SUPPLEMENTARY INFORMATION A tale of two cities: is air ... · Zone rurale Sud-Est - Forêt de FONTAINEBLEAU RUR_SE Background 59.6 PM 10,PM 2.5,NO X,NO 2 Quai des Célestins CELES

1

SUPPLEMENTARY INFORMATION

for the manuscript

A tale of two cities: is air pollution improving in Paris and London? Anna Font1,2*, Lionel Guiseppin3, Marta Blangiardo2,4, Véronique Ghersi3 and Gary W.

Fuller1,2

1 School of Population Health & Environmental Sciences, Faculty of Life Sciences & Medicine,

King's College London, UK 2MRC PHE Centre for Environment and Health UK

3Observatoire de l’air en Île-de-France, Airparif, France 4 Department of Epidemiology and Biostatistics, School of Medicine, Imperial College London, UK

*Corresponding author; [email protected]; 150 Stamford Street London SE1 9NH, United

Kingdom

2

List of Supplementary Tables

Supplementary Table 1. List of monitoring sites included in the trend analysis belonging to the Airparif network in

the Île-de-France region. Distance to central Paris is indicated. 5

Supplementary Table 2. List of monitoring sites included in the trend analysis belonging to the London Air Quality

Network or the Automatic Urban and Rural Network in London. Distance to central London is indicated. 7

Supplementary Table 3. Average Annual Daily Flows (AADF) (no. vehicles per day) for roads in inner Paris and for

roads in the ring-road (Périphérique). 21

Supplementary Table 4. Proportion of each vehicle category in the fleets of Paris and London; and average daily flows

in inner Paris and London. Proportions of each category in London calculated using those traffic counters near an air

quality monitoring site. 21

Supplementary Table 5. Trends (mean ± standard) deviation in Average Annual Daily Flow (AADF, vehicles day-1)

expressed in absolute and percentage change. *** significant at the 0.001 level; ** significant at the 0.05 level. 21

Supplementary Table 6. Trends in Average Annual Daily Flow (AADF, #vehicles day-1) expressed in absolute and

percentage change for the 2005 – 09 and 2010 – 16 periods in inner Paris roads and in the ring-road (Périphérique).

** significant at the 0.01 level; + significant at the 0.1; (blank) not statistically significant. 22


percentage change per each vehicle category for the 2005 – 09 and 2010 – 16 periods. Trends are expressed in

overall mean ± standard deviation as calculated by means of the REM for the roads in London near an air quality

monitoring station. 22

Supplementary Table 8. List of the assessed linear mixed-effects models of annual roadside increment (inc) as a

function of traffic variables: traffic flow (AADF) and traffic flow for vehicle categories and Euro norms. All are random

intercept models considering all roadside sites in Paris and London are included as random factors. β0 is the

intercept; us is the random effects at roadside site level; and Ɛ is the model residuals. 23

Supplementary Table 9. cAIC and R2 values for each of the linear-mixed effect models evaluated. The selected model

is indicated in bold. The number of observations (N) and the number of groups (random effect or monitoring sites)

are also indicated. 23

Supplementary Table 10. Model formulation for the selected linear mixed-model for each pollutant. 23

Supplementary Table 11. Statistical parameters of the selected linear-mixed effect models for incPM2.5; and also the

models without AADF heavy vehicles, AADF motorcycles and AADF Euro 4 light diesels. Values denote the 95%

confidence interval. Units for fixed effects are expressed per 1000 vehicles. Bold numbers indicate significant at 95%

level. 25

Supplementary Table 12. Statistical parameters of the selected linear-mixed effect models for incPM10; and also the

models without AADF Euro IV heavy vehicles, AADF Euro 4 light diesels, AADF Euro V heavy vehicles, AADF Euro 5

light diesels. Values denote the 95% confidence interval. Units for fixed effects are expressed per 1000 vehicles. Bold

numbers indicate significant at 95% level. 25

3

List of Supplementary Figures

Supplementary Figure 1. Localization of the monitoring sites in Île-de-France. 4

Supplementary Figure 2. Localization of the monitoring sites in London. 6

Supplementary Figure 3. Forest plots for the trends in NOX in urban background sites (A) and roadside locations (B)

in Île-de-France for the 2005-2009 period (expressed in µg m-3 year-1). *** significant at the 0.001 level; **

significant at the 0.01 level; * significant at the 0.05 level; + significant at the 0.1 level; (blank) not statistically

significant. Overall (RE) refers to the mean trend for all sites. 10

Supplementary Figure 4. Same as Supplementary Figure 3 but for NO2 trends. 11

Supplementary Figure 5. Same as Supplementary Figure 3 but for PM10 trends. 11

Supplementary Figure 6. Forest plots for the trends in PM2.5 in urban background sites for the 2005-2009 period

(expressed in µg m-3 year-1). 12


in Île-de-France for the 2010-2016 period (expressed in µg m-3 year-1). *** significant at the 0.001 level; **

significant at the 0.01 level; * significant at the 0.05 level; + significant at the 0.1 level; (blank) not statistically

significant. Overall (RE) refers to the mean trend for all sites. 12



Supplementary Figure 10. Same as Supplementary Figure 7 but for PM2.5 trends. 14


in London for the 2005-2009 period (expressed in µg m-3 year-1). *** significant at the 0.001 level; ** significant at

the 0.01 level; + significant at the 0.1 level; (blank) not statistically significant. Overall (RE) refers to the mean trend

for all sites. 14



Supplementary Figure 14. Same as but for PM2.5 trends. 16


in London for the 2010-2016 period (expressed in µg m-3 year-1). *** significant at the 0.001 level; ** significant at

the 0.01 level; * significant at the 0.05; + significant at the 0.1 level; (blank) not statistically significant. Overall (RE)

refers to the mean trend for all sites. 17



Supplementary Figure 18. Same as but for PM2.5 trends. 19

Supplementary Figure 19. Trends in roadside PM2.5 (incPM2.5) in two roads in London. 20

Supplementary Figure 20. Proportion of petrol and diesel cars (A) and LGVs (B) in Paris and London. 22

Supplementary Figure 21. Pairwise Pearson correlation values between the explanatory variables for the selected

model linear mixed-effect model for each pollutant. 24

Supplementary Figure 22. Trends in roadside increments in 2010-16 for each pollutant and the annual mean

concentration in 2010. Colour scale shows distance to the city centre. 26

Supplementary Figure 23. Trends in PM2.5 vs trends in PM10 for collocated measurements in Paris and London.

