The complications of winter forecasting

Image 1 of 4

You have likely been hearing about it for a week now, but our return to winter is finally here.  While social media has been hyping this storm since the middle of last week as the “superstorm”, reality shows us that it’s just another February snowstorm. 

While some will get nothing and others will get a lot, I am not here to discuss who will get what, so if you’re looking for a snowfall forecast, you’ve come to the wrong place.  I’m here to give you a little taste of why Fox 9 waits till about 36 hours ahead of time to share a forecast and why it can be complicated.

There are thousands of variables that go into a snowstorm, most of which take years of schooling and experience to understand.  But ultimately, these variables come together to develop a swath of snow across some part of the country.  In this case, it’s across parts of Minnesota.  We look to the forecasting computer models for the expected placement and intensity of these bands.  While these models help determine the placement of our forecast, they are just a guide because they often don’t agree with each other.

So in the pictures above, we have 4 different computer models from the same time period.  In theory, all 4 of these computer models should look pretty similar to each other because they are all forecasting for the same storm.  But they don’t.  They all have a storm pushing through the Upper Midwest because they CAN understand overall weather patterns really well.  The issue comes with the minutiae surrounding the storm which could make the storm curve a little left, or curve a little right, or produce a little more snow, or a little less.  But the truly difficult part can come when these forecasting models don’t line up at all… when one model has a specific location getting 1” of snow and the very next model has that same location getting 20”.  How do you forecast for that spot?  Do you believe one model over the other and lean that direction?  Do you go straight in the middle and go with 10”?  Do you believe none of them and go an entirely different direction?  This is the truly tough part of trying to predict the future… even our supercomputers don’t always know.

So take our current storm for example.  All 4 of these models look a bit different.  All of them place the “bullseye” of snow totals across parts of southern Minnesota.  But a couple are a little wider with that heavy snow band than the others.  Model number 1 Even takes the heavy stuff through much of the metro.  So which one is right?  Well, that’s the even tougher part, one could be right, none of them may be right, or it could be some combination of them.  Welcome to forecasting snow amounts!  And this is just ONE model run.  There is another one 6 hours later.  Most of the time, models shift a little bit in either direction or in their amounts.  So now what do you believe? See, this predicting the future stuff is harder than you thought isn’t it?

THIS is the exact reason why you shouldn't believe ANYTHING on social media hyping a potential storm that's days away, or even weeks away.  If the computer models can't even agree on a location or intensity 36 hours away, what makes you think they will be any better when a storm is a week away?  Exactly... they are just click bait.  So now that you are informed, stop biting and these tricksters will stop posting.