NOTE: The original forecast map from NWS Louisville has been replaced, therefore all discussion in this posting was for the original map and not the one currently posted. See the 4:00 pm update at the bottom for more details.
The NWS office in Louisville originally backed off on snow amounts early this morning (2″ for Louisville and 1″ for Lexington) but have since drastically raised amounts. I think this is a very optimistic forecast. Recent trends in the NAM hint that the smart bet would be to take the “under” with regards to this forecast.
Compare the Louisville NWS forecast (6-12″ north of the Ohio River; 3-8″ from E-Town to Louisville) to that of AccuWeather (mix to rain) and even the HPC winter warning page, which predicts only a 10% chance of >8″ of snow and only a 40-70% chance for more than >4″. Why is the Louisville NWS office so much higher than the others? What they have done is create a forecast that predicts the “optimal” snowfall forecast rather than the “probable” snowfall forecast. Here’s why…
The Louisville winter storm headline at the top of their webpage (seen here) states
Assuming a changeover does not occur too early, a total of 6 to 12 inches of snow is possible over parts of south central Indiana, with 4 to 8 inches possible just south of the Ohio River including Louisville. Up to 3 to 6 inches is possible in the Frankfort and Lexington areas if the changeover to freezing rain this evening does not occur too quickly.
Their forecast discussion, as seen here, states that they are following the NAM model since it does better with precipitation in an arctic air mass. While this is a true statement, they neglect to mention that the NAM model often has inflated QPF values which overestimates precipitation. All aspects of relative model performance must be considered, not just the aspects that can lead to high snowfall totals. They also mention the following…
NAM/GFS BUFR SOUNDINGS INDICATE HEAVY SNOW OVER THE NWRN CWA…WITH WARM AIR ALOFT PUSHING NORTHWARD THROUGH 12Z…CHANGING SNOW TO A SNOW/SLEET/FREEZING RAIN MIX TOWARD 12Z TUESDAY. WITH THE POTENTIAL FOR THUNDER SNOW…COULD BE QUICK 2-4 INCH SNOW ACCUMULATIONS (MAYBE MORE) OVER OUR FAR NRN INDIANA COUNTIES…WITH LESSER AMOUNTS FURTHER SOUTH.
They are basing their snowfall forecast on three factors, which if all coincide, will produce snowfall amounts similar to those in their forecast. The three factors are that 1) the NAM (which has higher QPF) is correct, 2) changeover to sleet/freezing rain/rain will be delayed, and 3) thundersnow will produce very high snowfall rates.
But what does the NAM actually say? I spent some time this afternoon looking through the four more recent model runs (12Z, 06Z, 00Z, 18Z) and comparing the model trends with the timing of the storm, 500 mb flow, and location of the 1000-500 mb thickness lines. I mentioned yesterday that the NW trend would be less for this storm than for recent storms due to the positive tilt of the storm, which will keep the surface low from rapidly intensifying and increasing warm air advection. However, that doesn’t mean the NW trend will not occur.
Here is a comparison of the 42 hr forecast from Sunday’s 18z model run compared to the 24 hr forecast from Monday’s 12z model run. Note that for the 12z run, the 500 mb vort max is stronger, located further west, and shows a sharper kink in the height field. This implies the NAM is predicting that this storm will be more intense at the mid-levels, which could in turn means more warm air advection at low levels, which could in turn intensify the surface low and lead to a faster changeover to rain in northern KY. If you look at the surface chart, you can clearly see a closed off 1012 mb isotherm on the 12z run, while the previous day’s 18z only has a 1016 mb low. You can also see that the location of the 540 dm thickness line is correspondingly further north into southern IN at 12z.
All of this implies that the “optimal” forecast, as currently represented by the Louisville NWS, most likely will not verify due to the increasing likelihood of a faster changeover to non-snow, especially for areas south of the Ohio river. Ironically, the recent changes in the NAM should lead to a better chance of verifying the 8-12″ snow amounts in southern Indiana, since a more intense system will mean greater isentropic lift north of the warm front boundary and an increased chance of convective snowfall. However, the gradient of snow amounts should be dramatic such that areas between Louisville and E-town may only get 2-3″ before the changeover.
Since the original map that was the focal point of this entire posting is no longer on the website, this is just a reminder that the original forecast was for 6-12″ north of the Ohio River (now reduced to 4-8″) and 3-8″ from E-Town to Louisville (now reduced to 2-6″). Since the warm-frontal snow band appears to be drifting to the north of Lexington, I can see accumulations of maybe only an inch or two in the Bluegrass before the changeover occurs. The NWS site is currently “locked up” so I can’t really look at current 5:00 pm observations.
7:00 am Tuesday update: Precipitation has changed to all rain in Louisville and Lexington as temperatures have nosed into the lower 30s. The changeover from snow to freezing rain occurred before midnight, which limited snowfall totals to 2-4 inches across northern KY and southern IN. Freezing rain accumulations have been significant in some areas.
In review of the forecast performance of this storm, only one of the three factors that were necessary to produce the 6-12″ snowfall that was predicted in northern KY and southern IN actually verified. The NAM was indeed the correct model to follow. Unfortunately, the changeover from snow to freezing rain occurred at 10 pm CST (instead of 12Z Tuesday) and convective snowfall did not occur at all in southern IN. In fact, the southern IN counties only received 1-2 inches of snow before changing over to light freezing rain. As I predicted Monday afternoon, the NAM trend was that warm air advection at low levels would lead to a faster changeover to freezing rain and that snowfall amounts would be more modest than originally predicted.