Jump to content
  • Member Statistics

    17,502
    Total Members
    7,904
    Most Online
    Weathernoob335
    Newest Member
    Weathernoob335
    Joined

March Threats/Snow Storms Pt. II


mitchnick

Recommended Posts

00z EPS for the next 10 days have <1" of the snow on the means for DC. Towards the end of the run, average to slightly above average temps showing up on the mean.

yeah - but that warmth is very temporary. It's that big low moving up the appalachians to our west and the high moving off the coast. Another high would be in the offing to reload the cold, though. 

Link to comment
Share on other sites

  • Replies 2k
  • Created
  • Last Reply

^ I wouldn't be surprised to see it snow again even IMBY.. Mar 25 is still a good ways off. :P

 

Could definitely have a better shot of a 'sizable' elevation event than normal. I wouldn't even totally rule out another mod event for the immediate area but it will take luck and then some. Probably some cold rain in our future either way. 

 

Get the right combo of +pna/-epo and we can easily get a -20+ airmass move through. The funny thing is even -20 won't get it done during the day. lol. March can have teens at night in the burbs and low 20s in DC through the end tho. 

 

My wag is we enter a favorable period during the last half but producing will be tough. Agree about elevation and N-W. They probably have a decent shot if things line up. Maybe it just wants to snow here now. Weird 2 year persistence. 

Link to comment
Share on other sites

Get the right combo of +pna/-epo and we can easily get a -20+ airmass move through. The funny thing is even -20 won't get it done during the day. lol. March can have teens at night in the burbs and low 20s in DC through the end tho. 

 

My wag is we enter a favorable period during the last half but producing will be tough. Agree about elevation and N-W. They probably have a decent shot if things line up. Maybe it just wants to snow here now. Weird 2 year persistence. 

 

Book it for Sunday March 22.

 

I was on a cruise during the first snowfall of the season (Thanksgiving week)

 

Then, left the state on Feb 20 to Albany, returning Feb 22,  missing the 10" on Feb 21st.

 

To bookend things, I figure the last snowfall of the season will begin as I'm getting off of a plane, in San Juan...

 

*shrugs* at least I got to enjoy the snow yesterday!

 

:yikes:  Eeeks, my bad, thought I was in the banter thread.   Delete if need be.

Link to comment
Share on other sites

I have a question for people who are knowledgeable about the models.  I'm not sure where to put this, so mods please move if necessary.

 

Looking back over the model runs for the last storm, it looks like all models had a pretty good handle on the storm on Tuesday, then the American models pulled a head-fake on Wednesday before recovering Wednesday night.  The foreign models were more consistent.  I've noticed the difference in run-to-run consistency before, especially when comparing the NAM and RGEM.  Even when the RGEM ends up with a worse solution, it's usually more consistent.

 

Is this primarily caused by the different data assimilation methods?  I would imagine that the 4DVAR, etc. assimilation methods used by foreign models might be less sensitive to errors / statistical noise in observations.  Could the head-fake on Wednesday have been caused, for example, by some misleading obs during the day that were effectively given more weight by the American models?

Link to comment
Share on other sites

Get the right combo of +pna/-epo and we can easily get a -20+ airmass move through. The funny thing is even -20 won't get it done during the day. lol. March can have teens at night in the burbs and low 20s in DC through the end tho. 

 

My wag is we enter a favorable period during the last half but producing will be tough. Agree about elevation and N-W. They probably have a decent shot if things line up. Maybe it just wants to snow here now. Weird 2 year persistence. 

why all the negativity about snow chances in march, it has happened before and will happen again. and considering the chances for a -nao to develop look better, then we still have at least 50-50 chance.That's pretty good in my book.

Link to comment
Share on other sites

AO at +5, eh. Is Judah Cohen in the witness protection program, yet?

There's a growing signal beyond that (somewhere around the 16th give or take) where we go back to a -epo/+pna and the AO drops from what appears to be an astounding +5 over the next few days. NAO progs show a dip to negative around that period but I'm not holding my breath.

The short story is it appears our milder period is going to be transient. But as you said, the second half of march becomes hostile for snow by default. Even the good setups need to be great or excellent to have a shot.

Link to comment
Share on other sites

I have a question for people who are knowledgeable about the models.  I'm not sure where to put this, so mods please move if necessary.

