36 Towards an Improved Method of Forecasting Snow Level in British Columbia

Monday, 18 August 2014
Aviary Ballroom (Catamaran Resort Hotel)
Mindy Brugman, EC, Vancouver, BC, Canada; and C. Emond, T. Smith, B. Snyder, M. Loney, J. Goosen, A. Chen, B. Kim, P. Joe, A. M. Macdonald, and D. Simpson

High quality snow level forecasts are needed for weather predictions and alerts to avoid serious consequences from a missed snowfall event, especially in British Columbia which is noted for rugged mountains. A challenge was given to our Coastal and Mountain Meteorology National Lab (CMML) to work with Pacific Storm Prediction Centre (PSPC) to propose a better snow level forecasting algorithm for use across the complex terrain of BC, for operational grid scales and for periods relevant to our weather operations (2.5 to 25 km; 0-10 days with a focus on the first two). Forecasting precipitation in complex terrain has many challenges. In the cool season, accurately forecasting the level where rain changes to snow (“snow level”) is needed before rain or snowfall warnings can be issued. In this study, the snow level is defined as where the ground turns white. Historically, forecasters at the PSPC have relied on simple operational guidance based on the upper freezing level for predicting snow level. Experienced forecasters have relied upon the “wet bulb” for a better estimate of snow level.

New snow level algorithms were developed by CMML during 2013-14 in consultation with PSPC forecasters and Canadian Meteorological Centre (CMC). These algorithms are based on wet bulb temperatures with modifications to the vertical profiles in the boundary layer. Model vertical profiles of temperature and dew point were transformed to the actual terrain and Updateable Model Output Statistics (UMOS) data were smoothed upwards to bring lowest levels of the model into better agreement with physical reality based on surface observations and radiosonde data. Modifications included diabatic cooling during snowfall as an enhanced algorithm.

A major challenge in assessing the best forecast method was to obtain representative observations of the snow level. A snow level observational data set was compiled across a myriad of environments from the milder wet south coast of BC into the dry cold Alberta Rockies. All wet bulb freezing and radar derived data sets were validated with actual snowfall data from observations and web cameras. Radar bright band characteristics and Doppler fall velocities were found to correlate with observed snow levels based on web camera observations. Freezing levels were represented by actual temperature profiles or the top of the radar bright band where fall speeds accelerated. The dip in snow level as one proceeds into the mountains increased with precipitation intensity as best shown along the Sea to Sky Corridor, hence snow level evaluations were carried out locally in elevation and distance, as much as possible close to each validation location. Our observational data is strengthened by including freezing level, melting layer thickness and precipitation amounts for validation of each model. Testing is underway to evaluate data sets, and choose the best snow level algorithm for operations.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner