Figure 1
Bias in mobility data due to normalisation by a single day (provided by Apple), namely Monday, 13 January, the first day in the dataset.
Source: https://covid19.apple.com/mobility/.
Figure 2
Changes in direction requests (driving, public transport, walking) from 13 January to the end of December 2020 (%).
Note: The mobility changes for lockdown periods (first and second) are shown in dark grey. The Apple mobility data are normalised by taking 62 days from 13 January to 14 March as the baseline.
Source: https://covid19.apple.com/mobility/.
Table 1
Average mobility reduction (for driving, public transport and walking) for the entire year 2020 and the two lockdown periods for the four cities.
| CITY | PERIOD | DATES | AVERAGE MOBILITY REDUCTION (%) | ||
|---|---|---|---|---|---|
| DRIVING | PUBLIC TRANSPORT | WALKING | |||
| Greater London | Year | 13 January–30 December | –23% | –45% | –43% |
| First lockdown | 23 March–10 May | –69% | –86% | –78% | |
| Second lockdown | 5 November–2 December | –36% | –58% | –59% | |
| Cardiff | Year | 13 January–30 December | –27% | –51% | –33% |
| First lockdown | 23 March–10 May | –72% | –84% | –76% | |
| Second lockdown | 23 October–9 November | –43% | –64% | –47% | |
| Edinburgh | Year | 13 January–30 December | –29% | –52% | –42% |
| First lockdown | 23 March–10 May | –75% | –89% | –82% | |
| Second lockdown | 20 November–11 December | –39% | –59% | –58% | |
| Belfast | Year | 13 January–30 December | –16% | –47% | –29% |
| First lockdown | 23 March–10 May | –62% | –83% | –68% | |
| Second lockdown | 27 November–11 December | –19% | –61% | –42% | |
[i] Note: Apple mobility data are normalised by taking 62 days from 13 January to 14 March as a baseline.
Source: Apple Mobility Reports.
Figure 3
Mean daily time-series analysis of the air pollutants (a) NO2 and (b) PM2.5 for 2020 and the period 2010–19 (10 years as the baseline) for each of the four cities.
Note: Dark grey highlighted columns show the periods of the first and second lockdowns.
Figure 4
Comparison between the yearly average concentration of NO2 for all stations for each city during the two lockdown periods and the previous 10 years.
Note: Dashed lines show the 10-year averages (2010–19); and broken lines show the five-year mean (trend for Greater London).
Table 2
Average air-pollution reduction (NO2 and PM2.5) for all ground-level monitoring stations as well as urban-background stations separately (in parentheses) for the entire year 2020 as well as for two lockdown periods for the four cities.
| CITY | PERIOD | DATES | AIR POLLUTION REDUCTION (%) | |
|---|---|---|---|---|
| NO2 (URBAN BACKGROUND) | PM 2.5 (URBAN BACKGROUND) | |||
| Greater London | Year | 13 January–30 December | –42% (–38%) | –31% (–28%) |
| First lockdown | 23 March–10 May | –47% (–39%) | –13% (–8%) | |
| Second lockdown | 5 November–2 December | –35% (–29%) | 0% (0%) | |
| Cardiff | Year | 13 January–30 December | –32% (–41%) | –35% (–30%) |
| First lockdown | 23 March–10 May | –36% (–50%) | –5% (–4%) | |
| Second lockdown | 23 October–9 November | –43% (–53%) | –10% (–10%) | |
| Edinburgh | Year | 13 January–30 December | –27% (–49%) | –42% (–42%) |
| First lockdown | 23 March–10 May | –54% (–53%) | –35% (–37%) | |
| Second lockdown | 20 November–11 December | –18% (–38%) | –24% (–24%) | |
| Belfast | Year | 13 January–30 December | –23% (–24%) | –33% (–33%) |
| First lockdown | 23 March–10 May | –43% (–35%) | –33% (–32%) | |
| Second lockdown | 27 November–11 December | –23% (–21%) | –14% (–14%) | |
Figure 5
Comparison between the yearly average concentration of PM2.5 for all stations for each city and for the two lockdown periods.
Note: Dashed lines show the 10-year average (2010–19); and broken lines show the five-year mean (trend for Greater London).
Table 3
Two modelled predictions—linear regression (LR) and gradient boosting regression (GBR)—for air pollutants in 2020 compared with ground-level concentrations.
| ANNUAL MEAN CONCENTRATION | 2020 GROUND-LEVEL MEASURED DATA (μg m–3) | LR PREDICTION (μg m–3) | MEAN ABSOLUTE ERROR, LR (μg m–3) | RELATIVE CHANGE (%), LR PREDICTION | GBR PREDICTION (μg m–3) | MEAN ABSOLUTE ERROR, GBR (μg m–3) | RELATIVE CHANGE (%), GBR, PREDICTION | |
|---|---|---|---|---|---|---|---|---|
| NO2 | Greater London | 25.3 | 30.6 | 5.88 | –18% | 31.9 | 5.31 | –21% |
| Cardiff | 17.8 | 24.1 | 8.95 | –26% | 21.9 | 8.75 | –19% | |
| Edinburgh | 19.4 | 42.9 | 9.63 | –55% | 33.2 | 7.96 | –41% | |
| Belfast | 27.7 | 38.9 | 8.05 | –29% | 37.8 | 7.93 | –27% | |
| PM2.5 | Greater London | 9.2 | 9.7 | 4.36 | –5% | 8.6 | 3.73 | +7% |
| Cardiff | 7.5 | 9.7 | 4.64 | –23% | 7.5 | 4.34 | –1% | |
| Edinburgh | 4.5 | 6.0 | 2.96 | –26% | 5.3 | 2.69 | –15% | |
| Belfast | 6.8 | 7.7 | 3.02 | –12% | 8.0 | 3.02 | –14% | |
[i] Note: Mean absolute errors and relative changes (%) are given for each model prediction.
