NOAA's weather modification database is riddled with errors
Summary
Two GAO reports, examined together, trace the same gap in weather modification oversight. A December 2024 assessment found that estimates of how much extra precipitation cloud seeding produces range from 0% to 20% in the scientific literature, and that federal reporting requirements don't collect enough information to narrow that range. A February 2026 follow-up found why: NOAA's own database of legally required weather modification reports is riddled with errors, missing maps, and activity that never got logged at all -- and NOAA has no written guidance for reviewing any of it.
Nobody can say for certain how well cloud seeding works
Cloud seeding has been practiced in the U.S. since the 1940s, and recent advances in radar and sensor technology have made it precise enough to show effects under certain conditions. As of July 2024, nine states -- California, Colorado, Idaho, Nevada, New Mexico, North Dakota, Texas, Utah, and Wyoming -- were actively running cloud seeding programs, almost entirely funded at the state level or below. Utah spends the most: $12 million in one-time funding approved in 2023, plus a $5 million annual budget. Ten other states have banned or considered banning cloud seeding or weather modification generally.
View data as table
| Actively using cloud seeding | 9 |
|---|---|
| Banned or considered banning it | 10 |
reviewed the scientific literature on cloud seeding's effectiveness and found estimates of the additional precipitation it produces ranging from 0% to 20% -- a spread wide enough that called it "difficult to evaluate the effects of cloud seeding due to limitations of effectiveness research." Existing studies suggest silver iodide, the chemical most commonly used, doesn't pose an environmental or health concern at current levels of use, but nobody knows whether that holds at a larger scale. identified federal reporting gaps as one specific driver of the uncertainty: the data collected isn't structured to support the kind of independent evaluation that could narrow those estimates.
The reporting system that was supposed to fix this doesn't work either
Fourteen months after that first report, went looking at 's actual database to see how well it was doing its one legally required job: collecting and sharing complete, accurate weather modification reports. 's online database held 1,084 files as of February 2025. sampled 111 of them and found the database to be incomplete, inconsistent, and unreliable -- 78% of initial reports had at least one error, and even interim and final reports, which have fewer requirements, had errors 21% of the time.
View data as table
| Initial reports with 1+ errors | 78 |
|---|---|
| Interim/final reports with 1+ errors | 21 |
The maps were often just as bad. Every initial report is required to include a map showing the target and control areas of the operation -- information operators and researchers could use to estimate costs, effectiveness, and whether an area had already been seeded. found 32% of initial reports were missing a map entirely, and another 36% had maps with at least one missing component. That leaves roughly a third of reports with a complete, accurate map.
View data as table
| Map missing entirely | 32 |
|---|---|
| Map present but with errors | 36 |
| Map complete and accurate | 32 |
Some activity never made it into the database at all. reviewed state agency websites describing weather modification activity from 2021 through 2025 and found that 4 of 10 states had publicly reported activity that never appeared in 's records. Two operators interviewed even produced emails and other evidence of reports they said they'd submitted, which couldn't locate in the database. The earliest complete report could find on file dated to 1999 -- even though, per , the reporting requirement itself took effect in 1971; officials said records from before the database went digital in 2019 may simply have been lost.
told it interprets its job as maintaining a public database for transparency, not conducting broader oversight of the activities themselves. In practice, found doesn't check reports for scientific accuracy, doesn't proactively search for operators who never file, and has never once imposed the fine the law allows for willful non-reporting. The office that reviews reports doesn't even host the database -- it hands reports off to the Library each quarter, with no written agreement covering how the process should work or how long records need to stay available.
The gaps extend to an emerging technology the 1972 law never anticipated: solar geoengineering, which aims to cool the planet at a regional or global scale rather than boost local rainfall. 's reporting forms, unchanged since 1974, require a map of a ground-based target and control area -- a requirement that made sense for a cloud-seeding operation over a specific mountain range, but that solar-geoengineering operators told is nearly impossible to complete for activities like marine cloud brightening or stratospheric aerosol injection, which may have no fixed location or comparison area at all.
made three recommendations: that write actual guidance for reviewing reports, improve its instructions for completing the forms (including for solar geoengineering), and start proactively telling operators and state and local agencies about its reporting requirements. agreed with all three, while asking to clarify that its authority covers the reporting itself, not oversight of the underlying weather modification activities.
The takeaway
- The science on cloud seeding's effectiveness is still genuinely unsettled. Estimates of the extra precipitation it produces span 0% to 20% in the literature reviewed, and pointed to gaps in federal reporting as one reason that range hasn't narrowed.
- The database meant to close that gap is itself unreliable. 78% of initial reports had errors, and two-thirds of required maps were either missing entirely or had at least one flaw -- leaving only about a third complete and accurate. has no written guidance for reviewing any of it.
- Some weather modification activity isn't in the federal record at all. 4 of 10 states checked had known activity missing from 's database, and the oldest complete report on file dates only to 1999 -- decades after the reporting requirement itself took effect.
The cloud-seeding effectiveness findings are from -25-107328, 'Cloud Seeding Technology: Assessing Effectiveness and Other Challenges' (December 2024), a Technology Assessment read directly and in full. The database-reliability findings are from -26-108013, 'Weather Modification: Should Strengthen Oversight to Ensure Reliable Information' (February 2026), also read directly and in full. The February 2026 report explicitly cites and builds on the December 2024 one -- both examine the same weather modification reporting system, but the earlier report focuses on whether cloud seeding works and the later one on whether the federal data needed to answer that question is trustworthy.
Sources(2) ▾
- U.S. Government Accountability Office, Weather Modification: NOAA Should Strengthen Oversight to Ensure Reliable Information (2026-02-23) — -26-108013, a report to the Ranking Member of the House Oversight Committee, examining whether is meeting its statutory duty to collect and share reliable weather modification reports -- the recordkeeping/oversight facet. 's own report explicitly cites and builds on doc-gao-107328 (footnotes 19, 25, 27), which examined cloud seeding's actual effectiveness and challenges -- a distinct, earlier facet of the same overall topic. Read in full directly from the PDF via the Wayback mirror (direct gao.gov fetch blocked with HTTP 403). gao.gov · original document
- U.S. Government Accountability Office, Cloud Seeding Technology: Assessing Effectiveness and Other Challenges (2024-12-01) — -25-107328, a Technology Assessment (report to congressional requesters) examining cloud seeding's real-world effectiveness, benefits, and challenges -- a distinct facet from doc-gao-108013 (which examines whether 's federal reporting/oversight system reliably tracks that activity in the first place). Read in full directly from the PDF via the Wayback mirror (direct gao.gov fetch blocked with HTTP 403). gao.gov · original document
Comments
Always open. Logged-in readers can annotate paragraphs in place.
Since the early 1970s, anyone conducting weather modification in the U.S. -- most commonly cloud seeding, which adds tiny particles to clouds to increase precipitation or suppress hail -- has been legally required to report it to , which is required to keep a public record. Two reports examine this system from different angles. A December 2024 report⧉ found that scientific estimates of how much extra precipitation cloud seeding actually produces range from 0% to 20%, partly because federal reporting requirements don't capture enough information to narrow that range. A February 2026 report⧉, which explicitly built on the first, examined 's reporting database directly and found 78% of initial reports had at least one error -- and that has no written guidance for reviewing any of it.