Durian Fiend - You seem to be downplaying what the article claims is " a decades worth of meteorological progress".
Never said anything of the sort - just noted that the improvements appear to be focused on short term 2 to 3 day accuracy, not longer time frames like 84+ hour predictions. Also, this is much more in the realm of old-school Machine Learning as opposed to Generative LLMs which most people believe to be "AI" so there are some semantics to sort through here as well.
Weatherman Harry Durgin is quite impressed by this AI development.
Neat, but based on the article and paper in Nature provided, what do you think?
The animations I saw available at tropical tidbits only run for 84 hours.
Here's a 354 hr (or ~15 day) gif animation of the GFS model centered on Hawai'i that Tropical Tidbits generated (click "Forecast GIF" at the bottom of the image and pick a Start and End hour - different models will have different options). It's too big to post on Punaweb, and largely pointless, partly because all the images of predicting what the storm is going to do beyond 3.5 days (84 hours) tends to be highly inaccurate, regardless of the source.
At least the spaghetti is now all lined up in our direction, but maybe now mostly to the north of us? Order up!
Never said anything of the sort - just noted that the improvements appear to be focused on short term 2 to 3 day accuracy, not longer time frames like 84+ hour predictions. Also, this is much more in the realm of old-school Machine Learning as opposed to Generative LLMs which most people believe to be "AI" so there are some semantics to sort through here as well.
Weatherman Harry Durgin is quite impressed by this AI development.
Neat, but based on the article and paper in Nature provided, what do you think?
The animations I saw available at tropical tidbits only run for 84 hours.
Here's a 354 hr (or ~15 day) gif animation of the GFS model centered on Hawai'i that Tropical Tidbits generated (click "Forecast GIF" at the bottom of the image and pick a Start and End hour - different models will have different options). It's too big to post on Punaweb, and largely pointless, partly because all the images of predicting what the storm is going to do beyond 3.5 days (84 hours) tends to be highly inaccurate, regardless of the source.
At least the spaghetti is now all lined up in our direction, but maybe now mostly to the north of us? Order up!