Figure 6
The 95% confidence intervals for the weather-corrected machine learning (ML) model for the four cities and for the predicted NO2.
Figure 7
Air pollution anomalies (relative change in percentage compared with the baseline) for NO2 (weather-corrected models) for each city in relation to changes (%) in driving, public transport and walking.
Note: For improved clarity, both mobility data and air pollution data are aggregated weekly to smooth out the curves.
Table 4
Spearman’s rank correlation coefficients (rs) between two predicted models for air pollutants (NO2 and PM2.5) and mobility (driving and public transport) for each of the four cities. The results are compared with the five-year mean daily and five-year mean weekly trends.
| CITY | POLLUTANT | SPEARMAN | FIVE-YEAR MEAN (WEEKLY) | LR (DAILY, AGGREGATED WEEKLY) | GBR (DAILY, AGGREGATED WEEKLY) | |||
|---|---|---|---|---|---|---|---|---|
| DRIVING | PUBLIC TRANSPORT | DRIVING | PUBLIC TRANSPORT | DRIVING | PUBLIC TRANSPORT | |||
| Greater London | NO2 | rs | 0.16 | 0.18 | 0.58 | 0.67 | 0.60 | 0.69 |
| p-value | 0.26 | 0.22 | <0.0001 | <0.0001 | <0.0001 | <0.0001 | ||
| PM2.5 | rs | –0.19 | –0.30 | –0.04 | –0.10 | –0.22 | –0.26 | |
| p-value | 0.17 | 0.03 | 0.80 | 0.49 | 0.13 | 0.06 | ||
| Cardiff | NO2 | rs | 0.31 | 0.33 | 0.34 | 0.40 | 0.32 | 0.34 |
| p-value | 0.02 | 0.02 | 0.01 | <0.01 | 0.02 | 0.01 | ||
| PM2.5 | rs | –0.10 | –0.16 | –0.28 | –0.31 | –0.22 | –0.25 | |
| p-value | 0.48 | 0.26 | 0.05 | 0.03 | 0.13 | 0.09 | ||
| Edinburgh | NO2 | rs | 0.29 | 0.24 | 0.50 | 0.53 | 0.54 | 0.57 |
| p-value | 0.04 | 0.10 | <0.001 | <0.0001 | <0.0001 | <0.0001 | ||
| PM2.5 | rs | –0.09 | –0.13 | –0.11 | –0.01 | –0.22 | –0.15 | |
| p-value | 0.52 | 0.37 | 0.47 | 0.93 | 0.14 | 0.33 | ||
| Belfast | NO2 | rs | 0.54 | 0.59 | 0.60 | 0.66 | 0.64 | 0.73 |
| p-value | <0.0001 | <0.0001 | <0.0001 | <0.0001 | <0.0001 | <0.0001 | ||
| PM2.5 | rs | 0.09 | –0.02 | –0.28 | –0.25 | –0.27 | –0.21 | |
| p-value | 0.55 | 0.91 | 0.07 | 0.10 | 0.07 | 0.16 | ||
[i] Note: GBR = gradient boosting regression; and LG = linear regression.
Table 5
Health impact assessments for the four cites based on attributable fraction (AF) of all-cause mortality associated with NO2 and PM2.5 (%) for 2020 and the previous 10 years.
| POLLUTANT | CITY | AF OF ALL-CAUSE MORTALITY ASSOCIATED WITH EACH POLLUTANT (%) | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2010 | 2011 | 2012 | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | ||
| NO2 | Greater London | 9.93% | 8.90% | 9.73% | 8.49% | 8.90% | 7.86% | 8.07% | 8.07% | 7.23% | 6.81% | 5.10% |
| Cardiff | 7.23% | 5.95% | 5.95% | 5.74% | 5.53% | 5.95% | 5.10% | 4.45% | 4.01% | 4.88% | 3.35% | |
| Edinburgh | 6.81% | 5.53% | 5.31% | 4.88% | 4.23% | – | 4.45% | 4.45% | 4.01% | 4.66% | 2.91% | |
| Belfast | 7.65% | 6.17% | 6.38% | 6.81% | 6.17% | 6.38% | 6.59% | 5.53% | 5.95% | 5.31% | 4.01% | |
| PM2.5 | Greater London | 8.37% | 9.43% | 8.90% | 7.83% | 8.37% | 6.21% | 6.21% | 6.21% | 6.21% | 6.21% | 5.11% |
| Cardiff | 7.83% | 7.30% | 6.75% | 7.83% | 6.75% | 5.66% | 5.66% | 5.11% | 5.66% | 6.75% | 4.00% | |
| Edinburgh | 5.66% | 6.75% | 6.21% | 4.55% | 5.11% | 3.44% | 3.44% | 4.00% | 3.44% | 3.44% | 2.30% | |
| Belfast | 7.30% | 7.83% | 5.66% | 6.75% | 6.21% | 5.11% | 5.66% | 5.11% | 5.66% | 6.21% | 4.00% | |