Beta.weather.gov

For posterity, here's the full text of the page:

---

beta.weather.gov Has Been Deactivated Until Further Notice

This page has been deactivated until further notice due to the loss of critical federal staff, which leaves this project without the resources required to continue its development or for routine monitoring and maintenance.

The National Weather Service remains committed to designing a more informative and user-friendly Weather.gov website, and we intend to reactivate this beta site as soon as resources are in place.

In the meantime, please continue to utilize Weather.gov as a source for official National Weather Service forecasts and warnings.

Thank you to everyone who has and continues to provide feedback to improve National Weather Service systems like beta.weather.gov!

---

It's been that way since at least March 20, 2025:

https://archive.ph/beta.weather.gov

How do you go bankrupt? Slowly, then suddenly.

How do you go bankrupt slowly? “This $10 subscription is only 1/400th of our finances; therefore it’s fine, and the kids really get a kick out of it.”

I could barely care less because NOAA weather models are garbage. The NOAA GFS and NAM models are basically dinosaurs at this point compared to what Meteoblue is running. I've been tracking this stuff for ten years and it's not even close anymore. GFS still limps along at 13km resolution and only updates four times a day. Like, are we still in 2010? It completely screws up thunderstorm timing and can't handle mountains for shit. NAM isn't much better at 12km resolution, constantly busting rainfall forecasts especially in complex terrain.

Meanwhile, Meteoblue's models are on another level. NEMS cranks at 3km resolution and nails thunderstorms like 30% better than NAM. The German ICON model with its weird icosahedral grid is crushing GFS by 15-25% on wind forecasts. Then there's the European ECMWF model everyone knows it consistently beats GFS by at least half a day in accuracy. AROME is insane for mountain forecasts with resolution down to 1.3km, making NAM look like a joke with 40% better precipitation predictions. The real game-changer though is Meteoblue's multi-model approach, which averages out the errors from individual models and boosts accuracy by 20-35% over the old school GFS/NAM combo. I don't know why anyone still relies on GFS or NAM except for historical consistency or if you need a really broad overview. For anything specific or local, the newer models demolish them.

Can we get a disclosure on your experience and who you work for, if you’re going to weigh in so heavily on important public vs private sector issues like this?

It’s an opinion - he/she is the source