Click on images for full-size versions.
I am one of the forecasters at the NWS Forecast Office in Norman Oklahoma. I was responsible for the afternoon public forecasts on December 8 and December 12. References to forecast challenges are mostly based on my own observations of the weather situation on those two days. The map presented above was created by Doug Speheger, another forecaster at this office.
The main concern on the 8th was the long-awaited arctic cold front that was predicted to plunge through the central and southern Plains on the 10th and 11th. The Medium-Range Forecast model (MRF) had suggested the possibility of such a front about 12 days ahead of time (including some extrapolation of the 10-day forecast). By the 8th, however, the MRF and several other models were showing only a glancing blow of arctic air. This is a common problem with the models and I elected to retain the major cool-down that was already in our public forecasts.
We also noted that the medium-range models showed a wave in the jet stream (a "short-wave" trough) approaching our forecast area on Tuesday, the 12th. If the arctic air behaved as we forecast, as opposed to how the models handled it, the precipitation associated with that wave would be either ice or snow, rather than relatively harmless rain. Most of our forecasts from Friday the 8th and over the weekend mentioned a chance of snow or freezing drizzle on the 12th and 13th.
As we now know, the arctic front did, in fact, sweep well past our forecast area - even farther than I had counted on (it even went through Brownsville TX!). The cold air was so well-entrenched that it would definitely not be eroded or otherwise removed from our area before the next weather disturbance began to influence the area. As the short-wave trough approached from the west, winds in the layer of air from 5 to 10 thousand feet, just above the arctic air mass, would turn southeasterly. Since this flow of air was from the Gulf of Mexico, ahead of the surface cold front, we expected the warm and humid Gulf air to ride up and over the arctic air. This would cause cloud cover and probably some precipitation. With the deep cold air, it appeared that snow was most likely over most of Oklahoma and western north Texas - and perhaps some sleet in the far south. Freezing drizzle or freezing rain would also be a possibility in the extreme south.
The weather elements came together generally as anticipated on the morning of the 12th. Due to a computer software upgrade at the Norman office, we didn't have much model data to go by, so we relied rather heavily on observed conditions. This was probably a good thing, since we later found out that most of the models did not handle critical changes in the vertical temperature profile very well during the day on the 12th.
Through most of the morning, doughnut-shaped low-intensity echoes prevailed on area radars. This suggested that snow was falling aloft, but sublimating (the technical term for evaporation directly from solid (snow) to gas (water vapor)) before reaching the ground. However, by noon, a few reports of snow and sleet were coming in from central Oklahoma. "Sleet?" The 12 UTC (6 a.m. CST) upper-air sounding showed a snow profile (temperatures well below freezing at all levels), as did the previous run of the short-range forecast models.
Several area forecast offices, including Norman, did a special 18 UTC (noon CST) upper-air sounding, which was particularly enlightening. The new sounding showed that significant warming had occurred in the layer between 5 and 10 thousand feet above the ground. The temperature at that level was almost up to freezing over Norman! As a result, we shifted our preliminary estimates of precipitation types and amounts about 75 miles farther north.
In the early afternoon, snow, sleet, and freezing rain began to fall in Norman. The various reports we received from other areas generally agreed with our revised assessment of the vertical temperature and moisture profiles. By this time we had already determined that there would be a Winter Storm Warning for part of our forecast area, but the details had yet to be worked out.
Several surrounding offices called to coordinate with our warning/advisory plans and we determined exactly which counties would be in what type of warning by about 3 p.m. From then until the end of the shift, we assembled the various public forecasts and advisories/warnings. A Winter Storm Warning was issued for most of our forecast area, with emphasis on freezing rain in the far south (Marietta to Durant), sleet over the rest of the south, and mainly snow forecast elsewhere. The heaviest snow band forecast was about 4 to 8 inches, and was expected to be roughly from Mangum to Watonga to Perry.
The main complicating factor to the snow forecast was to determine how much precipitation would fall as sleet or freezing rain. The liquid water content of 2 inches of sleet or one inch of freezing rain is roughly equivalent to between 10 and 12 inches of snow. So, if a site received an inch and a half inch of sleet and then an inch and a half of snow, the total observed accumulation at that site would be three inches. But, if the same site had received all of this precipitation in the form of snow, the total accumulation would be more than 10 inches! Obviously, there was a lot of uncertainty as to the proportions of snow versus other wintry precipitation. Since the latest model runs continued to indicate strong upward vertical motion caused by the continuous flow of warm air aloft overnight, the primary (and most difficult!) forecast challenge was to discern what the primary precipitation type would be across different locations of the forecast area. At the time the forecast was issued, it seemed apparent through the observations that a large component of the warm air flow aloft (roughly 5 to 10 thousand feet where precipitation was forming) was contributing to warming the atmosphere, rather than contributing to intense upward motion (which is why the moist layer warmed so rapidly). However, if an increase in upward motion actually occurred, this would suggest a decrease in the amount of warming and preference for snow. So, part of the forecast was an attempt to figure out what processes (and therefore what precipitation types) would dominate during the overnight hours in different parts of the forecast area. With that in mind, let's see how we did.
Our afternoon forecast from the 12th clearly underestimated the effects of sleet and freezing rain on the snow totals. As mentioned in the previous paragraph, that was a very difficult part of the forecast. Although it appears that the forecast did handle the geographic distribution of the various types of precipitation well, the snowfall forecast was overdone except in north-central Oklahoma and extreme south-central Oklahoma (where no snow accumulation was forecast - just ice). If the precipitation had fallen mostly as snow, as predicted by the morning runs of the computer models, snowfall amounts - at least across those areas where sleet and freezing rain occurred - would have been much greater, probably even greater than what was actually forecast.