February 2016 Forecasts/Disco/Obs

[SnowGoose69]

Wave 1 is interesting. Could be some snow for the area especially north and west. Para has it tracking over SNJ. EPS has it tracking under NYC.

 

I don't see that being much of an issue, I think it ultimately will trend north.  I don't like SWFE type stems here when there is ridging out west, they usually tend to amplify further in the short range and end up further north

[MJO812]

AO forecast looks good

 

[donsutherland1]

My concern about the possible February 24-25 storm is that blocking will not yet be in place during the event, even as it's developing. The latest GFS ensemble forecast has most members still showing a positive AO during the timeframe in question. While some significant snowfalls have occurred with an AO+ during the second half of February, 78% of NYC's 6" or greater snowstorms during that period occurred with an AO-.

 

 

There's still a good degree of uncertainty. First, it is possible that the AO dives more quickly, as shown by a few ensemble members. Second, if the storm develops somewhat later than the consensus forecast, that could also allow for a colder solution.

 

Right now, if I had to venture a guess, interior sections would have a greater likelihood of seeing appreciable or greater snowfall (4" or more) than New York City and eastward unless blocking develops more quickly and/or the storm's development/progression is slower than currently modeled. One still can't rule out a more significant impact either to the coastal plain or its focus being much farther inland e.g., if blocking develops even more slowly than shown by most of the ensemble members.

[tdp146]

FWIW (and I know you're just parroting what you read from real meteorologists and its them that should know better, not you), there is very little evidence for the importance of "seasonal trends".  I.e., there is very little evidence that if models are behaving in a certain way in a given pattern, that you should expect them to continue to behave that way in a given pattern.  This is not entirely true - certain patterns can interact with a known bias of a model in predictable ways - but big picture, the notion that you have to adjust a models output because, for instance, it's shown a SE bias with the last several storms, is not science-based.

 

You'll make better forecasts if you don't make handwave adjustments to model output.   That's why quantitative consensus forecasts outperform humans these days (to wit, the Weather Channel (quantitative) outperforming the NWS (qualitative/quantitative mix) for high temp forecasts; the model consensus (qualitative) outperforming the NHC (quantitative/qualitative mix) for hurricane track forecasts).

Serious question-

 

Do you have evidence that the weather channel performs quantitative forecasts vs NWS performing qualitative and is there actually something that 'scores' their performance similar to how there are model verification scores? Just curious

[MJO812]

12z Nam is far south with the 1st low for Monday

 

[UlsterCountySnowZ]

12z Nam is far south with the 1st low for Monday

The key will be the piece of energy to the north, how it steers the low, and effects the HIGH, models are hinting at a more western look to the energy allowing the cold to stay locked in longer

[uncle W]

I could see nothing significant until mid March if then...by then there could be a negative ao and a not as positive nao...

 

[donsutherland1]

Serious question-

 

Do you have evidence that the weather channel performs quantitative forecasts vs NWS performing qualitative and is there actually something that 'scores' their performance similar to how there are model verification scores? Just curious

There's a site that tries to compare forecasts (http://www.forecastadvisor.com/), but its methodology has limits.

 

For example, an accurate temperature forecast is defined as one that falls within 3° of the actual temperature.

 

Consider the following scenario of two forecasters:

 

Forecaster 1:

Errors for the last 3 temperatures: 3°, 2°, 3° -- Accuracy would be scored as 100%

 

Forecaster 2:

Errors for the last 3 temperatures: 1°,4°, 2° -- Accuracy would be scored as 67%.

 

Yet, if one scored each forecaster on average error, Forecaster 2 (2.3°) would be better than Forecaster 1 (2.7°). Sum of the squares error would yield the same result. This is just for illustrative purposes, as one would be dealing with many more forecasts (the site uses past month and past year).

 

Put another way, the issue of accuracy is more complex than what the site portrays. To some extent, it may well depend on what the users are seeking. Some might want forecasts within a set parameter. Others might want forecasts as close to the actual outcomes as possible.

