Jonathan Pleim and the CMAQ development Team, USEPA

A new major version of CMAQ is currently in the process of been released to the community through CMAS. The new version includes updates to three chemical mechanisms (CB05, SAPRC07, and RACM2) particularly to improve nitrogen cycling and halogen chemistry. The aerosol model has been revised to include additional sources and mechanisms for formation of secondary organic aerosols. There are also improvements to biogenic and sea salt emissions, updated aerosol nucleation, gravitation settling of coarse aerosols, bidirectional soil NO, and several improvements to meteorology modeling especially for high resolution applications. In addition, several modifications have been made to improve computational efficiency, including code restructuring and optimization, improved I/O, and faster PBL solver. The 2-way coupled WRF-CMAQ is also upgraded for both its direct and indirect aerosol feedback effects and updated to the latest version of WRF (v3.7). A beta version of CMAQv5.1 was made available to requesting members of the community for early testing and contribution of additional components. Major features of the new version will be summarized along with plans for developing the Next Generation Air Quality Model.

IOAPI_3 has been used to handle I/O in CMAQ for years. A library PARIO was built on top of IOAPI_3 to facilitate I/O operation while CMAQ is executing in parallel. However, this design ultimately uses a designated process to collect data from each sub-domain and then write it out to a file. The new implementation will be based on pnetCDF, a parallel I/O library, to utilize true I/O parallelism in a parallel file system. In addition, an application level data aggregation technique is adopted to enhance pnetCDF performance substantially. This new parallel I/O feature has been tested in CMAQ 5.1 and significant improvement in I/O, in particular writing out data, has been recorded.

Shunliu Zhao, Amir Hakami (Carleton University); Matt D. Turner, Shannon L. Capps, and Daven K. Henze (University of Colorado); Peter B. Percell (University of Houston); Jaroslav Resler (ICS Prague); Jesse O. Bash, Sergey L. Napelenok (USEPA); , Rob W. Pinder; Armistead G. Russell and Athanasios Nenes (Georgia Tech); Jaemeen Baek, Greg R. Carmichael, and Charlie O. Stanier (University of Iowa); Adrian Sandu (Virginia Tech); Tianfeng Chai (University of Maryland); Daewon Byun (NOAA)

Quazi Ziaur Rasool, Rui Zhang, Benjamin Lash, Daniel S. Cohan

Soil NO emissions comprise approximately 20% of the global NOx budget and can affect regional and urban air quality by contributing to ozone and particulate matter (PM) formation. Soil NOx is mainly emitted as NO released as a byproduct of microbial nitrification (Oxidation: NH₄⁺ NO₃^-) and denitrification (Reduction: NO₃^- N₂). Soil NO emissions estimates are highly uncertain, ranging from 4 to 15 Tg N yr¹ globally across different modeling studies using satellite and/or ambient observations. The current generation of air quality models (e.g. CMAQ in this case) has large uncertainty for soil NO due to underestimation attributed to use of the Yienger-Levy soil NO scheme. That scheme also has been shown to misrepresent the timing and spatial distribution of emissions due to inaccuracies related to soil temperature, soil water content, pH value, and mineral nitrogen availability in the soil.

In this study, soil NO estimates from CMAQ version 5.0.2 were compared as generated by: a) the Yienger-Levy scheme from the base model, and b) an adapted BDSNP (Berkley Dalhousie Soil NOx Parameterization) scheme. We further test the impact of alternate implementations of BDSNP with a soil N pool derived from either (i) static long-term average data for annual fertilizer application rate from Potter et al., 2010 or (ii) dynamic daily fertilizer N pool from Environmental Policy Integrated Climate (EPIC) model via the Fertilizer Emission Scenario Tool (FEST-C). The previous BDSNP module implemented by Rice in CMAQ used the soil biome map directly re-gridded from the GEOS-Chem global model, which is too coarse for regional model implementation. Hence, we developed a new soil biome spatial map based on 12km CONUS 40-category 2006 NLCD-MODIS land use classification (NLCD40) and climate zone definition to better represent LU/LC (Land Use/Land Cover) conditions. The impact of biogenic soil NO emission estimates using the above stated schemes on air quality modeling is quantified by a WRF-MEGAN-CMAQ simulation platform over the continental US with 12km horizontal resolution. The test simulation period coincides with the NASA DISCOVER AQ campaign from June 27 - July 31, 2011, over Maryland, where abundant meteorological, gaseous and aerosol pollutants measurements are available for comparisons.

Xinyi Dong^a, Joshua S. Fu^a*, Kan Huang^a

^a Department of Civil and Environmental Engineering,The University of Tennessee, Knoxville, Tennessee, 37996, USA

Inline calculation of dust plume rise was available with CMAQ yet model applications and validations of the scheme are limited over East Asia. Deserts in China and Mongolia are the main terrestrial sources of airborne dust, with major contribution from Taklamakan and Gobi which can travel to China, Japan, Taiwan and even across Pacific Ocean to US. In default CMAQ dust module, soil moisture was double-counted while analyzing the field measurement data for friction velocity threshold. In this study, we used the new parameterization of the thresholds based on the re-analysis work by Dr. Daniel Tong. We also revised the dust particle speciation profiles to be source-dependent for Taklamakan and Gobi respectively based on local measurements. By evaluating against observations, the revised dust module was found to significantly improve the model's performance for predicting PM10 and crust species concentrations. We also implemented heterogeneous chemistry for dust particles into CMAQ, and probe into their impact on ambient concentrations of O3, SO2 and PM2.5 at downwind areas over East Asia.

