Correction — September 18, 2026: The earlier article claimed statistical significance for a sinkhole cluster using an unsupported baseline and inconsistent geography. Its probability, confidence claim, causal conclusions and predicted high-risk locations have been withdrawn. The original URL and publication record are retained.
Several reported ground collapses do not, by themselves, establish an increasing regional rate or a common cause. The earlier article combined reports from Long Island and New York City without a complete historical inventory or a consistent definition of an event. That does not support a meaningful probability calculation.
What established research supports
The USGS sinkhole fact sheet describes how water can move subsurface material and notes that collapses can also occur outside karst areas, including around leaky pipes. General mechanisms do not establish the cause of a particular Long Island road failure.
A site-specific explanation requires evidence from the road owner, utility investigation or engineering inspection. Rainfall occurring near an event is not enough to attribute the collapse to a storm. We do not have a verified local dataset that can predict the next location.
What a defensible local study would require
- A fixed boundary and time period, with a documented inventory of comparable collapses.
- One record per event, separating repair updates from new failures.
- A consistent size and impact definition, plus information on missing reports.
- Site-specific findings on utility damage, subsurface conditions and other causes.
- An explicit model, checked assumptions and uncertainty estimates.
Until that evidence exists, the earlier confidence percentage should not be cited. For current road closures, use 511NY; this corrected analysis is not an active travel alert.