Jump to content
  • Member Statistics

    17,588
    Total Members
    7,904
    Most Online
    LopezElliana
    Newest Member
    LopezElliana
    Joined

The March 5-7 White Lion Obs/Nowcasting Thread


Ji

Recommended Posts

The thing is the only way we can really attribute statistical significance is over a large sample.  Hemispheric and regional verification over a month, for example, shows the EC is better than the GFS, and in the mean the difference is significant at a 95% confidence probably out to 5 days or so.  This does not automatically translate to a case-by-case or regional basis, however.  It is not such a superior model that it is simply always going to be better.  In fact, the GFS had better TC track forecasts in the Atlantic in 2012 (despite one very high profile case dragging the GFS skill down a bit).

Also, one of the big problems in attaining a statistically significant sample is that these models are being updated on a regular basis so it becomes rather difficult to compare any two models directly over a large period of time. You could do a general comparison independent of the updates by just looking purely at which one gets it right more often, and that would yield the necessary sample. But that wouldn't be especially useful because of the aforementioned update cycle. My hunch is that given a large enough sample for two comparable versions of each model you'd find that the differences between the Euro and the GFS to be fairly small - i.e., within 10% of one another in terms of verification rates. Perhaps higher differences with individual maps, but overall I wouldn't expect a staggering difference. 

Link to comment
Share on other sites

  • Replies 2.8k
  • Created
  • Last Reply

Thanks.

I believe that we are working on developing summary scorecards like the ECMWF has, such as this:

http://www.ecmwf.int/products/changes/ifs_cycle_38r1/scorecard.html

 

But this gives an idea in terms of the volume of stuff we actually look at, just in terms of mean scores.  We also look at case by case type stuff as well (and do our best to reply to customer observations/complaints/needs/etc. to address specific issues).

Link to comment
Share on other sites

I agree fully. It is always ebb and flow btw the models from a operational fcsting standpoint. Each pattern and scale are always weighed differently wrt model guidance.

I think that what gets to me at times. I'm just a weenie but I watch this stuff all winter. Usually every run of the gfs/euro unless it's really boring or an obvious long duration stretch of aoa.

I see way to much "if the euro doesn't show it the gfs is wrong" and "it's the gfs, throw it out". It's like the new hip thing to say. And it's narrow minded as heck. Relying on a single model for 80%+ of all your forecasting at all ranges is fools gold. But you can always spin how it was right all along in the end and retain your viewership so I guess it doesn't matter whether you're right or wrong anymore. As long as you are a good salesman actual accurate forecasting skill can take a back seat.

Link to comment
Share on other sites

I think that what gets to me at times. I'm just a weenie but I watch this stuff all winter. Usually every run of the gfs/euro unless it's really boring or an obvious long duration stretch of aoa.

I see way to much "if the euro doesn't show it the gfs is wrong" and "it's the gfs, throw it out". It's like the new hip thing to say. And it's narrow minded as heck. Relying on a single model for 80%+ of all your forecasting at all ranges is fools gold. But you can always spin how it was right all along in the end and retain your viewership so I guess it doesn't matter whether you're right or wrong anymore. As long as you are a good salesman actual accurate forecasting skill can take a back seat.

 

I see that on here all the time as well. Seems everyone wants to find the "right" model, when in fact, all models are wrong all the time to some degree. Model guidance is good to ascertain a general pattern evolution and to give a measure of confidence. However, they're not good in gaining a deterministic sensible wx fcst, which is painfully found out on many occasions.   

Link to comment
Share on other sites

I see that on here all the time as well. Seems everyone wants to find the "right" model, when in fact, all models are wrong all the time to some degree. Model guidance is good to ascertain a general pattern evolution and to give a measure of confidence. However, they're not good in gaining a deterministic sensible wx fcst, which is painfully found out on many occasions.   

 

Especially when they all reach a pretty good consensus as a storm's starting, only for the rug to be pulled out about 8 hours later.

 

:poster_oops:

Link to comment
Share on other sites

Who really cares what SNE gets.  It is climo, they get snow in March. Hell we got snow when i lived up there in Early April.   Im done with winter, bring on 55-65 degrees and sun

 

I normally don't.  At all.  But the whole situation is absolutely hilarious, so I can't help it.

 

There's a fairly small area getting f'ing blasted right now, but blasted they are.  :lol:

Link to comment
Share on other sites

I know DT tried to re-claim victory for his Euro after its crushing defeat, but isn't the on-going snows up in SNE proof that the GFS-led take on the evolution of this storm (that it would make the turn and impact SNE) was dead on and the Euro got overall schooled with regard the general track of  this system?

The Euro was locked in on a storm for VA for days while the GFS was driving the low down into GA and SC. It did that for days and days. It was awful in the medium range and then finally started catching on. I will give that it was the first to sniff out the snow for SNE, but hell its so erratic and sometimes downright hideous in the med-long range, it ought to get something right in the short range. The Euro is out of its wheelhouse inside about 75 hours...so yeah it shifted some and expanded the precip further north, but so what? It is a model and isnt perfect. Overall it sniffed out the threat first, and consistently showed a major snow event targeting the exact areas in the MA that ultimately got it. Kudos to the GFS for leading the way on the snow for the NE. 

Link to comment
Share on other sites

The Euro was locked in on a storm for VA for days while the GFS was driving the low down into GA and SC. It did that for days and days. It was awful in the medium range and then finally started caching on. I will give that it was the first to sniff out the snow for SNE, but hell its so erratic and sometimes downright hideous in the med-long range, it ought to get something right in the short range. The Euro is out of its wheelhouse inside about 75 hours...so yeah it shifted some and expanded the precip further north, but so what? It is a model and isnt perfect. Overall it sniffed out the threat first, and consistently showed a major snow event targeting the exact areas in the MA that ultimately got it. Kudos to the GFS for leading the way on the snow for the NE. 

 

You flipped those.  The euro had the ridiculous bowling ball moving SE off of SC solution while the GFS was at least realistic with the pattern.  The only thing the euro got right was that once it finally came around to the right 500 evolution, it put the precip max in a better position than the GFS and NAM which wanted to CCB the heck out of us in the DC area.

Link to comment
Share on other sites

Wife, baby, and I went to visit my in-laws who live out on 522S of Front Royal this evening.  I left my house in Stephens City where not a trace of snow still exists and discovered still a foot of snow on the ground at their house.  Incredible to see the difference, just remarkable.  Was able to play in snow with my daughter one final time this season (give it up, Ji).  Fun night.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.

×
×
  • Create New...