Jared H. Bowden, Kevin D. Talgo, Uma Shankar and Aijun Xiu

Institute for the Environment, UNC Chapel Hill

Increasing temperatures and shifts in the atmospheric circulation as a consequence of climate change will likely impact wildfire regimes. This study investigates potential changes in wildfire regimes over the Southeast US around the mid-twenty-first century. Here we use a combination of global and dynamically downscaled model output to examine these meteorological changes. Our analysis focuses on years with low and high fire potential for a historical decade (1996-2005) and a future decade (2041-2050) for both RCP4.5 and RCP8.5. The low and high fire potential years are selected based on the Haines Index, a measure of the moisture content and stability of the lower atmosphere, and used to dynamically downscale these years to a 12-km resolution over the Southeast U.S. Our analysis focuses on the summer season and relates the meteorological changes to the position of the North Atlantic Subtropical High (NASH). Relating the meteorological changes for selected years to the position of the NASH is important in a climate change context because the NASH has shifted westward in recent decades, and is projected to continue the westward migration as the climate warms. Discussing the high and low fire potential years in a climate context is important because selecting these years provide a cost-effective means of dynamical downscaling for assessing regional climate change impacts on smoke and fire, and the associated air quality.

Havala Pye, Deborah Luecken, Donna Schwede, Kirk Baker, Bill Hutzell

US EPA

Lu Xu, Christopher Boyd, Sally Ng

Georgia Tech

Karsten Baumann, Eric Edgerton

Atmospheric Research & Analysis, Inc.

Biogenic volatile organic compounds (BVOCs) contribute significantly to organic aerosol in the southeastern United States. During the Southern Oxidant and Aerosol Study (SOAS), a portion of ambient organic aerosol was attributed to isoprene oxidation and organic nitrogen from BVOC reaction with nitrate radicals. In this work, we compare observed values of biogenic aerosol with CMAQ v5.1 and updated model representations. We show that monoterpenes are predicted to account for the majority of particle-phase organic nitrates in Centreville, AL during SOAS, and a new mechanistic representation of particle-phase organic nitrates produce more organic aerosol than standard CMAQ v5.1. Hydrolysis of particle-phase organic nitrates decreases gas-phase alkyl nitrates, increases modeled HNO₃ and organic aerosol, but does not substantially increase particle nitrogen.

Shawn Urbanski, Robin Silverstein, Matt Reeves, and Wei Min Hao

We estimate daily pollutant emissions for multiple large wildfires using a newly developed wildland fuels map and an updated emission factor database. Fuel loading for forests is taken from a new forest fuel classification which was developed from a large set of USFS Forest Inventory and Analysis surface fuel estimates (n > 28,000). Rangeland fuel loading is estimated with a Normalized Differenced Vegetation Index (NDVI) based biomass product developed using a large set of field data from the USDA Soil Survey Geographic (SSURGO) database, NDVI from the MODIS sensor on the Terra satellite, and landscape attributes. The significance of the wildfire emissions as a source of NAAQS air pollutants is evaluated with respect to anthropogenic emissions reported in the US EPA National Emission Inventory.

Theran P. Riedel, Zhenfa Zhang, Kevin Chu, Joel A. Thornton, William Vizuete, Avram Gold, and Jason D. Surratt

The formation of epoxide products from isoprene photooxidation is known to be critical precursor of significant secondary organic aerosol (SOA) mass. Isoprene epoxydiols (IEPOX) have been shown to produce substantial amounts of SOA mass and are therefore considered major isoprene-SOA precursors. Heterogeneous reactions of IEPOX on atmospheric aerosols form various aqueous-phase components or "tracers" that contribute to the SOA mass burden. A limited number of the reaction rate constants for these acid-catalyzed aqueous-phase tracer formation reactions have been constrained through bulk laboratory measurements. Namely, only IEPOX tetrol and organosulfate formation have been characterized. While these tracers are responsible for a sizeable fraction of IEPOX-derived SOA, there are a number of other tracer formation reactions have yet to be examined and are of equal importance. To this end we have designed a chemical box model with numerous experimental constraints to explicitly simulate gas- and aqueous-phase reactions during chamber experiments of SOA growth from IEPOX uptake onto acidic sulfate aerosol. Specifically, the model is constrained by recent measurements of the IEPOX reactive uptake coefficient, the few aforementioned experimentally obtained aqueous-phase rate constants, chamber-measured aerosol mass and surface area concentrations, aerosol thermodynamic model calculations, and offline filter measurements of SOA tracer species. Through the use of the offline filter measurements collected during the chamber experiments, we are able to place estimates on the aqueous phase tracer formation rate constants that have yet to be measured for bulk solutions and assess how tracers vary with IEPOX levels. In this way we obtain valuable constraints on particle-phase species that have been quantified through offline techniques but lack formation rate information.

