Sunday, December 28, 2014

What is the Winter going to be like? You won't like the answer.

I always get a lot of questions this time of year from people I know about what kind of winter we are going to have. I hate to shatter their dreams of me knowing exactly how the next 3 months will play out, but the simple answer is I don't know and more importantly nobody really knows.

This doesn't stop various agencies putting out their seasonal predictions, here are examples of some of the forecasts for the upcoming winter (or at least December, January, and February) that I have seen:

Environment Canada:





The 2014-2015 AccuWeather Winter Forecast for Canada




The Weather Network:

 

The Weather Underground:




What I really don't like is that most of these agencies put out these forecasts without any mention of how much confidence they have in them or how well they have done in the past.  The exception to this is Environment Canada who also publish various measures of the confidence in their forecasts and how well their predictions have done in the past (see them in the images above).

Although a lot of media will just chose to publish the simple Environment Canada map showing the forecasted areas of above or below average temperature and precipitation, the other information is there for people who want to look.

The best place to see how the various models differ in their predictions is to look at the NMME forecasts.  This website compiles the forecasts from all the top models in the world and show their predictions for the following months.  Here are their predictions for the upcoming winter:

Temperature:


Precipitation:



You can see that for temperature, about half the models have colder than average temperatures for our area while the other half show warmer than average temperatures and that there is no consistent pattern for precipitation.  So you can take your pick of the forecasts based on what kind of winter you are hoping for and don't worry I won't judge you, unless of course you pick the GFDL, then we can no longer be friends (just kidding).

But more telling are the figures showing their predictions masked so that areas where the models have not shown skill in the past are greyed out:

Temperature;



Precipitation:



You can see that there are not many areas (especially for precipitation) where all the best models of the world have been able to accurately forecast 3 months into the future.

Ideally, what I would like to see is that anybody who publishes a long range forecast be made to also give an indication of how well their predictions have done in the past.  However, I think that most agencies wouldn't want to admit how much they don't know, but that is the state of seasonal forecasts worldwide at the moment.

In my opinion showing these long range forecasts alone leads the public to distrust all weather forecasts (short, medium, and long term) when they hear these various agencies make their predictions and the resulting season is totally different in the end.

So if you want my opinion of future weather I'll be happy to go out 10 days or so (and even that is sometimes pushing it) but after that your guess it truly as good as mine.



Frank

No comments: