The Cooperative Program for Operational Meteorology, Education and Training (COMET) has been developing small cases to facilitate NWS field forecaster training in the intelligent use of NWP models since early 2001.  A general overview of methodology of development of these so-called ?vignettes? has been submitted separately for this conference.

A vignette for training on use of short-range ensembles was developed for the 6-7 January 2002 snowfall over the interior northern Mid Atlantic States and New England.  The storm was poorly predicted by the operational numerical weather prediction (NWP) models and the newly developed NCEP short-range ensemble forecasts (SREFs). This poor performance can be traced back to difficulties in capturing the initial atmospheric state in a series of NWP and SREF forecasts. An example from 12 UTC 5 January 2002 analyses in the graphic below shows the comparison of the Global Data Assimilation System (GDAS) versus observed radiosondes at 500 hPa. Note that initial conditions for five of the 11 SREF ensemble members are derived from the GDAS.

In the poster, we will describe the genesis and evolution of the data assimilation problem, its impact on the SREF forecasts, and how it could have been detected and adjusted for by the forecaster, including satellite observations and comparisons to radiosonde observations.