Long Range Forcasting

Joined
Mar 5, 2010
Messages
340
Location
Cascade, CO
Hey friends!

Need a little advice here! I leave in a week for my "chase-cation" and am wondering where is the best place to look for long range forcasting? I have played with the GFS and have used the Climate Prediction Center, but didnt know if there was another good site for forcasting 7-14 days out?

Any thoughts?
 
Some websites like COD and HOOT have basic 500mb charts from the ECMWF. You can pay for a service called Accuweather Pro in order to get the full version of the European model, but the GFS is your only free option for long range numerical modeling.
 
In the long range time frame, the lack of free data from the ECMWF doesnt bug me as much. At that range, I am rarely looking at the exact parameters. Just give me a general location and strength of the low or the upper level pattern (which you can get for free). The accuracy of the GFS and ECMWF in the 1-2 week time is getting better, but its still not really reasonable to be looking at details that far out.
 
If you are doing this for amusement, then the above responses are valid. If you are actually trying to plan something, then it's a complete waste of time. There's a reason the SPC 4-8 day outlook is usually populated with "predictability too low";)
 
http://www.meteo.psu.edu/~gadomski/ewall.html

^^^ That's a neat link that I often use. Like Stan said, long range modeling doesn't really work unless the models are in fairly good agreement, and oftentimes they can be 180 degrees offset.

I obviously wouldn't recommend taking a chasecation for at least the next 7 days...but after that, who knows.
 
Long-range forecast discussions seem to involve a mild degree of embarrassment for many of us. We know that beyond day three we're entering into the realm of prestidigitation and reading animal entrails, so we shuffle our feet awkwardly, gaze at the floor, and mumble some kind of caveat when we talk about long-range: "Well, it's a long way out, and the models are going to change." Don't want anyone to think we're naive, after all. Nevertheless, a lot of us--I'll bet most of us--still look out all the way to the end of the road at 384 hours, if not always then at least some of the time. Why? Certainly not to forecast with any degree of certainty. We just want to get a sense of what may be coming down the pike and to start looking for a general consistency. We're particularly likely to do so when the short term looks bleak. What we're really looking for is hope, right? Plus, of course, people who have got to nail down a time frame for their chasecation need at least some sense that they're choosing a potentially productive time, not a death ridge.

Patterns often crash and burn as they move from far out to near in. That's understood. But the fact is, some of them verify. The burlier systems seem to broadcast themselves well in advance of day three. So it pays to watch the GFS and Euro. The big question is, what's going on with the 500 mb chart. For me, I also like to get a sense of how mid-levels may interact with instability, so I check out moisture, too, and CAPE on the GFS. I'm not looking for anything resembling detail or accuracy, but for shapes and possibilities. Last night's 00z GFS for May 23 looked horrible; this morning's 6z makes the northern plains look awesome. What can I extrapolate from that--a forecast? Heck no, just a sense of what to keep an eye on farther down the pike. Particularly on May 22, which, as I've said before, is my "lucky date." :) And a lucky date is reason enough to keep an eye on things and watch how they evolve as long-term moves into NAM territory, and from there into days 3-2-1.
 
Last edited by a moderator:
Jared,

That's my problem :) Us Indiana guys have to plan these chasecations months out.....vacation time etc!

UGH :)

Having started my chase career in Indiana, I strongly suggest you focus on local/regional events until you become comfortable with forecasting the risk involved in distant plains chasing. One solution is too bail on the Ohio valley grunge and move to one of 10,000 towns on the Great Plains.
 
The burlier systems seem to broadcast themselves well in advance of day three.

A very good point; it seems to me like all the biggest days the past few years have shown themselves WELL in advance. Just going by the SPC extended outlook:
-May 10, 2010 highlighted May 5
-April 27, 2011 highlighted April 23
-and of course April 14, 2012 highlighted all the way back on April 8

In addition, even though the day-to-day details weren't clear, it was apparent on May 19 of last year that something was brewing starting May 24.

Also an interesting note about May 22: it happens to be the day after May 21: a day that, for some reason, has an anomalously low number of tornado outbreaks. Even in the ten years since 2002 (the end of the data in the linked paper), only 2 tornado outbreaks have occurred on May 21 (both minor), and only 2 other days had more than one tornado report. Statistical anomaly? Probably. But being slightly superstitious I'd definitely take it into consideration :D
 
Last edited by a moderator:
It's all about the forecast funnel. Start with 250 hPa and work your way down :) Maybe even look at mountain torques and global atmospheric angular momentum if you want to. Looking at the MJO and convection of the Indian ocean can also clue on on a pattern change before the models do.

For me, I never look at the GFS because it's a piece of crap, and I don't mean that lightheartedly, the model just sucks. The ECMWF is vastly superior. There is a reason why it doesn't come in until about 20Z/7Z... its all the post processing that occurs after each synoptic run.
 
It's all about the forecast funnel. Start with 250 hPa and work your way down :) Maybe even look at mountain torques and global atmospheric angular momentum if you want to. Looking at the MJO and convection of the Indian ocean can also clue on on a pattern change before the models do.

For me, I never look at the GFS because it's a piece of crap, and I don't mean that lightheartedly, the model just sucks. The ECMWF is vastly superior. There is a reason why it doesn't come in until about 20Z/7Z... its all the post processing that occurs after each synoptic run.

I, too, have noticed that the ECMWF does far better in regards to long-range forecasting, and I even believe it to often be superior in the medium-range. The EPS (ECMWF Ensemble Prediction System) that runs out to 360 hours and the new monthly forecast system are great tools, if you have access. I wonder if NCEP would be better off modernizing the supercomputer system and doing away with the 18z and 06z runs of the American model guidance. If and when funds are available, several dozen more RAOB stations would also be beneficial, especially in the eastern Pacific region, but elsewhere too.
 
More raobs? Good luck with that! With the budget crisis and no longer maintaining the wind profilers.. that will not happen anytime soon. 06Z/18Z runs also aren't to be trusted because they don't have any raob data to ingest. My attitude towards the long range is look at the big picture. Focusing on small mesoscale details is foolish. Is there a trend? Also, the NAM should not be used past 24 to 36 hours due to non linear feedback mechanisms. It is a mesoscale model and should be used as such (mesoscale in the temporal sense). I do look at the EC ensembles. Usually the GFS and particularly the GEM are outliers.
 
Back
Top