Kenneth Craig¹, Sean Raffuse¹, Narasimhan Larkin², ShihMing Huang¹, Stacy Drury¹, and Kim Lorentz¹

¹Sonoma Technology, Inc., Petaluma, CA
²USDA Forest Service, Seattle, WA

Over the past several years, large high-intensity wildfires, or "megafires," have set records for the greatest burn area and most costly fires in several U.S. states. Megafires can release many tons of fine particles and other pollutants that are hazardous to human health over a short period of time. Under future climate scenarios, megafires may increase in some regions. The danger of smoke exposure from megafires in the future depends on several spatial factors, including the likelihood of megafire occurrence, emission rates, air transport patterns, and population density. We combined climatological transport modeling, fire emission rates, and population density to determine the areas within the U.S. where a megafire would result in the greatest human exposure to smoke. Climatological transport potential was assessed using 5-day HYSPLIT trajectories initiated throughout the U.S. on a 64-km² grid every 6 hours from 1979 to 2009, while smoke emission rates were developed based on FCCS-LANDFIRE fuels and Consume 4 consumption and PM_2.5 emissions. To corroborate results from this trajectory-based analysis, we also estimated probabilistic smoke impacts for select areas identified as high-risk for future megafires using a large ensemble of HYSPLIT dispersion model simulations. The dispersion simulations were linked to upstream vegetation, consumption, and emissions models through the BlueSky smoke modeling framework, and were conducted every third day from 1979 to 2009 using meteorological data from the North American Regional Reanalysis (NAYY). Initial results from the ensemble dispersion modeling analysis are consistent with results from the trajectory-based smoke impact analysis. Coupled with a synthesis of recent studies on the likelihood of megafire occurrence under future climate scenarios, these results provide a view of future smoke management and emergency response needs.

Jim Smith, Mark Estes and Jocelyn Mellberg, Texas Commission on Environmental Quality

Ou Nopmongcol and Greg Yarwood, Ramboll Environ

A pervasive issue with photochemical model application in Texas, particularly near the Gulf Coast, is over-prediction of ozone concentrations advected onshore from the Gulf of Mexico. To address this issue the Texas Commission on Environmental Quality contracted with Ramboll Environ to add halogen chemistry associated with oceanic emissions to the Carbon Bond version 6 (CB6) chemical mechanism implemented in the Comprehensive Air Quality Model with extensions (CAMx). Adding halogen chemistry is shown to reduce the bias at coastal sites, but still leaves substantial over-prediction. Additional modeling has shown that the bias is sensitive to boundary conditions over the Gulf and Atlantic waters, suggesting that the GEOS-Chem model may also over-predict maritime ozone. This paper describes the implementation of halogen chemistry in CAMx and shows results of several model runs aimed at improving the model bias along the Texas coast.

Radenko Pavlovic¹, Mike D. Moran², Paul-Andre Beaulieu¹, and Sophie Cousineau¹

¹Air Quality Modeling Applications Section, Environment Canada, Montreal, Quebec, Canada

²Air Quality Research Division, Environment Canada, Toronto, Ontario, Canada

The experimental FireWork North American air quality forecast system with near-real-time wildfire emissions was developed by Environment Canada (EC) in 2011. This system is identical to EC's operational regional air quality forecast system except for the inclusion of wildfire emissions. 48-hour air quality forecasts issued by FireWork have been available to EC operational forecasters since 2013 and certain FireWork products have been available to external users since 2014. At this time the system is being run twice daily from mid-May to November 1^st by the operational arm of EC's Canadian Meteorological Centre.

Near-real-time satellite-based information about active wildfires in both Canada and the USA is provided to FireWork via the Canadian Wildland Fire Information System (CWFIS), which is part of Natural Resources Canada (NRCan). Once processed, wildfire emissions from individual sources are injected into elevated model layers based on plume-rise calculations, after which transport and chemistry calculations are performed.

A detailed performance analysis of FireWork forecasts has been carried out for the summers of 2013, 2014, and 2015, with a focus on periods of intense fire activity. Case studies for both the Canada and the USA showed noticeable improvements in PM_2.5 forecasts for regions impacted by wildfires, and FireWork has also demonstrated skill in forecasting long-range transport from wildfires. In this presentation, a number of performance analyses will be presented for PM_2.5, such as objective and categorical scores, smoke dispersion and advection analyses, and case studies. We will also describe ongoing and future work planned for the FireWork system.

