Progetti CCM 2011
Metodi per la valutazione integrata dell'impatto ambientale e sanitario (VIIAS) dell'inquinamento atmosferico
OBIETTIVO SPECIFICO 5:Valutare a livello locale l’impatto ambientale delle
particelle ultrafini Environmental impact assessment of ultrafine
particles. Rome case study.
Giorgio Cattani, Alessandro di Menno Di Bucchianico, Alessandra GaetaDipartimento Stato dell’Ambiente e Metrologia Ambientale
Servizio Monitoraggio degli Impatti sull’atmosferamail: [email protected]
web: http://www.isprambiente.it/site/it-IT/
ISPRAIstituto Superiore per la Protezionee la Ricerca Ambientale
Background
• Research needs:– the burden due to chronic long-term effects is usually much larger
than that due to acute effects (Brook et al., 2010; Künzli et al., 2010)
– thus, studies on the long-term effects of UFP are necessary (Hoeket al., 2009; World Health Organization, 2006).
– No epidemiologic studies of long-term exposures to ambient UFPshave been conducted (HEI, jenuary 2013)
– Lack of spatially resolved exposure data
Ultrafine particles, particles number concentration and total particles number concentration
• Ultrafine particles: diameter less than 0,1 µm• Indicator for UFP exposure: particle number concentration
(PNC) in the <10 – 100 nm range• In urban setting ultrafine particles accounted for more than
80% of the total particle number concentration and the temporal correlation between ultrafine and total particle number concentration was very good
• Total PNC has been widely used as a more easily measured indicator for ultrafine particles (e.g. Aalto et al, 2005; Belleudi et al, 2010).
Land use regression model for ultrafine particles
Amsterdam (Hoeck et al, 2011)– Directly outside 50 homes. In each week seven 24 h average measurements
were made at one home and at a central urban background site in the city center, where measurements were during the entire study period. The average difference of the concentrations measured at the home outdoors and the continuous measurement site was used to develop the land use regression model. 67% variability explained
• Girona province, Spain (Rivera et al, 2012)– UFP for 15 minutes on the sidewalk of 644 participants’ homes in 12 towns
of Girona province (Spain) (Girona town 167 sites) during 7 week (june-july). Up to 72% variability explained
– Period averages estimated using NOx continuous measurements at an urban background urban city (Girona)
• Vancouer, Canada (Abernethy et al, 2013)– UFP for 1h at 80 sites – Period averages estimated (2 week) using 4 fixed point continuous
measurements
Study aim
• To explore the possibility to model the spatial variation of PNCconcentrations in the city of Rome with land use regression model
Study design
• measurements will be conducted at 31 monitoring sites from April 2013 until March 2014
• One week measurements per site repeated three times in different season
• Continuous total PNC measurements at one additional site, so that the discontinuous site specific measurements can be adjusted to the true long term average for the observation period
• Explorative use of alternative metrics to adjust the observed mean
Monitoring devices
• total PNC measurements using condensation particle counters • directly outside 30 homes: portable TSI CPC 3007
– The CPC3007 counts particles from 10 nm and above• One fixed sites: TSI CPC3022a
– The CPC3022a counts particles from 7 nm and above• Timing:
– Time resolution: 1 minute – three sampling period during the day, lasting two hours each:
00:30 – 02:30; 08:30 – 10:30; 16:30 – 18:30– One week measurement campaign; each sites will be visited three
times.• Sampling probe: conductive silicon tube less than 3 meter long
NOx Annual averages in Rome (2011) estimated using 6/24 hours vs 23/24 hours averages
y = 1.0027x + 1.1723R2 = 0.9953
0
20
40
60
80
100
120
140
160
180
0 20 40 60 80 100 120 140 160 180annual mean NOX - 23h/24h [µg/m³]
annu
al m
ean
NO
X - 6
h/24
h [µ
g/m³]
Siting of sampling points
• Sites spread over the city, within the Great Ring Junction (GRA)• Measurements at building façades • The ESCAPE (Eeftens et al, 2012) protocol have been followed
for macro and micro siting of sampling point for instance:– broadly distributed across the study area– selected to provide contrast in predictor variables (representative
of a range of traffic intensities and a range in urbanization) – that are representative for homes within the city (at least 10
meters from the roadside)– Unrestricted air flow around the sampler