Background sites are indicated with a white dot. Distance to the city centre is marked by the colour scale. 27

Supplementary Figure 24. Trends in traffic increments vs tends in AADF for different vehicle categories for 2010-16

in London. Colour dots indicate distance from the city centre. Blue solid line and grey area indicate the linear

regression model and the 95% confidence interval. 28

4

S1. Network of air quality monitoring sites in Île-de-France and

London Trends in nitrogen oxides (NOX), nitrogen dioxide (NO2) and particulate matter (PM10 and PM2.5)

have been calculated for background and roadside locations in Île-de-France (Supplementary Figure

1) and London (Supplementary Figure 2. Site information, pollutants measured and distance to the

city centre are indicated in Supplementary Table 1 and Supplementary Table 2.

Supplementary Figure 1. Localization of the monitoring sites in Île-de-France.

5

Supplementary Table 1. List of monitoring sites included in the trend analysis belonging to the Airparif network in

the Île-de-France region. Distance to central Paris is indicated.

Site name Site code Type Distance

(km)

Pollutants

measured

Paris_4 ème PA4C Background 0.7 PM10,PM2.5,NOX,NO2

Paris_1er PA1H Background 1.0 PM10,PM2.5

PARIS 13ème PA13 Background 2.9 PM10,NOX,NO2

Paris_12eme PA12 Background 3.8 NOX,NO2

Paris_7 ème PA07 Background 4.1 NOX,NO2

Paris 18ème PA18 Background 4.3 PM10,NOX,NO2

Ivry-sur-Seine IVRY Background 5.2 NOX,NO2

Neuilly-sur-Seine NEUIL Background 6.1 NOX,NO2

Aubervilliers AUB Background 6.2 NOX,NO2

Vitry-sur-Seine VITRY Background 8.9 PM10,PM2.5,NOX,NO2

La Défense DEF Background 8.9 PM10,NOX,NO2

St_Denis STDE Background 9.2 NOX,NO2

Bobigny BOB Background 9.4 PM10,PM2.5,NOX,NO2

Gennevilliers GEN Background 9.4 PM10,PM2.5,NOX,NO2

NOGENT NOG Background 10.0 PM10

Champigny CHAMP Background 12.9 NOX,NO2

Argenteuil ARG Background 14.3 NOX,NO2

Gonesse GON Background 16.8 PM10,NOX,NO2

Versailles VERS Background 17.0 NOX,NO2

MONTGERON MONT Background 18.2 NOX,NO2

Tremblay-en-France TREMB Background 20.1 PM10,NOX,NO2

LOGNES LOGN Background 21.0 PM10,NOX,NO2

EVRY EVRY Background 25.8 NOX,NO2

Zone Rurale Nord RUR_N Background 27.5 PM2.5

CERGY CERGY Background 31.1 PM10

MELUN MELU Background 41.6 NOX,NO2

Rambouillet RAMB Background 45.2 PM10,PM2.5

Mantes_La_Jolie MANT Background 49.8 NOX,NO2

Zone rurale Sud - BOIS-HERPIN RUR_S Background 55.2 PM10,PM2.5

Zone rurale Sud-Est - Forêt de FONTAINEBLEAU RUR_SE Background 59.6 PM10,PM2.5,NOX,NO2

Quai des Célestins CELES Roadside 0.8 NOX,NO2

Rue de Bonaparte BONAP Roadside 1.1 NOX,NO2

Place de l'Opéra OPERA Roadside 2.2 PM10,NOX,NO2

Boulevard Haussmann HAUS Roadside 2.6 PM10,NOX,NO2

Avenue des Champs-Elysées ELYS Roadside 3.2 PM10,NOX,NO2

Place Victor Basch BASCH Roadside 3.3 PM10,NOX,NO2

Boulevard Soult SOULT Roadside 4.7 NOX,NO2

Boulevard périphérique Est BP_EST Roadside 5.0 PM10,PM2.5,NOX,NO2

RN2-Pantin RN2 Roadside 6.2 PM10,NOX,NO2

Boulevard Périphérique Porte d'Auteuil AUT Roadside 7.1 PM10,PM2.5,NOX,NO2

Autoroute A1 A1 Roadside 8.0 PM10,PM2.5,NOX,NO2

RN20-Montlhéry RN20 Roadside 24.4 NOX,NO2

RN 6-Melun RN6 Roadside 42.6 PM10,PM2.5,NOX,NO2

RD934-Coulomiers RD934 Roadside 54.2 PM10

6

Supplementary Figure 2. Localization of the monitoring sites in London.

7

Supplementary Table 2. List of monitoring sites included in the trend analysis belonging to the London Air Quality

Network or the Automatic Urban and Rural Network in London. Distance to central London is indicated.

