jpeters3 Posted August 22, 2011 Share Posted August 22, 2011 Why do you find a model useful that misses by about 500 miles from 3 days out? It's like using the NAM QPF in the winter. I'd like to see some verification studies that support your opinion on the performance of this model. I'm not arguing that you are wrong, but this is a well known/used numerical mode, and there is no reason why someone should be bashed for posting its output in a thread dedicated to forecasting a tropical cyclone... Operational forecasters tend to develop these particular attitudes toward models without actually looking at verification statistics; however these are opinions, which are easily riddled with bias. Link to comment Share on other sites More sharing options...
Analog96 Posted August 22, 2011 Share Posted August 22, 2011 Historically, it's in the top 3 in accuracy. I agree it has been off so far, but we need to understand why. What is it seeing that others are not? 12Z (finally) initialized the storm pretty well, and it was still W, so the theory of poor initialization is out. It's been off this year, which is what matters most. Link to comment Share on other sites More sharing options...
Analog96 Posted August 22, 2011 Share Posted August 22, 2011 I'd like to see some verification studies that support your opinion on the performance of this model. I'm not arguing that you are wrong, but this is a well known/used numerical mode, and there is no reason why someone should be bashed for posting its output in a thread dedicated to forecasting a tropical cyclone... I'm just basing off what it's done this year. Link to comment Share on other sites More sharing options...
bobbutts Posted August 22, 2011 Share Posted August 22, 2011 Why are models so bad? Link to comment Share on other sites More sharing options...
jpeters3 Posted August 22, 2011 Share Posted August 22, 2011 I'm just basing off what it's done this year. What we don't understand is why. It could be an initialization problem, it could be something that has been changed within the model physics, and it could be simply "bad luck." Discounting the output of this model completely due to perceived poor performance over the course of several invests and tropical storms would be a bit rash. Link to comment Share on other sites More sharing options...
RaleighWx Posted August 22, 2011 Share Posted August 22, 2011 For those interested, my discussion http://www.examiner.com/weather-in-raleigh/hurricane-irene-to-be-a-big-threat-to-the-us Link to comment Share on other sites More sharing options...
gil888 Posted August 22, 2011 Share Posted August 22, 2011 The G-IV should help with the shorter-term stuff, but you can't expect that to totally carry through to day 5. You need the mid-latitude data to catch everything-- that's a trick too. I'm looking forward to the 00z runs Well, I'm not into this stalling out, but I like that the Euro isn't jetting it NNE like the GFS. Funny that the solutions are so different-- not just in terms of timing and speed, but really in the basic nuts and bolts. We need those high-res upper-air data. Link to comment Share on other sites More sharing options...
Southland Wx Posted August 22, 2011 Share Posted August 22, 2011 What we don't understand is why. It could be an initialization problem, it could be something that has been changed within the model physics, and it could be simply "bad luck." Discounting the output of this model completely due to perceived poor performance over the course of several invests and tropical storms would be a bit rash. I don't particularly like to throw models out. But in the tropics I think there's a strong correlation between intraseasonal performance to date and how well that model should do with the current storm. In other words, you need to play the hot hand w/ the tropics I think, and if a model has exhibited success season to date, it warrants more weight than others. Likewise, if it has performed poorly to date (and the GFDL has performed extremely poorly) it should be considered an outlier until it has support coming from elsewhere. That said, you can't with any certainty rule anything out at this point (I'm not 100% comfortable here in JAX right now), and trends/obs should be monitored until it becomes clear that you can rule someone or somewhere out. Link to comment Share on other sites More sharing options...
am19psu Posted August 22, 2011 Share Posted August 22, 2011 In 2010, the only models that were really worth anything were the Euro, GFS and UKMet. The rest were basically useless. 2010 NHC Verification Link to comment Share on other sites More sharing options...
HurricaneJosh Posted August 22, 2011 Share Posted August 22, 2011 The G-IV should help with the shorter-term stuff, but you can't expect that to totally carry through to day 5. You need the mid-latitude data to catch everything-- that's a trick too. I'm looking forward to the 00z runs Gotcha. Do you think the 00Z runs will give us much better clarity Re: landfall prospects? Link to comment Share on other sites More sharing options...
gil888 Posted August 22, 2011 Share Posted August 22, 2011 Also of note is that there are a whole lot of special soundings over the eastern United States at 18z-- looks like most of eastern Region NWS and eastern SR. Should have plenty of observations of the east coast weakness... Link to comment Share on other sites More sharing options...
meteorjosh Posted August 22, 2011 Share Posted August 22, 2011 That hot tower shows up very nicely on infrared (actually on visible also). goes Link to comment Share on other sites More sharing options...
jpeters3 Posted August 22, 2011 Share Posted August 22, 2011 I don't particularly like to throw models out. But in the tropics I think there's a strong correlation between intraseasonal performance to date and how well that model should do with the current storm. In other words, you need to play the hot hand w/ the tropics I think, and if a model has exhibited success season to date, it warrants more weight than others. Likewise, if it has performed poorly to date (and the GFDL has performed extremely poorly) it should be considered an outlier until it has support coming from elsewhere. That said, you can't with any certainty rule anything out at this point (I'm not 100% comfortable here in JAX right now), and trends/obs should be monitored until it becomes clear that you can rule someone or somewhere out. Again, while this may be the case (and may be good strategy to approach operational forecasting of tropical cyclones), i'd like to see some statistics to support such a theory. If what you say is true, particular models will, from time to time, have error bias that persists through a given season, or at least a portion of a given season (assuming that some cases of model biased are fixed as the season progresses). keep in mind, we haven't had any hurricanes this season before now. Many of the systems that have developed have had ill-defined centers, leading to potential initialization issues (which in turn may have been responsible for the error). These issues may be alleviated in a situation with a stronger system and more confident center fix. Link to comment Share on other sites More sharing options...