• Sampling points were placed at an height variable (2 – 30m above the ground)
• This issue will be taking into account during LUR development:– Using h above the ground as a model variable– Using measurement undertaken at different height in the same
place
•3 road side sites (< 30 m from a major road i.e. > 10,000 vehicles/day) in an high density population context (HD)
•1 road side sites in a low density population context (LD)
•7 traffic sites (30 – 100 m) from a major road in HD
•5 traffic sites in LD
•4 residential sites (> 100 m from a major road) in HD
•10 residential sites (> 100 m from a major road) in LD
Selected sites
Predictor variables
• distance to nearest road• distance to major road (>10,000 vehicles per day)• traffic density in several buffers size• Combination of traffic and distance to road • Population density in several buffers size• physical characteristics (e.g., elevation, latitude, longitude and
distance to coast)• Land use (urban green, high density residential land )
Model development
• Regression model will be developed using a supervised forward stepwise procedure
• Univariate regression analyses will be conducted for all possible predictor variables
• The model with the highest adjusted explained variance (R2) (and lowest RMSE) will be regarded as the start model
• Variables co-varying (r >0,6) with the first one will be removed• to this start model the remaining variables will be added
separately if the effect on the adjusted R2-value is greater than 1%, and the coefficient is conform the pre-specified direction, and the direction of the effect for predictors already included in the model do not change
• Standard diagnostic tests for ordinary least squares regression will be applied, such as checks on the normality of residuals, heteroscedasticity and influential observations
Scheduling
• Measurements:– First run: 6 April 2013 – 20 July 2013– Second run: 16 September – 2 December 2014– Third run: 7 jenuary – 25 march 2014
• Final model development:– By the end of October 2014
Total PNC time series. Example of co-located measurements
Day 2
0
20,000
40,000
60,000
80,000
100,000
0:30
0:40
0:50
1:00
1:10
1:20
1:30
1:40
1:50
2:00
2:10
2:20
8:30
8:40
8:50
9:00
9:10
9:20
9:30
9:40
9:50
10:0
0
10:1
0
10:2
0
16:3
0
16:4
0
16:5
0
17:0
0
17:1
0
17:2
0
17:3
0
17:4
0
17:5
0
18:0
0
18:1
0
18:2
0
#/cm
3
SeparatorVia Ermoli - Thursday 09/19/13Via Staz. di Tor Sapienza - Thursday 09/19/13Via Tarquinio Prisco - Thursday 09/19/13Via Tarquinio Prisco Ip - Thursday 09/19/13
Total PNC time series at 4 m above the ground (green line) and at 25 m from the ground (brown line). Traffic
oriented site, in an high density population setting. Thursday, 9/19/2013
Day 2
0
20,000
40,000
60,000
80,000
0:30
0:40
0:50
1:00
1:10
1:20
1:30
1:40
1:50
2:00
2:10
2:20
8:30
8:40
8:50
9:00
9:10
9:20
9:30
9:40
9:50
10:0
0
10:1
0
10:2
0
16:3
0
16:4
0
16:5
0
17:0
0
17:1
0
17:2
0
17:3
0
17:4
0
17:5
0
18:0
0
18:1
0
18:2
0
#/cm
3
Separator
Via Tarquinio Prisco - Thursday 09/19/13
Via Tarquinio Prisco Ip - Thursday 09/19/13
Day 5
0
20,000
40,000
60,000
80,000
0:30
0:40
0:50
1:00
1:10
1:20
1:30
1:40
1:50
2:00
2:10
2:20
8:30
8:40
8:50
9:00
9:10
9:20
9:30
9:40
9:50
10:0
0
10:1
0
10:2
0
16:3
0
16:4
0
16:5
0
17:0
0
17:1
0
17:2
0
17:3
0
17:4
0
17:5
0
18:0
0
18:1
0
18:2
0
#/cm
3
SeparatorVia Tarquinio Prisco - Sunday 09/22/13Via Tarquinio Prisco Ip - Sunday 09/22/13
Total PNC time series at 4 m above the ground (green line) and at 25 m above the ground (brown line). Traffic
oriented site, in an high density population setting. Sunday, 9/22/2013
Total PNC daily average at a traffic oriented site, in an high density population setting.
Brown box: 4 m above the ground Yellow box: 25 m above the ground.
Daily average
0
5,000
10,000
15,000
20,000
25,000
30,000
18/09/2013 19/09/2013 20/09/2013 21/09/2013 22/09/2013 23/09/2013 24/09/2013
Wednesday Thursday Friday Saturday Sunday Monday Tuesday
#/cm
3
Via T. Prisco 7nt floorVia T. Prisco 1st floor
First campaign: box plot of hourly data
a = road side, in HD
b = traffic in HD
c = traffic in LD
d = residential in HD
e = residential in LD
Total PNC (part/cm3) adjusted mean (April – July)
Thanks for your attention!Thanks for your attention!
Thanks are due to Marco Inglessis (ISS) for providing continuous measurements at the fixed site
and to the 30 study participants for their valuable contribution during the measurements campaign.
They hosted and provided daily care to the instrument for free, making possible this study