That Came Out of Nowhere

This one surprised us and caught us off guard.  "Surprised us", "off guard", normally phrases that make me cringe like the sound of nails on a chalk board as those phrases are often misused during big weather events in the media.  Some would call it sensationalism.  What happened last night was simply not the case.  The heavy snow totals of 8-12" truly surprised us.

Across the local area, snow totals were as low as 1" in spots of north Iowa to as much as 12" in rural Mower county, then again trailing back to 1" towards Winona.  Here's an interactive look at snow totals from last night.  Below shows a graphical look at those snow amounts, and a very steep gradient between low and high totals.

Going into Tuesday afternoon's forecast there were signs pointing to higher snow amounts than previous forecasts had indicated, so Tuesday evening the range was opened to 2-5" across the area.  This was our snow forecast on Tuesday afternoon.

Tuesday morning the expected range was 1-4".  The night before it was 1-3".  So the going trend was a little stronger over time.

In all, a 2-6" range would have been pretty good if there wasn't a narrow band, only about 20-30 miles wide at it's max, of enhanced snow that lined itself up with the southern tier of counties in Minnesota for several hours.  That band produced snow intensities of 1 to 2" per hour, in some places for six hours!  Simply put, our computer models did not see that enhanced snow band.

Below is an animated radar loop across southern Minnesota and northern Iowa.  Note the darkest purples and whites.  That's the band in question.  Towards the end of the loop you can really see how it parks itself right across Mower, Freeborn counties.  (Courtesy: National Weather Service - La Crosse, WI)

radar loop


Okay, here comes the technical part... trying to put it as simply as possible.

So here's what I was looking at heading into Tuesday's forecast.  We had a shortwave (upper level disturbance) moving in from South Dakota, arriving towards sunset.  The large scale dynamics with this disturbance didn't look to be overly strong, but the one thing that did stick out was frontogenesis, which is basically how upward motion is produced because of a temperature change across a front.  A stationary front was in place across central Iowa Tuesday afternoon, and the frontogenetic forcing was trending stronger.  As was isentropic lift.  Isentropic lift is basically explains how a parcel of air will naturally follow the plane at it's ambient temperature.  If that plane is tilted upwards, that pocket of air will naturally try to go up, producing lift. 

Moisture is also a big component.  You can have as much lift as you want, but if there's no moisture in the first place, you won't see any snow.  Surprisingly a decent fetch of moisture was coming in from Nebraska, where it had been pooling, south of that stationary front in the warmer airmass. 

And of course, temperatures were sufficiently cold in the atmosphere above the surface to support snow.  Lift + Moisture will produce precipitation.

So here's the raw output of snow from the models I was looking at Tuesday evening.  The middle scale (NAM) and large scale (GFS) computer models were pointing at snow amounts of 2-5" using different snow making processes.  All falling from about 6pm until 3am Wednesday morning.  If you do the math that's nine hours.  That amount of snow falling during that timeframe isn't overly intense.


A check of the small scale models (WSI, RAP) that have very high resolution were pointing at amounts in line with the other lower resolution models.  The RAP from 2pm for expected snow, suggesting anywhere from 1-4"...

The WSI run at 6pm with amounts of 2-5" depicted...

At 3pm, the time I'm hitting the forecast hard, there was a band of snow nudging closer to the area.  At the surface, snow (pink **) was falling in western Minnesota and back into SD.  There were a few pockets of heavier snow in there reducing visibility (pink numbers, depicted in miles) as low as 3/4 mile in Aberdeen, SD.  Considering the wind, that's a moderate to heavy pocket of snow.  Still it wasn't a widespread band of it.

dsm METAR plot

So far nothing earth shattering to suggest a change of 6-12" of snow.

The more course computer models, the NAM/GFS only run every six hours.  New model data for them wouldn't come in until 9PM.  Those new models still didn't hint at 6"+ with very little change, shown as the 00z options in the legend.

Only one model picked up on it, a little earlier, towards 7pm.  That was the higher resolution RAP model which was now suggesting some spots could get as much as 7 or 8".

With the radar growing stronger in intensity between 6-10pm, the snow reports coming in that timeframe and latest data coming in, I did opt to open the range up a bit.  The high end moved to 6" with isolated spots higher.  Obviously not high enough.  Did I even think of 10 or 12"?  At that point, absolutely not.

After 10 pm, additional ground truth began rolling in that this band of snow was intense in spots, so intense, it made me question the validity of some pictures that were coming in.  Because believe it or not, some people like to pull fast ones on us weather guys!  It was clear, we were being socked by some big snow.

Through reverse analyzing the data at hand, we can narrow it down to why the higher amounts came to fruition.  That narrow pocket of snow was a very focused area of that isentropic lift and/or frontogenetic lift, producing far more upward motion than anticipated.  Models were slow to pick up on it or never did pick up on it.  And given that the front to our south was stationary, there was no mechanism to move that line of heavy snow out.  It just sat there.

It's tough for meteorologists.  We take a great pride in our work to get it right.  But sometimes, this is the reality we get and we are reminded that meteorology is an in-exact science and it NEVER will be in our lifetime, maybe ever!  Well, not unless someone gets in the time machine and jumps ahead to get an almanac and bring it back to me! :)

It might not be perfect, but remember where things were 20 years ago in weather prediction.  Now think of 50 years ago.  Back then, you were lucky if you got a day or two heads up on impending weather.  In present day, we can give you a pretty good sense of what's to come in the next seven days, the upcoming three pretty darn accurately and severe weather lead time has grown over those years.

While I'm not writing this as an excuse, hopefully you can see the method behind the madness in weather forecasting.  If that narrow, heavy band never materialized, the 2-6" forecast would have been pretty good approximation for this storm.  The edges, where the lower amounts materialized prove it.

This one surprised us, as much as it pains me to say it.

Live through it, Learn from it, and move on to the next one.

Storm Tracker 6 Chief Meteorologist
Chris Kuball