Long-term simulation of photochemistry in the Grenoble urban area using a mesoscale model chain

including MM5, CHIMERE and MetPhoMod

Eric Chaxel*, Jean-Pierre Chollet*, Christophe Quiniou** and Olivier Couach***

* Laboratory of Geophysical and Industrial Flows, THEO, Grenoble (France)

** GIERSA, 44 avenue Marcelin Berthelot, Grenoble* LPAS, EPFL, Lausanne (Switzerland)

Contact: eric.chaxel@hmg.inpg.frWebsite: http://www.prevalp.fr.fm

Workshop CHIMERE, 23rd March 2005

2/23

Outline

• Framework of the study

• Description of the modelling chain

• Two highlights: – Evaluation of nested CHIMERE simulation for the summer

2003 (June-August) in Rhône-Alpes region– Impact of long range transport shown by CHIMERE in July

1999 on Grenoble photochemistry

• Discussion

3/23

Framework• In assistance to local air quality agency

ASCOPARG:– to assess emission reduction scenario– To develop a modelling chain able to correctly

predict ozone events

• In a scientific context: project PARAMET (Photochemistry and Aerosols in Alpine Region: Mixing and Transport– Model the processes responsible for high ozone

events in narrow valleys: slope winds, inversion layers, recirculation

– Evaluate the impact of regional transport in the valleys of Grenoble, Maurienne and Chamonix

4/23

Our use of CHIMERE• CHIMERE is used at boundaries of an other model:

MetPhoMod (LPAS, EPFL, Lausanne)

• Since CHIMERE has never been validated in very complex terrain our use of the model is limited to 2 regional scales:– Continental for calculate background concentration in the

Alps– Regional to account for main cities (Lyon, Geneva,

Marseille, Torino) contributing to regional ozone

• A “beta test” simulation at a 2-km resolution has been recently evaluated

5/23

Description of the modeling chain

NCEP/AVN forecastsΔx = 100 km, Δt = 6 hours

MM5Δx = 18 km

CHIMEREΔx = 18 km

MM5Δx = 6 km

CHIMEREΔx = 6 km

MetPhoMod

Δx = 2 km(meteo + chimie)

EMEP inventoryΔx = 50 km

CITEPA inventoryΔx = 6 km

GIERSA inventoryΔx = 1 km

Post-processingΔt = 1 hour

One-way nesting

Two-way nesting

Inputs/outputs

6/23

Technical schematic of the chain

Annual inventoryNOx, COV, CO, SO2

ASCII files

Annual inventoryNOx, COV, CO, SO2

ASCII files

Gridded hourly emissionsRACM or MELCHIOR

speciationNetCDF files

Gridded hourly emissionsRACM speciesNetCDF files

MetPhoMod

CHIMERE

MM5

ECMWF

GIS GIERSA

prep

Cad

astr

e

CA

DA

ST

RE

prepemis

Interfaces 3d modelsTemporary

files

Legend

TONETCDF + TOMETPH

prepemis_cdf

chimere2racm

7/23

Results for the summer 2003

• CHIMERE runs on 6 15-day periods:– 1 to 15 June

– 16 to 30 June

– 1 to 15 July

– 16 to 31 July

– 1 to 15 August (Heat wave period)

– 16 to 31 August

NCEP/AVN forecastsΔx = 100 km, Δt = 6 hours

MM5Δx = 18 km

CHIMEREΔx = 18 km

MM5Δx = 6 km

CHIMEREΔx = 6 km

MetPhoMod

Δx = 2 km(meteo + chimie)

EMEP inventoryΔx = 50 km

CITEPA inventoryΔx = 6 km

GIERSA inventoryΔx = 1 km

Post-processingΔt = 1 hour

One-way nesting

Two-way nesting

Inputs/outputs

8/23

Summer 2003: an high ozone summer

0

50

100

150

200

250

01-juin 08-juin 15-juin 22-juin 29-juin 06-juil 13-juil 20-juil 27-juil 03-août 10-août 17-août 24-août

Charavines Versoud Champ sur Drac Fontaine Les Balmes St Martin d'Heres

Grenoble Les Frenes seuil information seuil EU Casset

Daily ozone maximum (μg/m3) measured by ASCOPARG from 1st June to 31st August 2003.

