Jump to content
  • Member Statistics

    17,609
    Total Members
    7,904
    Most Online
    NH8550
    Newest Member
    NH8550
    Joined

Feb 12-13, 2014 Storm Model Forecasting Postmortem


WxUSAF

Recommended Posts

I think it's safe to say the Euro nailed this storm with long leadtimes.  Was quite consistent with surface track, 500mb and QPF for several days prior to the storm.  I think the NAM did quite well as well and had precip amounts that verified very near the actual totals.  BWI had about ~2", DCA had about 1.75" and IAD had ~1.35".  I think the GFS caught on with the 500mb pattern about 36-48hrs out and the surface track more like 24hrs out, but it never got the QPF in the right ballpark.  UKMET did pretty well also I think.  Thoughts?

Link to comment
Share on other sites

That looks right. I guess I am now differentiating when to trust the GFS and Euro most. Northern stream progressive flow? Lean GFS. Southern stream storms that ride the coast Miller A types? Euro and discount GFS.

 

Probably too simplistic, but man the GFS handled this one poorly.

Also, the totals the weather service and local mets went with ended up being way low. Pretty much 1 - 2 feet ended up being the swath from the cities, beltways, and north and west. I think all Balt TV mets stuck with a 5-9/6-10 range leading in, and LWX went into its usual severe catchup mode.

Link to comment
Share on other sites

That looks right. I guess I am now differentiating when to trust the GFS and Euro most. Northern stream progressive flow? Lean GFS. Southern stream storms that ride the coast Miller A types? Euro and discount GFS.

 

Probably too simplistic, but man the GFS handled this one poorly.

Also, the totals the weather service and local mets went with ended up being way low. Pretty much 1 - 2 feet ended up being the swath from the cities, beltways, and north and west. I think all Balt TV mets stuck with a 5-9/6-10 range leading in, and LWX went into its usual severe catchup mode.

I think that's a decent rule of thumb.  GFS has done quite well this winter I think up until this debacle.  Euro had been playing catch up on several scenarios.  But this was almost Sandy-esque in the beatdown the King put on the hometown team.  

 

LWX certainly was playing catchup.  I think they had a good initial forecast and then really backed down to the east which was really strange to me since it seemed like the overnight thump would get just about everyone to at least 6" (and it ended up working out that way).  

Link to comment
Share on other sites

Also, the NAM did so well, especially at 48 hours on in, I will give it more weight in the future. When the 12z suite was delivering disastrous runs Tuesday, it was the NAM to start 12Z and the NAM at 18Z that held and the others that joined back in thereafter (other than the GFS). It had a really good handle on this. 

Link to comment
Share on other sites

Also, the NAM did so well, especially at 48 hours on in, I will give it more weight in the future. When the 12z suite was delivering disastrous runs Tuesday, it was the NAM to start 12Z and the NAM at 18Z that held and the others that joined back in thereafter (other than the GFS). It had a really good handle on this. 

Yes it did.  It certainly was a little too cold (as is its bias) up until the last 24 hours, but the QPF was very good and did not end up 50% too high like often happens.  

Link to comment
Share on other sites

I would be curious to see what caused the GFS to have problems. I am sure these events are reviewed to try to improve the models future performance. Are the results of these reviews made public?

I hope dtk comments in this thread.  I think the convective issue certainly was a big cause of the QPF issue as WPC mentioned.  

Link to comment
Share on other sites

I think the gfs was having some sort of issue resolving precip fields. The track of the slp and 850 + the h5 patterns were very similar. It just wasn't resolving the nw side. Even run had me scratching my head. It just wasn't lining up. Dry on the nw side of a miller A is very unusual. This kinda made it easy to look the other way but the GFS really pissed me off. 

 

If this was some sort of miller b it would have been a major flag but not with a miller A. 

Link to comment
Share on other sites

I only have a hunch here, so I'm looking forward to the NWP experts. I think that level of frontogenetic forcing created a ridiculous secondary circulation. For all of you who aren't familiar with that, it is the ageostrophic component to the areas of strong uplift and subsidence. Behind this perfect marriage, there was an equally strong frontolysis signal (since gradient of potential temp was tightened by the perfect ingredients, now they could only weaken). It is quite possible that this signal weakened the QPF to the west of where the GFS was seeing the best lift. So, while it is sort of in the realm of "convective" issues, it's not quite that same concept.

The NAM / higher-res stuff will nail things like this and that is why a lot of experienced people disregarded the lower QPF of the GFS. Models are all tools and the GFS did fine with the placement of synoptic features.

Link to comment
Share on other sites

I think it's safe to say the Euro nailed this storm with long leadtimes.  Was quite consistent with surface track, 500mb and QPF for several days prior to the storm.  I think the NAM did quite well as well and had precip amounts that verified very near the actual totals.  BWI had about ~2", DCA had about 1.75" and IAD had ~1.35".  I think the GFS caught on with the 500mb pattern about 36-48hrs out and the surface track more like 24hrs out, but it never got the QPF in the right ballpark.  UKMET did pretty well also I think.  Thoughts?

The ECMWF went from an inland runner to coastal track from last Wednesday to last Thursday. Once it honed in on that, post-shift, it didn't budge with the general idea of a coastal storm. A lot of Miller A's have convection, strong isentropic lift/thermal advection and strong frontogenesis. All of these things can make for a QPF/precip type headache in a global model.

Before the models had something, the signal was in the west NAO regions post-stratospheric PV split. While subtle, the signal was there and we ended up having a classic 3 storm response:

1. First was precip loaded, warmer and widespread (+NAO to -NAO)

2. Second was weakest, not widespread and cold (peak -NAO)

3. Major winter storm/KU (-NAO to +NAO)

Link to comment
Share on other sites

I hope dtk comments in this thread.  I think the convective issue certainly was a big cause of the QPF issue as WPC mentioned.  

There were certainly precipitation related issues, but the blanket statements of "feedback" and "convective issues" are incomplete at best.  As a colleague of mine posted the other evening as the event was unfolding and we were having some of the discussion, the heavy precipitation that was being predicted by the GFS wasn't even "convective", at least in the eyes of the model and the use of the parametrization.  This has pretty significant implications.

 

I have not had a chance to do any sort of postmortem, but HM's comments regarding the predicted secondary circulation is as plausible as any I can think of.  Hopefully I'll be able to find some time to do some sort of analysis.  To be honest, I'm more interested in seeing how our 13km package handles the system.  It's unfortunate that we had to do a restart and weren't able to have the thing running in real-time.

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...