Southall Air Monitoring 2018–19

Weekly VOC results from laboratory samples at 29 monitoring points around the former gasworks, June 2018 to May 2019. They were extracted from all four disclosed attachments, and each value is checked against its source page. Residents' own odour reports from 2017 to 2019 are set against the hourly wind and the monitoring.

An independent residents' investigation, not peer-reviewed. It uses the developer's own monitoring results, released under the Environmental Information Regulations (request 19-1235). Corrections are welcome.

Most of this page is about two compounds. Both come from coal tar, which gasworks left in the ground.

Why compare them. Traffic raises benzene more than naphthalene. Near the soil hospital the opposite happened: naphthalene rose 3.5–4.8 times the network median, benzene only 1.8–2.8 times. That suggests the ground, not the roads.

Each dot is a monitoring point on the Blackwell C897 plan. Its colour shows how that station typically reads against the median of every station sampled in the same week. Comparing within a week removes the effect of weather and season. Use the buttons below to switch the plan between naphthalene and benzene.

Monthly medians for the five stations that read highest (Air 9, Air 8, Air 5, Air 1 and ES14) against the rest of the network. Months are placed by the end of each tube's exposure period. Air 8 and Air 9 start in late September 2018, so the June–August hot-spot line rests on Air 1, Air 5 and ES14 alone.

What it is. Trichloroethene (TCE) is a man-made solvent, used for degreasing metal and once for dry cleaning and for testing asphalt. The International Agency for Research on Cancer classes it as a known human carcinogen (Group 1, kidney cancer). Since April 2016 its use in the UK and EU has needed specific authorisation under REACH. It is not a normal product of coal-gas manufacture, though it was found in the site's ground (see below).

Why it matters. These are the highest readings against a health guideline in the whole dataset. One was off site, in a public park. We have found no record that the official explanation was tested. And PHE averaged the readings into a site mean of about 2 µg/m³, where they disappear.

Something released TCE repeatedly in spring 2019. The run begins the week the soil hospital stopped treating soil (end of March) and lasts through its dismantling: by 15 May, Ealing said, 90% of it had been removed. Cleaning and degreasing plant before it is taken away is a common use for a solvent. But TCE was also in the ground. Atkins' 2016 risk assessment puts the site's worst TCE at one borehole in the zone along the northern boundary with Beaconsfield Road, an old gasworks waste area: groundwater up to 1,008 µg/l, a hundred times the water standard. TCE was a clean-up target in that zone only, and Air 4 stood on that boundary.

The cis-DCE clue. When TCE breaks down in soil or groundwater, bacteria turn it mainly into cis-1,2-dichloroethene. Cis-DCE was not detected in any air tube all year; only the trans form was found (43 and 45). We first read that as a sign of a fresh solvent. It is not a reliable sign: soil-gas probes on the site in June 2017, before the main works, also held TCE (3 µg/m³) with no cis-DCE. So disturbing the north-east ground is now the most direct explanation for Air 4. It does not explain Air 19, a kilometre away in the park. This is an inference, not proof.

The certificates. Air 4's own screen of tentatively identified compounds gives TCE <1.0, against 400 in the main table. PHE's reports use 400.10 and 475.3, which are not the printed values.

Explanations. Atkins, followed by PHE, suggested road asphalting or off-site tarmac plants. Yet Atkins' own 2016 risk assessment had found TCE in the site's north-east groundwater and made it a clean-up target there; on the record we hold, that finding was not mentioned when the readings were explained. On-site surfacing between the two tubes would fit their positions better than a distant plant does. Hot asphalt itself is not a recognised TCE source, though TCE has been used to clean asphalt equipment and to test it. A laboratory or handling artefact cannot be ruled out for 4 April, but it does not explain the later weeks. Decommissioning, surfacing and solvent-use records for March–May 2019 would settle it. They are the subject of an information request. Hypothesis

The soil treatment hospital stopped treating soil at the end of March 2019 and was dismantled through April. If it was the only source, the southern-boundary readings should have fallen when treatment stopped. They did not.