Site name Site code Type Distance

(km)

Pollutants measured Nearest DfT

traffic

counter

Barking and Dagenham - Scrattons Farm BG2 Background 18.1 NOX, NO2 --

Camden - Bloomsbury BL0 Background 1.5 NOX, NO2, PM10, PM2.5 --

Barnet - Finchley BN2 Background 10.8 NOX, NO2 --

Bexley - Belvedere West BQ7 Background 18.3 NOX, NO2 --

Bexley - Belvedere West FDMS BQ8 Background 18.3 PM10 --

Brent - Kingsbury BT1 Background 13.8 NOX, NO2 --

Bexley - Belvedere FDMS BX0 Background 19.8 PM10 --

Bexley - Slade Green BX1 Background 22.1 NOX, NO2 --

Bexley - Belvedere BX2 Background 19.8 NOX, NO2 --

Bexley - Thamesmead BX3 Background 17.1 PM10, PM2.5 --

Bexley - Slade Green FDMS BX9 Background 22.1 PM2.5 --

Camden - St Martins College (NOX 1) CD4 Background 1.2 NOX, NO2 --

Camden - St Martins College (NOX 2) CD5 Background 1.2 NOX, NO2 --

Croydon - Thornton Heath CR3 Background 11.8 PM10 --

City of London - Senator House CT1 Background 2.1 NOX, NO2 --

City of London - Sir John Cass School CT3 Background 3.4 NOX, NO2 --

Ealing - Ealing Town Hall EA1 Background 12.7 NOX, NO2 --

Ealing - Southall EA7 Background 18.5 PM10, PM2.5 --

Enfield - Bush Hill Park EN1 Background 15.7 NOX, NO2 --

Enfield - Prince of Wales School EN7 Background 19.2 NOX, NO2 --

Greenwich - Eltham GR4 Background 15 PM10, PM2.5, NOX, NO2 --

Haringey - Priory Park HG2 Background 8.6 NOX, NO2 --

Haringey - Priory Park South HG4 Background 8.4 NOX, NO2 --

Hillingdon - Keats Way HI0 Background 23.3 NOX, NO2, PM10 --

Hackney - Clapton HK4 Background 7.4 NOX, NO2, PM2.5 --

Harrow - Stanmore HR1 Background 17.1 NOX, NO2 --

Hounslow - Cranford HS2 Background 20.1 PM10, NOX, NO2 --

Hounslow - Hatton Cross HS7 Background 21.6 NOX, NO2, PM10 --

Islington - Arsenal IS6 Background 5.6 NOX, NO2, PM10 --

Kensington and Chelsea - North Ken KC1 Background 6.3 NOX, NO2, PM2.5 --

Kensington and Chelsea - North Ken FDMS KC7 Background 6.3 PM10 --

Lambeth - Loughborough Junction LB3 Background 5.2 NOX, NO2 --

Lambeth - Streatham Green LB6 Background 8.9 NOX, NO2 --

Hillingdon - Harlington LH0 Background 22.1 NOX, NO2 --

Lewisham - Catford LW1 Background 10.1 NOX, NO2 --

Newham - Wren Close NM3 Background 9.7 NOX, NO2 --

Redbridge - Perth Terrace RB1 Background 15.9 NOX, NO2 --

Redbridge - Ley Street RB7 Background 16 NOX, NO2 --

Richmond Upon Thames - Barnes Wetlands RI2 Background 8.1 NOX, NO2, PM10 --

Southwark - Elephant and Castle SK6 Background 2.4 NOX, NO2, PM10 --

Sutton - Carshalton ST3 Background 16.2 NOX, NO2 --

Richmond Upon Thames - Ntl Physical Lab TD0 Background 17.9 NOX, NO2, PM2.5 --

Richmond Upon Thames - Bushy Park TD5 Background 17.9 PM10, PM2.5 --

Tower Hamlets - Poplar TH1 Background 7.3 NOX, NO2, PM10 --

Tower Hamlets - Victoria Park TH5 Background 7.3 NOX, NO2, PM10 --

Wandsworth - Wandsworth Town Hall WA2 Background 7.3 NOX, NO2 --

Wandsworth - Putney WA9 Background 7.9 NOX, NO2, PM10 --

Waltham Forest - Dawlish Road WL1 Background 10.3 NOX, NO2, PM10 --

Westminster - Horseferry Road WM0 Background 1.6 NOX, NO2, PM10 --

Westminster - Victoria WM8 Background 1.7 NOX, NO2 28513

Barking and Dagenham - North Street BG3 Roadside 14.3 NOX, NO2 79991

Barnet - Tally Ho Corner BN1 Roadside 12.3 NOX, NO2 47351

Brent - Ikea BT4 Roadside 10.4 NOX, NO2 74693

Brent - John Keble Primary School BT6 Roadside 9.1 NOX, NO2 8467

Bexley - Thames Road North FDMS BX6 Roadside 23 PM10 99907

Bexley - Thames Road North BX7 Roadside 23 NOX, NO2 99907

Bexley - Thames Road South BX8 Roadside 22.9 NOX, NO2 99907

Bromley - Harwood Avenue BY7 Roadside 15.2 NOX, NO2 8737

Camden - Swiss Cottage CD1 Roadside 5.3 PM10, PM2.5, NOX, NO2 48537

8

Camden - Shaftesbury Avenue CD3 Roadside 0.8 NOX, NO2 75096

Camden - Euston Road CD9 Roadside 2.2 PM10, PM2.5 56815

Croydon - Purley Way CR2 Roadside 16.2 NOX, NO2 36271

Croydon - George Street CR4 Roadside 15.1 NOX, NO2, PM10 28766

Croydon - Norbury CR5 Roadside 10.8 NOX, NO2 16252

Croydon - Purley Way A23 CR7 Roadside 16.3 NOX, NO2 36271

City of London - Farringdon Street CT2 Roadside 1.6 NOX, NO2 6751

City of London - Beech Street CT4 Roadside 2.4 NOX, NO2 26078

City of London - Walbrook Wharf CT6 Roadside 2.4 NOX, NO2 75400

City of London - Upper Thames Street CT8 Roadside 2.6 PM10 75401

Crystal Palace - Crystal Palace Parade CY1 Roadside 9.9 NOX, NO2 38568

Ealing - Acton Town Hall FDMS EA0 Roadside 9.9 PM10 37718

Ealing - Acton Town Hall EA2 Roadside 9.9 PM10, PM2.5 37718

Ealing - Hanger Lane Gyratory EA6 Roadside 11.9 NOX, NO2, PM10 58192

Ealing - Western Avenue EI1 Roadside 9.9 NOX, NO2, PM10 81091

Ealing - Southall Railway EI2 Roadside 17.7 NOX, NO2, PM10 47610

Enfield - Derby Road EN4 Roadside 12.9 NOX, NO2 74637

Enfield - Bowes Primary School EN5 Roadside 11.7 NOX, NO2 57115

Greenwich and Bexley - Falconwood FDMS GB0 Roadside 15.8 PM10, PM2.5 26102

