Using Forecast Ensembles for Long-Term Chase Planning

[Tom Dulong]

As occurs every year, we begin to look into our "crystal ball", read the moss on the trees, or use anything else to foresee when our first chase of the new season might be. I have seen several folks mention "ensemble" in our forum threads this winter. How useful have ensemble forecasts been for your chase planning (or wishing)?

Ensemble forecasts can be found from several web sources, but I prefer:
http://www.cdc.noaa.gov/map/images/ens/ens.html

COMET offers a free tutorial about ensemble forecasts at:
http://meted/topics_nwp.php

 

[Dan Robinson]

Extremely unreliable for long-term planning. They often flip-flop from one extreme to the other in a matter of days or even less.

Last year we thought we might have a western trough sometime after the 16th of May. We left for the Plains with some hope. The next day, we awoke at our hotel and saw that the upcoming trough had completely vanished from the ensembles. We turned around and went home! No real chase-friendly trough appeared until June.

I'm not saying they are completely useless. Anything that shows remarkable consistency has a good chance of happening, but there are never any guarantees and always the chance of a total failure up to the last minute.

 

[Jeff Snyder]

Ensemble spread can give an indication of how much confidence once can put in a particular model forecast. In the most basic, a high spread indicates that the forecast is highly sensitive to small changes in the initial conditions, which usually means that one should put less confidence in the operational model run. Likewise, a small ensemble spread would indicate that one CAN put more confidence in the ensemble mean... The actual operational model run may or may not be near the ensemble mean, so keep that in mind. Running ensembles is a way of dealing with the uncertainty in the initial conditions that are fed into the operational models.

You can check out SPC's Short Range Ensemble Forecast page at http://www.spc.noaa.gov/exper/sref/