The National Weather Service outage of 13 November 2026 explained: was the weather radar down, and what broke

Partly, and not in the way the screenshots suggested. On the afternoon of Friday 13 November, US time, the National Weather Service put up a notice saying it was having nationwide internet and communications issues, and that people might not be able to reach products on its web pages, or would find them out of date. Reuters and local news services reported the notice that afternoon, and a Los Angeles news service described people in Southern California, where rain was on the way, trying to check forecasts and radar pictures.

At the same time, images of an NWS page headed "Attention: National Radar Outage" were shared widely, often next to news of the FAA's nationwide ground stop, which had begun at 20:53 UTC. Put together, they made it look as if the country's weather radar had gone dark on the same day its air traffic control lost communications.

That evening the NWS said otherwise. At 23:01 UTC its Senior Duty Meteorologist, who watches the agency's national systems from College Park in Maryland, sent an alert administrative message. Reuters then reported the NWS as saying its radars were operating normally, its radar website was fully operational, and outdated content from isolated web pages had been circulating.

It helps to follow a forecast from start to finish. Radar sites scan the sky and send their data to the forecast offices. Each office writes its products, such as forecasts, warnings and forecast discussions. The NWS publishes those products in a central feed, which you can read at api.weather.gov, and both weather.gov and many outside weather apps are built on that feed. The web pages are the last step.

The evidence points to trouble at that last step. The NWS's own all-clear says the radar systems and the network were working. The forecast offices kept writing, as our count below shows. And the notice itself was about web pages: it said products might not be available on the NWS's pages, or might be old, and news reports at the time said watches, warnings and forecasts were still reaching people through other websites.

The "National Radar Outage" page is its own story. It belongs to the Anchorage office in Alaska. When we fetched it on 15 November it still carried that headline, with one line underneath: this page is no longer maintained or updated, and current radar status is at radar.weather.gov. A Hacker News reader linked an archived copy from October 2020. We could not open that archive ourselves because the Internet Archive was offline, but the NWS's own description, outdated content from isolated web pages, fits an old page being read as a new one.

Read next: Cache invalidation: making sure people stop seeing data that is no longer true

Nobody has said so. Both happened on the afternoon of 13 November, and the FAA's trouble was also described as a communications failure, so the question is natural. But the FAA's advisories describe a communications failure at its own air traffic control centers, and the NWS said its own network was working normally. Reuters reported at the time that it was not clear whether the two were related.

Until either agency publishes a cause, the honest answer is that we do not know. What we can say is that the NWS's data kept flowing while the FAA held every departure in the country.

Read next: The FAA nationwide ground stop of 13 November, step by step

The NWS has not said what caused the web trouble or how many pages were affected. If it publishes a service notice or a review, we will update this page with it. If you rely on NWS data in your own app, the API at api.weather.gov is the source to build on, and its products carry their own issue times, so you can always tell how old they are.

If you are checking the weather during a problem like this, look at the issue time printed on the forecast or warning, not at the page around it. Forecasts and warnings carry the time they were written, and that time is the fastest way to know whether what you are reading is current.

Read next: Graceful degradation: keep the useful part of a service running when a piece fails