J36.4 Automated Products for Forecasting Arctic Blizzard Conditions

Wednesday, 10 January 2018: 9:15 AM
Room 19AB (ACC) (Austin, Texas)
William R. Burrows, EC, Edmonton, AB, Canada; and C. J. Mooney

Blizzards in the Arctic are high impact weather events in the fall, winter, and spring months. Blizzard conditions can last from a few hours to several days. The Meteorological Service of Canada (MSC) criteria for issuing a blizzard warning are sustained wind ≥ 40 kph and visibility ≤ ¼ SM in snow, blowing snow, or concurrent snow and blowing snow for at least 4 hours south of the treeline and 6 hours north of the treeline. Environment factors important for blizzard conditions are: temperature below freezing; strong wind for a few thousand feet above the ground; snowpack depth at least 1 cm; a source of snow, either precipitation or lifted from the ground locally or advected from upstream. Secondary factors are: the age of the snowpack, the temperature, and the degree of convective instability near the ground. More than half the occurrences of blizzard conditions in the Arctic are “clear sky” events caused by blowing snow. In the MSC National Lab in Edmonton we run 3 automated forecast guidance products in real time for forecasting a favourable environment for blizzard conditions, driven by MSC’s operational regional weather forecast model hourly output from 1-48 hr and its operational global weather forecast model output every 3 hours from 51-120 hr. Forecasts are shown on an internal MSC website accessible to operational forecasters across Canada. These products are: 1) an expert-system set of forecast rules (the Blizzard Potential); 2) a perfect-prog forecast of the probability of visibility < ½ SM, derived by analysis of many years of radiosonde profiles (Baggaley-Hanesiak method); 3) a model-output-statistics forecast derived by applying a powerful machine-learning algorithm (RandomForest) to a database of observations matched with a set of 43 physically-related predictors generated from NWP model output. In this presentation we will outline the modelling methods and show output examples and verification results for the 2016-7 winter.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner