Methodology and data sources

1. Where the data comes from

Every river level and rainfall reading on this site comes from the Environment Agency real-time flood-monitoring API. We do not operate any sensors of our own, and we do not modify the readings we receive.

This uses Environment Agency flood and river level data from the real-time data API (Beta), made available under the Open Government Licence v3.0.

2. How often it updates

We fetch new readings from the Environment Agency every 15 minutes. Flood warnings are refreshed on the same cadence. Rainfall readings are fetched every 15 minutes on a slightly offset schedule.

The Environment Agency publishes most river level readings every 15 minutes, but the interval varies by station and some report less often. The “last updated” time shown on a station page is the timestamp of that station’s most recent reading, not the time we last checked, so you can always see how current the figure actually is.

3. What the figures mean

Typical range and highest on record come from the Environment Agency’s own station metadata. They describe that individual station, so a level that is unremarkable at one station may be exceptional at another. This is why we always compare a station against itself rather than against an absolute threshold.

Station condition is a three-way band derived from the current level and that station’s typical range:

• Calm — at or below the typical high.
• Elevated — above the typical high, but at or below the midpoint between the typical high and the highest level on record.
• High — above that midpoint.

Rate of rise fits a slope through a station’s recent readings to detect a river rising quickly, rather than one that is merely sitting high. A station can be high and steady, or low and rising fast; those are different situations and we treat them differently.

4. How alerting works

You choose a station and a level. We evaluate every watch every 15 minutes, shortly after each round of new readings arrives, and notify you by email or browser push when your level is crossed.

Each watch has a six-hour cooldown after it notifies you, so a single event produces one message rather than a stream of them. Notifications are recorded before they are sent and are never retried, which means we may occasionally miss one but will not send you duplicates.

5. Known limitations

The source data has real quality problems, and it is more useful to say so than to pretend otherwise:

• Faulty sensors. Some gauges stick, drift, or oscillate between implausible values. We filter readings that jump further than a station’s own typical range would allow before they can trigger an alert, but such readings may still appear on a chart.
• Multiple measures per station. Some stations publish several level measurements — for example both upstream and downstream of a lock. We use only the primary Stage series, because comparing a different measure against a station’s Stage metadata produces figures that look plausible and are wrong.
• Missing and delayed readings. Stations go offline and readings arrive late. Where a station’s most recent reading is more than 48 hours old, we do not display it as a current level at all — a stale number presented confidently is worse than no number.
• Rivers are not only rain. Levels are also affected by tides, sluice and lock operation, groundwater and upstream releases. Our analysis sees the gauge readings, not the cause.

6. This is not a safety-critical service

Flood Monitor is supplementary to official Environment Agency flood warnings. It must not be relied on to protect life or property.

Sign up for and follow official Environment Agency flood warnings as your primary source. In an emergency, or if there is an immediate risk to life, call 999.

If you find something on this site that looks wrong, please tell us at [email protected].