Greenwich and Bexley - Falconwood GB6 Roadside 15.8 PM10, PM2.5 26102

Greenwich - A206 Burrage Grove GN0 Roadside 14 NOX, NO2 47943

Greenwich - Plumstead High Street GN3 Roadside 15.5 NOX, NO2, PM10, PM2.5 6760

Greenwich - Fiveways Sidcup Rd A20 GN4 Roadside 15.5 PM10, NOX, NO2 73048

Greenwich - Trafalgar Road GR5 Roadside 9.1 NOX, NO2, PM10 26784

Greenwich - Blackheath GR7 Roadside 8.8 PM10, NOX, NO2 73556

Greenwich - Woolwich Flyover GR8 Roadside 10.2 PM10, PM2.5 46792

Greenwich - Westhorne Avenue GR9 Roadside 12.9 PM10, PM2.5 8576

Hammersmith and Fulham - Shepherds Bush HF4 Roadside 6.9 NOX, NO2 36822

Haringey - Haringey Town Hall HG1 Roadside 10.9 PM10, PM2.5, NOX, NO2 6639

Hillingdon - South Ruislip HI1 Roadside 19.9 NOX, NO2, PM10 38596

Hillingdon - Hillingdon Hospital HI2 Roadside 23.2 NOX, NO2, PM10 56742

Hillingdon - Oxford Avenue HI3 Roadside 20.9 NOX, NO2, PM10 36121

Hackney - Old Street HK6 Roadside 3.4 PM10, PM2.5 8475

Harrow - Pinner Road HR2 Roadside 18.7 NOX, NO2, PM10 57085

Hounslow - Chiswick High Road HS4 Roadside 9.3 PM10, NOX, NO2 16867

Hounslow - Brentford HS5 Roadside 13 NOX, NO2, PM10 27613

Hounslow - Heston Road HS6 Roadside 16.9 NOX, NO2, PM10 75069

Hounslow and Ealing - Gunnersbury Avenue HS8 Roadside 11.1 NOX, NO2, PM10 37112

Havering - Rainham HV1 Roadside 23 PM10, PM2.5 74514

Havering - Romford HV3 Roadside 22.3 PM10, NOX, NO2 28690

Camden - Holborn (inmidtown) IM1 Roadside 1.1 NOX, NO2 75102

Islington - Holloway Road IS2 Roadside 5.3 NOX, NO2, PM10 7213

Kensington and Chelsea - Cromwell Road KC2 Roadside 4 NOX, NO2 6121

Kensington and Chelsea - Knightsbridge KC3 Roadside 2.9 NOX, NO2 26115

Kensington and Chelsea - Kings Road KC4 Roadside 3.8 NOX, NO2 7606

Kensington and Chelsea - Earls Court Rd KC5 Roadside 5 NOX, NO2, PM10 48544

Lambeth - Christchurch Road LB1 Roadside 7.1 NOX, NO2 16743

Lambeth - Brixton Road LB4 Roadside 5 NOX, NO2 36269

Lewisham - New Cross LW2 Roadside 7 PM10, PM2.5, NOX, NO2 56093

Lewisham - Loampit Vale LW4 Roadside 9 NOX, NO2 7558

Merton - Morden Civic Centre ME1 Roadside 12.9 NOX, NO2 48548

Merton - Merton Road ME2 Roadside 11.3 PM10 16258

Westminster - Marylebone Road MY1 Roadside 2.6 NOX, NO2, PM10 27236

Westminster - Marylebone Road FDMS MY7 Roadside 2.6 PM10, PM2.5 27236

Newham - Cam Road NM2 Roadside 9.2 NOX, NO2 46201

Redbridge - Fullwell Cross RB3 Roadside 17.4 NOX, NO2 26670

Redbridge - Gardner Close RB4 Roadside 13.2 NOX, NO2 38683

Redbridge - South Woodford RB5 Roadside 14 NOX, NO2 17022

Richmond Upon Thames - Castelnau RI1 Roadside 8.4 NOX, NO2, PM10 26899

Southwark - A2 Old Kent Road SK5 Roadside 5.5 NOX, NO2 56089

Sutton - Wallington ST4 Roadside 16.7 NOX, NO2 6805

Sutton - Worcester Park ST6 Roadside 16.6 NOX, NO2 18493

Tower Hamlets - Mile End Road TH2 Roadside 6 NOX, NO2 7540

Tower Hamlets - Blackwall TH4 Roadside 8.2 PM10, PM2.5, NOX, NO2 6071

Wandsworth - Putney High Street WA7 Roadside 8 PM10, NOX, NO2 16764

Wandsworth - Putney High Street Facade WA8 Roadside 8 NOX, NO2 16764

Wandsworth - Battersea WAA Roadside 3.4 NOX, NO2, PM10 27666

Waltham Forest - Crooked Billet WL4 Roadside 12.8 PM10, NOX, NO2 7058

Westminster - Charing Cross Library WM4 Roadside 0.3 NOX, NO2 47874

Westminster - Oxford Street WM6 Roadside 2 NOX, NO2 36471

9

S2. Trends in air pollutants

S2.1 Trends calculations

Trends in air pollutants were calculated from hourly time series. From then, monthly means were first

calculated imposing a 75 % data capture threshold; and then de-seasonalized by applying a LOESS

smoothing function (Cleveland et al., 1990) after being linearly interpolating to fill missing monthly

values. Linear trends were calculated from the monthly means using the Theil-Sen method (Theil

1950; Sen 1968) available in the R-Openair package (version 2.1-0). Confidence intervals at 95% and

the p-values were calculated by bootstrap sampling. A statistically significant trend was assumed

when p < 0.1 (represented with a ‘+’ symbol) and different significance codes were assigned (p <

0.05, p < 0.01 and p < 0.001 marked by ‘*’, ‘**’ and ‘***’, respectively). The overall trend was then

calculated by fitting the linear Random-Effects (RE) model “DerSimonian-Laird estimator” (R-

metafor package version 1.9-9; Viechtbauer 2010). Trends at measurement sites with at least 75%

data capture in the monthly means for each time period were included in the calculation of the overall

trend independently of their significance. The graphical representation of the distribution of trends

across the network along with individual confidence intervals and the overall trend was done through

“Forest plots”. Examples of Forest plots are shown in Figure 2 and 3 in the main text; and

Supplementary Figure 3-Supplementary Figure 18.

