23 minutes ago, USAwx said:

Anyone have any model runs we can use from previous storms? it's fine to do this with re-analysis but would love to do it based on model projections 

 

It's interesting what you're doing because your system is a type of statistical forecasting system as opposed to being deterministic. 

 

This is exactly what they do with climate models, they feed them actual observations from the past, let's say 1950, run the model and see if it was able to predict the "future" accurately.

 

Which then we would be able to confirm by using historical data that occured after the starting point (1950) to the end point of the climate forecast (let's say 2020.)

 

I only bring climate up because it's statistical modeling like the type you're trying to engage in here. It can be a very useful tool. 

 

You're basically testing your system on real world historical data. Problem is I don't know if you can find past model forecasts going back that many years to find out how they did with all these snowstorms. 

 

 

24 minutes ago, Sundog said:

 

It's interesting what you're doing because your system is a type of statistical forecasting system as opposed to being deterministic. 

 

This is exactly what they do with climate models, they feed them actual observations from the past, let's say 1950, run the model and see if it was able to predict the "future" accurately.

 

Which then we would be able to confirm by using historical data that occured after the starting point (1950) to the end point of the climate forecast (let's say 2020.)

 

I only bring climate up because it's statistical modeling like the type you're trying to engage in here. It can be a very useful tool. 

 

You're basically testing your system on real world historical data. Problem is I don't know if you can find past model forecasts going back that many years to find out how they did with all these snowstorms. 

 

 

So is that good or bad. 

 

And you can't use this going back because models from 1995 would be totally different than now. 

 

 

20 minutes ago, USAwx said:

So is that good or bad. 

 

And you can't use this going back because models from 1995 would be totally different than now. 

 

 

 

I think your best bet is to stick with the re-analysis because you remove all the variables the different models and their changes over time introduce. 

 

Re-analysis is what actually evolved and occurred so that's the best way to test your system. 

 

Just pick a consistent time prior to the storm (5 days?) and roll forward and see how the system evolved to see if your numbers more often than not produce the results that are being predicted. 

55 minutes ago, Sundog said:

 

I think your best bet is to stick with the re-analysis because you remove all the variables the different models and their changes over time introduce. 

 

Re-analysis is what actually evolved and occurred so that's the best way to test your system. 

 

Just pick a consistent time prior to the storm (5 days?) and roll forward and see how the system evolved to see if your numbers more often than not produce the results that are being predicted. 

I'm gonna do it my way

 

Use the h5 day of those intervals (8 5 3) and also projected for the vort we are Watching

 

 

Euro winter 500.

Updated winter idea coming soon.

 

Soon we will be putting this to the test. 

Oh well, maybe for the 2025-26 winter you can try it. 

The system works. The issue is the shifting elements at various lead times. Unfortunately we have way to much volatility even at day 5. 

On 2/25/2024 at 8:03 AM, Andrew said:

Oh well, maybe for the 2025-26 winter you can try it. 

hard to believe the great winter of 1995-96 will be 30 years old by then

Beautiful day 

Still zero storms to test this

Maybe this year we can really test this out

40 minutes ago, USAwx said:

Maybe this year we can really test this out

1996...2006...2016...2026...

1 minute ago, uncle w said:

1996...2006...2016...2026...

just like ghost busters after life

On 9/14/2023 at 1:03 PM, USAwx said:

I have devised a scale to predict how likely it will be a megalopolis city sees a storm 10 inches or greater this winter. 

 

 

I am using the 5 core components to predicting a Northeast Snowstorm. 

 

1)  NAO

2) PNA

3) EPO

4) Cold Source

5) 50/50

 

The above 5 components get the following scale weight, with a wild card added for things such as lead energy in the trough, kickers, complexity of the pattern, something to that affect. 

 

1)  NAO 9

2)  PNA 4

3)  EPO 10

4)  Cold Source  8 

5)  50/50 16

6)  Wild Card 3  points 

 

Add them up divide by 50 multiple times 100 and that's your PERCENTAGE of seeing a 10 plus inch storm in one of the Megalopolis cities. 

 

The worse the component, the lower the score.