William T. Hutzell¹, David C. Wong¹, Frank Binkowski², Jesse Bash¹, and John Streicher¹

¹Atmospheric Modeling and Analysis Division, US Environmental Protection Agency, Research Triangle Park, North Carolina, USA

²Institute for the Environment, University of North Carolina, Chapel Hill, North Carolina, USA

How photolysis rates are calculated affects ozone and aerosol concentrations predicted by the CMAQ model and the model's run-time. The standard configuration of CMAQ uses the inline option that calculates photolysis rates by solving the radiative transfer equation for the needed actinic fluxes. The solution uses meteorological conditions and predicted concentrations at the current time step to determine extinction and scattering. In version 5.02 of CMAQ, the inline option has shortcomings in scientific and computational areas. The former includes how extinction and scattering are calculated for clouds and aerosols. The latter area deals with poor computational efficiency and a code structure that is difficult to follow. In the next version of CMAQ, 5.1, the inline option attempts to remove these and other problems. The scientific updates make extinction and scattering from clouds more consistent to meteorological inputs. They also provide runtime options to use more accurate methods for calculating optical properties of aerosols. Regarding the computational area, the source code of the inline option has been reorganized to improve legibility and has greater efficiency regarding run-time.

Kirk Baker

Previous photochemical model based source apportionment assessments of wild land fire impacts on O3 and PM2.5 provide an estimate of impacts from all wild land fires and do not differentiate the impacts from specific fire events on air quality. Here, a photochemical transport model is used to characterize the O3 and PM2.5 impacts from all fires and specific fire events using both source sensitivity and source apportionment approaches. Impacts are assessed for 3 different wild land fire events during 2011: Wallow (eastern Arizona), Waterhole (southeast Montana), and Big Hill (southern Idaho). A large prescribed fire event in the Flint Hills region of central and eastern Kansas is also tracked for contribution to O3 and PM2.5. Since downwind monitor locations are impacted by both these fire events and other sources it is difficult to directly evaluate the modeling system characterization of these events. Modeled O3 enhancements are compared to carbon monoxide from each event as a function of distance from the fire and compared to ambient estimates of fire events published in literature to provide some context regarding fire event representation in the model.

The contribution from all fires are tracked in CMAQ using source apportionment to model estimated O3 (Kwok et al., 2015)and PM2.5 (Kwok et al., 2013)and in a separate simulation all fires except the specific fire events of interest are tracked with source apportionment. This approach allows for estimating the contribution from these specific fire approaches using multiple methods: 1) brute force emissions zero-out of the fire event emissions, 2) source attribution of all fires, 3) and source attribution of the fire event estimated by difference between baseline source apportionment and subsequent source apportionment simulation where the fire even emissions are not included. The contribution using brute-force zero out and source apportionment are compared for episode peak primary and secondary pollutant impacts. Most of the PM2.5 impacts from these fire events are primary PM2.5 in this model assessment. The largest downwind component of fire PM2.5 is organic carbon mass, which is the largest component of the primary PM2.5 emissions based on the source profile used here. It is likely that the relative contribution from different forms of PM2.5 may change if PM2.5 organic aerosol emissions were treated as semi-volatile. Even though a relatively small amount of total VOC emissions from these fires is allocated to SOA precursors, a notable amount of PM2.5 SOA is formed downwind from these fires using the 2-product scheme employed here (Carlton et al., 2010). The range of O3:CO ratios modeled for these fires is generally consistent with those made in observation based assessments.

Carlton, A.G., Bhave, P.V., Napelenok, S.L., Edney, E.O., Sarwar, G., Pinder, R.W., Pouliot, G.A., Houyoux, M., 2010. Treatment of secondary organic aerosol in CMAQv4.7. Environmental Science and Technology 44, 8553-8560.

Kwok, R., Baker, K., Napelenok, S., Tonnesen, G., 2015. Photochemical grid model implementation of VOC, NO x, and O 3 source apportionment. Geoscientific Model Development 8, 99-114.

Kwok, R., Napelenok, S., Baker, K., 2013. Implementation and evaluation of PM2.5 source contribution analysis in a photochemical model. Atmospheric Environment 80, 398-407.

Patrick Hawbecker, Kiran Alapaty, Jimy Dudhia, Wei Wang, and Christopher Nolte

The microphysical impact of aerosols on cumulus convection and precipitation has been well documented for over half a century (Gunn and Phillips, 1957; HE Landsberg, 1970; Rosenfeld 1999; Rosenfeld et al. 2008). It has been shown that microphysical processes such as the freezing of cloud water lofted above the freezing levels can significantly increase vertical updrafts, resulting in cloud invigoration. In order for weather and climate models to simulate these processes, a suitable parameterization must be developed because the spatial and temporal resolutions are far too coarse to directly resolve these processes. However, most global and regional climate models simply parameterize convection by adjusting the vertical profiles for temperature and humidity to account for these processes, leading to overestimations of convective precipitation and underestimation of stratiform precipitation (e.g., Song and Zhang, 2011). These studies showed that incorporating convective microphysical processes directly in a cumulus convection scheme could significantly reduce precipitation errors in climate simulations.

