Forecasting Winter Weather Michael W. Cammarata Science Officer NWS Columbia, SC
Introduction • Forecasting winter weather is very challenging and complex • Small changes in temperature can have a significant impact (e.g. 33° F and rain 32° F freezing rain) • Small changes in liquid precipitation amounts can have a significant impact (10 to 1 ratio of snow to liquid) • Winter weather is relatively rare climatologically in the SC midlands
Factors we Consider • Climatology • Storm track • Agreement between models and run to run consistency. Model initializaion. • Soil/surface temperatures • Cold air
Climatology • Don’t forecast a climatologically unusual event unless the parameters contributing to the event are unusual and there is a high degree of confidence that they will occur as forecast
Uncertainty What conditions would lead to a certain forecast of snow meeting winter storm criteria? Entire sounding below freezing during event Cold air in place prior to event Cold soil temperature Categorical PoP Ample QPF High RH and UVV in snow growth region Good agreement among SREF members and other models Event conforms to climatologically favorable scenario MOS probabilities are high
Entire Sounding Below Freezing • Surfaced based warm layer • snow may melt as it falls into warm layer • snow may melt as it falls onto roads, ground etc. • Warm layer aloft • Snow may melt as it falls into warm layer aloft resulting in sleet and/or freezing rain
Cold air in place prior to event? If cold air is not in place, event becomes more uncertain Cold air can get held up by mountains or may not arrive as quickly as advertised by the models
Cold Soil Temperature Above freezing surface temperatures will result in snow melting as it falls on surface. Need heavy snowfall rates to produce accumulations. The warmer the ground/roads the heavier the precipitation must be in order to accumulate.
PoP/QPF • The various model parameters must indicate a high probability of precipitation and that the expected amount and intensity of the precipitation will be sufficient to exceed winter storm criteria
SREF Increased computer power has lead to increased use of ensemble forecasting in past few years Multiple runs of a model using different configurations and/or tweaking initial conditions then comparing forecasts gives an indication of uncertainty. The more the forecasts converge/diverge the greater the certainty/uncertainty
Climatologically favorable scenario Significant snow/ice storms are relatively rare in the midlands of SC and CSRA of GA Significant storms of the past have been associated with climatological patterns (storm track, upper level flow etc.) If a forecast event does not conform to climatologically favorable patters, uncertainty is greater
MOS event probabilities are high Model Output Statistics correlate past model forecasts with observed weather to come up with probabilities of precipitation and precipitation type. High MOS probabilities indicate greater certainty, however, since winter weather is relatively rare, the historical database upon which the statistical equations are derived is small.
If we don’t tell you, ask What is your level of confidence? Are all the ingredients in place? What could go wrong?