Dueling model biases

Everything from Monday’s discussion is on track. The NWS is starting to get on board with just how bitterly cold it will be across the eastern CONUS this weekend. The Duluth, MN forecast office noted Tuesday morning that some parts of northern MN may experience lows of -45 degrees F in the Sat-Tues time period. That same air mass will be coming our way and the amount of modification (see prior post) will determine whether or not BGKY will dip below zero or simply have lows in the single digits.

There is an interesting battle going on between known biases in the computer models that could affect the accumulation from Thursday’s snowfall.

Model Bias 1) Since the December 1, 2006 blizzard over OK, MO, and IL, all of the major computer models (GFS, NAM, ECMWF, UKMET) have made a consistent error of developing a surface low track too far to the southeast for the time period 72-96 hours away from the storm. The reason for this error has been the underestimation of the strength of the Southeast Ridge located over the western Atlantic. The strength of this ridge is related to the negative phase of the PNA, which teleconnects to a strong ridge over the Southeast. Note that the PNA was negative for the Dec 1 storm as well as during Jan 1-15.

Model Bias 2) All of the major computer models poorly handle arctic air. Arctic air is so dense that often the models underestimate the southward penetration of the air mass as well as how to translate the thermal structure of the air mass to surface temperatures (see the emergency lowering of Tuesday’s high for BGKY from 32 to 23 this afternoon by the NWS). Therefore, the models often will have a tough time determining the rain-snow line when arctic air is involved with a storm system.

In this case, the model biases work against each other. MB1 says that the actual surface track for Thursday’s storm will be farther NW than the models presently indicate. MB2 says that the actual surface track for Thursday’s storm will be farther SE than the models presently indicate. So which model bias wins?

Easy answer = MB2. Looking at the present phase of the PNA shows that the PNA is strongly positive, which teleconnects to a trough in the Southeast. Sure enough, examination of the models shows that the Southeast ridge is nowhere to be seen. Not coincidentally, the PNA turned sharply positive on 1/15, which is exactly when the Southeast ridge weakened and colder air began to move in the Southeast. Its amazing how meteorology works.

Additionally, the models are also now starting to accept just how cold these arctic air masses truly are, and the models have responded by bringing the surface low farther to the SE with each run (compare the 12z run to the 06z and 00z for Thursday afternoon). This portends to an all-snow or snow with some sleet situation for BGKY rather than the changeover to rain that some people have been talking about. I still like the idea of 1-2 inches for BGKY, but I’ll have a better idea of an amount Wednesday afternoon.

This entry was posted in Kentucky Weather. Bookmark the permalink.