In this study, the multi-scale Kain-Fritsch (MSKF) cumulus scheme (Alapaty et al., 2012; Zheng et al., 2015) is made microphysics-aware for the Weather Research and Forecasting (WRF) model to analyze the impact of such a scheme, along with global aerosol data, on convective precipitation. The implementation of a microphysics-aware convection scheme will allow for a more robust characterization of subgrid scale cloud process via interactions with aerosols as well as contributions to convective updrafts and downdrafts. We have introduced climatological global aerosol data (for 10 species) to the recently developed microphysics-aware MSKF scheme (MSKF-MP) where climatological aerosol concentrations were obtained from climate simulations by the Community Earth System Model (CESM) for the present as well as future climates. Here, we present results from weather and seasonal simulations for the eastern United States obtained from using the new MSKF-MP scheme compared against those obtained from using the current MSKF scheme and observational data.

S. Marcela Loria^1,2, Heather A. Holmes ¹, W. Patrick Arnott¹, and James C. Barnard¹

¹ Atmospheric Sciences Program, Department of Physics, University of Nevada, Reno, Reno, Nevada, USA

²Division of Atmospheric Sciences, Desert Research Institute, Reno, Nevada, USA

Health effects studies of aerosol pollution typically use surface particulate matter (PM_2.5) concentrations from ground-based monitoring stations. The reliability of these studies is affected by the lack of data available due to the limited number of monitoring stations. This issue can be improved using aerosol pollution observations from satellite retrievals. Satellite remote sensing provides the synoptic picture of air quality, including information about isolated events by measuring columnar aerosol optical depth (AOD). Ground level PM_2.5 concentrations have been successfully retrieved using AOD from the MODIS instruments onboard Aqua and Terra satellites in the dark, vegetated eastern U.S. However, the semi-arid western U.S. continues to be an unproven and infrequently explored area for remotely sensed atmospheric aerosol pollution retrievals. The study of aerosol transport and optical properties in this area is a challenge due to the complex terrain, bright surfaces, presence of anthropogenic and biogenic emissions, secondary organic aerosol formation, smoke from wildfires, and low aerosol concentrations during non-fire conditions.

The aim of the present work is to evaluate the uncertainty of MODIS AOD retrievals (Collections 5.1 and 6) through a comparison with columnar AOD and surface PM_2.5 measurements from AERONET and EPA networks, respectively. Data will be presented from multiple stations in California and Nevada from 2012-2014. This time period includes four major wildfires in the northern California and Nevada region; the Chips fire in 2012, the American and Yosemite Rim fires in 2013, as well as the King fire in 2014. Previous studies have shown that MODIS retrievals failed to estimate column-integrated aerosol pollution levels and particle size over Nevada and California in the summer months of 2012 and 2013 due to high surface albedo, heterogeneous vertical profile of aerosol concentrations, and incorrect parameterizations for surface reflectance. MODIS algorithms overestimated AOD by more than a factor of 3 in Reno, NV and more than a factor of 2 over Fresno, CA.

The presence of wildfire smoke impacts the correlation between surface and satellite measurements of AOD in this region. AOD from AERONET and MODIS showed low correlation for days with typical ambient pollutant concentrations (e.g. R²~0.46 for Terra-MODIS over Las Vegas, NV in August 2013). During fire periods the correlation between instruments showed an improvement (e.g. R²~0.87 for Terra-MODIS over Reno, NV in August 2012 and R²~0.82 for Aqua-MODIS over Fresno, CA in August 2013) because of the increase in columnar aerosol concentration. However, columnar AOD and surface PM_2.5 observations did not correlate over Fresno (R²~0) for this same time period. The transport physics of wildfire smoke plumes complicate the MODIS retrievals because the smoke plumes can travel at ground level or aloft with limited downward mixing to the surface. This transport is investigated using the planetary boundary layer height and the apparent optical height to characterize the extent of vertical mixing of the aerosol. Additionally, albedo products from MODIS will be evaluated to diagnose the reliability of the MODIS dark-target algorithm in the western U.S.

Carey Jang, Ph.D.,

Office of Air Quality Planning and Standards, USEPA,

MC: C439-01, 109 T.W. Alexander Drive, RTP, NC 27711,

Tel: 919-541-5638; Fax: 919-541-0044, E-mail: jang.carey@epa.gov

A series of joint efforts in the development of next-generation air quality decision support system, or "Air Benefit and Cost and Attainment Assessment System" (ABaCAS), by a team of U.S. and Chinese scientists have been initiated since 2012. These successful collaborative efforts have been undertaken to build a solid foundation toward the development this integrated air quality assessment framework. The objective of the ABaCAS system is to provide scientists and policy makers with a user-friendly system framework for conducting integrated assessments of air pollution emissions control cost and their associated health and economic benefits and air quality attainment. The "ABaCAS" system includes five key components: (1) Streamlined edition for integrated cost/benefit and attainment assessment for policy analysis (ABaCAS-SE); (2) Control cost estimate and analysis tool (CoST-CE); (3) Real-time air quality response to emissions control tool (RSM-VAT/CMAQ); (4) Air quality attainment assessment tool (SMAT-CE); (5) Health and economic benefit tool (BenMAP-CE). Initial efforts have been undertaken to apply the ABaCAS system over the Yangtze River Delta (YRD) in China to conduct assessment of emissions control strategies and their air quality and health benefits and air quality attainment, and the results of YRD pilot case studies (2017and 2030) will be presented.

