David John Gagne II
LightNews LightNews
djgagnedos.bsky.social
David John Gagne II
@djgagnedos.bsky.social
2.5K followers 460 following 300 posts
Machine Learning Scientist at NSF NCAR in Boulder, CO. Interested in the linkages among AI, weather, climate, disasters, and impacts.
Posts Replies Media Videos
Reposted by David John Gagne II
sickoscommittee.org
Don’t forget to turn your Texas Milk horns down this week.
A row of Texas brand milk with horns down
November 25, 2025 at 2:54 PM
djgagnedos.bsky.social
New paper provides more in depth proof of how and why ML emulators can outperform the model or data they were trained on. The inductive bias of the architecture choice can overcome the errors introduced by coarsening and discrete numerical solvers.
felix-m-koehler.bsky.social Felix Koehler @felix-m-koehler.bsky.social · 11d
Can your AI surpass the simulator that taught it? What if the key to more accurate PDE modeling lies in questioning your training data's origins? 🤔

Excited to share my #NeurIPS 2025 paper with @thuereygroup.bsky.social: "Neural Emulator Superiority"!
November 25, 2025 at 12:58 PM
Reposted by David John Gagne II
shutdownfullcast.bsky.social
OKAY OKAY FINE, GOD 🌝

submit your THANKSGIVING DISASTERS to [email protected]

do NOT submit it here, we won’t see it, sorry that happened to you (or happy for you)!!
November 25, 2025 at 1:19 AM
Reposted by David John Gagne II
edsbs.bsky.social
lolll Oklahoma has a horseshoe up their ass
November 22, 2025 at 6:14 PM
djgagnedos.bsky.social
I was today years old when I realized that the Swedish Chef is a parody of Julia Child, “The French Chef.”
November 19, 2025 at 2:37 PM
Reposted by David John Gagne II
cpsc.gov
Don't do any of this.
November 17, 2025 at 6:54 PM
Reposted by David John Gagne II
agitatedcu.bsky.social
Sunrise in 16 seconds. I sort of regret not doing the sculpted wave cloud to the west instead. Also, sorry for the small jump in the middle.
November 12, 2025 at 1:56 PM
djgagnedos.bsky.social
I could see the Aurora Borealis at this time of day in this part of the country while being entirely localized in my kitchen.
November 12, 2025 at 5:31 AM
Reposted by David John Gagne II
katemessner.com
Loving today's news that the mysterious "fedora man" outside the Louvre heist was actually a 15-year-old museum visitor who dresses like a 1940s French detective all the time, just because. apnews.com/article/louv...
Fedora man unmasked: Meet the teen behind the Louvre mystery photo
Fifteen-year-old Pedro Elias Garzon Delvaux has become an internet sensation after an Associated Press photo captured him outside the Louvre on the day of a crown jewels heist.
apnews.com
November 9, 2025 at 12:16 PM
djgagnedos.bsky.social
Applications are open for the @ncar-cisl.bsky.social SIParCS summer internship program! It's a great opportunity for computationally-inclined students to get experience with real-world research problems spanning computational and Earth System Science. Link: ucar.wd5.myworkdayjobs.com/en-US/UCAR_C...
2026 SIParCS Undergraduate Student Internships
Job Description Summary: The Summer Internships in Parallel Computational Science (SIParCS) program at the National Center for Atmospheric Research (NCAR) offers graduate students and undergraduates w...
ucar.wd5.myworkdayjobs.com
November 3, 2025 at 10:59 PM
Reposted by David John Gagne II
carlbergstrom.com
This morning my ChatGPT quota was inexplicably exhausted.

It took a while but I pieced it together. Voice mode somehow got activated when I went to bed.

