Skip to content

Without the hype

River flows and groundwater map

Environment
England’s river flows and groundwater levels each month since 1970, each compared with the same month in other years: a map of the Environment Agency’s river gauges and groundwater boreholes, and a tracker of groups of them with matching dry and wet months.
Published

14 November 2026

The map uses the Environment Agency’s records of 1,058 river flow gauges in England, which give a mean flow for each day, and of 3,273 groundwater boreholes (narrow wells drilled to measure the level of water in an aquifer: rock that holds enough water for wells to draw on and for springs to flow from), some with a logger that records the level through the day and many read by hand once a month. It colours each gauge and borehole by how its month compares with the same month in every other year of its record. A river with an October flow among the lowest 5% of all the Octobers on its record is exceptionally low; one in the middle 44% is normal, and the middle October of all is typical. These are the bands the UK Centre for Ecology & Hydrology uses in its monthly Hydrological Summaries.

Loading the gauges…

Map of gauges and boreholes

Press Play to step through every month since 1970, or drag the slider to a month. Point at a dot to see its name and band; click it to open its record beside the map (below it on a phone). The outlines are the Environment Agency’s river basin districts (basins for short), each a group of neighbouring catchments: the area of land from which rain flows into one river and its tributaries. Click a basin to see its gauges ranked. On the map, Tab and then the arrow keys move between gauges, and Enter opens one.

Gauges and boreholes with matching dry and wet months

The rivers in one basin can still differ. In the Thames basin, one group of gauges has dry and wet spells of about 8 months and another of about 2 months. Chalk streams fed by springs tend to stay low long after a dry winter, while the flow of streams on clay returns to typical within weeks of rain.

K-means clustering, a standard statistical method, splits a set of records into a chosen number of groups and puts each record in the group with the nearest average. Each record here is one gauge’s monthly departures from typical between 1995 and 2026, so the gauges in a group had their dry and wet months at mostly the same times, even when they are far apart. The boreholes are grouped separately in the same way. Most groups are named for roughly how long their dry and wet spells last and for the one or two basins that hold the most of their members; one group of boreholes is named for its rising trend instead. A group of boreholes is called chalk when most of its boreholes are known to be in chalk.

For each group, the chart shows its typical member month by month, which is the middle value among its members that month: a river’s flow as a share of its typical flow for the month, a borehole’s level in metres above or below its typical level. Click a group to see its members, and a member to compare it with its group.

Dry and wet months since 1970

Find a gauge or borehole

How it works

The data. The Environment Agency publishes every river flow gauge’s daily mean flow, and every groundwater borehole’s levels, through its online data service, the Hydrology API, under the Open Government Licence. Some records go back a long way: the Thames at Kingston starts in 1883. The map covers England only. Natural Resources Wales, the Scottish Environment Protection Agency and Northern Ireland’s Department for Infrastructure run their own gauges, and the National River Flow Archive brings all four together, but its licence doesn’t let another site republish its data.

A month’s figure. A river’s monthly flow is the average of its daily mean flows, kept when at least 80% of the month’s days have a reading that the Environment Agency hasn’t marked missing or suspect. A borehole’s monthly level is the average of its logger’s 9 am readings when there are at least ten that month; otherwise it is the average of the readings taken by hand, which for many boreholes is one a month. Some 880 borehole months lie far outside the range that holds 98% of that borehole’s months (from the level 1% of its months fall below to the level 99% fall below): further outside it than the range is wide, such as a level 25 m below anything in fifty years. They probably come from a reading taken while a pump was running in the borehole, or after a change to the reference height for its readings, so they are left out. The most recent months are provisional: the Environment Agency checks its readings later and may change them.

The bands. Each month is ranked against the same calendar month in every other year of that gauge’s record, Octobers against Octobers, once the record holds at least 15 of them. Its position is the share of those months below it, counting itself as half. Below 5% is exceptionally low, 5% to 13% notably low and 13% to 28% below normal; 28% to 72% is normal; above normal, notably high and exceptionally high mirror the low bands. A month that is the lowest or highest on record takes the exceptionally low or high colour, and its tooltip and the table name it as a record. Gauges with records too short to rank show as rings. Because each gauge is ranked only against its own record, a chalk stream fed by springs and an upland river that rises within hours of rain can share one map.

The groups. Each gauge’s record becomes its departure from typical in every month, where a typical October is the middle one of all the Octobers on its record, and likewise for each month. For rivers the departure is measured on a logarithmic scale, on which each doubling of flow adds the same step, so half and double typical flow lie the same distance either side of it. Only gauges with at least 70% of the months from January 1995 to the latest are included. Gaps of up to six months are filled with a straight line, boreholes are smoothed over three months because hand readings are noisy, and each series is scaled to the same standard deviation (the square root of the average squared gap between its months and its mean). Scaled like this, k-means puts gauges together by when they are high and low, ignoring how far they swing. That is close to grouping them by correlation: a number from −1 to 1, which is 1 when two records’ departures from their averages are in exact proportion, 0 when there is no straight-line relationship between them, and −1 when one is high exactly when the other is low. There are eight groups of river gauges and six of boreholes. Each run of k-means begins with random groups, and gauges are moved between them until each is in the group with the nearest average. Different starts end in different groups, so k-means was run 30 times and the run with the smallest total distance between gauges and their groups’ averages was kept. The data don’t fall into sharply separate groups: the timing changes gradually across the country, and a gauge near the edge of one group is nearly as close to the next. The spell length in a group’s name is roughly how long a high or low spell lasts in its typical member: how many months apart two months must be before the correlation between their departures from typical falls below 0.5. It ranges from one month for rivers in the North West to 16 months for one group of boreholes. In one group of boreholes the typical member has risen so steadily since 1995 that a straight line accounts for most of its variation, so the group is named for that trend instead; some of its boreholes are still below typical. A group’s typical member each month is the middle value among its members with a reading that month, once at least five have one. Gauges with shorter records have no group.

What it can’t show. A gauge measures what is in the river, after water companies and farms have taken water out and sewage works have put it back, so a low flow can come from what they take out as well as from dry weather. A borehole’s water level depends more on the rain of the winter before than on the last month’s; water takes months to soak into the chalk of southern and eastern England and months to flow out of it. Gauges and boreholes open and close, so the number behind each month on the history chart changes; the note under the chart gives it.

Files. prepare.py builds the page’s data from the downloaded records and the Environment Agency’s river basin district boundaries, groups.py finds the groups, and fetch.py downloads the records again.

Back to top