C. R. Lonsdale, M. J. Alvarado, R. J. Yokelson, K. Travis, J. C. Lin, D. R. Blake, D. W. T. Griffith, T. J. Johnson, S. Kreidenweis, T. Lee, A. May, G. R. McMeeking, S. Meinardi, J. Reardon, I. Simpson, A. Sullivan, S. P. Urbanski, D. R. Weise

Biomass burning is a major source of atmospheric trace gases and particles that impact air quality. The complex photochemistry within a smoke plume can cause large changes in the concentration, size distribution, composition, and optical properties of fine particles (PM_2.5), as well as significant formation of ozone (O₃) and organic nitrate species like peroxyacetyl nitrate (PAN). The Aerosol Simulation Program (ASP) is designed to simulate this chemical evolution of biomass burning plumes under a wide variety of conditions. Here we will present ASP simulations of South Carolina prescribed fires in October and November of 2011. This data set contains more detailed measurements of the non-methane organic compounds (NMOCs) in the smoke than the data sets previously used to develop and test ASP, allowing for a more detailed evaluation of the model's gas- and particle-phase chemistry. Additionally, we will discuss our work using the ASP model to build a sub-grid scale parameterization of the near-source chemistry of biomass burning plumes for use in regional and global air quality models. We will show examples of this model implemented in GEOS-Chem, and show how the parameterization alters estimates of the air quality impacts of biomass burning. Finally we will present preliminary work on the implementation of ASP into a stochastic Lagrangian air quality model, STILT-Chem.

Rick Saylor¹, Barry Baker¹, Pius Lee², Li Pan², and Youhua Tang²

¹NOAA Air Resources Laboratory, Atmospheric Turbulence and Diffusion Division, Oak Ridge, TN 37830

²NOAA Air Resources Laboratory, College Park, MD 20740

Dry deposition is a significant loss process for aerosol particles in three-dimensional air quality (3-D AQ) models. Process analysis results indicate that dry deposition is the single largest process controlling particle concentrations near the surface on days not impacted by wet deposition. Particle deposition parameterizations currently used in most 3-D AQ models have their roots in measurements and theoretical analyses made more than 30 years ago and have not changed substantially since, even though evidence has accumulated over time that the model parameterizations do not adequately mirror reality as reflected in particle flux measurements. In particular, model predictions and measurements have greatest divergence in the accumulation size range (diameter of 0.1-1.0 mm) for densely vegetated canopies, such as forests. In this work, several dry deposition parameterizations are implemented in a sectional version of the CAMx modeling system (Comprehensive Air quality Model with eXtensions) and the modal-based Community Multiscale Air Quality (CMAQ) model. Impacts of the dry deposition parameterizations on surface PM_2.5 concentrations are examined and compared between the two 3-D AQ modeling systems for a domain over the heavily forested Southeast U. S. Data from the 2013 Southeast Atmosphere Study and existing network measurements over the region are used to evaluate model results.

Hemming, et al.

The wildfire phenomenon is complex, with variable temporal and spatial signatures and variable pollutant emissions. Emissions from wildfire represent a potential confounding factor in efforts to understand the health and environmental impacts of emissions from industrial and other human activities, as well as in our ability to determine the effectiveness of air quality management plans. Warming climate adds another variable, particularly regarding the extent and possible frequency of wildfire events in the Western states.

The periodic review of the NAAQS, particularly for O3 and PM, and the assessment and planning required to ensure compliance with the standards, may need to account for the effects of wildfire emissions on ambient pollutant concentrations. As fire seasons grow longer, persistent background concentrations of wildfire-derived pollutants may influence the outcome of the air pollutant epidemiology studies needed for setting ambient standards. An accounting of the intermittent peaks due to wildfire is critical in the assessment of ambient air quality concentration patterns for urban and regional attainment planning. The effects of fire on air quality are tied to fire emissions characteristics, in terms of mass, chemical composition, particle sizes, temporal and spatial distribution, as well as plume rise and dispersion.

The basic science of wildfire emissions that is needed to inform air quality rulemaking at present, and as the global climate continues to warm, is incomplete. Furthermore, the analyses needed to quantify the contribution of smoke to ambient pollutant concentrations under present and changing conditions call for a wider array of data, models and analytical tools than are presently available. The intent of this presentation is to outline critical science needed to address key uncertainties in fire science that may influence the outcome of standards and policy analyses, now and in the future.

Disclaimer: The views expressed are those of the author and do not necessarily reflect the views or policies of the U.S. Environmental Protection Agency.