OSUmetstud Posted August 22, 2011 Share Posted August 22, 2011 this is a tough forecast, I think. Three separate shortwaves involved....one in the next 2 days...another in the 3 to 5 day range and possibly another one in the day to 6 to day 7 range. Link to comment Share on other sites More sharing options...
k*** Posted August 22, 2011 Share Posted August 22, 2011 DT says cat 4 Link to comment Share on other sites More sharing options...
Hugo Posted August 22, 2011 Share Posted August 22, 2011 Just a little historical tidbit: A Cat 4 has never made landfall north of Little River, SC. Link to comment Share on other sites More sharing options...
HurricaneJosh Posted August 22, 2011 Share Posted August 22, 2011 DT says cat 4 Predicting a Cat 4 for NC is climatologically unwise. Link to comment Share on other sites More sharing options...
HurricaneJosh Posted August 22, 2011 Share Posted August 22, 2011 Just a little historical tidbit: A Cat 4 has never made landfall north of Little River, SC. Bingo-- there is no historical precedent for it, so the bust potential is like 99%. Link to comment Share on other sites More sharing options...
yoda Posted August 22, 2011 Share Posted August 22, 2011 Predicting a Cat 4 for NC is climatologically unwise. I don't think he is predicting it.. I think he is trying to say thats what the 12z EURO model shows -- i think he is trying to equate the MSLP value to what Cat value on the hurricane scale Link to comment Share on other sites More sharing options...
HurricaneJosh Posted August 22, 2011 Share Posted August 22, 2011 I don't think he is predicting it.. I think he is trying to say thats what the 12z EURO model shows -- i think he is trying to equate the MSLP value to what Cat value on the hurricane scale But we all know that deriving a SS intensity from a global model five days out is kind of silly. Link to comment Share on other sites More sharing options...
gil888 Posted August 22, 2011 Share Posted August 22, 2011 Gotcha. Do you think the 00Z runs will give us much better clarity Re: landfall prospects? well if the GFS/ECMWF are still consistent after the G-IV, that would make me buy into their solutions a lot more. True the trough is still over western North America, but getting the short-term features correct is important too, along with the outer storm structure and size. Link to comment Share on other sites More sharing options...
Floydbuster Posted August 22, 2011 Share Posted August 22, 2011 Bingo-- there is no historical precedent for it, so the bust potential is like 99%. Note that DT is showing what the model shows, not a forecast I don't believe: *** alert *** 12z EUROPEAN MODEL show NO turn out to sea... LANDFAL at wilmington NC at 938 MB -- CAT 4 -- then tracking due North OVER RALIEGH - WILSON NC... into RIGHT OVER RICHMOND VA then over BALTIMORE MD... Link to comment Share on other sites More sharing options...
PSUBlizzicane2007 Posted August 22, 2011 Share Posted August 22, 2011 But we all know that deriving a SS intensity from a global model five days out is kind of silly. This. The best thing you can get out of the model solutions as a whole is that it is looking quite likely that Irene will have unusually rare favorable conditions for strengthening with respect to its position near the coast. Link to comment Share on other sites More sharing options...
OSUmetstud Posted August 22, 2011 Share Posted August 22, 2011 Link to comment Share on other sites More sharing options...
k*** Posted August 22, 2011 Share Posted August 22, 2011 I don't think he is predicting it.. I think he is trying to say thats what the 12z EURO model shows -- i think he is trying to equate the MSLP value to what Cat value on the hurricane scale it's DT....he hugs the euro no matter what so you might as well call it a forecast Link to comment Share on other sites More sharing options...
Typhoon Tip Posted August 22, 2011 Share Posted August 22, 2011 This could be an interesting late afternoon and night up coming... Recent IR imagery shows a very intense eruption of core convection (towering phenonenon); given to the size of Irene I am suspecting that the dry air we've been off and on concerned with is probably not making it too far into the actual core of the system. There is some drying evident in WV but this air appears to be neutralized once crossing within ~70 naut miles of the fix. Given to the system now being free from land, and situated over high oceanic heat contents, while also having the benefit of well-established U/A anticyclone there doesn't appear to be any reason to limit free fall with this system. Link to comment Share on other sites More sharing options...
HurricaneJosh Posted August 22, 2011 Share Posted August 22, 2011 well if the GFS/ECMWF are still consistent after the G-IV, that would make me buy into their solutions a lot more. True the trough is still over western North America, but getting the short-term features correct is important too, along with the outer storm structure and size. Gotcha. The GFS and Euro are quite different. Which do you think is more credible in this case? Link to comment Share on other sites More sharing options...
sea1skiff Posted August 22, 2011 Share Posted August 22, 2011 Bingo-- there is no historical precedent for it, so the bust potential is like 99%. Hazel? Link to comment Share on other sites More sharing options...
PSUBlizzicane2007 Posted August 22, 2011 Share Posted August 22, 2011 Gotcha. The GFS and Euro are quite different. Which do you think is more credible in this case? It's gotta be the Euro. The GFS can score one every once in a while, but the vast majority of the time the Euro will win out between these two... (not to mention the height progs make more sense on the Euro) Link to comment Share on other sites More sharing options...
28storms Posted August 22, 2011 Share Posted August 22, 2011 Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.