Heat wave

Many exceedance days of the treshold 180 μg/m3 (seuil information)

60

70

80

90

100

110

120

130

140

150

01-juin 08-juin 15-juin 22-juin 29-juin 06-juil 13-juil 20-juil 27-juil 03-août 10-août 17-août 24-août 31-août

Ozo

ne

(m

icro

g/m

3)

1-week average 2000

1-week average 2001

1-week average 2003

Background level of ozone measured at le Casset (alt. 1800 m) in summer 2000, 2001 and 2003

Background O3 level at le Casset (alt. 1800 m)

Heat wave 2003

A net increasing trend is observed over the summer 2003

General trend over the summer

9/23

Stations used for validation

Meteorological ground station

Ground station with O3/NOx meas.

CHIMERE regional domain40*40 cellsdx = 8 km8 vertical levels

MetPhoMod domain36*58 cellsdx = 2 km24 vertical levels

Lyon

Grenoble

45 AQ stations:O3, NO, NO2

35 meteo stations:U, V, RH, T, Ray

10/23

CHIMERE dx = 6 km

Hourly measurement

1 2 3 4 5 6 7 8 9 10 11 12 13 14

August 2003

Results of CHIMERE at 6 km resolutionfor hourly ozone

Hourly measurements are compared with model results at all stations

11/23

Some results of MM5 at 6 km resolution

1. MM5 at a 6-km resolution underestimates the daily maximum for 2-m temperatureat the sites located in valleys

2. This effect is observed at all sites during the heat wave (1st to 15th August 2003)

12/23

Results of CHIMERE at 6 km resolutionfor daily ozone maximum

0,00

2,00

4,00

6,00

8,00

10,00

12,00

14,00

16,00

18,00

1 to 16 June 16 to 30June

1 to 16 July 16 to 31 July 1 to 16August

16 to 31August

Eté 2003

Mea

n E

rro

r (m

icro

g/m

3)

Mean error over the summer 2003 at 45 sites: 12.7 μg/m3

0,00

5,00

10,00

15,00

20,00

25,00

Ann

ecy-

Nov

el

Ann

ecy-

Love

rchy

Bar

by

Cha

mbe

ry-P

aste

ur

St-

Jean

-de-

Mne

Cha

mbe

ry-le

-Hau

t

Cas

set

St-

Exu

pery

Alb

ertv

ille

Cha

mpa

gnie

r

Gre

nobl

e-le

s-F

rene

s

Rou

ssillo

n

Roa

nne

Tho

non

Cha

ravi

nes

Roc

hes-

de-C

ondr

ieu

Gen

as

Ann

emas

se

All

site

s

Lyon

-St-

Just

Ter

nay

Val

ence

-Cen

tre

Firm

iny

Ville

urba

nne-

Cro

ix-L

uize

t

Val

ence

-Sud

Cha

mon

ix

St-

Etie

nne-

Sud

St-

Etie

nne-

Cou

bert

in

Fon

tain

e-le

s-B

alm

es

Gai

llard

St-

Cha

mon

d

Vea

uche

Lyon

-Ger

land

Ver

soud

Sab

lons

Die

me

Ann

onay

St-

Mar

tin-d

-Her

es

Cha

mp-

sur-

Dra

c

Die

ulef

it

Mea

n E

rro

r (m

icro

g/m

3)

Mean bias over the summer 2003 at 45 sites : - 6.0 μg/m3 (model - obs)

13/23

Use of CHIMERE to constraint a other model: MetPhoMod

NCEP/AVN forecastsΔx = 100 km, Δt = 6 hours

MM5Δx = 18 km

CHIMEREΔx = 18 km

MM5Δx = 6 km

CHIMEREΔx = 6 km

MetPhoMod

Δx = 2 km(meteo + chimie)

EMEP inventoryΔx = 50 km

CITEPA inventoryΔx = 6 km

GIERSA inventoryΔx = 1 km

Post-processingΔt = 1 hour

One-way nesting

Two-way nesting

Inputs/outputs

?Why ?To provide realistic forcing compulsory for long term simulations (several months)

14/23

Use of CHIMERE to constraint a fine scale model: MetPhoMod

• Problems and questions:– How evaluate the quality of forcing by MM5 and CHIMERE ?– Model have different horizontal meshes

Lambert, dx = 6 km/ Lambert 2 Etendu, dx = 2 km– Model have different vertical descriptions

8 hybrid sigma levels/ 24 cartesian levels– Models have different chemical mechanisms

MELCHIOR/ RACM

• Solutions:– Perform comparison with 3D and ground data– Perform horizontal interpolation– Perform vertical interpolation– Using only common species to both mechanisms

15/23

Correspondance between MELCHIOR and RACM

Nom MELCHIOR

Description MELCHIOR

Nom RACM Description RACM

O3 ozone O3 ozone NO2 nitrogen dioxide NO2 nitrogen dioxide NO nitric oxide NO nitric oxide

