Status of the New Canadian Air Quality Forecast Model: GEM-MACH15 Michael Moran 1, Donald Talbot 2, Sylvain Ménard 2, Véronique Bouchet 1, Paul Makar 1, Wanmin Gong 1, Alexander Kallaur 3, Hugo Landry 2, Ping Huang 4, Sunling Gong 1, and Louis-Philippe Crevier 2 1 Air Quality Research Division, Environment Canada, Toronto, Ontario 2 Air Quality Modelling Applications Section, Environment Canada, Montreal, QC 3 Air Quality Research Division, Environment Canada, Montreal, Quebec 4 Independent consultant, Toronto, Ontario 8th CMAS Conference Chapel Hill, NC19 October 2009

DRAFT – Page 2 – October 21, 2015 Talk Outline Background GEM-MACH: General description GEM-MACH15: Configuration GEM-MACH15: Performance evaluation Current status

DRAFT – Page 3 – October 21, 2015 Environmental Forecasting at EC Daily weather forecasts plus Ocean waves and storm surges Sea ice Air pollution potential Volcanic ash UVB index and total column O 3 Surface O 3, PM 2.5, and PM 10 AQ Health Index (AQHI: based on O 3, PM 2.5, and NO 2 ) Hazardous releases (e.g., Chernobyl, FMD,...)

DRAFT – Page 4 – October 21, 2015 Chronology of EC Air-Quality Prediction Program 1998: Experimental CTM-based forecasts of ground-level O 3 begin for eastern Canada with CHRONOS (40-km grid spacing) 2001: Operational CHRONOS ozone forecasts begin; new national domain, 21-km grid spacing 2003: Bulk PM 2.5 /PM 10 forecasts added to CHRONOS output suite (4 chemical components)

DRAFT – Page 5 – October 21, 2015 Objective for GEM-MACH15 Project (Autumn 2005) To replace CHRONOS, the current EC operational off-line regional AQ forecast model for O 3, NO 2, PM 2.5, and PM 10, with a new GEM-based on-line operational AQFM that includes a science package equivalent to the one in the AURAMS CTM

DRAFT – Page 6 – October 21, 2015 Terminology What is GEM? – Environment Canada’s operational global/medium- range and regional/short-range weather forecast model What is GEM-MACH? – GEM with on-line chemistry from AURAMS CTM What is GEM-MACH15? – proposed operational limited-area configuration of GEM­MACH with 15-km grid spacing

DRAFT – Page 7 – October 21, 2015 Acronym “GEM-MACH” Stands For modèle Global Environnemental Multi-échelle –Modélisation de la qualité de l'Air et de la CHimie et / and Global Environmental Multiscale model – Modelling Air quality and CHemistry

DRAFT – Page 8 – October 21, 2015 "Multiscale" Examples: Three GEM Grid Configurations global regional limited area

DRAFT – Page 9 – October 21, 2015 GEM-MACH Data Flow (2005/6) CAN-USA-MEX Emission inventories SMOKE AreaMajor points GEM-MACH Emissions pre-processor GEM-MACH Main model PM 2.5, O 3, NO 2,... Temperature, pressure … GEM Meteorological pre-processor GenPhysX Biogenic info (Beld3, emis factor) ClimatologyGeophys Meteorological Analysis off-line system on-line system RPN standard file format Emissions binary files for each CPU tile Meteorological binary files for each CPU tile

DRAFT – Page 10 – October 21, 2015 Chemical Process Representations in CHRONOS vs. GEM-MACH15 PROCESSCHRONOSGEM-MACH15 Emissions PM 2.5 and PM c primary emissions are assumed to be bulk emissions; 17 gas- phase species emitted PM 2.5 and PM c emissions speciated to 7 species by primary source type (point, area, mobile); 17 gas-phase species emitted Gas-Phase Chemistry Mechanism ADOM-2 mechanism (Stockwell and Lurmann, 1989); 47 advected species ADOM-2 mechanism (Stockwell and Lurmann, 1989) with 1) p-SO 4 replaced by H 2 SO 4 +p-SO 4 2) N 2 O 5 + H 2 O “heterogeneous nitrate formation” rate enhancement switch=off. Aqueous-Phase Chemistry NoneADOM aqueous-phase chemistry PM Composition and size distribution 2 size bins: PM 2.5, PM 10 4 chemical species: sSO 4, sOC, H 2 O, primary PM 2 size bins: PM 2.5, PM 10 8 chemical species: SO 4, NO 3, NH 4, EC, pOC, sOC, CM, SS Aerosol DynamicsSedimentation Sedimentation, Nucleation, Condensation, Coagulation Secondary Organic (SOA) Yields Based on Pandis et al. (1992)IAY scheme Based on Jiang (2004) Wet DepositionDistribution of LWC is used to calculate the wet scavenging term by applying Sundqvist formulae for the rate of release of precipitation Transfer of tracers from cloud to rain water based on precipitation production. In- cloud and below-cloud scavenging of soluble gases and particles (size- dependent). Chemical boundary conditions Zero-gradient inflow, open-boundary out- flow Climatological profiles with Davies lateral boundary conditions