 

I will rate the storm at the following intervals:

 

8 days (192 HOURS)

5 days (120 HOURS)

3 days (72 HOURS)

 

The rating is not static over the 4 Megalopolis cities

 

Each may do better based on different variables.  I.E.   The cold source may be bunk for DCA, but serviceable for NYC.  The NAO may be too west for NYC, but good for DCA.  You get the point.

 

AS SUCH, each city must be scaled separately based on historical storms.

 

If you have access to old maps, we can try this out to see how it would work.

 

I think it will help us sort out the model windshield wiper effect. 

 

 

 

 

 

 

Right now we are focused on the 16th. It's far out for this, so let me use the end of the 18z Euro which is only 3 days before that period. These players would ALL be ingredients for the storm

NAO - There is some ridging into Greenland due to the 50/50 low, but it's pretty tame. This may be why models continue to show this for the most part as progressive. Out of 9, I can't go higher than 3 on this right now

PNA - Rock solid beautiful. The Ridge is building still at this our, and is very clean. Max out 4.

EPO - Not much here, it's more of a WPO, but that Ridging is pretty high into western Canada. I'll go 5.

Cold - There is deep cold North into Canada, it hasn't come into the Conus yet. This would improve in the next frames. Cold enough to SNOW though and we are in January now. I'll go 4 here.

50/50 - Our most important feature East Coast Snow. One low just moved out but you have low heights over this area. It's not amazing, but it's def. good enough to prevent a cutter. I'll go 10 here.

Wild Card - Seems like it wants to try to snow this winter at least, so let's give an extra 1.5 to the mix, and MLK weekend, as history has shown us, is historically pretty good time to snow

OVERALL SCORE:

27.5/50 *100 = 55% Chance at this large lead time of 10 inches or more for NYC area. That's a pretty strong signal.

goes down to the south and maybe up a little N and E.

1 hour ago, USAwx said:

Right now we are focused on the 16th. It's far out for this, so let me use the end of the 18z Euro which is only 3 days before that period. These players would ALL be ingredients for the storm

NAO - There is some ridging into Greenland due to the 50/50 low, but it's pretty tame. This may be why models continue to show this for the most part as progressive. Out of 9, I can't go higher than 3 on this right now

PNA - Rock solid beautiful. The Ridge is building still at this our, and is very clean. Max out 4.

EPO - Not much here, it's more of a WPO, but that Ridging is pretty high into western Canada. I'll go 5.

Cold - There is deep cold North into Canada, it hasn't come into the Conus yet. This would improve in the next frames. Cold enough to SNOW though and we are in January now. I'll go 4 here.

50/50 - Our most important feature East Coast Snow. One low just moved out but you have low heights over this area. It's not amazing, but it's def. good enough to prevent a cutter. I'll go 10 here.

Wild Card - Seems like it wants to try to snow this winter at least, so let's give an extra 1.5 to the mix, and MLK weekend, as history has shown us, is historically pretty good time to snow

OVERALL SCORE:

27.5/50 *100 = 55% Chance at this large lead time of 10 inches or more for NYC area. That's a pretty strong signal.

goes down to the south and maybe up a little N and E.

@USAwx Joshua! Now we are getting down to business. Kiss The Godfather's weather ring!

FYE

I am revising the predictive nature of this scale

i think 6 inches or more, not 10, is a better match

i scaled the storm on the 26th three days prior, it was 75% match for NYC to hit 10 inches, which was not really in the realm, but 6 was and that would have been a kick ass forecast at 3 days

41 minutes ago, USAwx said:

FYE

I am revising the predictive nature of this scale

i think 6 inches or more, not 10, is a better match

i scaled the storm on the 26th three days prior, it was 75% match for NYC to hit 10 inches, which was not really in the realm, but 6 was and that would have been a kick ***** forecast at 3 days

I like the 6" for the scale better. I would think if the % is high, like 90% for 6", then more than 6 is very possible

Just now, FrankPizz said:

I like the 6" for the scale better. I would think if the % is high, like 90% for 6", then more than 6 is very possible

agree strongly.