England River Levels: Measure-Day Reading Bundles

Export one UTC day of quarter-hour river-level readings for Environment Agency measures you already know, as one bundle per measure. Each bundle keeps every reading the source returned for that day (sorted oldest first), the published unit label (mASD, mAOD, m), the datum and qualifier the source states, missing values as null, and the exact source URLs with a SHA-256 of each raw response. Useful for a nightly refresh of a known gauge's previous day into a paddling-history page, an ecology notebook or a local river dashboard, without writing your own ETL.

Data comes from the Environment Agency real-time flood-monitoring API (Beta, no key, Open Government Licence). This uses Environment Agency flood and river level data from the real-time data API (Beta).

Not for safety decisions. The source is a beta service with no safety-critical guarantee, and readings can arrive late or in batches. This actor returns no flood warnings, thresholds or risk levels.

Input

{
    "measureIds": ["1029TH-level-stage-i-15_min-mASD"],
    "date": "2026-10-21",
    "maxItems": 100
}

• measureIds: 1 to 20 Environment Agency measure notations (the last part of a .../flood-monitoring/id/measures/<notation> link). Notation only, not URLs. Only quarter-hour (900-second) level measures are supported; rainfall, flow, temperature or other periods come back as free errors. Duplicates are removed and never cost extra. Default: 1029TH-level-stage-i-15_min-mASD.
• date: one UTC calendar day, YYYY-MM-DD, from 27 days before today (UTC) through today. Leave it out for the previous UTC day, which is resolved once when the run starts. Today's date returns a partial day so far. The 27-day window bounds the query; it does not promise the source still holds that day.
• maxItems: the most bundles to store (1 to 100, default 100). Measures past this limit are not fetched.

Find measure notations on the source itself, for example https://environment.data.gov.uk/flood-monitoring/id/stations/1029TH/measures. The actor never searches for stations.

Output

One dataset item per measure and day with at least one numeric reading (readings shortened here):

{
    "schemaVersion": 1,
    "recordId": "ea-level-day:1029TH-level-stage-i-15_min-mASD:2026-10-21",
    "deliveryId": "a737dc7c69173ec5fe9e1b2c3fe31e732c5e156368b4fa82190222f8c7aad535",
    "observedAt": "2026-10-22T21:37:32.041Z",
    "contentHash": "16671cb190390993ca342eaf852a9dd37a7fd0b14bec2bcbeccdeedd6f0fbc05",
    "provenance": [
        {
            "url": "https://environment.data.gov.uk/flood-monitoring/id/measures/1029TH-level-stage-i-15_min-mASD",
            "sha256": "af2ea76ed8d032f5a20b4c5a2b0bc051015211f7f63891774a06fd93add87875"
        },
        {
            "url": "https://environment.data.gov.uk/flood-monitoring/id/measures/1029TH-level-stage-i-15_min-mASD/readings?date=2026-10-21&_sorted&_limit=201",
            "sha256": "11e4db4106e162822fa0194497508c24819c22ea8e78832de424b73f55b34922"
        }
    ],
    "measureId": "1029TH-level-stage-i-15_min-mASD",
    "stationReference": "1029TH",
    "date": "2026-10-21",
    "parameter": "level",
    "qualifier": "Stage",
    "unitName": "mASD",
    "datumType": "http://environment.data.gov.uk/flood-monitoring/def/core/datumASD",
    "periodSeconds": 900,
    "coverage": "bounded_source_snapshot_not_guaranteed_complete_day",
    "reportedReadingCount": 96,
    "numericReadingCount": 96,
    "readings": [
        {
            "sourceReadingId": "http://environment.data.gov.uk/flood-monitoring/data/readings/1029TH-level-stage-i-15_min-mASD/2026-10-21T00-00-00Z",
            "dateTime": "2026-10-21T00:00:00Z",
            "value": 0.084
        },
        {
            "sourceReadingId": "http://environment.data.gov.uk/flood-monitoring/data/readings/1029TH-level-stage-i-15_min-mASD/2026-10-21T00-15-00Z",
            "dateTime": "2026-10-21T00:15:00Z",
            "value": 0.084
        }
    ]
}

What the fields mean, and what they do not:

• readings holds what the source returned for that UTC day, deduplicated by time and sorted oldest first. value is the source's number unchanged (negative stages are kept); a missing source value is null, never 0.
• reportedReadingCount counts the readings in the bundle and numericReadingCount the non-null ones. There is no "expected" count: 96 quarter-hours is common but not guaranteed, and a short day is not proof of a gap, a dry river or a closed station. coverage always says so.
• unitName, datumType and qualifier are copied from the measure's metadata. No datum arithmetic or unit conversion happens: mASD (above station datum) and mAOD (above ordnance datum) stay different labels.
• recordId is the same for a measure and day in every run, so you can upsert on it. contentHash is a SHA-256 of the bundle's source content (readings and unit context, not the timing fields), so a corrected historical reading changes it. deliveryId identifies this run's delivery of the bundle.
• observedAt is when the readings response finished; provenance lists both source requests in order with a hash of each raw response body.

Pricing

Pay per event: $0.002 per bundle stored plus $0.00005 per run start. One bundle costs $0.00205 per run, ten cost $0.02005, twenty $0.04005. Re-running the same measure and day bills it again: there is no change detection. Platform usage may be billed on top, as shown on the pricing tab.

Free (no item event): days with no readings or only null readings, unknown measures, unsupported measures, source errors, capped work and the run summary. Set a spending limit and the actor stops before a bundle the limit cannot cover.

Run summary and limits

Every run writes a free SUMMARY record to the default key-value store: the resolved input (including the default date), run status (complete, partial, failed, capped), one outcome per measure (delivered, empty, missing, error, not_attempted) with its error code, HTTP status and warnings, request and byte counts, and firstUndeliveredKey, the first measure left unfetched when a cap stopped the run, so you can start a manual rerun from there.

• Requests are sequential with at least 1.2 seconds between them, so 20 measures take about a minute or two.
• A day with 201 or more raw readings is refused as unit_limit rather than cut short.
• A 403, 429 or block page stops all source requests for the rest of the run; 5xx and timeouts are retried twice. Readings and metadata are separate requests, so a bundle is a snapshot and the source can change between them.
• Work stops after 300 seconds, 100 requests or 16 MiB of responses; anything not reached is listed as not_attempted.

Support

Open an issue on the actor's Issues tab with the run ID. Source documentation: Environment Agency flood-monitoring API reference.