Start here
Read this carefully
Each pairing below shows two series that genuinely track one another. Tracking together is not proof that one causes the other — and in every case here the honest reading is narrower than the eye-catching one. The three specific traps:
What these charts do — and do not — show
- (a) Spills → oxygen is inferential, not measured. Event Duration Monitoring records only the duration a storm tank discharged — never the volume or pollutant load, which are published nowhere. The oxygen record is monthly, daytime grab samples, which miss the acute, often pre-dawn dissolved-oxygen crashes a storm spill causes. So low oxygen and storm spills both cluster in the same June–October window, but that is co-occurrence, and warm water alone holds less oxygen — a confound we cannot separate out here.
- (b) Fines follow incidents; they do not yet reduce them. The penalties are
proven (an Ofwat civil penalty and Environment Agency criminal convictions) and are
imposed for past offences — they lag the pollution by years. A rise in recorded
incidents also partly reflects better monitoring and self-reporting. Keep the percentages
distinct:
+~33%(Thames Water total) is not+135%(serious) is not+29%(whole sector). - (c) The eel is collapsing across Europe, not just here. European eel recruitment has fallen by roughly 90%+ since the 1980s continent-wide — driven by migration barriers, shifting ocean currents, parasites and historical overfishing. Local warming is a plausible added stress and a genuine co-trend, but it is not a demonstrated local cause, and the eel counts mix survey gear (effort, not standardised catch-per-effort).
- The bottom line. Correlation ≠ causation. These pages are built to make the pressures visible, not to overclaim a mechanism the open data cannot support.
Correlation (a) · sewage & oxygen
Storm spills and dissolved oxygen — a seasonal coincidence
The tidal Thames below Mogden loses oxygen in summer, exactly when storm overflows are most frequent. Both charts below are true. Neither, on its own or together, proves that the spills cause the oxygen sag — see the reading note beneath them.
Tidal dissolved oxygen by month
Pooled grab samples, four tidal points, 2004–2026 · all years combined
Hover or focus a month for the full breakdown.
TH-PTTR0002 (Richmond), TH-PTTR0001 (Teddington), TH-PTLF0005/6
(low-flow); parsed by this programme. RAW-PULL.
Data table — tidal DO by month
| Month | Samples | Mean | Minimum | Below 5 mg/l | Mean temp (°C) |
|---|---|---|---|---|---|
| Jan | 7 | 12.17 | 10.50 | 0 | 7.2 |
| Feb | 6 | 11.47 | 11.20 | 0 | 8.1 |
| Mar | 9 | 11.59 | 10.90 | 0 | 9.0 |
| Apr | 7 | 10.71 | 9.46 | 0 | 12.3 |
| May | 8 | 9.32 | 8.15 | 0 | 15.8 |
| Jun | 16 | 7.65 | 4.92 | 2 | 18.8 |
| Jul | 10 | 7.06 | 2.53 | 2 | 20.9 |
| Aug | 5 | 8.31 | 4.24 | 1 | 21.2 |
| Sep | 6 | 7.18 | 2.84 | 2 | 19.1 |
| Oct | 7 | 8.76 | 4.68 | 1 | 14.7 |
| Nov | 9 | 10.32 | 8.18 | 0 | 10.5 |
| Dec | 8 | 11.45 | 10.70 | 0 | 8.7 |
Mogden storm-tank discharge by year
Hours of discharge; counted spills labelled · permit CNTD.0085, 2020–2025
Hover or focus a bar for detail. Duration only — no volume is published.
19f6064d…; XLSX returns parsed by this programme. EDM records
duration and a counted-spill number only. RAW-PULL.
See the full Mogden sewage page.
Data table — Mogden EDM by year
| Year | Duration (h) | Counted spills | EDM operational |
|---|---|---|---|
| 2020 | 412.8 | 42 | 99.9% |
| 2021 | 438.3 | 43 | 98.1% |
| 2022 | 164.3 | 20 | 100% |
| 2023 | 221.1 | 28 | 99.7% |
| 2024 | 471.7 | 50 | 100% |
| 2025 | 202.5 | 17 | 100% |
How to read pairing (a)
- The oxygen signal is real and seasonal. Mean tidal DO falls from about 11 mg/l in the cool half-year to 7.7 mg/l across June–October, and the lowest reading on record is 2.53 mg/l (30% saturation), Richmond, July 2006 — every sub-5 mg/l excursion sits in that summer window.
- But the spill link is inferential. The spill chart is annual and measures duration only; the oxygen chart is monthly daytime grabs. They are never overlaid on the same spill dates because the sampling cannot support it — the join from spill events to a continuous oxygen trace has not been made.
- Warm water is a confound. Summer water peaks near 21 °C here (see the table's temperature column); warmer water holds less oxygen regardless of any discharge, so part of the summer sag is temperature, not sewage.
- The spill series is volatile. Storm discharge tracks rainfall — the 2024 peak (471.7 h) and the fall to 202.5 h in 2025 reflect weather, not a steady year-on-year worsening and not any proven investment effect.
- And the worst readings are historical. The record minima are 2004–06; recent tidal oxygen has improved (Richmond post-2020 mean ≈ 10.9 mg/l). Do not read a present-day crisis from historical lows.
Correlation (b) · accountability
Escalating penalties, rising incidents
Regulatory penalties on Thames Water have climbed to a record, and its recorded pollution incidents have risen at the same time. Both are documented. But fines are imposed for past offences and land years later, so this is not evidence that penalties have yet cut pollution.
Thames Water pollution incidents
Total and serious (Category 1&2) incidents, 2023 vs 2024
Hover or focus a bar for detail. 2023 total is derived from the +~33% rise.
