Hurricane Irene Model and Forecast Discussion

[Dsnowx53]

He's using Edouard as an example though, and that wasn't a quick recurve either-- but an excruciatingly slow one. I do agree that it's premature to commit to any track 4 days out.

Edouard started its northward turn at a muuuuch further east Longitude. If you take Edouard's track, and shift it westward to where Irene's near-term Longitude would be, there would be a Long Island landfall. Of course that's not really how science works, but my point still stands.

[phil882]

Any chance that W ATL ridge can build in more than modeled? It's late August, for chrissakes.

It might not matter that much despite my earlier thinking... The newest ECWMF has a huge kicker trough at 144hrs that forces the second trough (the one that picks up Irene) to be more progressive. If the trough had more time to dig down it would allow Irene to move north, but with everything being so progressive, it doesn't give much time for a northward pull. In essence we want our mid-level features to trend slower so that it gives Irene a chance to move up the coast before it gets pushed out. The ECWMF run tonight was a step in the opposite direction.

[Bozart]

I know the standard error in the NHC track at 5 days is some 250 miles

I hear a lot about that 250-mile average 5-day track error, as if the storm could as easily go one way or the other. I'd be curious to know if that average error isn't in fact disproportionately to one side of the progged track. (And we all know which side, too.)

[A-L-E-X]

It might not matter that much despite my earlier thinking... The newest ECWMF has a huge kicker trough at 144hrs that forces the second trough (the one that picks up Irene) to be more progressive. If the trough had more time to dig down it would allow Irene to move north, but with everything being so progressive, it doesn't give much time for a northward pull. In essence we want our mid-level features to trend slower so that it gives Irene a chance to move up the coast before it gets pushed out. The ECWMF run tonight was a step in the opposite direction.

hmmm it sounds like this is the one associated with what was that former pac s/w we were all waiting for.

[Dsnowx53]

Because most deep troughs become negatively tilted before they reach the EC. Also, you have a level of mid-latitude cyclone phasing when you have a deep trough...so that also sucks it NW even if the trough is not severely tilted negative.

Oh okay, makes sense. Plus, if the trough was deeper, it would be a bit slower by default, so the pattern would become less progressive.

[ORH_wxman]

I hear a lot about that 250-mile average 5-day track error, as if the storm could as easily go one way or the other. I'd be curious to know if that average error isn't in fact disproportionately to one side of the progged track. (And we all know which side, too.)

For EC storms, its easily east-biased...storms will almost always trend east compared to guidance inside of 3 days. We aren't at that time range yet, so we probably shouldn't jump to conclusions, but we are certainly heading that direction. This is just for EC storms though...further south, I cannot claim any expertise.

[Thunderdude]

Edouard started its northward turn at a muuuuch further east Longitude. If you take Edouard's track, and shift it westward to where Irene's near-term Longitude would be, there would be a Long Island landfall. Of course that's not really how science works, but my point still stands.

Very true but I remember the case was that Edward was make a move to NNW but it shifted NE or NNE then eventually more E and the storm quickly fell apart when it brushed the Cape.

[ORH_wxman]

Suck? The fact that billions of dollars in property loss might be avoided is 'suck'?

Unbelievable.

Anyone who starts discussing morals of weather in here will be banned quickly. I assume you didn't know the rules since you just joined this year, so you will be given a pass. But do not do this again...make a separate thread in off topic, we have a forum just for this type of stuff.

[QVectorman]

Model error in track

Best 2; magenta line along the bottom is the AEMI, Purple CMC2

Model error intensity

Best 3; HWF2 greyish green HWRF pink SHIP orange

[SnowGoose69]

Because most deep troughs become negatively tilted before they reach the EC. Also, you have a level of mid-latitude cyclone phasing when you have a deep trough...so that also sucks it NW even if the trough is not severely tilted negative.

The below illustrate the above well...

Bob

http://www.meteo.psu.edu/~gadomski/NARR/1991.html

Gloria

http://www.meteo.psu.edu/~gadomski/NARR/1985/us0926.php

[phil882]

Nice Links QVectorman! The thing that strikes me the most from those images is how utterly horrible the 12-24 hour time-frame has been for these models. You see the error values sorta stabilize beyond that point. This storm will have lots of forecasters scratching their heads for sometime to come.

[Dsnowx53]

The below illustrate the above well...

Bob

http://www.meteo.psu.../NARR/1991.html

Gloria

http://www.meteo.psu...1985/us0926.php

Oh wow, that is some deep trough that we had with Gloria. Makes sense, though, since that hit in late September, where it's more typical for northern stream troughs to be more amplified and plunge further south. To be fair, Gloria started off NE of where Irene is now, but that's beside the point.

[Ed Lizard]

The magnitude of the shift Eastward has been reducing, leaving me some optimism that the models are close to converging, and a near miss on Hatteras could mean a bigger storm for Long Island/New England.

