« Marko's Halloween Party | Main | Operation Christmas Child »
October 31, 2006
Ensembles and More Ensembles
[On location in Laurel, Maryland]
There's not too much to report from Day 1 of the NCEP/NWS Ensemble User Workshop. Things got a bit interesting after the first session when some guys from the NWS and NOAA were having a rather heated discussion, with one NWS guy (Rich Grumm, State College WFO) wanting the resources in AWIPS to make more probabilistic forecasts available to the public (such as, "in five days it'll be in the 50s"), one of the NOAA guys responding that the superiors at the NWS recently voted down a proposal to do just that, at which point another NWS guy angrily responded that they were only providing for the government's stated requirements of NWS, i.e., deterministic forecasts (such as, "at 2pm five days from now the temperature will be 53 degrees"). Nothing like raised voices to snap you to attention from a drifting mind mid-morning. ;-) Basically one of the main themes out of today was that the "weather community" needs to communicate the uncertainties inherent in forecasting to the public much better than we are currently. Though it was somewhat amusing when one of the Air Force guys raised the point that for all this talk about uncertainty, none of the presenters had any error bars in any of their results. :-)
The second session of the morning was drastically less exciting, however. Not only was the subtopic rather dry, but all three presenters were Chinese and really weren't at all fluent in English. One of the guys was racing at breakneck speed through his broken English (couldn't understand him), another was speaking practically in a whisper (couldn't hear or understand him), and the other was just mostly unintelligible.
Fortunately the two afternoon sessions were much better. All six NCEP (Nat'l Center for Environmental Prediction) Service Centers (Hydrometeorological Prediction Center, Tropical Prediction Center/National Hurricane Center, Aviation Weather Center, Climate Prediction Center, Ocean Prediction Center, and the Storm Prediction Center) gave summaries of how they're using ensemble forecasts in their products, and I was impressed with a lot of what they do. SPC has really got their act together too, let me tell ya. When a severe weather watch box is issued, they even give the probabilities that damaging tornadoes, hail or wind will occur within the watch area. And they're also doing cool things like forecasting the probability of 3 inches of snow melting in 3 hours (a proxy for accumulation on roads). Hopefully tomorrow's sessions will be interesting.
With the midterm elections just around the corner, it's newspaper endorsement season! If you want to see what is quite possibly the most negative written endorsement ever, scroll to the bottom item at this page. It's absolutely hilarious! On opinionjournal.com's "Best of the Web Today" feature, I've been following along in their "Unenthusiastic Endorsement Watch" segment the past few days, and as good as the others have been, this one definitely took the cake!
Well, I guess I should stop wasting time and start maybe trying to work on my Grad Synoptic homework a little bit... Maybe...
Posted by Jared at October 31, 2006 09:15 PM
Comments
Please explain the concept of "ensemble" forecasting as opposed to the concept of an ensemble in performance music and as opposed to the concept of a woman wearing a chic ensemble.
Posted by: MERAmanuensis at November 1, 2006 06:00 PM
2nd try at posting a comment:
I know what a chic ensemble is in the clothing category. I know what an ensemble is in the music category. But I do not know what is meant by "ensemble forecasting" RE meteorology.
Expliquez vous, s'il vous plait.
Posted by: MERAmanuensis at November 1, 2006 06:04 PM
It's where they run lots of different models and see what the consensus is. :) Maybe.
Posted by: Nathan at November 2, 2006 05:33 PM
Yeah, ensemble forecasting makes use of several runs of the same model, with each run having slightly different initial and/or boundary conditions, internal physics or other parameterizations. Slightly perturbing some parameters in the model creates a different result since the atmosphere is a chaotic system. If the results of several of the ensemble members are similar, then you can have a reasonably high confidence in their predictions, but if the members are all over the map, then you have to rely on other things in order to make your forecast.
Posted by: Jared at November 5, 2006 11:46 PM