Amber J Soja a,b,*, Hyun-Deok Choi a, Mark Vaughan b, Thomas Duncan Fairlie b, David J Westberg c, David Winker b, Charles Trepte b, George Pouliotd, James J Szykmanb,d

a National Institute of Aerospace, Hampton, VA, United States.
amber.j.soja@nasa.gov; hyundeok.choi@nasa.gov
b NASA Langley Research Center(LaRC), Hampton, VA, United States. mark.a.vaughan@nasa.gov; t.d.fairlie@nasa.gov; charles.r.trepte@nasa.gov; david.m.winker@nasa.gov;
c Science Systems and Applications Incorporated, Hampton, VA, United States. david.j.westberg@nasa.gov
d Environmental Protection Agency, Research Triangle Park, NC, United States.
Pouliot.George@epamail.epa.gov; james.j.szykman@nasa.gov

There is a significant connection between biomass burning (BB) emissions, the terrestrial environment and the atmosphere, which has strong implications for feedbacks to the climate system and Air Quality. BB has the potential to alter numerous land and atmospheric processes that, in turn, feedback to and interact with the climate system (e.g., black carbon on spring Arctic ice; land-vegetation cover change alters albedo).

Specifically, the heights to which BB smoke is injected governs short- or long-range transport, which influences surface pollution, cloud interaction (altered albedo), and modifies patterns of precipitation (cloud condensation nuclei). We are working with the CALIOP science team and other Applied Science partners, primarily the Environmental Protection Agency and regional partners, to generate BB plume injection heights using multiple platforms, sensors and models (CALIOP, MODIS, NOAA HMS, Langley Trajectory Model) that will provide value to national and international scientific and air quality communities, the CALIPSO science and algorithm teams, and to public land, fire, and air quality management and regulations communities.

Specifically, we have: (1) developed a methodology that links BB injection height and CALIOP air parcels to specific fires, the variables that control these dynamics, which include ecosystems, fire-specific and meteorological variables; and (2) defined the daily evolution of smoke plumes for specific fires. Statistics that link fire behavior and weather to plume rise are crucial for verifying and enhancing plume rise parameterization in local-, regional- and global-scale models used for air quality, chemical transport and climate.

Nancy HF French, Michigan Tech Research Institute, Michigan Technological University, 3600 Green Court, Suite 100, Ann Arbor, MI 48105, USA. nhfrench@mtu.edu

Michael Billmire, Michigan Tech Research Institute, Michigan Technological University, 3600 Green Court, Suite 100, Ann Arbor, MI 48105, USA.

Michele Ginsberg, Previously with Epidemiology & Immunization Services Branch, San Diego County Health & Human Services Agency (Retired), 3851 Rosecrans Street, San Diego, CA 92110, USA.

Sumi Hoshiko, Environmental Health Investigations Branch, California Department of Public Health, Richmond, CA 94804, USA.

Justine Hutchinson, Environmental Health Investigations Branch, California Department of Public Health, Richmond, CA 94804, USA.

Jeffrey Johnson, Epidemiology & Immunization Services Branch, San Diego County Health & Human Services Agency, 3851 Rosecrans Street, San Diego, CA 92110, USA.

Benjamin Koziol, Michigan Tech Research Institute, Michigan Technological University, 3600 Green Court, Suite 100, Ann Arbor, MI 48105, USA. Currently with NESII/CIRES/NOAA Earth System Research Laboratory, 325 Broadway, Boulder, CO 80305, USA.

Vijay Limaye, Environmental Health Investigations Branch, California Department of Public Health, Richmond, CA 94804

Tatiana Loboda, Department of Geographical Sciences, University of Maryland, 2181 LeFrak Hall, College Park, MD 20742, USA.

R Chris Owen, Michigan Tech Research Institute, Michigan Technological University, 3600 Green Court, Suite 100, Ann Arbor, MI 48105, USA. Currently with US Environmental Protection Agency, Office of Air Quality Planning and Standards, Research Triangle Park, Durham, NC 27711, USA.

Brian Thelen, Michigan Tech Research Institute, Michigan Technological University, 3600 Green Court, Suite 100, Ann Arbor, MI 48105, USA.

Shiliang Wu, Department of Geological and Mining Engineering and Sciences and Department of Civil and Environmental Engineering, Michigan Technological University, Houghton, Michigan 49931, USA.

Particulate matter emissions from wildland fire smoke have been linked to a variety of acute human respiratory and cardiovascular health effects. In a project sponsored by the National Institute of Environmental Health Sciences Interagency Working Group on Climate Change and Health Initiative, physically based models of wildland fire emissions and atmospheric transport were linked to population health outcomes using syndromic surveillance data obtained during the 2007 San Diego County wildfires. The goal was to better forecast and prepare for air quality events caused by wildland fire under current and projected future climate conditions. The poster presentation includes a review of how smoke exposure maps were developed. Also included is a review of modeling methods used to connect emissions of particulate matter from wildland fires to respiratory health outcomes. The statistical model connecting particulate matter concentrations from wildland fire to syndromic surveillance data on respiratory health is based on a generalized additive modeling (GAM) approach, with the addition of environmental variables and lagged effects. The project allowed San Diego County to improve syndromic surveillance capacity and infrastructure through the addition of new hospitals to the local syndromic surveillance system. The project included development of a model to forecast future fires based on regional climate model predictions, which shows wildland fire risk for the next 30 years to be similar to the present; San Diego County will experience approximately two extreme fire seasons each decade to 2040. On-going work is underway to link the exposure maps to more comprehensive health data that will address respiratory, cardiovascular, and other health outcomes from California Department of Health Care Services (DHCS) Medi-Cal (California Statewide Medicaid) claims files from San Diego County for the study period of August 1, 2007 through December 31, 2007. The research team is hoping to expand this approach to other time periods and fire-affected regions to better understand how wildland fire events relate to health outcomes and to improve region-specific preparedness plans.

