transport mixing chemistry & deposition - polytechniquevmyrto/Files/MValPhDStuff.pdfOffice...

Preview:

Citation preview

Concentrations (CTM)

Concentrations d’exposition

Modèle Monte-Carlo

Expositions personnelles

Fonctions d’exposition-risque

déterministe/statistique

non-déterministe

statistique

Concentrations

Exposure

Risk

démographie

exposition

occupation du solémission

sfine échelle

Exposure-concentrations

Personal Exposures

Exposure-response Function

Sub-grid Model

deterministic/statistical

Monte-Carlo Model

non-deterministic

Regression/Health data

statistical

<1 km1-10 km

Chemistry Transport Modeling (CTM)

E

1-10 km

E

1-10 km

C E´

transport chemistry & depositionmixing

emissions

Desegregating emissions

∆x emissions [km]

Higher resolution

Model Error [ O3] ∆x [CTM]=6km∆x meteo=6km

Rel

ativ

e di

ffere

nce

with

mea

sure

men

ts (

%) ✕

✕✕ ✕✕

✘CTM

Direct ‘Downscaling’

✘CTM

Sub-grid modeling / urban environments

Statistical representation of E´

Concentrations per environnementEmissions per sector

2)Re-organisation1)Desagregation

+high resolution emission data

+statistical info(land-use)

Calcul dans des micro-environnements

•Deterministic CTM-modeling >1km:

•Sub-grid CTM-modeling <1km:

Local emission Sub-grid mixing

i = Park, residential, traffic, etc.

31/7 1/8 2/8 3/8 4/8

Background Stations

1/7 3/7 5/7 7/7 9/7

ResidentialGrid-averagedMeasurements

ResidentialMeanMeasurements

31/7 1/8 2/8 3/8 4/8

1/7 3/7 5/7 7/7 9/7

Traffic Stations TrafficGrid-averagedMeasurements

TrafficGrid-averagedMeasurements

Concentrations

Demography

Val d’Oise

Essone

Yvelines

Seine-et-Marne

Hauts-de-Seine

Seine-Saint-Denis

Val-de-Marne

>15%8-15%2-5%

Towards Paris

Out of Paris

Hauts-de-Seine

Seine-Saint-Denis

Val-de-Marne

Essone

Val d’Oise

Yvelines

Seine-et-Marne

10-12%5-6%

Essone

Val d’Oise

Yvelines

Seine-et-Marne

>10%8-10%5-8%1-5%<1%

>80%70-80%

Between Administrative Units

Demographic data Activity data

004

12

20

816

TransitOffice00

4

12

20

816 ☀12

16 8

004

12

20

816

Home00

20 4

=1% of the Paris population

Monte-Carlo Model

Demographic data

Monte-Carlo Model

004

12

20

816

TransitOffice00

4

12

20

816 ☀12

16 8

004

12

20

816

Home00

20 4

=1% of the Paris population

Hourly Concentrations

Exposure

Activity data

Tri-modal exposureO3 (µg/m 3)

June 2001

June 2002

June 2003

Number of persons

Exposure level

Low

Medium

High

Tri-modal exposureO3 (µg/m 3)

June 2001

June 2002

June 2003

Number of persons

Exposure level

Tri-modal exposureNO2 (µg/m 3)

HighMedium

Low

Number of Persons

June 2001

June 2002

June 2003

Exposure level

O3, 2004-07-07

Fre

qu

ency

HighMedium

Low

RuralPeriurban

Urban

Exposure Modes geographical area

Exposure to [O 3] (µg/m 3)

Pop

ulat

ion

Fra

ctio

n

σ2

f2

µ1 µ2 µ3

?

?

? ??

Exposure to [NO 2] (µg/m 3)

FaibleMoyenFortEntier

Pop

ulat

ion

Fra

ctio

n

Exposure to [PM 2.5] (µg/m 3)

FaibleMoyenFortEntier

Pop

ulat

ion

Fra

ctio

n

•One fit per day:

•Time series for 9 parameters

µ3, σ3, f3

tµ2, σ2, f2

µ1, σ1, f1

Generalized Additive Model (GAM)

Monitor data (concentrations)

concentrations CTM

Generalized Additive Model (GAM)

Monitor data (concentrations)This study

004

12

20

816

TransitOffice00

4

12

20

816 ☀12

16 8

004

12

20

816

Home00

20 4

Demographic/activity data

Emploi du temps

Monitor data (concentrations)

Generalized Additive Model (GAM)

‘Exposure’

Generalized Additive Model (GAM)

Monitor data (concentrations)

Generalized Additive Model (GAM)

ParametrizationMonitor data (concentrations)

Generalized Additive Model (GAM)

N number of monitors

Generalized Additive Model (GAM)

Area-aggregated surrogate Spatially resolved surrogate

Two-step comparison

1.Mono-pollutant model (one regression per pollutant)[APHEA Project]

...

...

...

Mono-pollutant model

Pa r

tial v

aria

bili

ty

[NO2] (µg/m 3)

[APHEA Project]

-Area-aggregated surrogate-Cette étude

[NO2] (µg/m 3)

-Area-aggregated surrogate-Spatially resolved surrogate

Mono-pollutant model

Pa r

tial v

aria

bili

ty

[O3] (µg/m 3)

-Area-aggregated surrogate-Cette étude

Mono-pollutant model

Pa r

tial v

aria

bili

ty

[APHEA Project]

[O3] (µg/m 3)

-Area-aggregated surrogate-Spatially resolved surrogate

Mono-pollutant model

Pa r

tial v

aria

bili

ty

[PM2.5] (µg/m 3)

-Area-aggregated surrogate-Cette étude

Mono-pollutant model

Pa r

tial v

aria

bili

ty

[APHEA Project]

[PM2.5] (µg/m 3)

-Area-aggregated surrogate-Spatially resolved surrogate

Mono-pollutant model

Pa r

tial v

aria

bili

ty

Spatially resolved surrogate•Less strong effects•Larger uncertainty•Effect of PM2.5 > NO2

Conclusions for the mono-pollutant model

Area-aggregated (monitor data)•Effect of NO2 > PM2.5

•Impossible to separate the effects of the co-pollutants

...

2.Tri-pollutants model (only one regression)

[NO2] (µg/m 3)

-Area-aggregated surrogate-Cette étudeEffect of NO2 over-estimated

Tri-pollutant model

Pa r

tial v

aria

bili

ty

[APHEA Project]

[NO2] (µg/m 3)

-Area-aggregated surrogate-Spatially resolved surrogate

Tri-pollutant model

Pa r

tial v

aria

bili

ty

[O3] (µg/m 3)

-Area-aggregated surrogate-Cette étude

[APHEA Project]

Tri-pollutant model

Pa r

tial v

aria

bili

ty

[O3] (µg/m 3)

-Area-aggregated surrogate-Spatially resolved surrogate

Tri-pollutant model

Pa r

tial v

aria

bili

ty

-Area-aggregated surrogate-Cette étude

[PM2.5] (µg/m 3)

Negative effect for PM 2.5

[APHEA Project]

Tri-pollutant model

Pa r

tial v

aria

bili

ty

[PM2.5] (µg/m 3)

-Area-aggregated surrogate-Spatially resolved surrogate

Tri-pollutant model

Pa r

tial v

aria

bili

ty

✘ ✘Spatially resolved surrogate•PM2.5 & NO2 indicators less correlated (0.72 0.6)•Positive effects for all 3 pollutants •Effect of PM2.5 > NO2

Conclusions for the tri-pollutants model

✔ ✔

Recommended