The tornado that hit Mayfield, Kentucky the night of December 10 had already compiled a history before entering the state of Kentucky. The authorities in Mayfield should have known that a tornado strike was very likely, but could not be certain that one would strike.
Mesoscale Convective Discussions (MCDs) are issued by the Storm Prediction Center for situations that may require a watch to be issued, extended or advised against. In a relatively new development, they are also issued for relatively local events within areas under weather watches and warnings. MCD 1995 predicted that Mayfield was in relatively high danger compared to other areas in the watch that were not warned.
SPC Mesoscale Discussion #1995 (valid 8:48 pm CST until 9:45 pm CST)
DISCUSSION...The long-track supercell in northwest Tennessee has shown 70+ knot Vrot nearly continuously for the past 90 minutes with a nearly continuous and deep TDS for over 80 miles. More storms have developed in the vicinity of this storm over the past 30 minutes which could disrupt its internal structure and bring an end to the steady state structure. However, the downstream environment is still very favorable with upper 60 dewpoints and backed surface winds and STP around 5 to 6. Therefore, even if the low-level circulation is disrupted, additional organization/cycling is likely and the intense tornado threat is expected to continue as this supercell continues northeast.
Some terminology:
Vrot means Rotation Velocity. This is the mean of the maximum outgoing and incoming winds associated with a tornadic supercell. 70+ is consistent with an EF3 tornado, verging into EF4 territory.
TDS: Tornadic Debris Signature. A TDS shows up on radar as a particularly strong signal, caused by the debris that is whirling around the tornado’s base. A continuous, deep signature means that the tornado has spent most of this time on the ground, tearing thing up in its path.
Downstream: in the direction the system is heading
STP: Significant Tornado Parameter: one indicator of how favorable conditions are that a tornadic supercell can develop. An STP above one indicated a favorable environment. (In my semiqualified opinion, an STP of 6 may indicate conditions are more favorable for a supercell to tear itself apart due to high values of atmosphere buoyancy, but such self-destruction is probably more of a concern during the warm season.)
Anecdotally, tornado chasers had converged on far western Kentucky, and actually proved useful by clearing the roads to Mayfield so first responders could get in.
A Tornado Warning was issued at 9:05 pm, one minute after the tornado had been detected near Cayce and near Clinton, about 10 -15 miles from Mayfield. Had the tornado continued on that track at 55 mph, it would have hit Mayfield 10-17 minutes after the warning was issued.
But did the factory really have only ten minutes’ notice of a tornado? The tornado had followed a path through Arkansas, Missouri and the northwest corner of Tennessee. The Mesoscale Discussion above strongly implied that Mayfield would be hit around 9:15 pm, plus or minus a few minutes, making for a 30-minute warning. On the other hand, a lady in the factory said that her phone’s emergency alarm gave little warning before the power went out and the storm hit.
In retrospect, it appears that the people at the factory should have been encouraged to seek shelter, either in a basement or in a fortified room within the facility. The problem is that short of a warning, a tornado strike could not be guaranteed. Even though all of Graves County was warned, most of Graves County was not hit — sadly the part that was hit was the most built-up and heavily populated part of the county.
Counterexample
MCD #1997 issued at 9:36 on December 10 placed Terre Haute, Indiana in the area of high likelihood of tornadoes. Around 10:00 pm CST as storms crossed I-57, storm motion in far eastern Illinois and Indiana turned out to be toward the NE, rather than the ENE. Of the two suspected tornadoes aimed at Terre Haute, one died with its severe thunderstorm surviving and the other backed northeastward, crossing US 41 north of the advised area and about 40 miles from Terre Haute. The approximate location of the couplet that triggered the tornado warning on the Indiana side of the state line is shown in green.
On the other hand, the red dot shows where a house 3 miles NW of Mattoon, IL was destroyed, trapping people inside. This would have been a close call and well within the danger zone. The Mattoon strike took place 30 minutes after the warning.
Upstream developments
Multiple storm reports had been received by authorities in northeast Arkansas starting at 7 pm Eastern. Tornado warnings would have been issued in Arkansas starting around 7 pm and for at least 90 minutes afterwards. These warnings should have been seen as the start of a multiple-tornado event heading for western Kentucky.
What should have been done?
When special discussions like MCDs 1995 and 1997 come out, they should be considered more urgent than watches, but not quite as urgent as warnings. Fortunately, the chance of actually getting hit within such areas of concern is far lower than it is within a warning proper, but taking evasive action on the basis of a discussion like this is guaranteed to mean a lot of false warnings, which means people are less likely to take action when they hear the warning.
For long-track tornadoes, it might appear that the local forecast offices could be a little more confident about their forecasting abilities, but this could be misleading. It turns out that they had warned the two counties ahead of Graves County (Fulton and Hickman Counties), and that the warning for Graves County was issued when the storm was miles from the Graves County line.
Saddest of all, it looks like significant time passed between the issuance of the warning by NWS Paducah and the receipt of the warning by people in the path of the storm.
Perhaps the strongest evidence of “things left undone should have been done” would be the recent history of the outbreak. Convective forecasts occasionally bust. Tornado watches are sometimes issued in cases where the tornadoes are few and weak. However, the local potential for severe weather was demonstrated by a rotating storm 11 miles south of Mayfield at 6:00 pm. By 8:00 pm, northeast Arkansas learned the hard way that the watches were justified, that damaging tornadoes were taking place.
With all this said, it was known all day that far western Kentucky had a high chance of experiencing an outbreak of tornadoes. When the tornadoes actually fired up in Arkansas, it should have been understood that an outbreak had actually begun. Schools close given such risks. Had the shift at the factory not started, it should have been cancelled. Had it started, the decision becomes more difficult: would people have been safer at home than at the factory? They would have been safer in the factory than on the road if a tornado hit.
Sources
James Hude, posting at www.ustornadoes.com/...: “Understanding Advanced Tornadic Radar Signatures