The bot then engaged in a 10 hour conversation with my snoring dog, answering questions the pup wasn’t asking and praising him for his insight.
October 18, 2025 at 9:04 AM
Reposted by David John Gagne II
kmalloy.bsky.social
An atmospheric general circulation model you can run on a Python Jupyter notebook?! Heck yeah! Check out this project- both an educational tool for undergraduates/graduate students, and a research tool for scientists interested in idealized climate modeling.
journals.ametsoc.org/view/journal...
journals.ametsoc.org
October 13, 2025 at 7:35 PM
djgagnedos.bsky.social
It's great to see the big crowd here in Boulder for the UCAR Members Meeting. I will be at the CISL booth during the reception this evening, so please come say hi and learn more about all the AI work we're doing at NSF NCAR.
October 7, 2025 at 3:17 PM
Reposted by David John Gagne II
profserious.bsky.social
Here are your 10 -essential- AI prompts for academics ... make your life easy with help from @profserious.bsky.social profserious.substack.com/p/10-ai-prom...
10 AI Prompts for Academics
making those hard jobs a little easier ...
profserious.substack.com
October 5, 2025 at 7:56 AM
Reposted by David John Gagne II
70dbz.bsky.social
Whistleblower report on NASA funding from the Senate Committee on Commerce, Science, and Transportation www.commerce.senate.gov/services/fil...
www.commerce.senate.gov
September 30, 2025 at 12:33 AM
Reposted by David John Gagne II
draaronhill.bsky.social
Extremely excited to announce that I am the incoming chair for the AMS AI STAC Committee! Looking forward to leading our committee alongside @zhonghuazheng.bsky.social and @tropmetpie.bsky.social

We are seeking new members to join the AI STAC, including both regular and student members.
September 17, 2025 at 7:02 PM
Reposted by David John Gagne II
tornatrix.bsky.social
I am so excited to announce that one of the radar meteorology courses that I developed with @comet-meted.bsky.social is now available! "Dual-Polarization Fundamentals" picks up where "Weather Radar Fundamentals" left off! learn.meted.ucar.edu#/online-cour... (Register on learn.meted.ucar.edu first.)
Screenshot of the lesson, showing a radar dish overlaid with text: "Online Course Dual-Polarization Fundamentals 1h 30m - 2h Description Once an experimental type of weather radar, dual-polarized (“dual-pol”) weather radar is now deployed across the United States and many parts of the world. This lesson will introduce the basic principles of dual-pol weather radar. By the end of this lesson, learners will understand physical meanings of several commonly used polarimetric radar variables, including differential reflectivity, correlation coefficient, differential phase, specific differential phase, and linear depolarization ratio. Examples will be presented demonstrating the distinctive polarimetric signatures of different scatterer types, both meteorological and non-meteorological. Note: This lesson has as a prerequisite the prior lesson, Weather Radar Fundamentals. It is strongly recommended that beginning learners complete that lesson before starting this one."
September 14, 2025 at 5:21 PM
djgagnedos.bsky.social
Checking my bkuesky feed this week/month/year.
September 12, 2025 at 11:05 AM
Reposted by David John Gagne II
plustssn.autumnsky.us
I've released autumnplot-gl v4.0! New features:

- Station plots
- Support for globe view in Maplibre (Mapbox is in progress)
- Support for mutiple colormaps on a contour fill plot (precipitation type, for example)
- A few performance optimizations

github.com/tsupinie/aut...
GitHub - tsupinie/autumnplot-gl: Hardware-accelerated geospatial data plotting in the browser
Hardware-accelerated geospatial data plotting in the browser - tsupinie/autumnplot-gl
github.com
August 28, 2025 at 10:17 PM
Reposted by David John Gagne II
faineg.bsky.social
I finally read computer scientist Joseph Weizenbaum’s 1976 classic “Computer Power and Human Reason.”

