Presentation on theme: "Development of a High Resolution Air Quality Prediction System for the 2015 PanAm and Para PanAm Games Craig Stroud, Sylvie Gravel, Balbir Pabla, Mike."— Presentation transcript:
1Development of a High Resolution Air Quality Prediction System for the PanAm and Para PanAm GamesCraig Stroud, Sylvie Gravel, Balbir Pabla, Mike Moran, Sylvain Menard,Paul Makar, Junhua Zhang, Alain Robichaud, Wanmin Gong,Heather Morrison, Veronique Bouchet
2Outline Objectives for AQ Forecasting during PanAm Summary of Prior AQ Studies in TorontoModel Development ProjectsCurrent ResultsFuture Work
3Objectives of the Air Quality (AQ) Science Showcase To develop an integrated environmental model to address scientific questions related to urbanization:To study the feedbacks between aerosol and weather in Canada’s largest cityTo perform an aerosol source apportionment study at the U of Toronto rooftop comparing both receptor and emission models, assess decadal changes in emission-based models and observations for Toronto.To improve urban spatial surrogate maps for allocating on-road mobile emissions and food cooking emissionsTo develop and validate a “high-resolution” objective analysisOpportunity to align projects to a common domain and time period to foster collaboration within EC and with our partnersTo accelerate the development of a “high-resolution” operational AQ prediction system for urban cities across Canadaopportunity
4Urban Heat Island and its Impacts on Air Quality and Weather Makar et al., (2006) Atmos. Environ.Used satellite visible radiation mapsto constrain anthropogenic heat inputImproved predictions of nighttime tempSignificant impact on PBL mixing heightPolandGEM-AQ5.Struzewska and Kaminski (2012)ACP, 12,Improved temp and wind speedIn most cases, primary pollutants were more vertically mixed with UHI.
5Literature Review on PM2.5 Observations For the Greater Toronto Area SouthernOntario2001 – 2003 NAPS PM2.5 Data30-45% of PM2.5 was from local sources and 55-70% from regional transport
6PMF Analysis 24-hr Filter Data 2004-2007 Secondary PM2.5 factors from sulfate and nitrate were 60%. Gasoline and diesel fuel combustion factors summed 26%. Biomass burning and road dust were small for Toronto. Food cooking was not identified as a unique factor in the analysis.
7Ontario NEI has POA area sources calculated at 64% of total POA sources Food cooking is the largest area source in summer.
8Interestingly, a factor analysis study, using highly time-resolved organic aerosol mass spectra and VOC mass spectra, yielded an organic aerosol budget as follows:Food cooking was identified as a significant source of PM1 organicaerosol. Comparable to the OOA-1 which is believed to consist largelyof SOA. Primary HOA was 15% and Biomass Burning was 14%.
9Model Development with GEM-MACH v2 Migrating chemistry routines to the latest version of EC’s operational weather forecast model (GEM v4) to benefit from improvements to the ongoing boundary layer dynamics and cloud physics parameterizations:New vertical coordinate (hybrid in log-hydrostatic-pressure); new vertical discretization (Charney-Phillips staggering) lowest layer depth is now 40-m; physics spin-up capability; piloting of LAM at the lid; global Yin-Yang grid; native vertical diffusion scheme possible for chemical tracers; new TKE scheme.Different sets of chemical lateral boundary conditions based on MOZART reanalysis (seasonal 1-D vertical profiles in GEM-MACH v1 vs. seasonal 3-D mean fields in GEM-MACH v2)Correction of previous errors in surface emissions input and gas-phase deposition in GEM-MACH v1New parameterization of below-cloud scavenging of size-resolved particles by both rain and snow6.
10Model Developments Continued Evaluated ADOM-II gas chemistry against SAPRC-07 gas chemistry for Canadian domainUpdate Maestro task sequencer suite for launching GEM-MACH v2New evaluation data sets (mobile labs, satellite obs) and evaluation tools (ValidatoR)Newer emissions inventories and projectionsUpdated 2006 Can inventory (projection to 2015), 2011 US inventoryNew U.S. spatial surrogates based on 2010 U.S. censusImproved Canadian and U.S. temporal profiles (e.g., diurnal NH3 profiles for livestock)Expanded library of PM speciation profilesTesting new spatial surrogate maps for high-resolution PanAm domain (for rail, road and food cooking)7.
