When Ocean Data Disappears: Science, Articles and AI

Short answer: when ocean data disappears, the numbers do not vanish from the internet. They stop being updated, and the articles written from them keep looking current. That is the real risk, for science, for travel writing like ours, and for the AI tools that answer questions from public data.

Buoys, satellites, ships, divers. In the United States, a big part of what they measure is published for free by NOAA, the national ocean and weather agency. Since 2025, some of it has been retired, frozen or handed over to others. Some changes are just the normal end of a satellite’s life. Others are simply not replaced. The details matter, so it is worth being precise, not dramatic.

What was switched off, and when

In spring 2025, NOAA’s data agencies published a list of products they would retire. Eos, the science magazine of the American Geophysical Union, reported on a batch of around 14 datasets and products about marine, coastal and estuary science, among other topics. The NESDIS “Notice of Changes” for 2025 lists examples such as the NCEI Coastal Water Temperature Guide and the AVHRR Pathfinder sea surface temperature series, which has no updates after 31 December 2023. The widely used “Billion-Dollar Weather and Climate Disasters” product was also retired, with no updates beyond 2024. NOAA says the reports and data for 1980 to 2024 stay archived and available.

It continued in 2026. On 23 March 2026, NOAA’s National Centers for Environmental Information (NCEI) announced that the World Ocean Circulation Experiment (WOCE) dataset would be discontinued, with no updates after 20 May 2026. The past data stays archived.

In August 2026, NOAA’s satellite service announced it will stop delivering the remaining science products from the Suomi NPP satellite at 13:00 UTC on 1 November 2026. That includes instruments such as ATMS, CERES, OMPS and VIIRS. NASA’s alerts show the same date in some places and 2 November in another, so check the current notice. NOAA points users to its newer NOAA-21 and NOAA-20 satellites for similar data.

Not all of this is a loss. A satellite at the end of its life has a successor. Where a dataset has a replacement, the story is about moving users across carefully. Where it has none, the story is a gap.

Why does a missing year of ocean data matter?

Coastal maps, seafloor data and sea records are used for fisheries management, coastal planning, climate research and protecting marine ecosystems. When a series stops, scientists lose the ability to compare this year with the years before. And slow change, like warming or reef decline, only becomes visible in long series. One year alone tells you very little. Many years in a row tell you a story.

For travellers this is not abstract. Water temperature guides can help divers and snorkellers choose when to go, and sea surface temperature records help scientists follow heat stress on reefs. When the feed underneath changes, the advice built on it can become less reliable, without anyone saying so.

What this means for articles and AI

This part is our own reasoning, not a finding from a study. Many online articles and AI answers rely on public databases. When a source stops updating, whatever was written from it keeps travelling. When a page disappears, links break. Then readers, or AI tools, may repeat a number with no way to check it.

Imagine a sentence like “the ocean in this region is the warmest on record”. If the series behind it ended in 2024, the sentence may be true for 2024 and silent about everything after. A careful reader can notice the date. A tool that summarises many pages at once may not. We cannot fix that for the whole internet, but we can write so that our own facts carry their date.

Who is helping

The Data Rescue Project coordinates volunteers who back up federal datasets, and it has a tool where anyone can nominate a dataset for backup. Copies go to DataLumos, a public repository for at-risk government data hosted by the University of Michigan. As of April 2026, the project said it had paused recruiting new volunteers for a while, to simplify its workflow, so check its site for the current status. It also says restricted or identifiable federal data are outside what it can legally preserve.

Some communities have also taken over decommissioned products, and in 2025 scientists urged colleagues to download data before it was removed. These are quiet, practical jobs. They do not make headlines, and they keep the long series alive.

How we handle it at Ocean Calling

In our own work, we put a date and a source next to every fact, link the original document in a collapsed list at the end, and keep copies. When a number changes, we update the article and say so. When we cannot confirm something, we keep searching, and if it is still open, we tell you what we checked and what is still unknown, instead of a confident guess. It is a small habit, and it matters more as official sources shift.

If you keep your own travel notes, date them. A camera records its own date, and a small underwater camera makes a handy visual record of a reef as it looked on one day. A roll-top dry bag keeps a paper notebook safe on a boat. Tiny records, yes. But science has always grown from tiny records kept with patience.

As a reader, you can help too. Check the date of the data, not only the date of the article. Look for a link to the original source. And be kind to writers who say “we don’t know yet”. It is often the most honest sentence on the page. If this made you curious to see the ocean behind the numbers, you can explore our retreats. For the bigger picture on reefs, read what is really causing coral reef loss and how US ocean protections for whales and turtles are changing. We are honest about the cost too: flights to the places we visit have a carbon footprint, which is one reason we prefer small groups and slow itineraries.

Questions people ask

Wildlife encounters are never guaranteed and never forced. Ocean Calling works only with vetted, ethical local operators, and every animal is free to approach, or not, entirely on its own terms.