Trends in roadside increments were calculated (incNOX, incNO2, incPM10 and incPM2.5) by

subtracting a background concentration from each hourly roadside measurement. The reference

background in central Paris was PA1H (Paris_1er) until 2010; and PA4C (Paris_4eme) from 2011

onwards. Three roadside sites were located outside inner Paris (A1, RN2 and RN6) and their paired

background sites were: STDE for NOX and NO2; and GEN for PM10 and PM2.5 (A1); AUB for NOX

and NO2; BOB for PM10 and PM2.5 (RN2); MELU for NOX and NO2 (for RN6 no background location

was available for PM). For London, one background location was used for all roadside sites: KC1

(Kensington and Chelsea – North Kensington; KC1) as in Font and Fuller (2016). KC1 was chosen

for three reasons: i) the use of a single background site allowed roadside increments to be directly

compared between different roadside locations; ii) it is the urban background site with the longest

complete time series for all pollutants; iii) trends observed at KC1 were the same (within 2σ

confidence interval) of the overall trends observed for all urban background sites in London, with the

exception of trends in NOX and NO2 in 2005-09 when a faster decrease (-2.90 and 2.06 μg m-3 year-1

for NOX and NO2, respectively) was observed compared with the overall urban background trends (-

1.28 and -0.50 μg m-3 year-1 for NOX and NO2, respectively). Given the overall positive trend in

roadside increments in NO2 in 2005-09, the use of other background location instead of KC1 would

have led to more positive trends in the roadside increment.

For measurement site locations see Supplementary Figure 1 (Île-de-France) and Supplementary

Figure 2 (London).

Trends were expressed in absolute values (µg m-3 year-1) and in percentage (% year-1). This latter was

calculated differently in Paris and London. In Paris, Theil-Sen percentage trends were calculated for

individual sites as 100* (Cend/Cstart-1)(end year – start year) where Cstart and Cend were the

concentrations at the start and end date. Overall percentage trends were calculated by means of the

RME estimator. In London, given that some of the increments were very small in some roads, this

method provided some very high percentage changes. Therefore, the overall absolute trend was

divided by the overall annual mean increment in the first year, this latter also calculated using the

RME fit. Individual variances for the annual mean increment were calculated using the propagation of

10

errors methodology using the uncertainty of the instruments used in the calculation of increments: 4%

for NOX and NO2 measurements; 6.2% and 17% for PM measurements done by BAM and TEOM-

FDMS, respectively (Font and Fuller, 2016).

S2.2 Trends in air pollutants in Île-de-France


in Île-de-France for the 2005-2009 period (expressed in µg m-3 year-1). *** significant at the 0.001 level; ** significant

at the 0.01 level; * significant at the 0.05 level; + significant at the 0.1 level; (blank) not statistically significant.

Overall (RE) refers to the mean trend for all sites.

11

Supplementary Figure 4. Same as Supplementary Figure 3 but for NO2 trends.

Supplementary Figure 5. Same as Supplementary Figure 3 but for PM10 trends.

12

Supplementary Figure 6. Forest plots for the trends in PM2.5 in urban background sites for the 2005-2009 period

(expressed in µg m-3 year-1).


in Île-de-France for the 2010-2016 period (expressed in µg m-3 year-1). *** significant at the 0.001 level; ** significant

at the 0.01 level; * significant at the 0.05 level; + significant at the 0.1 level; (blank) not statistically significant.

Overall (RE) refers to the mean trend for all sites.

13



14

Supplementary Figure 10. Same as Supplementary Figure 7 but for PM2.5 trends.

S2.3 Trends in air pollutants in London

Supplementary Figure 11. Forest plots for the trends in NOX in urban background sites (A) and roadside locations

(B) in London for the 2005-2009 period (expressed in µg m-3 year-1). *** significant at the 0.001 level; ** significant at

15

the 0.01 level; + significant at the 0.1 level; (blank) not statistically significant. Overall (RE) refers to the mean trend

for all sites.


16


Supplementary Figure 14. Same as but for PM2.5 trends.

17

Supplementary Figure 15. Forest plots for the trends in NOX in urban background sites (A) and roadside locations

(B) in London for the 2010-2016 period (expressed in µg m-3 year-1). *** significant at the 0.001 level; ** significant at

the 0.01 level; * significant at the 0.05; + significant at the 0.1 level; (blank) not statistically significant. Overall (RE)

refers to the mean trend for all sites.

18



19

Supplementary Figure 18. Same as but for PM2.5 trends.

20

S.2.4 Trends in roadside increments in PM2.5 (incPM2.5) in London

Supplementary Figure 19. Trends in roadside PM2.5 (incPM2.5) in two roads in London.

21

S3. Traffic data

S3.1 Traffic counts in the Parisian network

Supplementary Table 3. Average Annual Daily Flows (AADF) (no. vehicles per day) for roads in inner Paris and for

roads in the ring-road (Périphérique).

