Office of Research and Development Atmospheric Modeling and Analysis Division, NERL AQMEII Phase 2: Overview and WRF/CMAQ Application over North America Christian Hogrefe 1, Stefano Galmarini 2, Shawn Roselle 1, and Rohit Mathur 1 1 Atmospheric Modeling and Analysis Division, NERL/ORD, U.S. Environmental Protection Agency, RTP, USA 2 European Commission Joint Research Centre, Ispra, Italy ITM Miami, August 29, 2013

Office of Research and Development Atmospheric Modeling and Analysis Division, NERL AQMEII Overview Established in 2009 by researchers from NA and EU with support from EPA, Environment Canada, and the European Commission’s Joint Research Center AQMEII’s goals are to –Exchange expert knowledge in regional air quality modeling –Identify knowledge gaps in air quality science –Develop innovative methods for evaluating the uncertainty in air quality modeling –Establish methodologies for model evaluation to increase the knowledge on atmospheric processes and to support the use of models for policy development –Prepare coordinated research projects and model inter- comparison exercises 2

Office of Research and Development Atmospheric Modeling and Analysis Division, NERL AQMEII Activities Phase 1: 2010 – 2012 –Annual regional air quality simulations over North America (NA) and Europe (EU) for 2006 –Results published in July 2012 special issue of Atmospheric Environment –Phase 1 database available for ongoing research Phase 2: 2012 – 2014 –Focus on evaluating coupled meteorology/chemistry models –Common emission inputs (TNO, U.S. EPA, and Environment Canada) and chemical boundary conditions (MACC) –Include dynamic model evaluation aspect: simulate both 2006 and 2010 –~20 participating groups from EU and 4 from NA 3

Office of Research and Development Atmospheric Modeling and Analysis Division, NERL AQMEII Phase 2 Timeline May 2012: Kick-off workshop in Utrecht January 2013: Inputs and experiment specifications available Fall 2012 – Spring 2013: collection of observational datasets and preparation of TSD at JRC Summer 2013: Most simulations are completed and data have been / are being uploaded to JRC’s ENSEMBLE system August/September 2013: Workshops in Miami (in conjunction with ITM) and Ispra (in collaboration with COST) Fall 2013 – Spring 2014: Joint and individual analyses, sensitivity simulations, preparation of manuscripts May 2014: Submission of manuscripts to a special issue of Atmospheric Environment 4

Office of Research and Development Atmospheric Modeling and Analysis Division, NERL AQMEII Phase 2 Inputs – 2006 and 2010 Emissions 5 -20% -35% U.S. Domain Total NO x U.S. Domain Total SO 2

Office of Research and Development Atmospheric Modeling and Analysis Division, NERL 6 SO4, Spring 2006 Dust, Summer 2006 Dust, Spring 2006 OC, Summer 2010 AQMEII Phase 2 Inputs – Boundary Conditions ECMWF provided MACC fields that are based on MOZART-IFS with chemical data assimilation MACC examples: selected 750 mb seasonal average PM Concentrations for NA

Office of Research and Development Atmospheric Modeling and Analysis Division, NERL U.S. EPA WRF/CMAQ Simulations Used for Illustrative Analyses WRF3.4/CMAQ5.0.1 two-way coupled model (Wong et al., 2012) Feedback: shortwave direct effects only WRF options RRTM radiation –Morrison 2-moment microphysics scheme –Kain-Fritsch (new ETA) cumulus scheme –Pleim/Xu ACM2 PBL scheme and land/surface scheme –Nudging of temperature, wind speed and water vapor above the PBL –Deep soil nudging CMAQ: CB-05 chemistry, aero6 (3 aerosol modes – nucleation, accumulation, coarse), in-line photolysis calculations Simulations: –2006 and 2010 annual simulations at 12 km over the continental U.S. with direct feedback effect –Two week sensitivity in July 2006 without feedback 7

WRF/CMAQ Sensitivity Simulations With and Without Feedback 8 AOD (Feedback Run) 2m Temperature PBL Height PM 2.5 ConcentrationsO 3 Concentrations Average Differences for Land Cells with AOD >= 0.1: 2m Temperature: K PBL Height: -15 m PM 2.5 : ug/m3 O 3 : ppb What is the strength of the direct feedback effect for other models? Differences Feedback – No Feedback, July 1 – 17, 2006, Daytime Hours

2006 PM Model Predicted Changes in PM 2.5 and Radiation Between 2006 and 2010 May – September, All Hours 2010 PM – 2006 PM – 2006 Clear-Sky Shortwave Radiation at the Surface Modeled surface PM 2.5 concentrations decreased substantially between 2006 and 2010 in many areas These decreases generally lead to increases in modeled clear-sky shortwave radiation at the surface Future analysis will consider the modeled PM 2.5 vertical distribution and composition and will compare the effect simulated by WRF/CMAQ to other models

