Methodology
How we turn raw data into a risk rating — and what it can and can't tell you. Risk model v0.4.
The risk rating
Each beach gets a guidance rating, recomputed continuously (live overflow status every ~15 minutes, a full data sweep daily):
- RED — High concern A storm overflow is discharging within 5 km, South West Water reports storm sewage affecting the bathing water, or a recent sample showed high bacteria.
- AMBER — Caution An overflow discharged nearby in the last 48h, a South West Water pollution-risk warning is in place, recent rainfall exceeded the beach's threshold, or the beach is classified Poor (the EA advises against bathing all season).
- GREEN — Lower concern No recent overflow, no pollution warning, low rainfall and no high samples.
- GREY — Not enough data No nearby monitoring, rainfall data or classification to rate against.
Each input is shown on the beach page so you can see why a rating was given: active/recent overflows within 5 km, official pollution-risk forecasts, rainfall over the last 24–48h, tide state, the latest official bacteria sample, independent samples, and nearby citizen reports.
Rainfall thresholds vary by beach
Rainfall triggers AMBER at 8 mm in 48h by default — but beaches classified Poor use a stricter 4 mm (they are more rain-susceptible) and beaches classified Excellent a more tolerant 12 mm. This mirrors how the Environment Agency's own pollution-risk forecasting treats each beach's rain response individually.
Tide context
When a rating is AMBER or RED, the current tide state (from Open-Meteo Marine) is added to the explanation: a falling tide can draw discharge along the shore, while an incoming or high tide can carry it closer in. Tide never changes the rating on its own — it explains what an existing risk may do.
Model changelog
- v0.4 (July 2026) — added South West Water pollution-risk forecasts (RED when storm sewage is reported, AMBER for risk warnings), per-beach rainfall thresholds from the EA classification, an AMBER floor for Poor-classified beaches, and tide context in explanations.
- v0.3 — precomputed ratings refreshed continuously; overflow proximity (5 km), rainfall (8 mm/48h) and high bacteria samples.
Why risk, not certainty
Event Duration Monitoring (EDM) tells us whether an overflow is active and for how long — but not the volume spilled, and monitors can record false positives. Tides, currents and time since a spill all affect what actually reaches the water. So Glanmor presents evidence-based guidance, never a guarantee that water is safe or unsafe. Always check official alerts and use your own judgement.
Illness pollution-correlation score
Illness reports get a transparent score out of 100 across pollution proximity, water-test support, rainfall, illness timing, medical support and corroboration. This is a collation aid only — correlation, never a determination of legal causation.
Data sources & update frequency
Full source list and licences are on the about page. In summary:
- Environment Agency bathing-water samples & classifications (Open Government Licence) — ~7 years history.
- South West Water WaterFit Live — live overflow status & precise spill events (~15-minute refresh).
- Rivers Trust EDM annual returns 2020–2025 — per-outfall spill counts/durations.
- Open-Meteo — rainfall & tide context. GDELT — news. POOPy/sewage-map — modelled downstream impact.
Known limitations
- EDM shows duration, not volume; false positives are possible.
- The 5 km radius is a heuristic — local geography, currents and tides vary.
- Citizen reports are unverified unless marked otherwise.
- Data can be delayed or incomplete; absence of a record is not proof nothing happened.
Risk model v0.4. Changes to the model are reflected here. See also the raw data.
Do not rely solely on this website for health decisions. Check official alerts and use your own judgement. Seek medical advice if you are unwell.