• Hello, please take a minute to check out our awesome content, contributed by the wonderful members of our community. We hope you'll add your own thoughts and opinions by making a free account!

Pattern November 2023

I came across a great article last night by accident and lost the link. They have AI doing forecast using the euro initialization. Say its smoking all the wx models which are built to generate forecast with math,physics. Has another way of doing. Anyone else see this or related info please pass along. Thanks
 
JHS said:
That one and Dec 2017 missed Union county in SC for the most part. Had some sleet but no snow in 2018. Only rain in 2017. Our best December snow here was in 1993.
Click to expand...
Amazing what an hour or so drive NW will do,
I've lived in my home since 06.
The power lines are buried here and Miles from here.
Only time the power went out was during that storm.
We had heavy wet snow falling here for 5 or 6 hours.
Snowed over 8 inches here.
With plenty of trees down.
 
Last edited:
A look we will be seeing a lot in Dec, Jan, Feb and March.
gfs_asnow_us_65.png
 
Wild Weather Monger said:
Wow. That was really interesting.

As an old weather enthusiast geezer, I'm not sure I like the continued improvements in weather forecasting. In some ways, knowing what the weather will do with near certainty takes some of the fun out of it.

Edited to add, AI forecasting seems headed to making a met degree about as financially useful as basket weaving.
Click to expand...
At 6 days it's still just a crap shoot, with AI modeling or traditional ones. Still sitting at three days for any real degree of accuracy. There is only so much you can do unless you have better real-time data coverage.
 
weatherfide said:
At 6 days it's still just a crap shoot, with AI modeling or traditional ones. Still sitting at three days for any real degree of accuracy. There is only so much you can do unless you have better real-time data coverage.
Click to expand...
Yes, as of now. And of course, it's good to remember today's 6-day accuracy is roughly 20 years ago's 3-day accuracy.

What's intriguing about the AI approach is it's essentially short and medium-term analogs on steroids which would seem to improve medium and long-term forecasting the most.
 
Wild Weather Monger said:
Yes, as of now. And of course, it's good to remember today's 6-day accuracy is roughly 20 years ago's 3-day accuracy.

What's intriguing about the AI approach is it's essentially short and medium-term analogs on steroids which would seem to improve medium and long-term forecasting the most.
Click to expand...
Ok, this is interesting. It's basically running tons of nested if/then statements, using limited historical data to predict a short distance in to the future, but it is not modeling how the physics actually work. This is all AI is which means it "understands" nothing. What are the implications for relying on it? Seems like we devolve in to idiocracy.
 
weatherfide said:
Ok, this is interesting. It's basically running tons of nested if/then statements, using limited historical data to predict a short distance in to the future, but it is not modeling how the physics actually work. This is all AI is which means it "understands" nothing. What are the implications for relying on it? Seems like we devolve in to idiocracy.
Click to expand...
I suspect this method coupled with dynamic modeling is the future. And by the future, a few years at most.

"The new AI models skip the expense of solving equations in favor of “deep learning.” They identify patterns in the way the atmosphere naturally evolves, after training on 40 years of ECMWF “reanalysis” data—a combination of observations and short-term model forecasts that represents modelers’ best and most complete picture of past weather. When fed a starting snapshot of the atmosphere based on the same combination of observations and modeling, GraphCast can outperform the ECMWF forecast out to 10 days on 90% of its verification targets, including hurricane tracks and extreme temperatures."
 
They should add a D5 on the drought monitor to highlight all-time record-breaking droughts.

Total precip for Chattanooga currently stands at 0.48" for meteorological fall.
The driest met fall on record (going back to 1879) is 2.72" back in 1939.
Driest ever 3-month period is Aug-Oct 1963 with 2.07".

So, even if we get 1.59" of precip the next two weeks, it will still end up being the driest 3-month period ever recorded, lol, that's how deep this drought is. A relative deluge of 2.23" still yields the driest met fall on ever.
 
cyclogent said:
They should add a D5 on the drought monitor to highlight all-time record-breaking droughts.

Total precip for Chattanooga currently stands at 0.48" for meteorological fall.
The driest met fall on record (going back to 1879) is 2.72" back in 1939.
Driest ever 3-month period is Aug-Oct 1963 with 2.07".

So, even if we get 1.59" of precip the next two weeks, it will still end up being the driest 3-month period ever recorded, lol, that's how deep this drought is. A relative deluge of 2.23" still yields the driest met fall on ever.
Click to expand...
Thankfully this hit in the fall when the days are short and cool. We’d be on the news if it was summer.
 
EmersonGA said:
Thankfully this hit in the fall when the days are short and cool. We’d be on the news if it was summer.
Click to expand...
Yeah, just like we were in 1986 when we had to get hay from the midwest to prevent a major disaster. That was all over the news that summer. 1993 was another rough one too, along with 2002 and especially 2007.
 
You must log in or register to reply here.
Back
Top