Year AADF inner Paris AADF Périphérique

2005 29,521 104,103

2006 28,648 103,418

2007 28,080 103,780

2008 27,281 104,285

2009 27,137 103,461

2010 26,746 101,361

2011 26,504 102,951

2012 25,858 101,669

2013 25,057 101,267

2014 24,470 101,071

2015 23,791 100,609

2016 24,148 100,498

S3.3. Proportion of vehicle categories in Paris and London

Supplementary Table 4. Proportion of each vehicle category in the fleets of Paris and London; and average daily

flows in inner Paris and London. Proportions of each category in London calculated using those traffic counters near

an air quality monitoring site.

Inner Paris London

Year 2005 2010 2012 2014 2005 2010 2012 2014

Motorcycles (%) 6.4 7.4 7.4 7.7 3.6 3.4 3.3 3.5

Buses & coaches (%) 0.4 0.5 0.5 0.6 3.1 3.4 3.6 3.3

HGVs (%) 6.6 6.2 6.1 5.7 4.8 4.5 5.3 5.0

LGVs (%) 15.3 16.0 16.0 15.4 12.8 13.3 13.6 15.3

Cars & taxis (%) 71.1 69.9 70.0 70.6 75.8 75.4 74.2 73.0

AADF

(vehicles/day) 29521 26746 25858 24470 34004 32236 30908 31706

S3.4. Trends in traffic in Paris and London

Supplementary Table 5. Trends (mean ± standard) deviation in Average Annual Daily Flow (AADF, vehicles day-1)

expressed in absolute and percentage change. *** significant at the 0.001 level; ** significant at the 0.05 level.

Inner Paris London

Period Trends vehicles

(AADF year-1)

Trends vehicles

(% year-1)

Trends vehicles

(AADF year-1)

Trends vehicles

(% year-1)

2005 – 09 -613.5 ± 70.4** -2.1 -564.7 ± 73.1*** -1.5

2010 – 16 -521.7 ± 62.3*** -1.9 -53.9 ± 19.4** -0.2

22

S3.5. Trends in vehicles Supplementary Table 6. Trends in Average Annual Daily Flow (AADF, #vehicles day-1) expressed in absolute and

percentage change for the 2005 – 09 and 2010 – 16 periods in inner Paris roads and in the ring-road (Périphérique).

** significant at the 0.01 level; + significant at the 0.1; (blank) not statistically significant.

Inner Paris Ring-road (Périphérique)

Period Trends vehicles

(AADF year-1) Trends vehicles

(% year-1) Trends vehicles

(AADF year-1) Trends vehicles

(% year-1)

2005-09 -613.5 ± 70.4** -2.1 -41.7 ± 137.9 -0.04

2010-16 -521.7 ± 62.3*** -1.9 -281.7 ± 113.8+ -0.3

Trends in AADF per vehicle category were calculated in London using the counts on each vehicle

category for those traffic counters near an air quality monitoring station. Overall trends were

calculated using the REM for each of the time periods. Results are summarized in Supplementary

Table 7.


percentage change per each vehicle category for the 2005 – 09 and 2010 – 16 periods. Trends are expressed in overall

mean ± standard deviation as calculated by means of the REM for the roads in London near an air quality

monitoring station.

2005 - 2009 2010 - 2016

Category

Trends vehicles

(AADF year-1)

Trends vehicles

(% year-1) N

Trends vehicles

(AADF year-1)

Trends vehicles

(% year-1) N

All vehicles -564.7 ± 73.1*** -1.5 79 -53.9 ± 19.4** -0.2 79

Cars & taxis -435.2 ± 52.8*** -1.5 79 -75.9 ± 14.4*** -0.3 79

Bus & coaches 20.0 ± 2.7*** 1.7 79 -6.9 ± 2.4** -0.6 79

Motorcycles -3.8 ± 3.1 -0.3 79 10.7 ± 2.0*** 0.9 79

HGVs -15.3 ± 4.1*** -0.8 79 -9.0 ± 3.2** -0.6 79

LGVs 4.7 ± 9.1 0.1 79 6.0 ± 8.5 0.1 79

S.3.6 Fuel type distribution in Paris and London

Supplementary Figure 20. Proportion of petrol and diesel cars (A) and LGVs (B) in Paris and London.

23

S.4 Linear-mixed effect models

Supplementary Table 8. List of the assessed linear mixed-effects models of annual roadside increment (inc) as a

function of traffic variables: traffic flow (AADF) and traffic flow for vehicle categories and Euro norms. All are

random intercept models considering all roadside sites in Paris and London are included as random factors. β0 is the

intercept; us is the random effects at roadside site level; and Ɛ is the model residuals.

model Formulation

null inc ~ β0 + us + Ɛ

1 inc ~ β0 + β1 · AADF + us + Ɛ

2 inc ~ β0 + β1 · AADF + β2 · LEZ + us + Ɛ

3 inc ~ β0 + β1 · AADF + β2 · LEZ + β3 · city + us + Ɛ

4 inc ~ β0 + β1 · aadf.LightDiesels + β2 · aadf.HeavyVehicles + β3 · aadf.Motorcycles + us + Ɛ

5 inc ~ β0 + β1 · aadf.Euro4.LightDiesels + β2 · aadf.Euro5.LightDiesels + β3 · aadf.HeavyVehicles +

β4 · aadf.Motorcycles + us + Ɛ

6 inc ~ β0 + β1 · aadf.LightDiesels + β2 · aadf.Euro3.HeavyVehicles + β3 ·

aadf.Euro4.HeavyVehicles + β4 · aadf.Euro5.HeavyVehicles + β5 · aadf.Motorcycles + us + Ɛ

7 inc ~ β0 + β1 · aadf.Euro4.LightDiesels + β2 · aadf.Euro5.LightDiesels + β2 ·

aadf.Euro3.HeavyVehicles + β3 · aadf.Euro4.HeavyVehicles + β5 · aadf.Euro5.HeavyVehicles +

β6 · aadf.Motorcycles + us + Ɛ

8 inc ~ β0 + β1 · aadf.Euro4.LightDiesels + β2 · aadf.Euro5.LightDiesels + β2 ·

aadf.Euro3.HeavyVehicles + β3 · aadf.Euro4.HeavyVehicles + β5 · aadf.Euro5.HeavyVehicles +

