Storm post-mortem 

Posted: 12:48 pm Sunday, December 10th, 2017

By Kirk Mellish

The biggest snow in parts of the Metro area in at least 25 years.

Clearly my forecast snow amounts were a big bust fail in the West and some North suburbs, whereas it WAS correct elsewhere… including the rain-snow line, and where there would be mostly rain and little or no snow. I realize that is no comfort to those impacted by how wrong I was in the W/NW suburbs where my 1-3 inches became a ton more.

Here’s a quote from my blog of last Wednesday morning:

Screen Shot 2017-12-10 at 10.20.50 AM


I hope you checked back often. I updated my Friday forecast at 11:30am to 3-6 inches and accumulation on roads. Still not good enough but still before roads got dicey.

But another failure on my part was one of communication. I should have pointed out all along that if and whenever THE RATE of snowfall becomes fast it can overcome warm ground and warm air temperatures and accumulate on roads despite those roads being above 32F. This is something I and all forecasters are aware of, sometimes we are so (duh) aware of things that we (I did here) forget that the general public and even other broadcasters probably are not aware of these obvious things. (obvious to experienced forecasters that is). The snow fall RATE was expected to be low enough to keep most roads wet or just slushy but that was one of the keys that went wrong.

The RATE of snowfall was an inch an hour in a few counties NW of 285, whereas rates of only a tenth to quarter inch/hour was what was expected.

People get hung up on 32 degrees, but heavy snow has been observed at surface temps as high as about 39F. Calculations suggest an upper limit of 46-51. The math here.

Often I forget or leave some important CAVEATS out because I only get 20-90 seconds on the radio to explain a forecast for 24 hours and/or 5 days covering a Metro area larger than some states. And because I don’t want people to fall asleep reading a 20 page forecast blog I am limited.

Another MAJOR thing that I forget to point out is that we can find out what CURRENT pavement temperatures are with GDOT sensors, but that tells us nothing about FUTURE pavement temps. Meteorologists do NOT have models that predict future pavement temps! We can project how much snow will fall from clouds but not how much will melt or stick at a given point, that’s a stab in the dark.

There is NO universal agreement on what constitutes “Metro Atlanta” but it’s not just where you or I live or work and Metro doesn’t just mean Atlanta city limits. I forecast for most but not all of these counties in yellow, my FIRST forecast (Thursday and early Friday Morning) was garbage in terms of amounts West and North of the perimeter (left of blue dots) but fine in the rest of the metro:

Screen Shot 2017-12-10 at 6.18.00 PM



Suffice to say weather is in constant flux its not static and its infinitely complex and complicated, it can be hard to translate that: physics, kinematics, thermodynamics, synoptics, dynamics and 40 years of experience into something short and sweet.

This was a remarkable and very rare storm. I’ve been here 30 years and the snow in my own backyard is the most I’ve seen (above the Blizzard of March 1993) at 9 inches. Although 1993 was a much worse storm in many ways for the whole Southeastern region of the U.S.


TRULY an unusual abnormal and REMARKABLE system producing snow in northern Mexico and Texas BEFORE Boston, and with greater snow amounts in the deep south than Minneapolis Chicago and many other northern cities have had so far through Friday!

Screen Shot 2017-12-11 at 6.27.51 AM



There were flurries in New Orleans, a dusting on the ground in Mobile, AL, snow in the air at Destin and Panama City with snow falling 30 miles into the Gulf of Mexico!!

The more abnormal a weather system the more unpredictable it is for obvious reasons. They fall outside the distribution curve and thus become outliers. These are terms from statistical analysis for anyone not familiar. At the extreme they become a “Black Swan Event”.

Screen Shot 2017-12-10 at 10.26.27 AM

The fact that no forecaster locally, regionally, or nationally nailed the forecast belies that fact. It is NOT uncommon for different forecasters to have different forecasts, just like why you get second opinions from doctors. Did any app or automated web site get it right in advance? NOPE.

People think there is just ONE forecast delivered by different forecasters or apps etc. But that is NOT true. My forecast OFTEN differs from others and apps and automated web sites.

This was the type of winter storm that always gives Georgia it’s best chance at significant snow or ice, they are called a “Miller-A”. As opposed to a “Miller B” type system. The rain-snow/ice line varies from system to system and no two are ever exactly alike.

Screen Shot 2017-12-10 at 10.31.15 AM

From NCEP Here are the basic weather features on Friday December 8th at 7am:

sfcplot_sm_20171208Screen Shot 2017-12-10 at 10.36.06 AM

You can see the “long-wave trough” in the 500mb jet stream from Hudson Bay Canada to Mexico with a potent “short-wave” vorticity max in Texas responsible for cyclogenesis in the Gulf of Mexico.

It’s worth noting that some models did have “scary” amounts of snow projected (which I mentioned in one of the blogs or tweets prior to storm) but that is NOT uncommon which is why forecasters don’t just say, “welp that’s the forecast then”. Otherwise every time the forecast would be ‘2-10 inches, that’s what models show’.

It’s should also be noted those heavy snow outputs did NOT get the LOCATION right, so even if I believed those models verbatim the forecast would have been off the mark.

