2024 Global Model Runs Discussion (Out thru day 16)

[Steve]

https://i.imgur.com/TSUwZZU.png
Just like that

August 29th. Katrina, Ida.

[Steve]

13 weeks gets us into the first week of September, CSU calling for 23 NS and we have had 5. 18 storms in 13 weeks, not buying it, my 19 NS might be high but we will see. The math is the math . I would take my 19 down to 17 if I had a redo, less storms but same ACE.

Gonna take some simultaneous systems to get there. I'm not sure where we are as far as named storm days or hurricane days so far, but CSU called for 120 named storm days and 50 hurricane days. I'm assuming we're roughly 15-20 into NS and probably 5 or 6 into H days. If that assumption is close, there are 108 days until November 30th though we know that most of November won't feature activity (might be one or two ocean systems or in the Caribbean due to the La Nina). We would have 100 named storm days to go which is about a constant system between now and the end of the season adjusted by however many simultaneous storms there are to offset the days without any named storms.

I don't have an opinion one way or the other on this. There are times when bursts bring on 2, 3 or even 4 systems. But there are usually only a couple of these a season that even features those. So it's all a big ? to me until we hit 4-5 weeks from now and see what we are dealing with numbers wise.

As of 03:00 MT this morning .... climatology norm 1991 - 2020 in ()
18.5 NS days (11.9)
6.5 H days (2.2)
42.2 ACE (14.1 AVG)
4.5 MH days (0.4)

Thanks Toad. In other words we ain’t seen nothing yet. Also glad I was somewhat in the ranges and not just typing out of my ***.

[Category5Kaiju]

https://i.imgur.com/TSUwZZU.png
Just like that

Merely a single day ago, there were talks of the season busting with models not detecting anything in the future.

Alarm clock effect in play, as per usual.

[AnnularCane]

And there's the switch flip coming, just like every year

https://i.imgur.com/gQjTGAq.png

Are they really coming off that far north though? That doesn't look right.

[skyline385]

And there's the switch flip coming, just like every year

https://i.imgur.com/gQjTGAq.png

Are they really coming off that far north though? That doesn't look right.

Nah, its correct. This is a good thread on the topic

 https://x.com/WxTca/status/1823493478892122501

[Beef Stew]

I’ve noticed in recent years it really seems like the models do a very poor job at picking up the onset of peak season. I’m basing this off nothing but memory, but most of the activity we’ve seen happen in late August/early September wasn’t really picked up on until ~5 days in advance- and then suddenly there’s multiple areas of interest and even some significant storms in the making. This time of year, things tend to happen fast.

While it looks like ensembles have perked up a bit today, even if the signal subsides I personally would bet that any modeled “calm” in late August/early September is likely an illusion. Ultimately, if I had to bet on it, I’d say that someone- be it the islands, GOM, or east coast- will be staring down the barrel of an approaching storm come two weeks from today.

[chaser1]

The general public (and fair extent of the media) simply do not understand the concept that Global Models do not have a memory. They are good at picking up strong signals and typically better forecasting with "all of the players on the field". The GFS had forever been solid in sniffing out genesis but with too many false negatives. That is why following the more subtle increase/decrease in the ensembles offers better hints toward an upswing in activity. Even still.... one day the ensembles see nothing, and next model run suddenly picks up the scent. Not much different from hunting dogs bred to hunt. Forecast models are our generations' Weather AI bloodhounds

[skyline385]

The general public (and fair extent of the media) simply do not understand the concept that Global Models do not have a memory. They are good at picking up strong signals and typically better forecasting with "all of the players on the field". The GFS had forever been solid in sniffing out genesis but with too many false negatives. That is why following the more subtle increase/decrease in the ensembles offers better hints toward an upswing in activity. Even still.... one day the ensembles see nothing, and next model run suddenly picks up the scent. Not much different from hunting dogs bred to hunt. Forecast models are our generations' Weather AI bloodhounds

Well said, numerical models are basically that - just numbers and equations. They do not know that climatologically conditions in the NATL will suddenly flip near August 20th to be very favorable and they are only able to calculate based on what was used in their initialization which is 10-15 days before the switch flip depending on the model which is in most cases generally unfavorable in the NATL. I expect this to change as we switch to hybrid and AI models in the future which are trained on past data but for now, we can't really do anything about the limitations of the model arising from them being completely numerical in nature.

