Winter Will Take a Holiday, but “Ensembles” Show It’s Not Done.January 4th, 2013 at 11:52 am by Don Paul under 4 Warn Weather
You often hear meteorologists talk about computer models, blog newcomers. But over the last decade a new set of important tools are being used by professionals who have been keeping up with this aspect of the science. A few words about “Ensembles”: because we cannot know the precise state of the atmosphere at any given moment (far too many variables, far too much unavailable or missing information) supercomputers with enormous crunch power run models with 20 or more different-but-reasonable-variations on initial conditions at the start of model runs. That is, an NWS model may have over 20 different runs, each a member of an ensemble. When these 20 ensemble members show only relatively small differences between one another, confidence in the use of that model and its ensembles increases. But if the spread between those ensemble members is large, and the ensemble shows a huge range of outcomes, then confidence shrinks. The NWS global model, called the GFS (Global Forecast System) has a counterpart from Environment Canada, and from the European Centre for Medium Range Weather Forecasting, in England. The latter, the European model and ensembles, is generally thought of as the most reliable global model. But NO model or ensemble can be counted on to automatically be the best on a daily basis (some models run 4x per 24 hours), so we need to look at all of them. One might assume that the Mean of the many ensemble members is most likely to be the most accurate, but it doesn’t always work out that way. Sometimes members close to being “outliers” turn out to do best, but it’s very difficult to know when that’s going to be the case. The best we can do is to combine basic meteorology (looking at surface and upper air observations, satellite imagery, buoy data, doppler radar products, etc) and combine that current reality with the performance of models and ensembles in the near term. If a model is already way off the mark early in its forecast, the odds favor it being even more way off further out in time. Ensemble forecasting is the wave of the present and the future. Meteorologists who fail to jump in to this new and complex statistical set of tools do so at their own peril. The NWS and academia have led the way in putting these tools into everyday use, and those of us in the private sector have caught on as well. Ensembles, particularly those from the European model, are what made the forecasting of Superstorm Sandy so incredibly accurate before the disaster unfolded.
As for what those ensembles are showing now…our snowpack will take a significant hit during the week of January 7th, with airmasses of Pacific origin replacing the arctic air. Temperatures will likely move well above average by midweek and late week, and to make matters worse for ski resorts and snow mobile trails, some rain is likely to hasten the thaw around Thursday and possibly again by Saturday (European ensemble). However, it had been looking until yesterday (the 3rd) as if that mild trend would persist most of the time over the next 16 days. Today’s ensembles are now showing seasonably cold air (not brutally cold) which had been slated to stay over the northern plains and Canadian prairies will be able to make it back to the eastern Great Lakes by around the 17th-19th…possibly a bit sooner, future runs will determine that. Right now there are no signs of truly harsh, bitter cold getting to us, but the temperature scheme I’m seeing in these ensembles will be returning closer to average or a bit below average and, in any case, it will be chilly enough for rain events to be replaced by snow events.