In July 2018 Ealing's council leader said the smells had "become worse as a result of current high temperatures and unusual wind directions" (MyLondon, 13 July 2018). Atkins later put the summer readings down partly to winds that were "unusually from the north east". Residents' own records show what that meant: when the wind blew from the site, people downwind smelled it.

Residents logged odour in Facebook posts, e-mails to the council and the Environment Agency, tweets and Nextdoor threads. Each timed report was set against the hourly wind at Northolt. A report "fits the site" when the wind was blowing from the site toward the person reporting.

In autumn 2018 residents were told summer was the worst period. Atkins' technical notes (October and November 2018) called summer the time of "potentially higher VOC emissions". Its FAQs said the yearly average would be lower because "temperatures over the course of a year are lower". PHE's first report said concentrations "during cooler weather periods will decrease".

At the ten stations monitored in both periods, the opposite happened:

µg/m³. Stations Air 1–5, ES06, ES09, ES14, ES15, ES16. The rise sits mainly in December–April; October–December benzene shows none. The result holds with non-detects at half the reporting limit and without undated June–July tubes.

Then PHE's argument changed. Its second and third reports said colder weather would lower exposure, because people stay indoors. The fourth says nothing about seasons.

Caution: winter air disperses pollution less well, and traffic and home heating add benzene and naphthalene everywhere in winter. So higher winter readings prove the forecast wrong, but not on their own that the site emitted more. The same-week comparisons on the map allow for that, because every station shares the same weather.

The highest single week for each compound, set beside the nearest legal limit or health guideline. All values are µg/m³.

Other high weeks: naphthalene 189.5 (Air 5), 177.7 and 172 (ES14); benzene 42 and 41 (Air 9 and Air 8, March 2019); TCE 400 (Air 4).

These standards are averages over a year or a lifetime. One week above them is not a breach of the law, nor proof of harm. But the high weeks were not isolated: Air 5, Air 8, Air 9 and ES14 read high week after week. PHE's highest benzene for Air 8 is 56.3, which does not match the certificate (46.4).

Residents were told something simpler. In September 2018 Berkeley told them emissions were all "within legal limits". Yet that same month, at Ealing's air quality scrutiny panel, councillors asked whether the benzene standard had ever been breached and "heard that it had been breached" (official minutes, 27 September 2018). Where, when and over what period was not recorded. In August 2019 PHE accepted that benzene and toluene had exceeded WHO levels in summer 2018, but expected "no significant harm". We have found no assessment that considered the compounds together, or repeated high weeks over two years.

• Dates. Chemtest's "Date Sampled" field is inconsistent. Some reports print the tube's start date, some its end date, and one prints a date before the exposure began. This analysis uses the exposure period printed for each sample. 32 samples, mostly from June–July 2018, have no printed period and use an approximate date.
• A report of zeros. Report 18-18978 (June 2018) prints every VOC as "0", including benzene and naphthalene, while its TPH fractions are high (Aromatic TPH >C10–C12 = 550 µg/m³ at Air 1). A bare zero is not a valid result for this method. These entries are excluded and flagged.
• Detection limits. Naphthalene's reporting limit is <0.3 µg/m³ in almost every report. The low June–August 2018 medians are real non-detects at a sensitive limit, mixed with large single-week spikes (174 µg/m³ at ES14 in early July).
• Coverage. Air 1–5 and the ES stations run from June 2018, Air 6–16 from late September 2018, and Air 17–22 only in 2019. December 2018 to January 2019 has few samples.
• Resident reports. They record experiences, not measurements. Reporters are mostly campaign members; some reports lack a time or place and are left out of the wind tests; place is coded only to a street, and bearings are approximate. Airport wind is not on-site wind.
• Unresolved values. 4.3% of values read from page images are still awaiting manual review and are not used here.