Mark Poxton, independent analytical systems consultant

CLIENT WORK / GAP REVIEW ยท MPA-SOM-001

Storm overflows: monitoring gaps and repeat offenders

In one line

Every storm overflow in England, five years of returns joined up and set against local rainfall. 583 overflows do 40% of the spilling among reliably monitored sites (a third of all spill hours in England), year after year, and 306 spilled as much or more in a much drier year.

The problem

Storm overflows release diluted sewage into rivers and the sea when the network is overwhelmed. Water companies face fines, public anger and multi-billion-pound investment programmes, so the money has to go to the right assets. Three things get in the way: - monitors that do not work, so an overflow reports little because nobody is measuring it - a national total that hides how concentrated the problem is - rain gets the blame for everything, when some overflows spill whether it rains or not

The question

Which overflows have unreliable monitoring? Which ones spill heavily every year? And which ones spill out of proportion to the rain that falls on them?

The data

The test rules were written down before the run.

What I found

Chart for Storm overflows: monitoring gaps and repeat offenders
Chart for Storm overflows: monitoring gaps and repeat offenders

1. 583 overflows do 40% of the reliably monitored spilling, every year

What it means: a programme aimed at these 583 assets addresses about 40% of the problem among reliably monitored overflows, and a third of it nationally. They are the same assets every year, so this is not bad luck with the weather.

2. 306 overflows spilled as much or more in a much drier year

2025 was much drier than 2024: the median gauge recorded 24% less rain. Spill hours nationally fell by 48%.

What it means: when an overflow keeps spilling with less rain, the likely causes are groundwater getting into the sewers, blockages, pumps or screens failing, or a network simply too small for its catchment. All of these can be fixed, and most cost far less than new storage. These are the first assets an operations team should visit. Most of them sit on the sewer network or at treatment works inlets and storm tanks.

3. Monitoring has improved a lot, but 901 overflows were unreliable in two of the last three years

What it means: a low spill count from an overflow with a poor monitor is not good news. It should be treated as unknown until the monitor is fixed.

By company (anonymised)

Overflows Spill hours 2025 Hours per overflow Repeat offenders Dry-year spillers Unreliable 2 of 3 years
Company A 2,452 200,361 82 41 47 94
Company B 2,270 327,453 144 90 78 130
Company C 2,187 285,931 131 90 88 206
Company D 1,561 123,803 79 28 6 148
Company E 1,442 105,806 73 32 14 62
Company F 1,365 407,006 298 143 43 36
Company G 1,346 190,666 142 102 19 18
Company H 977 114,868 118 0 3 183
Company I 581 107,822 186 57 7 21
Company J 121 14,156 117 0 1 3

The companies have very different problems: - Company F's problem is a group of heavy, persistent spillers. - Company C and Company B have the most dry-year spillers. - Company H shows no repeat offenders at all, but 183 overflows with unreliable monitoring. Its true position is the least certain of the ten.

What it means for a water company

  1. Rank the overflows by repeat behaviour, not by last year's total. The 583 persistent offenders are the core of any investment list.
  2. Send operations teams to the dry-year spillers before designing anything. Failure that rain does not explain is often fixable for a fraction of the cost of storage.
  3. Treat poor monitoring as unknown, not as low risk. Fix the monitors on the 901 before trusting their totals.

How I would run this for a client

Limits

Hostile self-audit

Sources

Ask me about this work All seven client case notes