There might have been a case for CIG2 for hail in Illinois given the environment and warm front setup, but the guidance products are arguably not reliable enough for >2 inch hail to make that determination consistently.
Looks like the southern storm is producing an above-anvil cirrus plume.
Most detailed reporting I've seen on what might happen with NCAR. FYI, the president of Lynker is a former UCAR exec and was in the W. Bush administration.
CCs cater pretty heavily to non-traditional students and usually don’t have strict entrance criteria to enroll except for high-demand professional programs. Just make sure credits transfer to 4-year schools in your area. Also look at continuing education options for picking up specific skills.
The fear of being chastised for bringing up potential dangers in low signal to noise ratio environments is hard to overcome alone. It requires training and major culture changes to encourage sticking one‘s neck out when one sees something wrong.
Avoiding false alarms leads to hesitation and lost lead time around critical decision thresholds. It is visible in reliability diagrams for convective outlooks and hurricane RI too. The social penalty is higher for false alarms b/c more people are complaining afterward.
The Inspector General for NSF must launch an investigation into the Trump administration’s alleged attempt to sell-off parts of NCAR — immediately!
👇🏽👇🏽👇🏽
Observational evidence for the use of a semi-slip lower boundary condition in idealized tornado models.
Thank you!
Some exciting news to share: very soon, NOAA Research will be hiring new federal employees into science positions across the agency! A couple of notes:
1) They will only be open on USAJobs for seven days.
2) Multiple positions at different sites may be in one posting.
(1/2)
According to the governor’s spokesperson, no action has been taken re: Peters. His office adds that Polis is not considering a pardon, but commutation remains on the table.
The National Science Foundation is systematically being converted to the National AI and Quantum Research Foundation.
“I see it as the administration exerting political control over what has traditionally been NSF’s ability to fund the best science.”
If I wanted to know what ChatGPT thought I'd fucking well ask ChatGPT!
I'm asking YOU because you're a human and you have experience!
S.3926, the Weather Research and Forecasting Innovation Reauthorization Act of 2026, will be marked up in committee next week. It’s a bipartisan act that should be able to pass the Senate without much issue. A comprehensive summary is available below. (1/2)
www.commerce.senate.gov/2026/2/cantw...
Rust is still growing in popularity but does tend to be used on the backend and is more of a C/C++ replacement rather than something you would use as the high level language for data analysis.
A still satellite image of the blizzard of 2026 in the Northeast United States. This image contains the GeoColor product from the GOES-19 weather satellite from the evening of 2026-02-23.
A view we will not soon forget.
The Blizzard of 2026.
My team has a postdoc working on this so that we can benchmark the different coding LLMs and hopefully produce some useful code. A lot of interest in porting to Jax although I am also curious about more torch physics since our MLWP code is in torch and it has better Mac support.
I could see Julia being a good agentic AI language just from being more token efficient by not having so many imports for array operations. It’s also matured a lot from groups like Clima probing performance limits and fixing bugs.
That’s because all the really spicy software drama happens in person. Lots of debate about what to do with Fortran model codes right now among the C++, Python, Julia, and modern Fortran factions.
You often hear "software development is changing, and we must adapt" because of Claude Code.
But for my entire career, code has been the *easiest* part of making shit people care about, and we still haven't adapted.
The easier coding becomes, the less often we ask "do users actually NEED this?"
If you’re wondering how the NCAR breakup is playing out, a key update: www.wyomingnews.com/news/local_n...
Relevant news to people who depend on government science funding and not just health. www.nytimes.com/2026/02/18/u...
NCAR hosts the Improving Scientific Software Conference in April. Abstract submission deadline just passed but you’re welcome to attend. sea.ucar.edu/iss/2026/
All the Denver airport trains broke down so everyone had to walk across the bridge and somehow go from there. Looks like trains are starting to run again but there’s a big line down the escalators.
The I-70 midwest corridor (Indy, Columbus, Cincinnati, Pittsburgh) all have more developed downtowns and can handle winter weather and sometimes can be more mild in January.
Lots of satellite mosaics in Houston at #AMS2026 but only one made of real tiles.
I'm providing an update on CREDIT at 4:30 PM today in 330A at #AMS2026. Here's a taste of some of the changing views on ML weather prediction I will be providing. (Thanks @plustssn.autumnsky.us for the last minute bug fix!)
Bridging Physics & AI in Climate Modeling 🌍🤖
Introducing JCM v1.0: A fully differentiable, intermediate-complexity atmospheric model built in Python/JAX.
Training hybrid models is hard because legacy code lacks gradients. JCM solves this.
My UCSB colleague Gurusha Juneja wrote a python tool to auto-check your bibtex for hallucinated citations: pypi.org/project/harcx/
It supports papers, books, and
URLs.
Usage:
pip install harcx
harcx references.bib
When flying in winter, it is reassuring to have a pilot with an upper Midwest accent. See y’all in Houston!