Data table — incidents
| Year | Total (TW) | Serious (TW) | EPA rating |
|---|---|---|---|
| 2023 | ≈393 | 14 | 2-star |
| 2024 | 523 | 33 | 1-star |
| 2024 sector | 2,801 | 75 | — |
Proven financial penalties
Ofwat civil penalty (2025) and cumulative EA criminal fines
Hover or focus a bar for detail. All figures are proven, not alleged.
Data table — penalties
| Penalty | Amount | Type / date |
|---|---|---|
| Ofwat — wastewater ops | £104.5m | Civil penalty · 28 May 2025 |
| Ofwat — dividend penalty | £18.2m | Civil penalty · 28 May 2025 |
| Ofwat — total | £122.7m | Record for a water company |
| EA criminal fines | ≥ £30m | Convictions, cumulative since 2017 |
How to read pairing (b)
- Fines lag the pollution. Penalties punish offences already committed — the Ofwat action and EA prosecutions cover historical conduct — so rising fines alongside rising incidents is expected, and is not evidence that penalties have reduced (or failed to reduce) pollution.
- Keep the percentages apart.
+~33%is Thames Water's total-incident rise (≈393→523);+135%is the serious-incident rise (14→33);+29%is the whole-sector rise. Conflating them overstates the case. - Recording effort matters. Some of the increase in recorded incidents reflects improved monitoring and self-reporting, not only more pollution.
- Language discipline. The Ofwat penalty and the EA fines shown here are proven (a civil penalty and criminal convictions). Where a matter is only charged or alleged, that is not the same as convicted and is not counted here.
Correlation (c) · wildlife & climate
A Critically Endangered eel, a warming river
Across a 32-year netting series at Richmond, the European eel has all but vanished from the tidal catch — while the upper-tidal river has warmed measurably in summer. The two trends run in opposite directions over the same decades. That is a juxtaposition, not a proven mechanism.
European eel counted per year vs upper-tidal summer warming
All-gear eel counts, Richmond/Teddington tidal reach, 1993–2025
Hover or focus a bar for the eel count. The warming line is an independent ZSL trend, not plotted on the eel scale.
GB530603911403, parsed by this programme
RAW-PULL. Warming — ZSL State of the Thames 2021, upper-tidal
summer temperature +0.19 °C/yr (2007–2020, statistically significant)
RAW-PULL. Counts mix survey gear (effort, not standardised).
Data table — eel counts by survey year
| Year | Eels counted |
|---|---|
| 1993 | 12 |
| 1994 | 3 |
| 1995 | 2 |
| 1996 | 1 |
| 1997 | 3 |
| 1999 | 1 |
| 2002 | 1 |
| 2003 | 1 |
| 2005 | 1 |
| 2007 | 1 |
| 2010 | 1 |
| 2012 | 1 |
| 2015 | 1 |
| 2019 | 1 |
| 2020–2025 | 0 |
How to read pairing (c)
- The decline is real but the driver is not local. European eel is IUCN Critically Endangered and recruitment has collapsed across its entire range since the 1980s. Attributing the Thames decline to local warming or discharge would ignore the continent-wide picture.
- Warming is a co-trend, shown honestly. The +0.19 °C/yr line is the ZSL upper-tidal summer trend for 2007–2020; it is drawn as an independent annotation, not on the eel axis, precisely so the chart does not imply a fitted relationship. ZSL also finds dissolved oxygen improving long-term (p=0.015) — the river is not uniformly degrading.
- The counts are sparse and gear-mixed. Surveys were not annual, and seine, kick and elver records are pooled; totals are indicative of presence, not a standardised catch-per-unit-effort. NBN fish records for the reach separately stop around 2017.
- What it does support. A clean-water indicator has effectively disappeared from the local record while the river warmed — a signal worth monitoring, stated as context, not causation.
Method, sources & provenance
Charts are drawn from data objects inlined directly from the programme datastore; every value is also in the data table beneath each chart. Two measures on different scales are never forced onto a single dual axis — they are shown as separate panels, or (for the warming trend) as a labelled annotation. Provenance tags: RAW-PULL data this programme downloaded and parsed; PUBLISHED a figure quoted from an official source; DERIVED our inference from a stated figure or rule.
- RAW-PULL Dissolved oxygen & temperature: EA Water Quality Archive (WIMS), determinand 9924/76, tidal points
TH-PTTR0002/0001,TH-PTLF0005/6; pooled DO n=98./tmp/thames_datastore/water_quality/. - RAW-PULL Storm spills: EA Event Duration Monitoring Storm Overflow Annual Returns, Mogden WWTW permit
CNTD.0085, 2020–2025. Duration and counted spills only — no volume anywhere. - PUBLISHED Incidents & rating: EA Environmental Performance Report 2024 (gov.uk): 523 incidents, 33 serious, 1-star. 2023 total ≈393 is DERIVED from the +~33% rise.
- PUBLISHED Penalties: Ofwat, 28 May 2025 — £122.7m (£104.5m + £18.2m). EA criminal fines DERIVED as ≥£30m (the £24.4m figure is a Feb-2021 snapshot).
- RAW-PULL Eel: EA Ecology & Fish Data Explorer, Richmond Seine/Kick + Teddington Weir,
GB530603911403./tmp/thames_datastore/fish_ecology/. - RAW-PULL Warming: ZSL State of the Thames 2021 — upper-tidal summer +0.19 °C/yr (2007–2020, significant); Middle Tidal not significant; a Jan-2026 update quotes +0.13 °C/yr for the whole tidal river (different scope).
Full narrative, every figure and every caveat: /tmp/thames_datastore/FINDINGS-v1.md.
Sibling evidence pages: sewage & storm overflows,
water quality, fish & ecology,
accountability.