Hurricane season 2026

[Durian Fiend]

Durgin must be otherwise occupied today.  Jeana Jones at Puna weather posted this image recently, showing a steering ridge just north of the islands.  So, I would guess at this point that if the storm doesn't go south then it will go quite close to the islands- if not through them.

[Punatang]

Edge Posted: You may say what you like.


Why, thank you.  How gracious.  The following hurricane witticisms are more "dad joke" style since you're concerned that, "Not sure everyone reading Punaweb gets your joke".  Because, that would be a real disaster.


[*]Why don’t hurricanes ever get lost? They always follow the current.

[*]How do hurricanes see? With their eye.

[*]What’s a hurricane’s favorite game? Twister.

[*]Why was the hurricane a terrible cook? It always stirred up trouble.

[*]Why did the hurricane break up with the tornado? It said, “I can’t handle all your twists and turns!”

[*]What do hurricanes and marriage have in common? They both start off mild and warm, but before you know it, you’ve lost your house.

[HereOnThePrimalEdge]

Google Deepmind projection for this Sunday:

   

[ironyak]

Have you figured out a way for the Google WeatherNext Cyclone layer to actually provide a storm track? The yellow and orange tracks in your image are just the ECMWF models, which is useful for comparison, but all I can display of Google's predictions is a grid of dots and very opaque squares for wind, rain, and pressure predictions.

Something worth playing around with until Wednesday I guess when we get to see if Google's 3 day prediction for Saturday is any better than the other options.

[Punatang]

Wait, AI slop is now approved?

[HereOnThePrimalEdge]

a way for the Google WeatherNext Cyclone layer to actually provide a storm track?

That’s a good question.  I haven’t dabbled with the settings much, yet.  Or checked to see if Deep Mind runs its own simulations, or acts more like a computational spaghetti model, feed in the available public data, crunch the info and create their own (or “own”) model?

[Durian Fiend]

Durgin posted this a couple hours ago:

“Monday 10th August 2026

POSSIBLE HURRICANE THIS WEEKEND


• Today through early Friday humid with light to moderate easterly trade winds and widely scattered showers.

• Late Friday through Sunday possible storm conditions!

This week all eyes are on Invest 93C which is likely to strengthen into a tropical cyclone as shear begins to relax over the system in a day or two. Currently models and satellite animations show about 20kts of westerly shear over 93C. As the Invest moves slowly north or northwest shear turns to northerly, then drops to near zero Wednesday or Thursday. With very warm sea surface temperatures along its path, it is likely to strengthen as it approaches. The main limiting factor is mid-level dry air along the same path, which will hopefully keep the storm from reaching its fullest potential.

For the past few days we've been wondering if another low to the east of 93C would help to pull the storm north, but models have been showing strong northerly shear over the more easterly low, preventing it from strengthening much, and in turn making it less of an influence on 93C's path.

Various model runs are showing the storm jogging north just before passing by us... I am suspect of this idea as the steering flow for a 985mb low is around 500mb and that is from the northeast, so we're not out of the woods yet.

The low hasn't formed yet, and models could change their tune over the next few days, in the mean time you should be preparing for a possible repeat of Iselle or stronger, possibly significantly stronger”

[Durian Fiend]

Still hasn't formed yet. Just sitting there...

[ironyak]

Yeah - boil, boil, toil, and trouble...


and yet Google's WeatherNext cyclone models are now kicking out spaghetti paths (in green, usual ECMWF model in orange) with the storm passing south / over us and 70 knot / 80 mph max winds by late Saturday / early Sunday. Is it actual AI or just ML? Slop or sashimi? Only time will tell, I guess.

   

[HereOnThePrimalEdge]

kicking out spaghetti paths (in green, usual ECMWF model in orange)

Google WeatherNext shows their models in blue, either as dots or paths. There are two ECMWF models shown, as Google says for reference. One is multiple individual probabilities based on factors that may or may not occur over time, the other “a single, deterministic best-estimate forecast track.”

At this point, this far out in time, the variations in the multiple model runs are wildly divergent, so the single forecast track leaves much to be desired. In terms of accuracy.

The most troubling element I noticed this morning is that now, when you run the time forecast slider on the bottom of the page, both low systems converge over the Big Island at the same time.

Anyone know what that effect might generate?