β6 · aadf.Euro2.Motorcycles + β7 · aadf.Euro3.Motorcycles + us + Ɛ

Supplementary Table 9. cAIC and R2 values for each of the linear-mixed effect models evaluated. The selected model

is indicated in bold. The number of observations (N) and the number of groups (random effect or monitoring sites)

are also indicated.

incNOX incNO2 incPM10 incPM2.5

N groups N groups N groups N groups

464 67 463 67 390 59 45 12

model cAIC R2 cAIC R2 cAIC R2 cAIC R2

null 3931.1 0.971 3241.4 0.934 1830.6 0.785 209.4 0.570

1 3921.0 0.972 3222.3 0.943 1827.7 0.785 208.3 0.598

2 3922.4 0.971 3222.0 0.944 1828.6 0.784 208.1 0.633

3 3922.5 0.973 3222.2 0.945 1828.1 0.799 208.1 0.633

4 3911.9 0.970 3228.3 0.931 1837.1 0.750 209.1 0.621

5 3839.3 0.974 3140.0 0.946 1808.2 0.776 182.8 0.796

6 3846.5 0.975 3088.3 0.960 1830.1 0.764 195.7 0.738

7 3840.1 0.975 3098.7 0.954 1800.9 0.781 186.7 0.795

8 3843.5 0.975 3094.8 0.957 1798.6 0.801 190.7 0.779

Supplementary Table 10. Model formulation for the selected linear mixed-model for each pollutant.

model Formulation

#5 incNOX ~ 93.8 + 0.3 · aadf.Euro4.LightDiesels – 2.4 · aadf.Euro5.LightDiesels + 3.5 ·

aadf.HeavyVehicles + 4.8 · aadf.Motorcycles + us + Ɛ

#6 incNO2 ~ 17.4 + 0.6 · aadf.LightDiesels + 1.7 · aadf.Euro3.HeavyVehicles + 2.9 ·

aadf.Euro4.HeavyVehicles – 4.1 · aadf.Euro5.HeavyVehicles + 1.8 · aadf.Motorcycles + us + Ɛ

#8

incPM10 ~ 4.2 + 0.1 · aadf.Euro4.LightDiesels – 0.4 · aadf.Euro5.LightDiesels + 0.7 ·

aadf.Euro3.HeavyVehicles – 0.5· aadf.Euro4.HeavyVehicles + 1.2 · aadf.Euro5.HeavyVehicles +

3.5 · aadf.Euro2.Motorcycles + 0.2 · aadf.Euro3.Motorcycles + us + Ɛ

#5 incPM2.5 ~ 1.2 – 0.002· aadf.Euro4.LightDiesels – 0.3 · aadf.Euro5.LightDiesels + 0.4 ·

aadf.HeavyVehicles + 1.2 · aadf.Motorcycles + us + Ɛ

24

The Pearson correlation value between all the explanatory variables used in each model were

calculated to explore possible collinearity. A threshold of r>|0.70| was suggested to indicate possible

collinearity (Dormann et al., 2013). Each pollutant had a different model formulation and then

different explanatory variables. The explanatory variables included in each model is shown in

Supplementary Table 10. Correlation values are shown in Supplementary Figure 21. Note that NOX

and PM2.5 had the same model formulation but they do not share the same roads and the same length

of years and hence the different correlation values between explanatory variables. The majority of

correlation values were all below r< |0.7| with few exceptions: AADF Motorcycles - AADF Heavy

Vehicles (r = 0.82) ; AADF Euro 4 light diesels - AADF Heavy vehicles (r = 0.8); and AADF Euro 4

light diesels - AADF Motorcycles (r = 0.77) for the PM2.5 model; and AADF Euro 5 light diesels -

AADF Euro V heavy vehicles (r = 0.81); and AADF Euro 4 light diesels - AADF Euro IV heavy

vehicles (r = 0.76) for the PM10 model (Supplementary Figure 21).

Supplementary Figure 21. Pairwise Pearson correlation values between the explanatory variables for the selected

model linear mixed-effect model for each pollutant.

In order to evaluate the impact of correlation between those explanatory variables with r >0.7, linear-

mixed effect models were re-run excluding one of the variables at a time and coefficients estimated

(Supplementary Table 12 for the PM2.5 model and Supplementary Table 12 for the PM10 model).

The magnitude of the coefficients did not change in any of the different models for incPM2.5 with the

exception of AADF Heavy Vehicles unadjusted for AADF motorcycles. In this case, the coefficient

increased as well as its confidence interval indicating positive effect on incPM2.5. The results indicate

that the positive effect of AADF motorcycles is transferred to AADF heavy vehicles when the model

is not adjusted by the former. Therefore, the adjustment for AADF motorcycles in the PM2.5 model is

therefore crucial.

25

Supplementary Table 11. Statistical parameters of the selected linear-mixed effect models for incPM2.5; and also the

models without AADF heavy vehicles, AADF motorcycles and AADF Euro 4 light diesels. Values denote the 95%

confidence interval. Units for fixed effects are expressed per 1000 vehicles. Bold numbers indicate significant at 95%

level.

Model

formulation

incPM2.5

#5

#5 without

AADF Heavy

vehicles

#5 without

AADF Euro 4

light diesels

#5 without

AADF

Motorcycles

R2 0.796 0.800 0.799 0.776

Intercept

1.2 [-0.5, 2.8] 2.2 [1.2, 3.2] 1.2 [-0.5, 2.9] -0.1 [-2.7, 3.0]

AADF heavy

vehicles 0.4 [-0.1, 1.0] 0.4 [-0.1, 0.9] 1.1 [0.4, 1.8]

AADF

motorcycles 1.2 [0.9, 1.6] 1.4 [1.1, 1.7] 1.2 [0.6, 1.6]

AADF Euro 4

light diesels 0.0 [-0.2, 0.1] 0.0 [-0.1, 0.2] 0.0 [-0.1, 0.2]

AADF Euro 5

light diesels -0.3 [-0.5, -0.2] -0.4 [-0.5, 0.3] -0.4 [ -0.5, -0.2] -0.3 [-0.4, -0.2]

The regression coefficients did not change of magnitude in the unadjusted incPM10 models for those v

ehicle categories showing high correlation values. Therefore, the results from the base model (model

#8) are robust.