 

Looking back over the model runs for the last storm, it looks like all models had a pretty good handle on the storm on Tuesday, then the American models pulled a head-fake on Wednesday before recovering Wednesday night.  The foreign models were more consistent.  I've noticed the difference in run-to-run consistency before, especially when comparing the NAM and RGEM.  Even when the RGEM ends up with a worse solution, it's usually more consistent.

 

Is this primarily caused by the different data assimilation methods?  I would imagine that the 4DVAR, etc. assimilation methods used by foreign models might be less sensitive to errors / statistical noise in observations.  Could the head-fake on Wednesday have been caused, for example, by some misleading obs during the day that were effectively given more weight by the American models?

 

It was a very anomalous setup. One that the Nam will never do well in. The bigger the system the worse the nam does with synoptics. You see it all time with strong miller A's and B's. The nam will be very jumpy an unreliable beyond a day or 2. This setup was a ridiculous moisture feed + arctic boundary + multiple waves to resolve. Tricky stuff. 

 

GFS is hard to say. Seemed to key on a stronger lead and weaker second but bounced back to reality after a hiatus. None of us really know the gfs's strengths and weaknesses. It will take a few years to better understand what it does and doesn't do well with. Overall it's still a pretty good improvement over the old gfs through 4 days or so. 

Link to comment
Share on other sites

why all the negativity about snow chances in march, it has happened before and will happen again. and considering the chances for a -nao to develop look better, then we still have at least 50-50 chance.That's pretty good in my book.

 

I didn't think I was being negative at all. I showed that we are moving back to a colder pattern but pointed out late March requires a lot more help than early March. 

 

I can extrapolate this 16 day plot and say the -epo/-nao/+pna and 50/50 low would not let the bowling ball cut.

 

gfs_z500a_nhem_53.png

Link to comment
Share on other sites

NAO was supposed to be headed negative by early march after it was supposed to in Feb and after it was supposed to in Jan. Skeptical.  Not sure why we should be optimistic about snow after Mar 15 either way. 

Link to comment
Share on other sites

I didn't think I was being negative at all. I showed that we are moving back to a colder pattern but pointed out late March requires a lot more help than early March. 

 

I can extrapolate this 16 day plot and say the -epo/-nao/+pna and 50/50 low would not let the bowling ball cut.

 

gfs_z500a_nhem_53.png

bob i respect you greatly for your expertise and attitude, I did not mean to be disrespectful, i just had a different interpretation of what you said, My apologies did not mean to offend, your right about the above graphic thanks 

Link to comment
Share on other sites

bob i respect you greatly for your expertise and attitude, I did not mean to be disrespectful, i just had a different interpretation of what you said, My apologies did not mean to offend, your right about the above graphic thanks 

I'd definitely trust the d16 op GFS. The models are so good at range.

 

4pwGXgU.gif

Link to comment
Share on other sites

I have a question for people who are knowledgeable about the models.  I'm not sure where to put this, so mods please move if necessary.

 

Looking back over the model runs for the last storm, it looks like all models had a pretty good handle on the storm on Tuesday, then the American models pulled a head-fake on Wednesday before recovering Wednesday night.  The foreign models were more consistent.  I've noticed the difference in run-to-run consistency before, especially when comparing the NAM and RGEM.  Even when the RGEM ends up with a worse solution, it's usually more consistent.

 

Is this primarily caused by the different data assimilation methods?  I would imagine that the 4DVAR, etc. assimilation methods used by foreign models might be less sensitive to errors / statistical noise in observations.  Could the head-fake on Wednesday have been caused, for example, by some misleading obs during the day that were effectively given more weight by the American models?

ECMWF's higher resolution at 2m helps it detect and avoid headfakes. Heads tend to be located between 1m and 2m, where headfake formation typically occurs. 

Link to comment
Share on other sites

^ I wouldn't be surprised to see it snow again even IMBY.. Mar 25 is still a good ways off. :P

 

Could definitely have a better shot of a 'sizable' elevation event than normal. I wouldn't even totally rule out another mod event for the immediate area but it will take luck and then some. Probably some cold rain in our future either way. 