PAN peroxyacetyl nitrate PAN peroxyacetyl nitrate ans higher saturated PANs HNO3 nitric acid HNO3 nitric acid SO2 sulfur dioxide SO2 sulfur dioxide CO carbon monoxide CO carbon monoxide CH4 methane CH4 methane C2H6 ethane ETH ethane

NC4H10 n-butane HC5 alkanes, alcohols, esters and alkynes with HO rate

between 3.4*10-12 and 6.8*10-12 cm3 s-1 C2H4 ethene ETE ethene C3H6 propene OLT terminal alkenes OXYL o-xylene TOL toluene and less reactive aromatics C5H8 isoprene ISO isoprene

APINEN α-pinene API α-pinene and other cyclic terpenes with one double

bonds HCHO formaldehyde HCHO formaldehyde

CH3CHO acetaldehyde ALD acetaldehyde and higher aldehydes GLYOX glyoxal GLY glyoxal

MGLYOX methyl glyoxal MGLY methylglyoxal and other α-carbonyl aldehydes CH3COE methyl ethyl ketone KET ketones

Table : correspondance between MELCHIOR and RACM species

20 species from CHIMERE are related to RACM species

16/23

Validation of CHIMERE forcing using 3D data from 25-27 July 1999

The CHIMERE model is used at boundaries of MetPhoMod

Validation of this forcing using data from POI 1 of GRENOPHOT 99

UHF radar and Lidar are located in Vif (south of Grenoble)

CHIMERE domain

MetPhoMod domain

UHF radar

windprofiler

Grenoble

Vif

17/23

Chartreuse rangeDévoluy range La Mure

Grenoble

Synoptic wind

Valley wind

Night valley wind in Grenoble by northerly synoptic wind

27 July 1999 00UT

N

S

Typical stratified atmosphere:#1: 0 - 400 m agl: ground-influenced layer#2: 400 - 2000 m agl : Thermic layer #3: 2000 m agl - ~: Synoptic regime

18/23

Validation of the MM5 forcing using UHF radar

Both synoptic and valley winds are correctly described by MetPhoMod forced with MM5

Strong synoptic southerly wind blows on 27th July 1999

Altitude amsl

Altitude amsl

Altitude amsl

1

2

3 ?

Low speeds = max. ozone production

Wind shears = transition zones with low speeds

19/23

Validation of the chemical forcing using LIDAR data

Long range transportassociated with

southerly wind

Long range transport as described by CHIMERE

• CHIMERE provides a good description of the synoptic situation (above 2500 m amsl)• The long range transport is well detected by CHIMERE but slightly underestimated (140

μg/m3 modelled versus 160 μg/m3 measured) • MM5 forcing provides southerly wind reaching too early the MetPhoMod domain• The valley bottom is more affected by sourtherly winds in the simulation than on radar

profiles

Altitude amsl Altitude amsl

Transition zone with max.

ozone production

20/23

Long range transport of ozone on July 27th 1999

Marseille (H-12)

Turin (H-36)

Backward trajectories reaching Grenoble on July 27th at 12UT shows that:

• the switch from northeasterly to southerly wind occured during the evening on July 26th

• air mass travelled over Marseille on July 26th

• air mass travelled over the Pô valley on July 25th

Very particular circulation leading to long range transport of ozone

Grenoble(July 27th, 12UT)

21/23

A try at a 2-km resolution

windprofiler

MM5 (dx = 2 km) correctly describes the local wind system on the period 25-27 July 1999

but.. CHIMERE underestimates ozone production compare with measurements

MM5

Grenoble

22/23

Discussion (1/2)• For our applications CHIMERE at a 6-km resolution

provides good synoptic/regional forcing (mean error of 10 μg/m3 on daily maximum)

• For a use at small scale several points should be/are in development in MM5:– Better impact of the topography on flows in MM5 such as

impact on slopes on incoming solar radiation (improvements in version 3.7)

– Description of the PBL in MM5: actual parameterisations are not adapted to complex terrain (problems on slopes)

– Vertical mesh should be refined near the PBL top to well detect the thermic layer (#2)

23/23

Discussion (2/2)

• For a use at small scale several points should be/are in development in CHIMERE:– Use a domain top higher than 500 hPa in CHIMERE– The number of layers should be increased in our simulations

(> 8)– Vertical mesh has to be refined near the ground to well

account for transport by valley winds (layer #1 is 400 m agl at night and 2000 m agl during the day)

– The thickness of the layers near the transition zones (wind shears) are of major importance

– In Grenoble case the reservoir layer is formed at 1800 m agl

Thanks for your attention!

Further information available at:

http://www.legi.hmg.inpg.fr/~Alpes/Internet