The Canadian Air Quality Forecasting System

1
The Canadian Air Quality Forecasting System

• V.S. Bouchet, R. Moffet, M. Moran, A. Robichaud,
  L.-P. Crevier, S. Ménard, S. Cousineau, S.
  Gaudreault, M. Sassi, H. Landry, W. Gong, P.
  Makar, J. Pudykiewicz, A. Kallaur

2
AQ Modelling Application Group - MANDATE

• Air Quality Forecasts
• Prepare models tools for operation and support
  real-time operations (models)
• Prepare model-ready emission files
• Participate in real-time evaluation (user group)
• Air Quality Scenarios
• Support development of policies, national and
  international assessments
• Prepare model-ready emission files
• Real-time and case studies (AQ models
  trajectories)
• Support to Regions
• Support modellers in Regions with implemetation
  of air quality models and analysis tools (SPI,
  tool box, trajectories)
• Support to field campaigns
• Provide tailored products for field campaigns

3
HISTORY OF THE FORECASTING PROGRAM

• Summer 1999 Pilot Air Quality Forecasting
  Program for eastern Canada
• Limited to ground-level ozone
• Numerical guidance CHRONOS, eastern domain,
  21km spatial resolution
• Summer 2001 Air Quality Prediction Program
  extended to all subarctic Canada
• CHRONOS, continental domain, 21km spatial
  resolution, limited to surface ozone
• Summer 2002
• Introduction of bulk aerosol scheme in
  experimental CHRONOS version
• Operational version ozone, sulphates and SOA
• Beginning of experimental real-time scenarios
• Summer 2003
• Implementation of bulk aerosol scheme in
  operational CHRONOS version (public)
• Forecasts and experimental real-time scenarios
  for ozone and PM
• Experimental Objective Analysis of ground-level
  ozone
• Summer 2004
• Continuation of ozone and PM forecasts,
  experimental real-time scenarios, OA (op)

4
MODEL DESCRIPTION OPERATIONAL VERSION

• Meteorological driver GEM, the Canadian
  regional/forecast model
• Chemistry Gas- and heterogeneous- phase
  chemistry
• No aqueous-phase chemistry
• PM representation sectional approach, 2 bins
  (fine/coarse)
• Advection, diffusion and dry wet deposition
  (sedimentation only for PM)
• Emissions 1995 Canadian national inventory and
  1996 US NEI
• Domain specifics
• 21km grid covering North America
• 24 levels to 6 km
• Performance
• Run from 00 UTC for 48 hours
• Less than 3 hours of computing time for whole
  North-America
• Forecast available at 08 UTC

5
SAMPLE CHRONOS FORECAST - SUMMER
6
SAMPLE CHRONOS FORECAST - WINTER
TEOMs measurements in the vicinity of
Montreal (courtesy Environ. Canada, Quebec region)
PM2.5 (ug/m3) February 5, 2005 (6h avg)
Jan. 31
Feb. 1
Feb. 2
Feb. 3
Feb. 4
Feb. 5
Feb. 6
Feb. 7
7
SUMMARY STATISTICS FOR SUMMER 2004
8
SYSTEMATIC POINT TO POINT COMPARISON
CHRONOS versus AIRNOW real-time data Aug. 20th,
2004 1800Z
CHRONOS tends to overpredict ozone afternoon
values (1h)
9
CATEGORICAL ANALYSIS
10
2005 OPERATIONAL VERSION

• Changes in anthropogenic emissions dated
  emission information
• 1995 Canadian NEI to 2000 Canadian NEI
• CEPS to SMOKE
• Changes in biogenic emissions afternoon ozone
  over-prediction
• BEIS2/USGS framework to BEIS3/BELD3 framework
• BEIS3/BELD3 has limitations for Canada
• Changes in heterogeneous chemistry options
  timing of model run
• HETV, based on ISORROPIA (Nenes et al., 1998)
  optimized for vector machine, fast on scalar a
  well
• Initially, full representation of
  suphate-nitrate-ammonium-water system
• Opted for metastable option for 2005 operational
  version
• Very little impact on model with limited aqueous
  phase representation
• Decreased execution time by 20 (2h for 48h
  forecast of whole continent)
• Now in position to consider 00Z and 12Z runs

11
CHANGES IN ANTROPOGENIC EMISSION FILES
Old emission file
New emission file
12
CHANGES IN BIOGENIC EMISSIONS
Isoprene emissions (ug/m2/hr) at standard T (30C)
and standard PAR
USGS/BEIS2
BELD3/BEIS3
13
Validation of 2005 operational version parallel
run

• Operations to be part of the evaluation of new
  versions of AQ models as they are installed
  officially in parallel mode
• Operational meteorologists at CMC and in the
  regions are identified to evaluate the parallel
  model for acceptance to go operational.
• April/May 2005 1 month long parallel run
• Tools to compare parallel and operational runs
  are being developed.
• At the end of evaluation period, CPOP review

14
OBJECTIVE ANALYSIS

• Presently no observation data is used for
  forecast
• One avenue to improve the forecast is to correct
  initial conditions to take into account observed
  data
• Currently, OA of surface O3 is performed but not
  fed to CHRONOS
• First step in assimilation cycle
• Every hour, 24/7, year round
• Operational since June 2004
• Multi-year analysis since summer 2002

http//www.msc.ec.gc.ca/aq_smog/analysis_e.html
15
OTHER REAL-TIME APPLICATIONS

• Experimental real-time scenarios with CHRONOS
• Testing of emerging developments
• assimilation of surface ozone in CHRONOS
• More complete parameterization of PM AURAMS

• Support for development of health base AQI for
  Canada

16
EXPERIMENTAL REAL-TIME SCENARIOS
6 on/off scenarios 1 base case BC, Prairies,
Ontario, Quebec, Maritimes, US
February 5, 2005 Quebec off scenario
February 5, 2005 Base simulation/forecast
17
REAL-TIME SURFACE OZONE DATA ASSIMILATION

• Run experimentally during ICARTT campaign
• Using AIRNow real-time data
• 3h of assimilation on 12Z forecast (limited by
  timeliness of data)
• - Success technologically
• - Scientific evaluation in progress definitely
  promising

• Other data assimilation projects at MSC (R.
  Ménard)
• Online chemistry with operational meteorological
  model GEM
• - CO assimilation GEM with chemistry (CO) MSC
  3D VAR-Chem
• - Coupled chemical-climate data assimilation
  (statosphere) looking at the benefits and
  drawbacks of chemistry-dynamics coupling in data
  assimilation

18
AURAMS REAL-TIME SIMULATIONS

• Initiated for the ICARTT campaign has been
  running ever since
• Part of ensemble forecast during ICARTT (Stu
  McKeen, NOAA)
• AURAMS was developed as comprehensive PM model
• - 12 bins sectional model
• - 7 aerosol chemical species (internally mixed)
  SO4, NO3-, NH4, OC, EC, CM, Sea-salt
• - Aqueous-phase chemistry represented (including
  evaporation, but no convective mixing yet)
• 5h for 48h run at 42km resolution versus CHRONOS at 21km res

19
AURAMS REAL-TIME SIMULATIONS

• Will be providing flight guidance in real-time
  for PRAIRIE 2005
• Part of ensemble forecast at UBC

20

• Support for development of health base AQI for
  Canada

21
INCLUDING FOREST FIRE IN AIR QUALITY FORECAST

• Development lead by MSC - Quebec region
• Inclusion of forest fires for specific
  simulations in CHRONOS (past cases)

2002 Quebec forest fires