 

It would take a lot of things to line up just right to get a moderate event in the general area after March 15.  And since it looks like we're in a normal/slightly above period for the next week or so, I think that's the timeframe we're looking at.  That said, it wouldn't surprise me at all if at least a few things came together to threaten the immediate area with some slop but gave the favored areas a 3-5" slushing in the March 16-23 period.

Link to comment
Share on other sites

I have a question for people who are knowledgeable about the models.  I'm not sure where to put this, so mods please move if necessary.

 

Looking back over the model runs for the last storm, it looks like all models had a pretty good handle on the storm on Tuesday, then the American models pulled a head-fake on Wednesday before recovering Wednesday night.  The foreign models were more consistent.  I've noticed the difference in run-to-run consistency before, especially when comparing the NAM and RGEM.  Even when the RGEM ends up with a worse solution, it's usually more consistent.

 

Is this primarily caused by the different data assimilation methods?  I would imagine that the 4DVAR, etc. assimilation methods used by foreign models might be less sensitive to errors / statistical noise in observations.  Could the head-fake on Wednesday have been caused, for example, by some misleading obs during the day that were effectively given more weight by the American models?

I have long suspected that the NAM (and frankly the RAP/HRRR as well) because of the use of partial cycling.  Essentially, twice a day the NAM goes backward in time 12 hours and restarts from a GFS IC, performs a sequence of DA/short forecast cycles, catches up to real time, and goes from there.  Since the GFS and NAM are very different models and very different bias characteristics, I suspect this leads to various issues.  On the other hand, this helps avoid model drift, accumulation of serious biases, and aids in the use of satellite data assimilation (since it can use the global bias correction coefficients).

 

In terms of the GFS, it's hard to say and would require significant digging.  As an aside and I've harped on this before....while it's true that all things being equal, 4DVAR > 3D assimilation, there is a lot going on under the hood that can make 4DVAR inferior.  For example, 4DVAR requires the linearization (and transposing, the adjoint) of the model.  This almost impossible for components such as moist physics.  Furthermore, running 4DVAR is insanely expensive from a computational standpoint, and most centers that do it have to make the tradeoff to run the assimilation at much lower resolution.  I think people would be surprised to learn the effective resolution of the ECMWF and UKMO analysis systems.

 

Another thing worth mentioning is that the Canadians no longer use 4DVAR for their operational initialization, but instead utilized an ensemble-based method called 4D EnVar.  This method has other issues but does not need the linearized model to do the assimilation.  NCEP is in the final stages of preparing their own 4D EnVar for implementation within a year (along with more GFS model changes....sorry Bob!).

 

The ECMWF model is typically less jumpy for the simple reason that they have the best modeling and assimilation system in the world, by far.  This isn't simply "just because they have 4DVAR".  The navy has 4DVAR, and they are one of the poorer scoring operational centers in the world.  The details are very important.  They also make use of a longer assimilation window compared to the other centers.  This really helps constraint the model trajectory (and not just a single snapshot in time).

 

I think the Europeans do a much better job in terms of observation handling, bias correction, and quality control.  While the US has actually been a bit of a leader and pioneer with respect to satellite bias correction, they have lagged behind in terms of more general quality control and bias correction for more conventional observations.  In other words, I think the US probably uses observations a bit too aggressively. 

 

The model also matters, since data assimilation is an incremental (and cumulative procedure).  The better the model, the less "work" the data assimilation has to do.

 

To summarize, it's complicated.

Link to comment
Share on other sites

I have long suspected that the NAM (and frankly the RAP/HRRR as well) because of the use of partial cycling.  Essentially, twice a day the NAM goes backward in time 12 hours and restarts from a GFS IC, performs a sequence of DA/short forecast cycles, catches up to real time, and goes from there.  Since the GFS and NAM are very different models and very different bias characteristics, I suspect this leads to various issues.  On the other hand, this helps avoid model drift, accumulation of serious biases, and aids in the use of satellite data assimilation (since it can use the global bias correction coefficients).

 

In terms of the GFS, it's hard to say and would require significant digging.  As an aside and I've harped on this before....while it's true that all things being equal, 4DVAR > 3D assimilation, there is a lot going on under the hood that can make 4DVAR inferior.  For example, 4DVAR requires the linearization (and transposing, the adjoint) of the model.  This almost impossible for components such as moist physics.  Furthermore, running 4DVAR is insanely expensive from a computational standpoint, and most centers that do it have to make the tradeoff to run the assimilation at much lower resolution.  I think people would be surprised to learn the effective resolution of the ECMWF and UKMO analysis systems.

