Britain's clay just had its two driest summers in 26 years, back to back. Expect a subsidence-claims surge.

Every year, same instrument, same clay

The clay-drought index: mean soil-moisture z-score across the 392 BGS Significant shrink–swell hexes, 1 June - 5 August window, 2001-2020 baseline. Lower is drier. Surge years marked from the industry's own record.

Where the dry clay is

The 20 driest districts of the 110 holding Significant shrink–swell clay, by this summer's index. A caution that keeps us honest: z-scores run large where summers are normally wet, so compare within the table, not against other datasets — and note the same districts were already dry in 2025.

Download: the full index by year (CSV) · the district table (CSV).

Method — so the forecast can be checked, now and next year

1. Soil moisture: NASA POWER root-zone wetness (GWETROOT, MERRA-2 — satellite-informed model reanalysis), daily, 0.5° grid over England and Wales, 2001–2026. The window is 1 June - 5 August because that is where 2026's data currently ends — every year is scored on the identical window.
2. Clay: the 392 hexes where BGS GeoSure rates shrink–swell susceptibility Significant (the same layer our property reports use). The index is the mean z-score over those hexes; z against the 2001-2020 baseline.
3. Hindcast before forecast: the documented surge summers must rank dry (2022: 3, 2006: 5, 2018: 6) and the quiet wet summers must rank wet — they do. The exception is 2003, whose drought was an autumn event (2003's desiccation was an autumn event - documented, not hidden).
4. The claim-surge record is the insurance industry's own: ABI, March 2023 — “23,000 subsidence claims and a £219m expected bill - the highest since 2006”. Quoted, not re-tabulated.
5. The receipt: this page is dated 2026-08-07. When the industry publishes the coming period's subsidence figures, the outcome gets appended here — right or wrong.

Related: Sinking Britain — where the ground is already moving, measured; the paid UK property report carries measured ground motion and clay susceptibility for a single address. We did not produce the measurements; we produced the join.

Is my house about to crack?

Almost certainly not — most homes, even on clay, come through dry years fine. This is an area-level forecast about how many claims the insurance industry should expect, not a prediction for any building. Clay-related damage develops over months and shows up as new or widening cracks, typically around openings; if you see that, the sensible steps are photographs, dates, and your insurer — not panic. Our paid report includes measured ground motion and clay susceptibility for a specific address.

Why should anyone trust a forecast?

Because we tested it backwards before publishing it forwards. The same index — satellite-era soil moisture over the same high-risk clay, the same summer window — scores the documented surge summers of 2022 (rank 3), 2006 (rank 5) and 2018 (rank 6) among the driest, and scores the quiet wet summers as wet. One known surge year, 2003, does not light up in the summer window — because its drought arrived in the autumn (August–October, where it ranks 4th driest). We publish that exception rather than hiding it.

What could make this forecast wrong?

Rain, mostly. A wet autumn can rehydrate clay before damage compounds — it blunted the aftermath of 2018 — while a dry autumn deepens it, which is how 2003 produced a surge on its own. Claims also lag: cracks from this summer surface over winter and into next year, and insurer thresholds and reporting behaviour move the numbers. That is why the forecast is about direction and scale — an elevated claims period — never a precise count.

What exactly did you measure?

NASA POWER root-zone soil wetness (GWETROOT) — a satellite-informed model reanalysis, not a probe in the ground — averaged over 1 June - 5 August each year from 2001, scored against the 2001-2020 baseline, then averaged over the 392 five-kilometre hexes where BGS rates shrink–swell susceptibility Significant. Model reanalysis is the honest label: it is the same instrument for every year, which is what a fair comparison needs.