Wednesday, August 9, 2017

GFE Grid Work at the C&V desk

Day 2 in the testbed brought the first full day of forecasting with the GFE grids at the Clouds and Visibility desk.

After experiencing some issues with the AWIPS servers the day before, which would cause CAVE to crash and/or get hung up on occasion, it was decided to focus on one half of the US each day as opposed to tackling both East and West every day.  Today's focus region was over the Eastern half of the CONUS.

The weather of the day consisted of a front moving over the southeast, and a tropical system moving up into the southeastern part of Texas and southwestern part of Louisiana later in the TAF forecast time.  Forecasters working the desk compared the satellite observations along with the derived probabilities of flight rules to determine which model was handling the frontal ceilings and visibility the best.

Consensus was to use the AWCRAP for the beginning of the forecast period, with AWCHRRR to finish it out.  Some periods of missing data in the feeds also drove the reasoning to use multiple models for the forecast.  Of course this created some "jumps" between the two models which required some tweaking by the forecasters to keep the forecast smooth and consistent in time.

The above screen shot of the forecast Sky grid illustrates the issues when trying to interpolate between two models.  Forecasters decided to use a different approach to branch between two different forecast time grids, this involved using a tool available that can advance highlighted areas in time given current wind speeds and direction. 

Final grids were sent off to OPG to edit and create TAFS in their three specific WFO areas; Atlanta, New York, and Greenville-Spartanburg.  Lots of good discussion took place during the afternoon debrief regarding the detail in the resultant TAFs from the AWC/OPG grid driven ones using the TAF formatter and the ones produced operationally in the field. The TAF formatter tends to give lots of detail by producing several lines, and the overall consensus from the discussion was that more detail is desired in the early part of the TAF (by the end user), where the formatter should be structured to smooth out some of the unnecessary detail in the later part of the TAF period.

No comments:

Post a Comment