 

Another thing worth mentioning is that the Canadians no longer use 4DVAR for their operational initialization, but instead utilized an ensemble-based method called 4D EnVar.  This method has other issues but does not need the linearized model to do the assimilation.  NCEP is in the final stages of preparing their own 4D EnVar for implementation within a year (along with more GFS model changes....sorry Bob!).

 

The ECMWF model is typically less jumpy for the simple reason that they have the best modeling and assimilation system in the world, by far.  This isn't simply "just because they have 4DVAR".  The navy has 4DVAR, and they are one of the poorer scoring operational centers in the world.  The details are very important.  They also make use of a longer assimilation window compared to the other centers.  This really helps constraint the model trajectory (and not just a single snapshot in time).

 

I think the Europeans do a much better job in terms of observation handling, bias correction, and quality control.  While the US has actually been a bit of a leader and pioneer with respect to satellite bias correction, they have lagged behind in terms of more general quality control and bias correction for more conventional observations.  In other words, I think the US probably uses observations a bit too aggressively. 

 

The model also matters, since data assimilation is an incremental (and cumulative procedure).  The better the model, the less "work" the data assimilation has to do.

 

To summarize, it's complicated.

 

Thanks for the thoughtful and informative reply, but paulythegun nailed it.

 

ECMWF's higher resolution at 2m helps it detect and avoid headfakes. Heads tend to be located between 1m and 2m, where headfake formation typically occurs. 

 

Seriously, what you said makes a lot of sense.  I'm looking forward to the next GFS.  What's the reason for initializing the NAM from the GFS initial conditions 12 hours ago? 

Link to comment
Share on other sites

Thanks for the thoughtful and informative reply, but paulythegun nailed it.

 

 

Seriously, what you said makes a lot of sense.  I'm looking forward to the next GFS.  What's the reason for initializing the NAM from the GFS initial conditions 12 hours ago? 

 

Regional models have difficulty maintaining the largest scales/longest wavelengths (think continental scale and beyond even) when they are allowed to cycle on their own states.  The NAM, RAP, and HRRR all use some form of partial cycling for this very reason.  It also helps with satellite radiance assimilation, which requires bias correction, which requires large (global) samples of data to do correctly/consistently.

Link to comment
Share on other sites

Euro ensembles took it a step further this afternoon. Pretty strong cold/-epo/+pna/-nao signal showing up beginning on the 18th or so. NAO goes neg on the 15th but eastern trough doesn't build for a few days. Still way out there but staying put in time and getting stronger as we move forward. 

Link to comment
Share on other sites

The ensembles have been pretty darn good at long leads this winter so I'm fairly optimistic even for being mid to late March. Good stuff Bob.

huh? we've seen them fail on an -nao repeatedly.. they still have a greenland vortex till like d10. 

Link to comment
Share on other sites

They've been horrible with the nao start to finish. Good with the epo/pna though. We have some warmer days ahead to ponder the colder days further ahead. 

My dog could probably have predicted the overall epo/pna state given its persistence. Chasing an -NAO d 12-16 seems borderline ridiculous at this point, esp using selective memory to do so. It might happen.. but we've done fine without.. so who cares other than needing magic late march. 

Link to comment
Share on other sites

My dog could probably have predicted the overall epo/pna state given its persistence. Chasing an -NAO d 12-16 seems borderline ridiculous at this point, esp using selective memory to do so. It might happen.. but we've done fine without.. so who cares other than needing magic late march. 

 

Just pointing out agreement. If it holds firm in time and we get inside of 7 days we can discuss more seriously. 

 

18 ens members have 2" or more of snow through d15. Couple nice hits....[ducks and runs]

Link to comment
Share on other sites

Just pointing out agreement. If it holds firm in time and we get inside of 7 days we can discuss more seriously. 

 

18 ens members have 2" or more of snow through d15. Couple nice hits....[ducks and runs]

Ugh,  if so, I'll have another article to write.  I was hoping I was done. 

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...