Office of Research and Development Atmospheric Modeling and Analysis Division, NERL 10 Soil 2010 – 2006Organic Carbon 2010 – 2006 Elemental Carbon 2010 – 2006Sulfate 2010 – 2006 Model Predicted Changes in PM 2.5 Species Between 2006 and 2010 May – September, All Hours Reductions in SO 2 Emissions Changes in Wildfire Emissions Changes in Mobile Emissions Changes in Wildfire Emissions Changes in Windblown Dust Emissions

Observed (CERES) and WRF/CMAQ Changes in Clear-Sky Shortwave Radiation at the Ground, May – June, 2010 Minus WRF/CMAQCERES WRF/CMAQ CERES WRF/CMAQ generally captures the spatial pattern of changes in clear-sky shortwave radiation at the ground between 2006 and 2010 but underestimates the magnitude of the change

12 Observed (MODIS) and WRF/CMAQ Changes in AOD, May – September, 2010 Minus 2006 WRF/CMAQMODIS WRF/CMAQ MODIS WRF/CMAQ generally captures the spatial pattern of changes in AOD between 2006 and 2010 but underestimates the magnitude of the change

13 Spatial Average All Sites Observed (AERONET) and WRF/CMAQ AOD at 500nm, 2006 Modeled – Observed AOD, Annual Mean Observed AOD, Annual MeanModeled AOD, Annual Mean WRF/CMAQ captures spatial and temporal AOD patterns, underestimates magnitude especially during summer – other models?

Example Observed and WRF/CMAQ AOD Time Series at Individual AERONET Sites Use AOD time Series at individual AERONET sites to identify locations and time periods for in-depth diagnostic analyses Utilize observed and modeled total and speciated PM 2.5 to interpret AOD results Also perform offline diagnostic analysis (e.g. using FlexAOD) to separate the effects of PM 2.5 composition, vertical distribution, and optics calculations on modeled AOD Long-Range Transport Wildfires Urban Environment

Observed and WRF/CMAQ Total PM 2.5 at IMPROVE Locations Spatial Average Time Series and Monthly Average Statistics In both years, WRF/CMAQ shows an overestimation of total PM 2.5 during winter and an underestimation during summer at these predominantly rural monitors

16 Observed and WRF/CMAQ Seasonal Average PM 2.5 Composition at CSN Locations, All Sites, 2006 Winter 2006 ObservedWRF/CMAQ Summer 2006 ObservedWRF/CMAQ Seasonal average SO 4, NO 3, and NH 4 concentrations are generally captured by WRF/CMAQ in both winter and summer WRF/CMAQ shows a significant overestimation of total carbon, non-carbon organic matter, and other primary PM 2.5 during winter at these predominantly urban monitors

Office of Research and Development Atmospheric Modeling and Analysis Division, NERL 17 Dynamic Evaluation: Changes in Summer Average PM 2.5 Concentrations, Relative Changes (%)

Office of Research and Development Atmospheric Modeling and Analysis Division, NERL 18 Dynamic Evaluation: Changes in Summer Average O 3, NO x, and SO 2 Concentrations, Observed Ozone Change WRF/CMAQ Ozone Change

Office of Research and Development Atmospheric Modeling and Analysis Division, NERL Dynamic Evaluation: Differences in Summertime Meteorological Fields; 2010 minus m TemperaturePBL Height10m Wind Speed Cloud Fraction Changes in meteorology likely counteracted some of the emission effects on ozone in parts of the Eastern U.S. and enhanced them in the Western U.S. Future analyses to quantify the impact of changes in meteorology on pollutant concentrations may include synoptic typing, development and application of meteorological adjustment techniques, etc.

20 Dust, 750 mb, Summer Dust, 750 mb, Spring OC, 750 mb, Summer Dynamic Evaluation: Impact of Boundary Conditions on Regional-Scale Simulations: Differences in Selected MACC 750 mb Seasonal PM Fields SO4, 750 mb, Spring Future analyses will be geared towards understanding the impact of boundary conditions on simulated interannual variability for different species and AOD and how these impacts vary by region and season

Office of Research and Development Atmospheric Modeling and Analysis Division, NERL Summary and Outlook (1) AQMEII Phase 2 focuses on evaluating and intercomparing online coupled meteorology/chemistry modeling systems By preparing inputs for 2006 and 2010 and by collecting a large observational dataset, it will be possible to perform operational, diagnostic, dynamic, and probabilistic evaluation of these models WRF/CMAQ simulations were used to illustrate some of the planned analysis approaches Joint analyses will be performed by JRC, and additional analyses will be performed by participating groups Process-based modules like FlexAOD (see presentation by G. Curci) will be used for diagnostic analyses of feedback effects 21

Office of Research and Development Atmospheric Modeling and Analysis Division, NERL Summary and Outlook (2) Case studies of interest for diagnostic analysis include Russian wildfires, Saharan dust transport events, and CALNEX/CARES Some participating groups plan to perform sensitivity simulations with different feedback and/or nudging options for selected episodes Through all of these analysis and the AE special issue, it is envisioned that AQMEII Phase 2 will document the state of science in coupled regional-scale modeling and help to establish the value of including feedback effects to better account for the effects of changing emissions and meteorology in regional-scale modeling applications 22