Here and Gone: The Vanishing Seismic Records That USGS Won't Explain
The United States Geological Survey operates what is arguably the most comprehensive earthquake monitoring network on the planet. The Advanced National Seismic System feeds data from hundreds of sensors into public-facing databases that scientists, emergency managers, and independent researchers rely upon daily. The premise is straightforward: the American taxpayer funds this infrastructure, and the data it produces belongs to the public.
Except when it doesn't.
A Network Built on Transparency
The USGS Earthquake Hazards Program publishes real-time seismic data through its publicly accessible feeds at earthquake.usgs.gov. The ComCat catalog—the Comprehensive Earthquake Catalog—is the primary repository, updated continuously as new events are detected and processed. Researchers worldwide use it. Emergency planners depend on it. The agency has, for decades, positioned this openness as a point of institutional pride.
The system also generates something less frequently discussed: high-resolution waveform data, the raw sensor output that underlies the processed event records visible in ComCat. This waveform data is accessible through the Incorporated Research Institutions for Seismology, or IRIS, now operating under the umbrella of EarthScope Consortium, and through direct USGS data services. It is substantially more detailed than the summary records most users consult, and it is substantially more revealing.
It is also, in documented cases, ephemeral.
The Disappearance Pattern
Independent seismologists and citizen researchers monitoring USGS data feeds have reported, with increasing frequency over the past several years, instances in which waveform data appears in public servers and then ceases to be accessible—sometimes within a matter of hours, occasionally within minutes of initial posting. The events in question are not always minor. Several documented cases involve readings associated with significant seismic activity in the western United States and Alaska.
The pattern first gained sustained attention within amateur seismology communities, where participants routinely archive data pulls as a matter of practice. When archived snapshots diverged from what the official record later showed—or showed nothing at all where data had briefly existed—the discrepancies were logged and shared. Those logs now constitute a distributed, informal record of anomalies that the USGS has never formally acknowledged.
The agency's standard response to inquiries about missing or altered data is to attribute discrepancies to processing pipeline updates, sensor calibration revisions, or the routine practice of removing preliminary data once refined analysis is complete. These are legitimate explanations for certain categories of data revision. They do not adequately account for cases in which high-resolution waveform records disappear entirely, without replacement, and without notation in the event changelog.
What the Timestamps Reveal
A cross-reference of documented disappearance incidents with concurrent federal activity produces a set of correlations that are difficult to dismiss as coincidental, though causation cannot be established from available evidence.
Several instances of data removal occurred within hours of seismic events in geographic areas that overlap with known Department of Defense testing ranges, including portions of the Nevada Test and Training Range and areas adjacent to Dugway Proving Ground in Utah. Others coincide with time windows during which the Department of Energy's National Nuclear Security Administration was conducting activities that are either classified or not publicly disclosed in sufficient detail to permit independent verification.
This does not prove that missing seismic data is being suppressed to conceal government activity. It establishes that the timing of certain data anomalies is consistent with that hypothesis—and that the absence of a transparent explanation from the USGS makes alternative hypotheses impossible to rule out.
The Calibration Explanation and Its Limits
When pressed on specific disappearance incidents through FOIA requests and direct media inquiries, USGS representatives have most frequently invoked the calibration and quality-control explanation. Preliminary data, the agency notes, is posted automatically by processing systems and may not reflect the accuracy standards required for the official catalog. Removal of such data, in this framing, is routine housekeeping.
The explanation is internally coherent. It is also, in critical respects, unverifiable. The USGS does not publish its quality-control criteria in sufficient detail to permit external assessment of whether specific removed datasets fell below applicable thresholds. It does not maintain—or does not disclose—a public log of removed preliminary records. And it does not explain why, in cases where preliminary data is removed, the standard practice of replacing it with revised data is not followed.
Data quality management in scientific institutions typically involves versioning: the original record is preserved, flagged as preliminary, and superseded by the corrected version, with both versions available and the revision documented. What is described by researchers who have tracked these anomalies is not versioning. It is deletion.
The Independent Archive Problem
The informal archiving networks that have documented these anomalies operate entirely outside official channels. They are composed of hobbyists, retired geologists, and researchers from outside the United States who have no institutional standing to demand explanations from American federal agencies. Their methods are sound—timestamped data pulls, hash verification of file contents, systematic cross-referencing—but their findings carry no regulatory weight.
This is not an accident of circumstance. The absence of a formal, independent archiving requirement for federal scientific data means that the government is under no legal obligation to preserve preliminary records, to log their removal, or to explain discrepancies to anyone. The Federal Records Act applies to agency records, but its application to real-time scientific data feeds is ambiguous, and enforcement has not been pursued in this context.
Legislative proposals to establish mandatory independent archiving for federally generated scientific data have been introduced in Congress on multiple occasions and have not advanced. The scientific agencies that would be subject to such requirements have not publicly supported them.
The Question That Remains Open
There are two possible explanations for the pattern documented here. The first is bureaucratic: data management at a large federal agency is imperfect, quality-control processes generate collateral deletion of valid records, and the absence of a transparent logging system makes normal operational noise look like intentional suppression.
The second is more uncomfortable: certain seismic records are being removed from public access because they document events that the government prefers not to explain.
The USGS has the information necessary to resolve this question. It has not provided it. Until it does—through a public, auditable log of all data removals, with stated justifications and retained originals—the pattern will continue to accumulate, and the public will continue to fund a monitoring network whose outputs it cannot fully trust.