11“Zoom” of New Canadian Railroad Shapefile over Toronto Area: The Shapefile is Used to Construct Spatial Surrogate Fields for Allocating Canadian Railroad Emissions
12Link-Based On-Road Pollutant Emissions (McMaster University Traffic Flow Model)
13Evaluating New Traffic Emissions Traffic Flow Modellingon Road NetworkEvaluating New Traffic EmissionsAverage Modeled NOx Mixing RatioJune 18 – July 9, 2007Little bias in Modeled NOxmixing ratiosLink-basedtraffic emissionsinput to model2.5-km model grid spacing
14Residential, August, Weekday, V2 Commercial, August, Weekday, V2 Food Cooking PM2.5 Inventory and Spatial DistributionMcMaster University StudyResidential, August, Weekday, V2Commercial, August, Weekday, V2Needs evaluation – PM source apportionment project, in collaboration with Prof Arthur Chan at U of Toronto
15Real-Time AQ Forecasts for PanAm are now Running Real-Time AQ Forecasts for PanAm are now Running ! Nested GEM-MACH v2 at 2.5-km, 250x300 gridHigh Resolution DomainIncludes Detroit/Windsor,Cleveland, Buffalo, PittsburghIncludes 3 Great Lakes to capture lake breeze effectMet will be driven by operational 2.5-km GEMv4GMv2 just compiled, currently under evaluation, PanAm first project to use v2.Differences in v2 vs v1, vertical coord and discretization, thinkness of lowest layer increases from 20m to 40m. Native vertical diffusion scheme to GEM now used for tracers, TKE equation for PBL dynamics changed. Cloud physics scheme more physically based (M&Y). For regional GMv2, the driving GEMv4 model currently is GDPS (25km) for met initialization and met piloting files. Operations will switch to global variable at 15-km soon which will match 10-km resolution better.24-hr forecast, initiated once a dayReal-time predictions of AQHI at air quality stations
16GEM-MACH v2 Aerosol Mass Average for July 2014, Weekdays Seasonal PM2.5 CompositionNAPS Filter Data4 ug/m3 pSO4
17GEM-MACH v2 Aerosol Composition Average for July 2014, Weekdays NitrateBlack CarbonPOASulfate
18O3 Evaluation for GEM-MACH v2, Summer 2008 2.5-km10-kmModelO3ModelO310-km StatsO3 Bias = 16.4 ppbvR = 0.67RMSE = 21.5 ppbv2.5-km StatsO3 Bias = 9.3 ppbvR = 0.75RMSE = 14.9 ppbvImproved O3 bias with 2.5-kmgrid spacing, likely due to higher resolution and new emission data sets.
19Sensitivity Runs to Diagnose Ozone Over-prediction Evaluated isoprene mixing ratios with NAPS data – no systematic bias was observed for eastern CanadaEvaluated photolysis rate of NO2 at one location – model J-values were biased high for periods with observed cloudy conditions; may contribute to the ozone over-prediction; improvements needed.Reviewed literature for OH+NO2→ HNO3 reaction rate – our kOH in mechanism is reasonable.Lowered vertical diffusivity lower limit for rural land use and increased for urban land use – no significant change in ozone predictionsSensitivity test with doubled ozone dry deposition showed a strong sensitivity to assumed deposition velocity for land use type – review of ozone dry deposition parameters is underway.
20Future Work with High Resolution AQ Modeling Complete the evaluation of the high resolution system and emission data sets against measurement data sets:PM speciation dataMobile data sets (CRUISER and MAPLE)Science Code Revisions (this fall):Port the 2-way interaction code for GEM-MACH v2Semi-volatile organic aerosol parameterizationAdd several toxic species (benzene, benzo-a-pyrene)Updates for outstanding issues (advection, O3 gas deposition)Non-linear Post-Processing, develop with summer 2008, test for 2014Objective Analysis, develop with summer 2008, test for 2014During Games, upload forecast products to data portals (WISDOM, DATAMART) and work with data users (EC forecast desk, Toronto and Kingston health units)After Games, develop case studies for impacts of aerosol indirect effect on weather and AQ in an urban environment – evaluate model against rich met mesonet. Interested in joining WMO working groups.
21Air Quality Mobile Laboratories – CRUISER and MAPLE for Summer 2008NOxNO2
22Acknowledgements Prof. Greg Evans, University of Toronto Prof. Pavlos Kanaroglou, McMaster UniversityDavid Henderson, AQHI MSC ProgramDr. Jeff Brook, ARQP SectionDr. Bob Vet, Natchem data, ARQM SectionDr. Sylvie Leroyer and Dr. Stephane Belair, MRDNAPS, IMPROVE, AIRS Data NetworksThank you for your attention!