Methodology and sources
Every figure on WatertightHub traces to a primary source, and we are deliberate about which numbers are measured and which are modeled. This page lays out how each dataset is built and, just as important, what each one does not capture. If a number can't be trusted, it isn't worth publishing.
Cost ranges (modeled)
- Source:
- A national cost band per service, scaled by a single per-state regional index.
- Method:
- Each service’s national low/high is multiplied by one hand-set index per state to produce a regional planning range.
- Known limit:
- These are estimates, not quotes. They do not read your specific soil, permit, labor, or completed-project prices. Never present them as measured or surveyed. For a real price, get contractor quotes.
Water table depth (179 cities)
- Source:
- USDA NRCS SSURGO via Soil Data Access.
- Method:
- For each city, wtdepannmin (annual-minimum, i.e. seasonal-high, water-table depth) sampled on a 17-point grid and aggregated to a median across rated points.
- Known limit:
- It is a seasonal high, not a constant level — the water table ’rises to’ this depth, it does not sit there. It is soil-survey modeled, not a reading of a specific lot. Arid and dense-urban map units carry no rating and are excluded.
Freeze-thaw cycles (200 cities)
- Source:
- ERA5 reanalysis via the Open-Meteo archive API, 2014–2023.
- Method:
- Count of days per year with daily minimum temperature below 0°C and daily maximum above 0°C (the ground surface crosses freezing), averaged over ten years.
- Known limit:
- A surface-air, temperature-crossing count — a proxy for frost stress, not a lab freeze-thaw test of a foundation. Deep soil and a heated wall cycle less than the air. Read it as a relative ranking between cities.
Expansive clay soil (197 cities)
- Source:
- USDA NRCS SSURGO linear extensibility (LEP).
- Method:
- Share of a city’s 17 sampled points whose dominant soil component has LEP ≥ 6% in the top 100cm — the threshold at which shrink-swell clay routinely damages foundations.
- Known limit:
- LEP measures soil, not bedrock. Where expansive ground is deep claystone bedrock (e.g. Denver’s Pierre Shale), this reads low even though the risk is real. A low score means the soil isn’t expansive, not that the ground beneath it isn’t.
Basement flood claims (156 counties)
- Source:
- FEMA National Flood Insurance Program (NFIP) redacted claims data.
- Method:
- Basement flood claims aggregated to the county (FEMA redacts city), split by whether the property was inside or outside the mapped Special Flood Hazard Area.
- Known limit:
- County-level, always — ’the county containing X’, never the city’s own figure. NFIP covers only insured properties, and insurance is effectively mandatory only inside the floodplain, so the outside-the-zone share is a floor, not a population rate.
Building-permit fees (118 cities verified)
- Source:
- Each jurisdiction’s own published fee schedule.
- Method:
- For each verified city, the permitting authority and how it prices foundation work (flat, valuation, tiered, area, mixed, or unpublished), read from the city’s document and checked a second time, with the source linked and dated.
- Known limit:
- Only cities we have verified carry an authority and fee; unverified cities fall back to generic guidance rather than a guessed fee. Schedules are reissued yearly, so each carries the date it was checked.
Corrections
We correct errors when we find them or when a reader points one out. Questions or corrections: admin@watertighthub.com. See all five datasets on the research page.