Aucun titre de diapositive

1
An operational system for air quality monitoring
and forecasting over Europe
Honoré C.(1), Menut L. (2), Bessagnet B.(1),
Meleux F.(1),Rouil L. (1) , Vautard R.(2),
Beekmann M.(3), Poisson N.(4), Peuch
V-H(5) 1Ineris 2 LMD/IPSL 3 LISA 4 ADEME 5
Meteo-France This platform is proposed by the
PREV AIR consortium - http//www.prevair.org
Questions about PREV AIR? Please send an e-mail
to frederik.meleux_at_ineris.fr
2
PREVAIR www.prevair.org

• Built for operational purposes
• In France public information related to pollution
  episodes (threshold exceedances) can be driven by
  observations but also by forecasts since 2003
• The system
• a cooperative system for air quality monitoring,
  forecasting and mapping over Western Europe and
  France
• set up in 2003 by

Builds up and hosts the prev air system Model
development CHIMERE AQ expertise for the
Ministry
Public organization under the supervision of the
Ministry of Ecology
INERIS
National Agency of Environment
ADEME
Builds up and hosts the NRT databaseAQ expertise
for the Ministry
Model development CHIMERE
National Research Centre
CNRS
Model development MOCAGE Meteorological
forecasts
National Meteorological services
Météo France
3
Other objectives

• ? Complement the French air quality monitoring
  network
• Complete areas little or not at all covered by
  fixed measurement stations
• ? Technical support to the French Ministry for
  Ecology in the framework of international
  negotiations about transboundary air pollution
  (implementation of integrated policies aiming at
  reducing pollutant emissions in Europe)
• ? Science Validation of the models and of our
  knowledge
• ? Provide operational boundary conditions for
  smaller-scale forecasting systems

4
Architecture of the system

• 3D CTM
• CHIMERE (CNRS/INERIS)
• MOCAGE (Météo-France)
• Input data
• Meteorological forecasts
• AVN / NCEP MM5
• ARPEGE, ALADIN
• Emissions inventories
• EMEP
• EDGAR
• Boundary concentrations
• LMDzINCAGOCART
• MOCAGE
• BASTER / ADEME NRT obs. data
• Post processing tools
• Statistically adapted forecast
• Analyses
• Evaluation
• Extraction of numerical data

5
The CHIMERE Model Set-Up in PREVAIR

• Domains
• Over Europe Over France
• Horizontal resolution 0.5x 0.5 0.15
  x 0.1
• Vertical resolution 8 levels from surface
  pressure up to 500 hPa
  
• Meteorological forecast data
• AVN / NCEP for initialisation and boundary
  conditions
• MM5 higher resolution forecasts (36km - 18km)
• Chemical scheme
• MELCHIOR reduced (45 species, 120 reactions)
• Aerosol module
• Dust, PPM, SOA, nitrates, sulphates,
• ammonium, Water Contents
• 25 reactions (aqueous and heterogeneous phase)
• Other model configurations
• meteorology ECMWF, ARPEGE / ALADIN
• BC MOZART, LMDZ-INCA, GOCART
• See http//euler.lmd.polytechnique.fr/chimere

6

• The MOCAGE Model Set-Upin PREVAIR

• Domains
• Global Model / Over Europe /
  Over France
• Horizontal resolution 4x 4
  0.5 x 0.5 0.1 x 0.1
• Vertical resolution 47 levels from surface
  pressure up to 5hP
• Meteorological forecast data
• ARPEGE and ALADIN
• Chemical scheme
• RACMOBUS (118 species, 381 reactions)
• No aerosol module

7
PREVAIR Outputs Daily Forecasts

• Available at D0, 01-06 h LT
• Daily peak and averaged concentration maps
• D0, D1, D2
• Pollutants O3, NO2, particulate matter
• Global scale, Europe, France

8
The Observation Network
41 Air Quality related organisms (AASQA)
680 monitoring stations O3, NO2, SO2, PM10, PM2.5
Collected in real time to the national AQ
database BASTER
Downloaded for being used within PrevAir for
evaluation and correction processing
9
Ozone peak, rural stations, RMSE (µg/m3) D0
(15/06/2006 - 15/09/2006)

• Statistics and Scores

-Each summer and winter, scores are estimated.
Ozone peak, rural stations, correlation () D0
(15/06/2006 - 15/09/2006)
-Summer 2006 from April to September CF Chimere
over France OI with Optimal Interpolation

• -For ozone the model slightly overestimates the
  peak values
• -The correlation decreases but slowly
  highlighting the robustness of the model.

10
PREVAIR monitoring Analyses

• Analyses are combination of obervations and
  model outputs (NRT)
• Kriging method
• Outputs
• Ozone peak concentration
• PM peak concentration
• Daily data

July 25, 2006
11
Statistical adaptation

• Modelled concentrations corrected with
  climatology - over the last past years
• From 2003, 2004, and 2005 a linear regression is
  built for the ozone concentration error for each
  station
• Statistical adaptation kriging
• method
• Available at D0 for D0 D1
• and D2

O3errorO3sim-O3obs over 240 stations
(rural and suburban)
O3error ?station Tsim ?station O3sim
12
Finer grids
13

• User Information Exchange

• Since the information is daily published on
  Internet and filed in a database, the Prev Air
  system is opened to all persons and organisation
  that would like to, personally or professionally,
  be informed on the continent-wide evolution and
  trends of pollutant behaviour in the short and
  medium term.
• PREV AIR numerical data
• Extraction over user-defined domains
• Available through internet for Air Quality
  Institutions (user accounts)
• Applications
• 3D data Boundary Conditions for Air quality
  simulations at a local scale
• 2D data for Air Quality monitoring and
  forecast
• ? Feedback from Prev Air Users

14
Ozone map on French public television

• For the summer 2006, 17 days with ozone
• forecast broadcasting.
• In 2005, there were 18 days with ozone forecast
  on TV