I think it is almost coming down to wobbles West versus North in the next day as far as threat goes, with a slow bend Northeast seen.

/half full.

[Sunny and Warm]

Model error in track

Best 2; magenta line along the bottom is the AEMI, Purple CMC2

Model error intensity

Best 3; HWF2 greyish green HWRF pink SHIP orange

So, if I'm reading the track error chart correctly, the mean is actually closer to 125 miles and not 250 miles, and most of that occurs in the first 48 hours and flattens thereafter instead of occuring at days 4-5. Is that the correct interpretation? If so, why would models have most of its error front loaded, and not back loaded as we all surmise? I must be looking at this with 4am eyes.

[HurricaneJosh]

The magnitude of the shift Eastward has been reducing, leaving me some optimism that the models are close to converging, and a near miss on Hatteras could mean a bigger storm for Long Island/New England.

I think it is almost coming down to wobbles West versus North in the next day as far as threat goes, with a slow bend Northeast seen.

/half full.

Ha ha, I sort of agree with this. I guess I'm graspin' for them straws.

[Ed Lizard]

GFS individual petrurbations together might suggest a frustrating miss, but its close enough to stay optimistic.

[Dsnowx53]

Ha ha, I sort of agree with this. I guess I'm graspin' for them straws.

This is what I've been feeling as well. I've already told several people within the past two days that I expect a Hatteras grazing, and an essentially due north track, or just east of due north until it's just south of Central/Western Long Island, then the east turn starts to become a bit more dramatic, and it hits eastern Long island...maybe somewhere around the Hamptons or something. Since we know we don't have the 1938 LI express type of synoptics where the storm enters on a NNW heading, it might be "better off" for LI weather enthusiasts if the storm grazes or even just "misses" Hatteras.

Of course, I'm FROM Long Island and am currently residing here until school starts, so...

[SnowGoose69]

The magnitude of the shift Eastward has been reducing, leaving me some optimism that the models are close to converging, and a near miss on Hatteras could mean a bigger storm for Long Island/New England.

I think it is almost coming down to wobbles West versus North in the next day as far as threat goes, with a slow bend Northeast seen.

/half full.

Its not unusual to see a bit of a converging at various cycles during these systems. We saw it with Isabel in 2003 where every model had NYC/LI getting hit and then after a good 36 hours of near steadfast agreement they all began swinging back to the SW. I still have a feeling there may be a very strong eastward run by the models sometime around 36 hours from now. I'm continuing to lean towards something between Bob and Edouard as far as landfall which would put it into SE MA.

[QVectorman]

So, if I'm reading the track error chart correctly, the mean is actually closer to 125 miles and not 250 miles, and most of that occurs in the first 48 hours and flattens thereafter instead of occuring at days 4-5. Is that the correct interpretation? If so, why would models have most of its error front loaded, and not back loaded as we all surmise? I must be looking at this with 4am eyes.

This is only for Irene...model performance varies from storm to storm. There is never one model that has the right answer every time.

[Bozart]

For EC storms, its easily east-biased...storms will almost always trend east compared to guidance inside of 3 days.

Wouldn't that imply, then, that for an EC storm, we can take the 5-day forecast position and just tack on that extra distance east as a matter of routine?

Of course, in this case, we'd have still been way too far west.

[Dsnowx53]

Because most deep troughs become negatively tilted before they reach the EC. Also, you have a level of mid-latitude cyclone phasing when you have a deep trough...so that also sucks it NW even if the trough is not severely tilted negative.

Additionally, this also begs the question of how exactly (or at all) the GFS bias even comes into play. The GFS can certainly have a progressive bias with many features, including being fast with the northern stream, which would obviously help it in forecasting a further east track. The Canadian has a progressive bias, too. However, doesn't the GFS also sometimes have a bias to dig northern stream features a bit too much? If that's the case, and we want a more amplified trough, then maybe the "bias" card of the GFS can't be played as much.

Or maybe since the trough is so far north to begin with, the trough being a tad "diggier" doesn't really matter, and simply its progressive bias with that trough would aid it in being further east. The Euro is often a much slower model than the GFS with timing to begin with, so that could explain why the Euro has continually been further west than the GFS. The fact that the Euro seems to be picking up on a kicker of some sorts is certainly not a good sign, but there is plenty of time for a feature like that to change.

[ORH_wxman]

Wouldn't that imply, then, that for an EC storm, we can take the 5-day forecast position and just tack on that extra distance east as a matter of routine?

Of course, in this case, we'd have still been way too far west.

4-5 days is too far out to do this IMHO. I think this is appropriate for inside of 3 days. I say this because we've seen storms at 5 days out go west then go east after we hit the 2-3 day threshold.