[jlauderdal]

Gonna take some simultaneous systems to get there. I'm not sure where we are as far as named storm days or hurricane days so far, but CSU called for 120 named storm days and 50 hurricane days. I'm assuming we're roughly 15-20 into NS and probably 5 or 6 into H days. If that assumption is close, there are 108 days until November 30th though we know that most of November won't feature activity (might be one or two ocean systems or in the Caribbean due to the La Nina). We would have 100 named storm days to go which is about a constant system between now and the end of the season adjusted by however many simultaneous storms there are to offset the days without any named storms.

I don't have an opinion one way or the other on this. There are times when bursts bring on 2, 3 or even 4 systems. But there are usually only a couple of these a season that even features those. So it's all a big ? to me until we hit 4-5 weeks from now and see what we are dealing with numbers wise.

As of 03:00 MT this morning .... climatology norm 1991 - 2020 in ()
18.5 NS days (11.9)
6.5 H days (2.2)
42.2 ACE (14.1 AVG)
4.5 MH days (0.4)

Thanks Toad. In other words we ain’t seen nothing yet. Also glad I was somewhat in the ranges and not just typing out of my ***.

20+ is certainly not off the table but I look at the math and the idea its difficult to get everything to align for a NS.

[MetroMike]

According to Yaakov Cantor on X he replied to me saying the EPS has a bias of moving waves too far north off the African coast in the long range forecasts. So that posted graphic from earlier shows said bias.

[skyline385]

According to Yaakov Cantor on X he replied to me saying the EPS has a bias of moving waves too far north off the African coast in the long range forecasts. So that posted graphic from earlier shows said bias.

EPS has a bias of sometimes having waves form too north from brown water intensification inland but as we have seen this year in the thread started by TCA, the AEJ is abnormally north resulting in waves dropping much higher that what is normal for this time of year. When the EPS does it as a result of its bias, you can tell as waves drop down at TS strength or even hurricane strength sometimes, that is not the case for a majority of members on the run posted earlier so it is very likely that it not showing much bias this run especially since we know that the AEJ is abnormally elevated this year.

Am posting the same thread here as well, I would seriously recommend going through it as it provides a good explanation for the issues we saw in July and early August this year.

 https://x.com/WxTca/status/1823493478892122501

[Stratton23]

18z GFS starting to slowly see the switch now, has a little gulf mischief in the 10–12 day range lol

[Hammy]

And there it is



Probably wont be there on the next run, GFS is off and on with this until it gets closer, but we have a low center and a fairly tight, lower-latitude wave at the end of today's run, something that has been completely absent for weeks of model runs

It's the first sign the flip is near

[Ubuntwo]

MDR signal now showing up on the GFS ensembles as well.

[toad strangler]

And there it is

https://i.imgur.com/GcB84rF.png

Probably wont be there on the next run, GFS is off and on with this until it gets closer, but we have a low center and a fairly tight, lower-latitude wave at the end of today's run, something that has been completely absent for weeks of model runs

It's the first sign the flip is near

There is real ACE ticking off since yesterday and will go on for the rest of the week into the weekend and beyond most likely. I’m not sure if the switch hasn’t already gone north.

[AutoPenalti]

Some people here seem to forget that models, especially the ensembles, runs 4 times a day with each run having a completely different solution. One run may have 3 different storms at once vs the prior one, merely 6 hours apart, which had none. Each run we get throughout the day is an opportunity to light up the basin.

Edit:
GEFS starting to show some noodles.

[skyline385]

And there's the switch flip coming, just like every year

https://i.imgur.com/gQjTGAq.png

Similar support today

[Teban54]

And there's the switch flip coming, just like every year

https://i.imgur.com/gQjTGAq.png

Similar support today

https://i.imgur.com/nLeXsgF.png

Give how recurve-happy the EPS has been this year, already seeing members heading west through the Caribbean is

[Category5Kaiju]

And there's the switch flip coming, just like every year

https://i.imgur.com/gQjTGAq.png

Similar support today

https://i.imgur.com/nLeXsgF.png

Give how recurve-happy the EPS has been this year, already seeing members heading west through the Caribbean is

Iirc, during the very early stages of models first detecting Beryl, the isolated ensembles had it running through the Caribbean like that too

[Stratton23]

Euro hinting at weak development in the NW gulf