Just like in hurricane track forecasts there is “a range” of possibilities for snow (or rain/t-storm) amounts and locations in every forecast and a normal and expected margin of error in all forecasts, including temp and wind outlooks.

It’s not unusual for the rain-snow line for example to be AT LEAST least 20 miles in a different direction from forecast. (In this case though THAT part of the forecast was right on target).

Reducing the normal margin of error will require more data and better physics in the computer algorithms and higher resolution models.

Atlanta is quite often right on the line between rain or snow, it’s always a battle between these guys around here:


Here’s just ONE example of a snow forecasting formula:

Screen Shot 2017-12-10 at 1.11.07 PM

For those who like to learn and really want to understand…

Here is some of the meteorology science of what happened:

That strong upper-level disturbance swinging east/NE from Texas over a strong baroclinic zone (tight temperature gradient) resulting in Gulf of Mexico cyclogenesis (storm formation as a wave of low pressure on the stalled cold front) as seen above.

Typically the axis of the heaviest snow is found about 90 miles North of the track of the 850mb (about 5,000 feet) low center. In this case the 850mb level had a low move just south of Atlanta (black L). In theory the textbook says this would have put the heaviest snow well North of us! Studies (Younkin) show the heaviest snow axis close to the 850mb low pressure center happens only 30% of the time.

Screen Shot 2017-12-10 at 11.18.37 AM

Using an 850mb temp technique would have had the forecast heaviest snow band too far south from I-20 to south Cobb and South Hall. Again, showing how this particular system went rogue.

Another method (not shown) uses 700mb (about 10,000) feet vertical velocities to project snow amounts. In this case most models showed the strongest vertical motion velocities (UVV) well north or well south of where it actually occurred!

The R/S line was quasi-stationary near I-85 wiggling a little north and south during the day, here it is depicted at 7pm Friday:

Screen Shot 2017-12-10 at 11.41.36 AM

On of the primary reasons for the narrow heavy snow band was an unexpectedly strong persistent and nearly stationary feature at around 10,000 feet as discussed by Papin was 700mb frontogenesis (creation of a front aloft in a tight baroclinic temperature gradient zone) which causes strong ascent or vertical motions able to lift moisture into the dendritic growth zone aloft (DGZ) creating the heavy snow band. This ‘frontogenic forcing’ (purple lines first map, colors in 2nd map) depicted here at 2pm and 7pm Friday:

Screen Shot 2017-12-10 at 12.23.23 PM

Screen Shot 2017-12-10 at 12.27.12 PM

Pettersson Frontogenesus equation:

Screen Shot 2017-12-11 at 8.58.04 AM

This process was aided by strong warm air advection (WAA) at both 850mb and 700mb as winds at those levels transported warm moist air into the cold air farther north (not shown).

Our area was also in a favorable right rear (RRQ) quadrant of the jet stream wind jet streak (or jet max) supporting upper level divergence and low level convergence of air streams, a region supportive of vigorous upward air motion:

Screen Shot 2017-12-10 at 5.01.19 PM

As described by Haby here:

Screen Shot 2017-12-10 at 5.13.22 PMScreen Shot 2017-12-10 at 5.10.49 PM

Screen Shot 2017-12-10 at 5.17.38 PM

Interestingly the mesoscale forecasters at SPC expected mostly rain south of Rome even as a band of heavy snow rates was approaching:

Screen Shot 2017-12-10 at 12.14.39 PM

Screen Shot 2017-12-10 at 12.32.17 PM

In most areas temps cooled from around 39 early morning to 32 or so the rest of the day and most of the night.

But early that evening they saw this:

Screen Shot 2017-12-10 at 12.33.34 PM

Screen Shot 2017-12-10 at 12.34.29 PM

It did not finally move all the way South and East through the area until the wee hours of Saturday morning as the wave of low pressure moved into the Atlantic and the 500mb vorticity max (short-wave) shown in the colors and with Xs in charts below at the base of the jet stream (“baroclinic leaf” signature) trough swung across Georgia:

Screen Shot 2017-12-10 at 11.51.04 AM

Screen Shot 2017-12-10 at 11.48.02 AM


The system like most Miller As rode NE up the coast as a Nor’easter:


Screen Shot 2017-12-11 at 6.04.09 AM

If it was simple and easy we’d always be right.

Forecasters must work with just a minuscule sampling of the atmosphere and then extrapolate from that woefully inadequate data to simulate the future state of the atmosphere. We’ve come along way since the first attempts at Numerical Weather Prediction but mother nature will always have the upper hand over mankind, for we are mere mortals.


Screen Shot 2017-12-10 at 12.42.52 PM

I am reminded of what Teddy Roosevelt said about the “Man in the arena”, but I am also reminded of the old Breton Prayer:Screen Shot 2017-12-10 at 12.45.43 PM

Thanks for reading, thanks for listening, thanks for following, thanks for understanding.

Hope you enjoyed the beauty of the snow if you got some of it, and Merry Christmas, everyone.

Follow me on Twitter @MellishMeterWSB.