DRAFT – Page 11 – October 21, 2015 Simplifications to GEM-MACH15 for Operational Use Perform meteorological calculations every time step (450 s) but AQ calculations every 2nd time step (900 s) Used 58 vertical levels to 0.1 hPa rather than 80 levels to 0.1 hPa used by GEM15 Used metastable option in heterogeneous chemistry Switched from 12-bin to 2-bin representation of PM size distribution [reduces number of advected tracer fields by 80 from 137 to 57, i.e., by ~60%, but had to implement new sub-bin calculations to account for size dependence in some PM processes (sea-salt emissions, dry deposition, intersectional transport)]

DRAFT – Page 12 – October 21, 2015 GEM15 and GEM-MACH15 Grids GEM15 core grid (red) ; GEM-MACH15 grid (blue) GEM15 employs a variable grid, but uniform core has 15-km spacing GEM-MACH15 employs a limited-area grid (LAM), also with 15-km spacing and co-located with GEM15 grid points GEM15 supplies meteorological initial conditions and hourly lateral boundary conditions to LAM

DRAFT – Page 13 – October 21, 2015 CHRONOSGEM-MACH x 250, polar stereographic proj’n horizontal grid spacing of 21 km 24 Gal-Chen vertical levels to ~ 6 km Δt = 3600 s Emissions: 2000 Canadian & 2001 US with adjustments to 2007 EGU levels 348 x 465, rotated lat-long projection horizontal grid spacing of 15 km 58 eta hybrid levels to 0.1 hPa Δt = 450 s for meteorology Δt = 450 x 2 = 900 s for chemistry Emissions: 2006 Cdn, 2005 US, 1999 Mx CHRONOS and GEM-MACH15 Setups

DRAFT – Page 14 – October 21, 2015 Summer 2009 "Parallel Run" Evaluation GEM-MACH15 was run “in parallel” alongside CHRONOS from May 27 to Aug. 17, 2009 Subjective evaluation – CMC and regional AQ forecasters (6 offices) – comparison to GEM15, to local AQ measurements, and to CHRONOS Objective evaluation – discrete statistics (with significance testing) – categorical statistics

DRAFT – Page 15 – October 21, 2015 Example of Day 2 Hourly O 3 Comparison (Full Domain)

DRAFT – Page 16 – October 21, 2015 Example of Day 1 Hourly NO 2 Comparison (Full Domain)

DRAFT – Page 17 – October 21, 2015 Example of Day 1 Hourly PM 2.5 Comparison (Eastern North America)

DRAFT – Page 18 – October 21, 2015 Summer 2008 O 3 and NO 2 Monitor Locations

DRAFT – Page 19 – October 21, 2015 CHRONOS vs. GEM-MACH15: Discrete Statistics June-July-Aug Not statistically different 0 % yellow CHRONOS better 43 % blue GEM-MACH better 57 % red

DRAFT – Page 20 – October 21, 2015 CHRONOS vs. GEM-MACH15: Categorical Statistics– June-July 2009  Percent Correct (PC) is in CHRONOS’s favour for this period: O 3 = 0.77 (0.80); PM 2.5 = 0.73 (0.77)  Probability of Detection (POD) is slightly in favour of GEM-MACH15  False Alarm Rate (FAR) and Critical Success Index (CSI) are comparable for two models for this period

DRAFT – Page 21 – October 21, 2015 Outcome of "Parallel Run" Evaluation Request at the Sept. 3, 2009 meeting of the Committee on Operational and Parallel Runs at the Canadian Meteorological Centre (CMC) for GEM-MACH15 to be elevated to operational status was accepted Final modifications to GEM-MACH15 operational suite are now being completed’; the switch from CHRONOS to GEM-MACH15 is expected the week of Oct. 26 th GEM-MACH15 is a better platform for further development and improvements Entire effort has taken about 4 years

Thank you for your attention!