and other decision makers in major East Coast population centers that created a sense of urgency in the public’s mind. Internal communications, such as the HPC guidance document, may have been misinterpreted by some as deterministic forecasts as opposed to forecaster guidance. Some of the language used created a building sense of inevitability about the event. There was an over-anticipation by the public that the storm would hit the major metropolitan areas, and when it failed to materialize, the public and media were very critical of the forecasts and of forecasters.

Analysis of the Case

By February 26, nearly a week before the first precipitation fell, NWS forecasters recognized the potential for an East Coast storm. Confidence was not high 6 to 7 days out, however, as there was significant disagreement among the medium-range forecast models. By March 1, a number of forecasters were indicating increasing confidence in the possibility of a major East Coast storm. By the afternoon of March 2, the medium- and short-range models (MRF, ETA,4 AVN) were coming into much better agreement, which was reflected in forecast discussions of high confidence of a major East Coast event. On Saturday, March 3, even though the storm had yet to form and new guidance indicated that it would be slower to develop than had been indicated 24 to 48 hours earlier and perhaps be more of a rain event rather than a snow event, there was still wide-spread action by public and private sectors in many major East Coast cities to prepare for an impending snowstorm all along the East Coast from Virginia to New England. Confidence of a major snowstorm was lowered even further later that day because warmer air was holding along the coastal plain. On Sunday, March 4, predictions were continually refined as the storm slowly intensified. By Sunday afternoon there was increasing expectation within the forecasting community that the storm would primarily bring rain rather than snow from Washington, D.C., to New York.

Lessons Learned

During the workshop, participants suggested several lessons learned from this case.

  • It is possible to overstate the implied accuracy of longer range forecasts, even though the synoptic scale outlooks of the storm reflected the state of the science for a forecast 2 to 7 days in advance.


The ETA model is a mesoscale numerical weather prediction model run at NCEP. Its name derives from the eta coordinate system mathematical formulation used in the model.

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement