🚧👷This is an initial rough draft.
The study protocol has gaps, but serves as guidance for the initial exploration of what is possible using readily available open data.
Quantify the correlation between aircraft landing activity and pollution at existing ground-level Air Quality Monitors (AQMs).
To what extent do ground-level AQMs near Heathrow flight path detect pollution attributable to aircraft landings?
We hypothesise that periods of high aircraft landing traffic are significantly correlated with elevated ground-level concentrations of PM2.5 and PM10, controlling for road traffic pollution and meteorological factors.
Primary variables
The study will first look into the effect of the primary variables below. If the correlation coefficient is low, we will use utilise the secondary variables in study.
| Primary variable | Type | Description |
|---|---|---|
| Aircraft altitude | Independent | Affects pollution dispersion and influences how much pollution reaches the ground |
| AQM location | Control | Distance (vertical and horizontal) to flight path determine how much pollution is measured |
| Road traffic | Control | Affects pollution readings independently of aircraft. Needs to be accounted, so it's not attributed to aircraft emissions |
Secondary variables
These are all weather measurements that can affect how pollution dissipates in the atmosphere and influence how it is measured on the ground.
| Variable | Type | Description |
|---|---|---|
| Atmospheric pressure | Control | Can affect emissions dispersion |
| Rain | Control | Can affect pollutant removal from atmosphere |
| Wind direction/speed | Control | Can affect direction of dispersion (e.g., if an AQM is located "upwind" it won't likely measure emissions) |
Wind is a challenging variable to control because speed and direction can change depending on altitude. This dissipates pollution horizontally and vertically in ways that is complex to predict at ground level.
Uncontrolled variables
There are other variables that may be impossible to control due to scarcity of data. The following are likely factors in the amount of pollution generated by aircraft, but are not controlled in this study:
- Aircraft/turbine model, age and state of maintenance
- Aircraft total weight
- Quality of fuel
We are interested in measuring the amount of pollution in aircraft landings, not take-offs, because:
- all aircraft follow a predetermined path as they approach the airport
- aircraft fly over London in relatively low altitude for a prolonged amount of time
However, aircraft will not always approach the airport from the same "side" and land at the same runway. Wind direction (at 600m) and a runway schedule dictates which runway is in use at any give time. So we need to model Heathrow's runway operations.
Runway operations at Heathrow
Heathrow has two runways (north and south) that are used as follows.
- Easterly operations: when wind blows from the east, aircraft approach from the west. The South runway is always used.
- Westerly operations: when wind blows form the west, aircraft approach from the east (flying over London). In this configuration, either the North or South runways is in use, depending on a schedule.
The airport operates a Westerly service 70% of the time, as follows:
| 05:00-15:00 | 15:00-23:00 | |
|---|---|---|
| Westerly week 1 |
Northern | Southern |
| Westerly week 2 |
Southern | Northern |
| Easterly | Southern | Southern |
It is possible to identify which runway was operating based on the official schedule programme and confirming it based on historical wind directions.
- AQMs historical data is free and retrievable from the London Air network
- Horizontal distance to landing flight path
- Horizontal distance to airport
AQMs to the west of Heathrow:
- There are no AQMs that fit the selection criteria.
AQMs to the east of Heathrow:
- Richmond upon Thames - Richmond
- Wandsworth Battersea
| AQM Name | AQM Height | Species | Horizontal distance to LHR¥ | Horizontal distance to flight path¥ | Aircraft altitude¥ |
|---|---|---|---|---|---|
| Wandsworth - Battersea | 2.5 meters | NOx PM10 PM2.5 |
20 km | north runway: 0km south runway: 1.3km to the north |
1.2 km |
| Richmond Upon Thames - Richmond | 1.5 meters | NOx PM10 PM2.5 |
9 km | north run: 2.2 km to the south south run: 0km |
0.6 km |
¥ Measurements collected on 20/10/2025 at 10:00am on a sample of 2 aircraft
Road traffic follows a cyclical approach depending on day (e.g., Christmas), day of the week (e.g., Sunday or Monday), and hour of day.
Within day, road traffic follows a regular pattern around commuting time. For example, we can use the openly published peak times of the London underground as proxy for road traffic and compare to the airport operating times. For example:
| Events | Road Traffic | Air Traffic | |
|---|---|---|---|
| 00:00 - 05:00 | low🔹 | --🔹 | |
| 05:00 - 07:00 | 05:00 first landing | low🔹 | high🔹 |
| 07:00 - 09:15 | 07:00 peak commute | high | high |
| 09:15 - 17:45 | mid | high | |
| 17:45 - 19:45 | 17:45 peak commute | high | high |
| 19:45 - 23:00 | 23:00 last landing | low🔹 | high🔹 |
| 23:00 - 23:59 | low | -- |
This makes it possible to compare pollution measurements, for example, at 4am (low road, no aircraft) against 5am or 6am (low road, high aircraft).
This is an observational, time-series correlational study using secondary data from multiple public sources.
Compare pollution data from the closest AQM to the flight path, against different times of the same day, to identify aircraft as significant source of pollution.
We are controlling for:
- Road traffic: variation within the same day
- AQM location: directly under the flight path
Compare pollution data from the furthest AQM to the flight path downwind, against different times of the same day, to identify aircraft as significant source of pollution. We'll control for wind direction at ground level.
We are controlling for:
- Road traffic: variation within the same day
- AQM location: downwind from flight path
Work In Progress
Sample output data structure:
| PM10 | 04:00 | 05:00 | 06:00 | 07:00 | 08:00 |
|---|---|---|---|---|---|
| 2025-03-01 | 30 | 35 | 41 | 59 | 55 |
| 2025-03-02 | 27 | 33 | 37 | 34 | 49 |
| 2025-03-02 | 31 | 34 | 38 | 40 | 50 |
A future research could look at comparing pollution data across different days; for example, a workday Monday against a school holiday Monday whilst controlling for cofounding variables (i.e., only days with similar meteorological conditions)