation amounts and intensities to mid latitudes.  Each major AR event is accompanied with a major Above Freezing Layer (AFL) and strong winds.   Using this new graphic, results show that the lower limit of Melting Snow Layer Thickness increases with Precipitation Intensity, but surprisingly the upper limit decreases with precipitation intensity. The upper limit also increases with distance from the coast.  Results are explained because an AR (warm moist conveyor belt) event typically delivers not only the highest precipitation rates, but also the biggest AFL heating aloft with strongest AR jets near the coast.   A sensitive SL interplay can result with a strong AR, as the snow levels plummet then skyrocket.  This study demonstrates SL forecasts are improved by solving for diabatic effects within the Melting Snow Layer, and this EWBZ method better simulates the sensitive response of SL to Precipitation Intensity than current operational guidance.