 

Also, the site doesn't examine strictly quantitative output e.g., MOS, vs. the short list of forecasting entities, so an evaluation of the net gain/loss from the human skill-human bias interaction can't be made. In theory, human input should add value otherwise, there's little point in going beyond automated output.

 

NHC did a study that covered the 2007-11 and 2012 hurricane seasons. In terms track, only the Florida State University Superensemble outperformed NHC (p.31). In terms of intensity, the NHC typically outperformed all the guidance, including the FSU Superensemble (pp.33-34). That's for the Atlantic basin. Interestingly enough, the FSU tool does not cover the eastern North Pacific basin. For that basin, the NHC's track forecasts were superior to all of the available guidance (p.42), as was the case with its intensity forecasts (pp.44-45).

 

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.363.4690&rep=rep1&type=pdf

 

So, at least with the limited data available (and I'm sure NWS has a lot of internal data), I don't think one can conclusively argue that automated guidance has reached the point where human input no longer provides value. With advances in computing and modeling, the value proposition may well change in the future.

[tdp146]

Thanks Don. I will have to bookmark and read in my spare time. 

[IsentropicLift]

http://videowall.accuweather.com/detail/videos/trending-now/video/2430839568001/east-coast-storm-to-form-next-

Want a good explanation

watch Bernies morning video....

For the record, he's calling for significant snows NW of 95, more towards I-81 in PA and up into upstate NY and New England.

[IsentropicLift]

The Western ridge is much stronger on this GFS run than 06z, should make for a more amplified solution down the road.

[IsentropicLift]

Digging to the Northern gulf instead of the TX panhandle. Major differences.

[IsentropicLift]

Still multiple pieces of energy running out ahead of the main system. 

[IsentropicLift]

The Northern stream is dropping down into the system. This is what Bernio was referring to in his video. It will help to steer the system up the coast instead of OTS.

[IsentropicLift]

This still looks nothing like the Euro. It's a somewhat in between. 

[IsentropicLift]

Light precip moving in early Wednesday morning. Looks too warm for snow.

[IsentropicLift]

Large swath of light rain over the area with pockets of freezing rain.

[IsentropicLift]

Good hit for interior New England, perhaps even Boston with the first surface low tracking east of the BM.

[IsentropicLift]

Then some light snows on the backside assoicated with the ULL as it slides OTS. This was a step towards a better outcome and I think we're still pretty far from reality.

[IsentropicLift]

You can see here how the GFS attempts to drop in some Northern stream energy which would help to drive it up the coast. It just didn't quite get the job done this run.

 

 

As compared to 06z where the Northern stream stayed up in Canada, helping to push the system OTS.

 

[IsentropicLift]

Some potential snow for the interior or some ice with the initial surface low

 

[IsentropicLift]

Then a flip to light snow further SE as the SLP slides east and the ULL moves through. As I mentioned, this was a good run for interior New England, perhaps even Boston.

 

[snowman19]

Some potential snow for the interior or some ice with the initial surface low

Looks like a brief mix on the front end NW of NYC before it flips to rain

[UlsterCountySnowZ]

Looks like a brief mix on the front end NW of NYC before it flips to rain

NW never flips to rain, what r u looking at?

[IsentropicLift]

The GGEM has a sneaky storm for Sunday night into Monday. Would be a quick few inches North of the low track.

[UlsterCountySnowZ]

Gfs was. 3-6" for interior... Grain of salt, 6 days out

[IsentropicLift]

GFS total snowfall. LI really cashes in with the ULL.

 

[IsentropicLift]

The GGEM looks like it will be West and more wrapped up.

[PB GFI]

GFS total snowfall. LI really cashes in with the ULL.

 

 

 

That second swath is what the PARA is picking up on

 

2nd SW .

 

[IsentropicLift]

The GFS is a long duration of light, wet snow. Ratios would be poor until the ULL swings through and we got into the better dynamics on the backside.