The ETA Model in 2000 is The Nam today. Has trouble with these storms deepening off the coast. Good read on model error back then , that took everyone by suprise.
Model Error
In general, forecast models preformed very poorly during this event. An example of this is shown in the image below which compares the 48 hour, observed liquid equivalent precipitation with the Eta model 48 hour forecast valid at the same time (click on the image to enlarge, image provided by Michael Brennan, NC State University). The model failed to capture the significant precipitation over the Carolinas and Virginia; in fact the Eta forecast had no precipitation over Raleigh and Richmond where over an inch of liquid equivalent was observed.
As forecasters scrambled to adjust forecasts to reflect the unfolding snowstorm, there was little time to analyze the full reason why the forecast had gone bad. In the wake of the storm, many were left wondering how nearly two feet of snow could have fallen in less than 24 hours with very little warning. Was there something that could have been analyzed ahead of time that would have lead to questioning of model precipitation forecasts? Or was there a physical mechanism that was misunderstood, or missed by the models? Below is a brief summary of some of the forecast errors and research topics that arose from this case.
Incipient Precipitation and Latent Heat
An area of precipitation developed along a frontal system over southern Alabama and southern Mississippi at around
06Z on January 24, 2000. The area of precipitation expanded and intensified as it
moved into the cold airmass north of the front in Georgia by 12Z on January 24. Analysis of surface observations and radar imagery show that
this area of precipitation was poorly forecast by the Eta model. Research into this storm indicates that this under forecasted area of precipitation that fell into a cold and somewhat dry airmass across the Deep South had a significant impact in how the models handled the developing storm system.
Based on quasi-geostrophic (QG) theory, height rises (falls) occur above (below) a mid-level latent heating maximum, due to changes in the density of the air above and below. This response also creates a low-level maximum in cyclonic potential vorticity (PV), as low-level static stability is increased. The
effects of the induced cyclonic PV max are manifest in wave amplification and enhanced rotation around the PV center. The precipitation across the Deep South induced a PV anomaly that enhanced the easterly flow downstream, creating stronger westward moisture advection over the Carolinas from off the Atlantic (Brennan and Lackmann 2005), thereby extending the heavy snowfall further inland. The inability of the models to accurately predict the precipitation across the Deep South was a major reason that the model forecasts of this storm were so inaccurate.
The combination of increased moisture advection, enhanced dynamics, and a deep subfreezing column of air (see GSO soundings below), created an environment that was primed for intense winter weather