Tim Supinie
banner
plustssn.autumnsky.us
Tim Supinie
@plustssn.autumnsky.us
Hydrometeor connoisseur 💧❄️🌪️ / Electron herder 💻 / Rhythmic airbender 🎸🎶 / Amateur edible chemist 🍞🍳 / @climbingfox86.autumnsky.us's worse half 👫 / OU SoM alum 👨‍🎓
Reposted by Tim Supinie
The EF-2 tornado in Purcell, Oklahoma on January 8, 2026 is the earliest January EF-2 on record in the State of Oklahoma and the first occurrence of a January EF-2 in the NWS Norman County Warning Area (CWA) in 53 years (January 18, 1973 F2 tornado near Seminole) #okwx
January 9, 2026 at 11:27 PM
Reposted by Tim Supinie
WFO Norman has confirmed four tornadoes from this morning’s storms
OUN issues Damage Survey PNS (Max: EF1) at Jan 8, 8:27 PM CST ...NWS Damage Survey for 1/8/26 Tornado Event... Link
January 9, 2026 at 2:34 AM
Reposted by Tim Supinie
Our first candidate of the year for "was it a tornado or straight-line winds?" Oklahoma weather sure doesn't waste any time. #okwx #okmesonet
January 8, 2026 at 3:15 PM
Well, I wasn't planning on checking off "first thunder of 2026" and "first tornado warning of 2026" within an hour of so of each other on January 8, but here we are.
January 8, 2026 at 1:51 PM
I don't know who needs to hear this*, but in C++,

delete[] a, b;

is equivalent to

delete[] a;
b;

NOT

delete[] a;
delete[] b;

*Yes I do. Me from a couple months ago needs to hear this.
January 7, 2026 at 4:49 PM
Reposted by Tim Supinie
Unfortunately such conflating of LLMs with other uses of AI is seemingly already pretty well entrenched in the minds of many. This sort of incident where the NWS is using LLMs and not checking for gaffes is only going to muddy the waters further, I fear.
*inhales*

AI is a blanket term. It's not the same thing as using generative AI (images/videos from prompts) or large language models (LLMs; text from prompts).

It's misleading to say "AI" when genAI or LLMs are the actual tools being used.

The IMAGE was generated here, not the FORECAST.
The National Weather Service used AI to create a forecast for Idaho. Low chances of wind for the nonexistent towns of Cocrerrireod, Orangeotild and Whata Bod www.washingtonpost.com/weather/2026...
January 7, 2026 at 4:09 PM
Reposted by Tim Supinie
New forecasts are forthcoming this morning, but a new AI forecast system my group is developing depicted a broad day 5 signal for severe weather hazards across the Southeast U.S. on Friday. Something to keep an eye on
January 6, 2026 at 3:46 PM
Checking out the OKC Thunder game tonight with @climbingfox86.autumnsky.us and her parents.
January 6, 2026 at 1:00 AM
Reposted by Tim Supinie
Well, that ended up being easier than I anticipated!

Summary pages for both SPC Convective and Fire outlooks are up on my website. There may be some bugs with summary and preview graphic generation that will be solved as they come up.

www.stormscale.io/spc/convective
www.stormscale.io/spc/fire
January 4, 2026 at 11:08 PM
Reposted by Tim Supinie
Remember how appalled we all were in 2022 when Russia completely unprovoked attacked Ukraine? That is what we’re doing tonight with Venezuela. We are the bad guys here. Everyone who supports this is supporting unprovoked murder. Fuck every last one of them
January 3, 2026 at 6:49 AM
</2025>
<2026>
January 1, 2026 at 6:00 AM
And here's satellite data in autumnplot-gl as well. Improvements to the grid system I made for the radar meant this was not nearly as terrible as I was afraid it was gonna be.
December 31, 2025 at 2:31 PM
Reposted by Tim Supinie
Since SPC isn’t on BlueSky yet, I’ll share these fun “Wrapped”-style graphics here. Here’s a look back at SPC product issuances in 2025. Stay tuned for a more thorough review of the severe storm season coming in the new year.
December 30, 2025 at 7:01 PM
Got NEXRAD single-site radar plotting in autumnplot-gl (@jaredpbostic.bsky.social). There are some optimizations I could do before calling it good, but not sure they're worth it in the grand scheme of things.

Pictured is butt-clench 2 of 3 in spring 2023.
December 29, 2025 at 4:04 AM
Perfectly normal map for Oklahoma after midnight in late December ahead of a dry cold front.
December 28, 2025 at 6:45 AM
Post a perfect album from the 90s that isn't Nirvana, Pearl Jam, Soundgarden or Alice In Chains.
December 28, 2025 at 5:40 AM
Christmas Eve brunch. Got a full day of meal prep ahead.
December 24, 2025 at 5:07 PM
Do you think Mary and Joe stopped by the KFC for some food after getting into town?
December 24, 2025 at 3:50 PM
Reposted by Tim Supinie
Even after the news last week, we are still doing great research here at NCAR. My group has released a new paper, led by John Schreck, on arxiv about an efficient method for generating calibrated AI weather ensembles with reproducible scale specific perturbations. arxiv.org/abs/2512.18815
Controllable Probabilistic Forecasting with Stochastic Decomposition Layers
AI weather prediction ensembles with latent noise injection and optimized with the continuous ranked probability score (CRPS) have produced both accurate and well-calibrated predictions with far less ...
arxiv.org
December 23, 2025 at 7:30 PM
Reposted by Tim Supinie
I'm struggling to understand the causal link from non-renewal of contracts between the City of Norman and CCEDC to the CCEDC pulling funding for the NWM.

Do the contracts involve money going from the City to CCEDC, and the CCEDC no longer has the cash to fund the NWM (plus other stuff, presumably)?
December 23, 2025 at 4:34 AM
Mother nature is apparently subscribing to the theory that Jesus of Nazareth was actually born in the spring.
December 23, 2025 at 3:24 PM
I'm struggling to understand the causal link from non-renewal of contracts between the City of Norman and CCEDC to the CCEDC pulling funding for the NWM.

Do the contracts involve money going from the City to CCEDC, and the CCEDC no longer has the cash to fund the NWM (plus other stuff, presumably)?
December 23, 2025 at 4:34 AM
The Silos climbing gym in OKC closed this evening, so @climbingfox86.autumnsky.us and I went to the going-away party, where I took some pics. We've been climbing there (her more often than me) for a bit over a decade now, and it's sad to see such a cool and unique facility go.
December 22, 2025 at 5:37 AM
Help me, fantasy-range GEFS member 15, you're my only hope.
December 20, 2025 at 5:17 AM
This post sent me down a bit of a rabbit hole. Overall, for the entire year, 77% of fires are human-caused (for fires whose cause is known, it's 84%). But for December specifically, it's 89% (99% for fires whose cause is known). Fires this time of year are way more likely to be human-caused.
December 19, 2025 at 5:23 PM