This book deserves a massive revival in our current age of grotesque and largely thoughtless AI creep into everything:
JOSEPH WEIZENBAUM COMPUTER POWER AND HUMAN REASON FROM JUDGMENT TO CALCULATION
August 28, 2025 at 10:34 PM
Reposted by David John Gagne II
erichfischer.bsky.social
Numerical weather prediction models (ECMWF IFS) still outperform AI weather models in forecasting record-breaking hot and cold extremes and unseen wind extremes

Preprint
arxiv.org/abs/2508.15724

with Zhongwei Zhang @erichfischer.bsky.social @zscheischlerjak.bsky.social and Sebastian Engelke
Numerical models outperform AI weather forecasts of record-breaking extremes
Artificial intelligence (AI)-based models are revolutionizing weather forecasting and have surpassed leading numerical weather prediction systems on various benchmark tasks. However, their ability to ...
arxiv.org
August 22, 2025 at 5:43 AM
Reposted by David John Gagne II
carlbergstrom.com
5. Today I read a paper by @sabinaleonelli.bsky.social and Alexander Mussgnug that I think illustrates this point perfectly.

philsci-archive.pitt.edu/24891/1/Phil...
Convenience AI Sabina Leonelli & Alexander Martin Mussgnug12 Abstract: This paper considers the mundane ways in which AI is being incorporated into scientific practice today, and particularly the extent to which AI is used to automate tasks perceived to be boring, “mere routine” and inconvenient to researchers. We label such uses as instances of “Convenience AI” — that is situations where AI is applied with the primary intention to increase speed and minimize human effort. We outline how attributions of convenience to AI applications involve three key characteristics: (i) an emphasis on speed and ease of action, (ii) a comparative element, as well as (iii) a subject-dependent and subjective quality. Using examples from medical science and development economics, we highlight epistemic benefits, complications, and drawbacks of Convenience AI along these three dimensions. While the pursuit of convenience through AI can save precious time and resources as well as give rise to novel forms of inquiry, our analysis underscores how the uncritical adoption of Convenience AI for the sake of shortcutting human labour may also weaken the evidential foundations of science and generate inertia in how research is planned, set-up and conducted, with potentially damaging implications for the knowledge being produced. Critically, we argue that the consistent association of Convenience AI with the goals of productivity, efficiency, and ease, as often promoted also by companies targeting the research market for AI applications, can lower critical scrutiny of research processes and shift focus away from appreciating their broader epistemic and social implications.
August 19, 2025 at 5:11 AM
Reposted by David John Gagne II
davidimiller.bsky.social
🎤 BOOM: Federal judge just ruled to *reinstate* the suspended ~300 NSF grants at UCLA.

Determines that NSF violated the previous court order by attempting to swap "termination" with "suspension".

storage.courtlistener.com/recap/gov.us...
NSF’s actions violate the Preliminary Injunction. Though there may be situations where a “termination” and “suspension” are not the same thing, there is no principled difference between a “termination” and the immediate, indefinite, and “final” “suspension” of funding in this context. The suspensions have the same effect, and are based on the same type of deficient explanations, as the original terminations. NSF communicated the suspensions by means of a form letter that failed to provide the requisite grant-specific reason for halting funding, and that failed to adequately consider grant-specific interests, including the reliance interests of the researchers. Therefore, pursuant to the Preliminary Injunction (Dkt. No. 55), NSF’s suspension of the grants at issue here is VACATED. For avoidance of doubt, the Court also clarifies that grant “termination,” as the term is used in the Preliminary Injunction, encompasses circumstances where grant funding is cut off on a long-term or indefinite basis, like the suspensions carried out by NSF on July 30
August 13, 2025 at 1:12 AM
Reposted by David John Gagne II
samlmontano.bsky.social
Do not get on this internet on the 20th anniversary of Katrina and rehabilitate George Bush I swear to god.
August 5, 2025 at 4:04 PM
djgagnedos.bsky.social
PSA: the wind components in the GFS and GEFS cube sphere tile netCDF files are already in Earth-relative coordinates. Figured this out by making this plot after two days of wondering why the transformed wind vectors wouldn't line up across tiles.
August 4, 2025 at 9:05 PM