How we measure ground movement from space
The measurement
Radar satellites (like ESA's Sentinel-1) pass over every city every few days and bounce microwave pulses off the ground. By comparing the phase of these pulses across hundreds of passes over multiple years — a technique called InSAR (interferometric synthetic aperture radar) — scientists can measure whether any patch of ground is rising or sinking, with millimetre-per-year precision. This is peer-reviewed, widely used science: the same technique monitors dams, mines, and volcanoes worldwide.
Our data
We build on the published, peer-reviewed vertical-land-motion datasets of the Ohenhen/Shirzaei group (Virginia Tech): the Nature Cities 2025 dataset covering the 28 most populous US metros (VT Figshare doi:10.7294/27606942, CC0 public domain), the denser 2025 New York City–Long Island deposit (5.1M points, Zenodo doi:10.5281/zenodo.17644691, CC-BY-4.0), and the US Atlantic-coast dataset spanning every East-Coast state from New Hampshire to Florida (VT Figshare doi:10.7294/19350959, CC0). In total we aggregate over 45 million InSAR measurement points to a ~100 m screening grid of ~9 million cells. Contains modified Copernicus Sentinel-1 data.
Shoreline change (coastal addresses)
For addresses within a mile of a measured shoreline, paid reports include the long-term shoreline change rate from the USGS National Assessment of Shoreline Change (public domain) — tens of thousands of transects whose rates are regressed over a century-plus of mapped and satellite shorelines. We report the nearest transect's rate and its distance; retreat is measured history, not a prediction for any single storm or decade.
The risk bands
- Green — worst measured motion near the address is slower than 3 mm/yr downward.
- Amber — 3–10 mm/yr downward measured within 150 m.
- Red — faster than 10 mm/yr downward measured within 150 m.
- No data — no reliable measurement at that spot (vegetated lots, new construction, water). This is not a sign the ground is stable.
We assess the worst measured cell within 150 m of the address — ground risk is local, and a fast-moving patch nearby matters. We also show how that value compares with every measured location in the metro (the percentile), because "2 mm/yr" means more with context.
Hazard context layers (paid reports)
- Wildfire — USFS Wildfire Hazard Potential (2023, 30 m). We sample the address pixel and a ring ~1 km around it: developed lots read "non-burnable" even where surrounding vegetation rates Very High, and it is the surroundings that carry fire to a home.
- Superfund proximity — the most significant NPL-relevant site within 5 km from EPA SEMS: final and proposed listings first, then deleted or part-of-NPL, then the closer one. EPA-reported coordinates can be approximate; we say so on the report.
- Dam proximity — the most significant dam within 15 km from the USACE National Inventory of Dams: poor or unsatisfactory condition first, then High hazard class, then the closer one. The report labels it with its actual class (High or Significant). Hazard classification describes the consequence of failure, not its likelihood.
- Sea-level exposure — the NOAA tide-gauge trend nearest the address combined with measured local ground motion into one effective relative-rise number.
- Drought (farmland) — US Drought Monitor weekly county history since 2000: the share of the county in severe-or-worse drought now, and how often half the county has been there in 25 years.
United Kingdom engines
- Coastal erosion — Environment Agency NCERM 2024 official projections: cumulative metres of expected recession by 2055 and 2105, with and without maintained defences, for the frontage nearest the postcode. Scotland via NatureScot Dynamic Coast historical change rates.
- Coal-mining legacy: Mining Remediation Authority Development High and Low Risk Areas, queried live per postcode, and its coal mining reporting area, read from our copy of the Authority’s published file; recorded mine entries (shafts/adits) probed and mapped from the National Coal Mining Database view services. A screening before the official CON29M search, never a replacement for it.
- Flood zones — Environment Agency (England), Natural Resources Wales, and SEPA (Scotland) official flood maps, routed automatically by the postcode's nation.
What a screening is — and is not
A screening tells you whether the ground around an address has been measurably moving. It does not tell you whether a specific building is damaged, whether its foundation is sound, or what will happen in the future. Satellite radar measures reflective surfaces (roofs, pavement) — not every parcel has a measurement, and line-of-sight motion is an approximation of pure vertical motion. Treat any amber or red result as a reason to commission a licensed geotechnical engineer — never as a verdict.
How we check it
Every data source a paid report depends on is asked a hand-verified question every morning, every paid report records which sections came back empty as it is built, and every correction we have made is public: how we check our answers.