How a Network of Climate Researchers Is Secretly Backing Up Government Data Before It’s Deleted

Cathy Richards is always near her phone. She might receive a message on any given night informing her that a tranche of federal climate data is expected to arrive by morning, not because she’s checking social media at odd hours or waiting for a call from a friend. When those messages arrive, sometimes at eleven o’clock, sometimes later, she and her coworkers begin downloading. She works for a nonprofit organization called the Open Environmental Data Project out of Hudson, New York. Not the following day.

You get a message at 11 o’clock at night saying, ‘This is going down tomorrow,'” Richards told the BBC that evening. “You make an effort to enjoy your day, but then everything goes wrong. You simply download data all night long. She claimed that some of the messages are devastating. Fearing that the data they have dedicated their professional lives to gathering—years of fieldwork, measurement, and meticulous archiving—will soon vanish from the internet with no assurance that it can be retrieved, scientists reach out. “You hear the urgency,” she remarked. “You understand that this is someone’s X amount of years of research and this is their baby.”

Government Climate Data Deletions & Archiving Efforts: Key Facts

From the inside, the data rescue movement appears like this. Hundreds of volunteers are dispersed throughout the nation and the world, keeping an eye on government websites, tracking deletions, downloading datasets, and rebuilding tools that agencies have taken offline. This is not dramatic in the sense of a movie, but it is urgent in a way that builds up subtly over weeks and months. Since its formation the week following Trump’s first election in 2016, the Environmental Data and Governance Initiative, or EDGI, has been carrying out this work in one way or another. Researchers took Trump’s declared animosity toward climate science seriously. After eight years, their organization had improved. The data losses this time, however, were worse than almost anyone anticipated.

By the end of January 2025, roughly 2,000 records had disappeared from Data.gov, the central federal repository that indexes government datasets. By spring, the Climate and Economic Justice Screening Tool — a Biden-era resource used to direct federal climate investment to disadvantaged communities — had been taken offline within 72 hours of Trump’s inauguration. Early in February, the EPA released its EJScreen environmental justice mapping tool. NOAA announced in April that ocean monitoring datasets would be removed in May. On June 30, 2025, the US Global Change Research Program website was dismantled and all five previously published National Climate Assessments — the authoritative scientific reports on how climate change is affecting the United States — disappeared from federal servers. NOAA’s Billion-Dollar Weather and Climate Disasters database, which had tracked extreme weather costs since 1980, stopped being updated in May. NOAA’s climate.gov stopped publishing new content in July after its entire staff was fired.

There’s a particular quality to the CEJST story that captures what this effort actually involves. The Climate and Economic Justice Screening Tool had been built under Biden’s Justice40 initiative, designed to identify which communities should receive 40 percent of federal climate investment. It was open-source from day one, meaning the underlying code was publicly accessible. So when it came down, three people working with the Public Environmental Data Partners coalition rebuilt a functional version within 24 hours and hosted it on their own servers. That’s the optimistic version of the data rescue story. The EJScreen narrative is not as neat. That tool was not open-source and had been in use since the Obama administration, even during Trump’s first term. It took seven workers over three weeks to rebuild it, essentially attempting to recreate a recipe from an ingredient list without assembly instructions. It’s still being refined.

The extent of what transpired in the first year of Trump’s second term, according to Eric Nost, a geographer at the University of Guelph in Canada who has been working with EDGI since 2016, was radically different from the first. He and his colleagues recorded the removal and modification of webpages, the softening of climate language, and the disappearance of particular documents between 2017 and 2021. The majority of raw datasets survived. The deletion has moved more quickly and deeper this time. Nost simply referred to what he was witnessing as propaganda and censorship. This is an evaluation of what occurs when public information is removed in deceptive ways without justification; it is not a partisan assessment.

As Nost pointed out, the largest publisher in the world is the US government. For generations, people from all over the world have relied on it as a source of scientific knowledge. USDA climate tools are used by Illinois farmers to decide what to plant. Environmental hazards in communities of color were documented and grant applications were written by nonprofit organizations using CEJST and EJScreen. University researchers have developed entire research programs using NOAA datasets that are currently in a state that is somewhere between deleted and just inaccessible. Paul Bierman, a geomorphologist at the University of Vermont who has studied Arctic ice sheets and landscape change for almost 40 years, told the BBC that he had never seen anything like it in his whole career. For the first time, he was doubting the security of the National Science Foundation data repositories where he had stored his research.

The irony that permeates the entire situation is difficult to ignore. Since 1958, the Mauna Loa Observatory in Hawaii has been continuously monitoring atmospheric CO2 concentrations. It saw the biggest increase in CO2 levels in a single year since measurements started last year. According to reports, the Trump administration was thinking of terminating the support office’s lease. The infrastructure for gathering and storing the data is being dismantled at the exact moment when it is most important.

As they observe all of this, the researchers involved feel that they are carrying out a role that democratic societies were meant to incorporate into their institutions from the start: redundancy, resilience, and backup. According to Gretchen Gehrke of EDGI, there is a structural failure of public information infrastructure in the digital age when decades of taxpayer-funded scientific data are now primarily protected by a few hundred volunteers working on pro bono cloud storage agreements. The books are slowly burning. It is these individuals who arrive carrying boxes.