[A-L-E-X]

Its not unusual to see a bit of a converging at various cycles during these systems. We saw it with Isabel in 2003 where every model had NYC/LI getting hit and then after a good 36 hours of near steadfast agreement they all began swinging back to the SW. I still have a feeling there may be a very strong eastward run by the models sometime around 36 hours from now. I'm continuing to lean towards something between Bob and Edouard as far as landfall which would put it into SE MA.

SG, I remember when you posted those Isabel charts that showed a Cat 3 landfall around Belmar, NJ..... scary stuff indeed!

[ORH_wxman]

Additionally, this also begs the question of how exactly (or at all) the GFS bias even comes into play. The GFS can certainly have a progressive bias with many features, including being fast with the northern stream, which would obviously help it in forecasting a further east track. The Canadian has a progressive bias, too. However, doesn't the GFS also sometimes have a bias to dig northern stream features a bit too much? If that's the case, and we want a more amplified trough, then maybe the "bias" card of the GFS can't be played as much.

Or maybe since the trough is so far north to begin with, the trough being a tad "diggier" doesn't really matter, and simply its progressive bias with that trough would aid it in being further east. The Euro is often a much slower model than the GFS with timing to begin with, so that could explain why the Euro has continually been further west than the GFS. The fact that the Euro seems to be picking up on a kicker of some sorts is certainly not a good sign, but there is plenty of time for a feature like that to change.

The GFS bias I think has little play here IMHO...the Euro is a better scoring model WRT Tropical Cyclones at this time range, but its still pretty flawed at D4-5. The Euro has plenty of viable reasons that it could be wrong, but because its a worse model is not one of them...its better.

[SnowGoose69]

Wouldn't that imply, then, that for an EC storm, we can take the 5-day forecast position and just tack on that extra distance east as a matter of routine?

Of course, in this case, we'd have still been way too far west.

Basically if you live in LI, CT, RI, or MA you generally want to find yourself near or west of the forecast landfall point at days 3-5. If you had 100 storms forecast to come ashore in this region, probably 85 or so would trend eastward in some form from the initial track at 96 or 120 hours. As a result, places like LGA, JFK, Farmingdale, Norwalk and Greenwich I'd feel pretty good right now if I lived there...you may have a very windy and rainy day on Sunday into Sunday night but the odds you're going to see total destruction or devastation solely based on past storms and how these usually come up the coast is very low.

[CharlestonNYC]

Wouldn't that imply, then, that for an EC storm, we can take the 5-day forecast position and just tack on that extra distance east as a matter of routine?

Of course, in this case, we'd have still been way too far west.

Seriously...the whole "it sucks to be in the bullseye x days out" theory I truthfully believe has some merit. Be it with a computer model in the winter, or NHC in the summer with TCs, for the mere fact that weather forecasting is still inherently prone to such great error, you can almost guarantee what a model or TC track prediction shows 4-5 days out will NOT be what will happen. It's sad but funny and true at the same time. The second NHC put Charleston in Irene's bullseye 5 days before landfall, I stuck a fork in this storm.

[HurricaneJosh]

Seriously...the whole "it sucks to be in the bullseye x days out" theory I truthfully believe has some merit. Be it with a computer model in the winter, or NHC in the summer with TCs, for the mere fact that weather forecasting is still inherently prone to such great error, you can almost guarantee what a model or TC track prediction shows 4-5 days out will NOT be what will happen. It's sad but funny and true at the same time. The second NHC put Charleston in Irene's bullseye 5 days before landfall, I stuck a fork in this storm.

Good points.

[SnowGoose69]

Seriously...the whole "it sucks to be in the bullseye x days out" theory I truthfully believe has some merit. Be it with a computer model in the winter, or NHC in the summer with TCs, for the mere fact that weather forecasting is still inherently prone to such great error, you can almost guarantee what a model or TC track prediction shows 4-5 days out will NOT be what will happen. It's sad but funny and true at the same time. The second NHC put Charleston in Irene's bullseye 5 days before landfall, I stuck a fork in this storm.

I'd like to go back and see how many storms that the initial Day 5 landfall spot in the U.S. ever verified within 50 miles. I'm sure even the few which did the track probably shifted several times in between. Hugo I think could be one of the few where the Day 5 forecast was almost correct.

[Bozart]

Seriously...the whole "it sucks to be in the bullseye x days out" theory I truthfully believe has some merit.

It only has merit to the extent that the error could as likely go one direction as the other. But that doesn't seem to be the case with EC hurricanes. If (and I emphasize if) the average error is known to be heavily biased in one direction, and if that bias is quantified, it seems to me it becomes useful empirical data to be weighted legitimately into forecast thinking.

[SnowGoose69]

FYI, new NHC landfall point is around the twin forks of LI...about 50 miles west of Bob.