Wintry April Fools? (1-2) Surprise Storm

[GaWx]

I think the best thing to do is to assume your area will get nothing like most models show for most areas (and this is with a cold bias that prevails on most models) and then just hope for a surprise as the thread title says. There's no excuse for whining if this doesn't pan out, especially since it is early April. Keep expectations low or else.
My personal expectation as of now is that there will likely be some areas of the Carolinas and N GA that see big wet flakes falling, but that most of these areas would have little or no accumulations although a small portion could see more than a dusting and up to as much as a few inches similar to what happened with the early March storm. So, for example, maybe only the area around Charlotte gets more than a dusting. Or perhaps only around GSP or maybe even both. But be prepared for no accumulations anywhere.

 

[Webberweather53]

12Z EPS ? for your viewing pleasure: Charlotte is in sweet spot with mean of 1”, but keep in mind EPS cold bias meaning take this with a grain:
View attachment 18230

Not sure there's really a "cold bias" here given that we're well inside the medium range now and other models like the ICON that are actually biased warm in the BL are showing virtually the same solution.

 

[Broken024]

NAM is running, getting the feeling of disappointment already

Not sure. The front genesis schmensis is looking good.

 

[Webberweather53]

Here's a close-up of the 0z EPS on NC, SC, and NE GA.

As noted earlier, Charlotte is in the "sweet" spot w/ 1" mean although on one hand this assumes 10:1 SLRs (biases accums too high) but there's also a lot of internal and member-to-member variability (which biases the mean too low because the spread is still fairly large).

I'll take whatever I can get.

 

[pcbjr]

schmensis

LMAO, Webb ...

 

[Myfrotho704_]

NAM slowed the energy down just a tad based off this Z500 map

 

[Myfrotho704_]

Not sure. The front genesis schmensis is looking good.

Nani

 

[GaWx]

Not sure there's really a "cold bias" here given that we're well inside the medium range now and other models like the ICON that are actually biased warm in the BL are showing virtually the same solution.

The EPS cold bias has been well documented by Radiant to exist even just 1-2 days out. Does that mean the mean will be too cold in this case, too? Not necessarily as we know that's not how biases work. I'm talking about the general VERY persistent cold bias that starts well before the medium range.

 

[Webberweather53]

The EPS cold bias has been well documented by Radiant to exist even just 1-2 days out. Does that mean the mean will be too cold in this case, too? Not necessarily as we know that's not how biases work. I'm talking about the general VERY persistent cold bias.

This cold bias is really pronounced in the extended range (day 7+), inside the medium range not so much and the EPS is virtually in line w/ other guidance like the ICON that's usually way too warm, I'm just urging you not to broadbrush a NWP bias in every situation and at every time step because there are a multitude of confounding factors that mask or invigorate any bias at a given point in time. I think using prior knowledge about how storms like this often turn out in addition to taking into account how advantageous it often is to have a few extra thousand feet of elevation allows one to arrive at the conclusion that the axis of the EPS snow mean is probably too far SE verbatim but that doesn't necessarily mean it's cold biased in this particular situation, even if a more NWly solution verifies.

 

[GaWx]

Here's a close-up of the 0z EPS on NC, SC, and NE GA.

As noted earlier, Charlotte is in the "sweet" spot w/ 1" mean although on one hand this assumes 10:1 SLRs (biases accums too high) but there's also a lot of internal and member-to-member variability (which biases the mean too low because the spread is still fairly large).

I'll take whatever I can get.

View attachment 18232

I don't know about internal variability biasing the mean too low. But let's just assume the total snow shown were to turn out right when adding up the entire shaded area. IF SO, there'd likely be a small portion of the shaded area getting a good bit more than 1" (say perhaps as much as 3-5" or so) while most of the shaded area gets less than what is shown here with many spots getting a dusting or less. So, whereas a few areas would get much greater than the max that is shown here, most would get less than what is shown for their area on this map. So, the maximum shown here of 1" would likely be biased too low while at the same time the mean isn't biased low. That's how these things seem to work out a lot of times.

