Can Numerical Models Even See Cold When It’s Coming?
One has to wonder if standard modeling is capable of predicting any major cold. Here are the three most glaring reversals in temperatures that the U.S.-based CFSv2 model had no idea were coming, and by failing to do so caught people off guard. But the background question is: Can models even see reversals to cold?
One has to wonder if standard modeling is capable of predicting any major cold. Here are the three most glaring reversals in temperatures that the U.S.-based CFSv2 model had no idea were coming, and by failing to do so caught people off guard. But the background question is: Can models even see reversals to cold?
The first example comes via what is happening now.
Look at the CFSv2 outlook for December back in November:
Month so far:
Look at its forecast for Nov. 2014 from October:
Look what happened:
But Asia truly takes the cake. This has to be one of the worst forecasts I have seen from a model. We used the cold that developed there to forecast the cold December here, a little-known old-timers method called the “bathtub slosh” — the idea that cold in the fall in Asia would “slosh” over to our side of the world in winter. This was shown as you will see below in our analog-based Weatherbell Pioneer Model in September. The three-month forecast from the numerical model the U.S. puts out, the CFSv2, looked like this:
You can’t get any more opposite.
Fittingly, global warming fanatics are pounding away at how warm it is in the Arctic, yet where people are living it’s brutal, for one, and two, these examples make one wonder: Even if a cooling trend were to develop for the globe, how would the models even see any of it if they can’t see such major cold when it’s in front of them?
We have a long, long way to go this season. But the fact is that so far this winter the analog-based Weatherbell.com Pioneer Model has done well.
At the very least it raises questions about what a numerical model is capable of doing.
I am happy we had this out…
…rather than this:
Particularly given the actual result so far:
One must ask: If it’s warm, but it’s going to get cold, can high-powered models see it over good old-fashioned forecast techniques and analog procedures that we use in our model?
At the very least, food for thought.
Joe Bastardi is chief forecaster at WeatherBELL Analytics, a meteorological consulting firm, and contributor to The Patriot Post on environmental issues.