Supplementary Table 12. Statistical parameters of the selected linear-mixed effect models for incPM10; and also the

models without AADF Euro IV heavy vehicles, AADF Euro 4 light diesels, AADF Euro V heavy vehicles, AADF Euro

5 light diesels. Values denote the 95% confidence interval. Units for fixed effects are expressed per 1000 vehicles. Bold

numbers indicate significant at 95% level.

Model

formulation

incPM10

#8

#8 without

AADF Euro IV

heavy vehicles

#8 without

AADF Euro 4

light diesels

#8 without

AADF Euro V

heavy vehicles

#8 without

AADF Euro 5

light diesels

R2

0.800 0.799 0.796 0.794 0.801

Intercept

4.2 [2.8, 5.7] 4.3 [2.8, 5.7] 4.5 [3.2, 6.0] 5.1 [3.7, 6.6] 4.0 [2.6, 4.4]

AADF Euro 4

light diesels 0.1 [-0.0, 0.3] 0.0 [-0.0, 0.2] 0.1 [-0.1, 0.39] 0.1 [-0.1, 0.2]

AADF Euro 5

light diesels -0.4 [-0.5, -0.2] -0.3 [-0.5, -0.2] -0.3 [-0.5, -0.2] -0.2 [-0.3, 0]

AADF Euro III

heavy vehicles 0.7 [0.1, 1.2] 0.6 [0.1, 1.2] 0.5 [0.0, 1.0] 0.3 [-0.3, 0.8] 0.8 [0.2, 1.3]

AADF Euro IV

heavy vehicles -0.5 [-1.3, 0.3] 0.1 [-0.4, 0.6] -0.3 [-1.0, 0.6] 0.3 [-0.4, 1.0]

AADF Euro V

heavy vehicles 1.2 [0.6, 1.8] 1.2 [0.6, 1.8] 1.2 [0.6, 1.8] 0.5 [0.0, 1.0]

AADF Euro 2

motorcycles 3.5 [1.9, 5.1] 3.8 [2.2, 5.3] 4.0 [2.6, 5.5] 3.3 [1.7, 5.0] 3.9 [2.3, 5.5]

AADF Euro 3

motorcycles 0.2 [-0.6, 0.9] 0.2 [-0.6, 0.9] 0.4 [-0.4, 1.1] 0.2 [-0.6, 1.0] -0.4 [-1.1, 0.4]

Also, possible collinearity was explored using the Condition Number (CN) and the maximum

variance inflation factor (VIF). A threshold of 30 and 10, respectively, have been proposed to indicate

26

collinearity (see references in Dormann et al., 2013). Values were: 5.8 (CN) and 1.3 (VIF) for the

NOX model; 7.2 (CN) and 3.4 (VIF) for NO2; 13.0 (CN) and 6.3 (VIF) (PM10); 12.6 (CN) and 3.3

(VIF) (PM2.5); all well below the recommended thresholds.

S.5 Discussion graphs Supplementary Figure 22 relates the trends in roadside increments for incNO2, incPM10 and incPM2.5

in 2010-16 versus the annual mean concentration in 2010. The graph shows that the faster downward

trends in roadside increments took place in the roads with the highest concentrations in the start of the

period.

Supplementary Figure 22. Trends in roadside increments in 2010-16 for each pollutant and the annual mean

concentration in 2010. Colour scale shows distance to the city centre.

27

Supplementary Figure 23. Trends in PM2.5 vs trends in PM10 for collocated measurements in Paris and London.

Background sites are indicated with a white dot. Distance to the city centre is marked by the colour scale.

28

Supplementary Figure 24. Trends in traffic increments vs tends in AADF for different vehicle categories for 2010-16

in London. Colour dots indicate distance from the city centre. Blue solid line and grey area indicate the linear

regression model and the 95% confidence interval.

29

References Cleveland, R.B., Cleveland, W.S., McRae, J.E., Terpenning, I., 1990. STL: A seasonal-trend

decomposition procedure based on loess. J. Off. Stat. https://doi.org/citeulike-article-id:1435502

Dormann, C.F., Elith, J., Bacher, S., Buchmann, C., Carl, G., Carré, G., Marquéz, J.R.G., Gruber, B.,

Lafourcade, B., Leitão, P.J., Münkemüller, T., Mcclean, C., Osborne, P.E., Reineking, B.,

Schröder, B., Skidmore, A.K., Zurell, D., Lautenbach, S., 2013. Collinearity: A review of

methods to deal with it and a simulation study evaluating their performance. Ecography (Cop.).

36, 027–046. https://doi.org/10.1111/j.1600-0587.2012.07348.x

Font, A., Fuller, G.W., 2016. Did policies to abate atmospheric emissions from traffic have a positive

effect in London? Environ. Pollut. https://doi.org/10.1016/j.envpol.2016.07.026

Sen, P.K. Sen, 1968. Journal of the American Statistical. J. Am. Stat. Assoc. 63, 1379–1389.

Theil, H., 1950. A Rank-Invariant Method of Linear and Polynomial Regression Analysis. Ned. Akad.

Wetenchappen 53, 386–392. https://doi.org/10.1007/978-94-011-2546-8

Viechtbauer, W., 2010. Conducting Meta-Analyses in R with the metafor Package. J. Stat. Softw. 36,

1–48. https://doi.org/10.1103/PhysRevB.91.121108

Documents

SUPPLEMENTARY INFORMATION A tale of two cities: is air ... · Zone rurale Sud-Est - Forêt de FONTAINEBLEAU RUR_SE Background 59.6 PM 10,PM 2.5,NO X,NO 2 Quai des Célestins CELES