 

[Myfrotho704_]

@ForsythSnow should like the new NAM, I personally don’t like it lol

 

[Webberweather53]

NAM slowed the energy down just a tad based off this Z500 map
View attachment 18233

Really can't afford much, if any slowing along & NW of the I-85 corridor in NC, little bit more wiggle room for those in SC & GA. Events like this w/ modest mid-level warm advection often start on time or earlier than anticipated as the precipitation on the NW side of the coastal low is more expansive/intense than modeled due to the aforementioned mishandling of weak WAA (among other things)

 

[Webberweather53]

I don't know about internal variability biasing the mean too low. But let's just assume this mean were to turn out right. IF SO, there'd likely be a small portion of the shaded area getting a good bit more than 1" (say perhaps as much as 3-5" or so) while most of the shaded area gets less than what is shown here with many spots getting a dusting or less. So, whereas a few areas would get much greater than the max that is shown here, most would get less than what is shown for their area on this map. So, the maximum shown here of 1" would likely be biased too low while at the same time the mean isn't biased low. That's how these things seem to work out a lot of times.

Internal variability amongst ensemble members biases the mean low because more spread = less signal, that's just basic statistics... A probability matched mean product would likely capture the amplitude correctly wrt 10:1 SLR snow in the models but after accounting for lower SLRs, warm ground, etc. that do have some effect (which is often overstated) on SLR, the actual mean here may be pretty close to reality.

 

[Webberweather53]

@ForsythSnow should like the new NAM, I personally don’t like it lol

Precipitation type products may be a bit deceiving in this case w/o seeing soundings etc. but the total snowfall shows you're in the thick of it. P-type algorithms & most probable p-type output can be pretty wonky in a very marginal setup like this esp when p-type is even more dependent on p-rate that's extremely hard to predict beyond the e-folding time of mesoscale features (several hours or so)

 

[Myfrotho704_]

Really can't afford much, if any slowing along & NW of the I-85 corridor in NC, little bit more wiggle room for those in SC & GA. Events like this w/ modest mid-level warm advection often start on time or earlier than anticipated as the precipitation on the NW side of the coastal low is more expansive/intense than modeled due to the aforementioned mishandling of weak WAA (among other things)

Yeah, also 3km seems to be doing the same thing slowing it down a tad but not crazily like the 12km

 

[GaWx]

Internal variability amongst ensemble members biases the mean low because more spread = less signal, that's just basic statistics...

The mean for any location is the sum of the 52 members divided by 52. So, let's say, for example, that Charlotte is given 10" by one member, 5" by 2 members, 1" by 32 members, and none by 17 members. The 52 members add to 52". Then 52" divided by 52 = 1". Where is the low bias?

 

[Webberweather53]

Internal variability amongst ensemble members biases the mean low because more spread = less signal, that's just basic statistics... A probability matched mean product would likely capture the amplitude correctly wrt 10:1 SLR snow in the models but after accounting for lower SLRs, warm ground, etc. that do have some effect (which is often overstated) on SLR, the actual mean here may be pretty close to reality.

More or less what I'm saying is the mean snowfall products may be right (more aptly less wrong) for the wrong reasons.

 

[Tarheel1]

#EXCITED! 18z NAM

 

[NickyBGuarantee]

I think there will probably be more precip in this area where there is that forcing band over SC/NC line

 

[Webberweather53]

The mean for any location is the sum of the 52 members divided by 52. So, let's say, for example, that Charlotte is given 10" by one member, 5" by 2 members, 1" by 32 members, and none by 17 members. The 52 members add to 52". Then 52" divided by 52 = 1". Where is the low bias?

The low bias emerges from the standard deviation amongst the members, the higher the standard deviation/spread between members, the lower the mean is going to be in this case, the less significant/meaningful the results, and this also says the mean is going to deviate collectively more from each member in the ensemble, which = low bias or more generally, you actually get less signal. In this case, our "signal" is positive (> 0) snow output, thus more spread = less signal, less signal usually = less snow. But again, there are other confounding variables that are going to skew this figure back in the other direction towards less snow & perhaps completely overcompensate for this, thus we arrive at a similar conclusion anyway, just not for the right reason(s).

What's stated above is among the primary reasons why global ensemble mean snow in cases where a storm is actually coming to fruition, are practically useless for precise numbers inside a specific interval of time (usually inside 48 hours) because they're not properly dispersed inside the medium range which leads to a low bias in snow vs reality (Jan 16-18 2018 & Jan 3-4 2018 events are great case studies).

Again, this is basic statistics.