Ho-Chun Huang^1,2, Susan O'Neill³, Barbara Stunder⁴, Perry Shafran^1,2, Jeff McQueen², Mark Ruminski ⁵, Shobha Kondragunta⁵, Jerry Gorline⁶, Jianping Huang^1,2, Ariel Stein⁴, Robert Solomon^3,*, Ivanka Stajner⁷, Sikchya Upadhayay^7,8, Narasimhan Larkin³

1 I.M. Systems Group, Inc.

2 NOAA NWS/National Centers for Environmental Prediction

3 US Forest Service, PNW Research Station

4 NOAA Air Resources Laboratory (ARL)

5 NOAA National Environmental Satellite, Data, and Information Service (NESDIS)

6 NOAA Meteorological Development Laboratory (MDL)

7 NOAA NWS/Office of Science and Technology Integration

8 Syneren Technologies

The particulates generated from forest fires often severely impact the air quality and human health in the nearby and downstream areas. The National Weather Service Operational HYSPLIT smoke modeling system (NWS/HYSPLIT smoke) uses a Lagrangian Dispersion Model (HYSPLIT) to forecast the smoke concentration resulting from fire emissions. The NWS/HYSPLIT smoke fire emissions are generated using the US Forest Service BlueSky Smoke Modeling Framework (BlueSky) with the input fire locations based on the NOAA NESDIS's Hazard Mapping System (HMS) fire and smoke detection system. Experienced analysts inspect satellite imagery (two GOES, five NOAA-AVHYY and NASA EOS Aqua and Terra), identify the location, size and duration of smoke emissions and provide this information for the model. The National Air Quality Forecasting Capability (NAQFC) uses CMAQ to provide air quality guidance available to state and local air quality forecasters for their daily operation. The processed fire emissions created for NWS/HYSPLIT smoke are currently being tested by NAQFC to provide fire emissions for CMAQ predictions. NWS/HYSPLIT smoke is being updated to use a newer version (v3.5.1) of USFS BlueSky. The new BlueSky incorporates the Fuel Characteristic Classification System version 2 (FCCS2) over the continental US and Alaska, which includes a more detailed description of the fuel loadings with additional plant type categories. It also utilizes an improved fuel consumption model and fire emission production system. For the period of July-August 2014 and April-May 2015, NWS/HYSPLIT smoke simulations show that the fire smoke emissions of new BlueSky are stronger than that of current BlueSky in NWS/HYSPLIT smoke in the Northwest US. For the same comparisons, weaker fire smoke emissions of new BlueSky were observed in the middle to the eastern part of the US. A statistical evaluation of NWS/HYSPLIT smoke predicted total column concentration (0-5000m) compared to NOAA NEDSIS GOES EAST Aerosol/Smoke Product (GASP) retrievals is underway. Preliminary results show that using the newer version of BlueSky emissions leads to improved performance of NWS/HYSPLIT-smoke for the Spring 2015. The impact study of NWS/HYSPLIT-smoke using new BlueSky on NAQFC CMAQ PM prediction is in progress.

Alexandra Larsen, Brian Reich, Mark Ruminski, Ana Rappold

We examine the impact of wildfires and agricultural/prescribed burning on regional air pollution and Air Quality Index (AQI) between 2006 and 2013. We define daily regional air pollution using monitoring sites for ozone, PM_2.5 collected by Federal Reference Method, and constituents of PM_2.5 (fine particulate matter) from the Interagency Monitoring of PROtected Visual Environment (IMPROVE) network and use satellite image analysis from the NOAA Hazard Mapping System (HMS) to determine days on which visible smoke plumes are detected in the vertical column of the monitoring site. To examine the impact of smoke on regional air pollution we use a two stage approach, accounting for within site (1^st stage) and between site (2^nd stage) variations. At the first stage we estimate a monitor-specific plume day effect describing the relative change in pollutant concentrations on the days impacted by smoke plume while accounting for confounding effects of season and temperature. At the second stage we combine monitor-specific plume day effects with a Bayesian hierarchical model and estimate a pooled nationally-averaged effect. HMS visible smoke plumes were detected on 6% of ozone (number of stations: n=1595), 8% of PM_2.5 (n=1058) and 6% of IMPROVE (n=264) network monitoring days. Our preliminary results indicate that the long range transport of air pollutants from wildfires and prescribed burns increase ozone concentration by 11% and PM_2.5 mass by 34%. On all of the days where monitoring sites were AQI code Green for ozone, 6% of those days experienced smoke plume cover. For ozone, HMS-defined plume days accounted for 18% of AQI code Yellow days ('moderate'), 25% of Orange days ("unhealthy for sensitive groups"), 29% of code Red days ("unhealthy") and 28% of code Purple days ("very unhealthy"). Similarly, for PM_2.5, HMS-defined plume days accounted for 4% of code Green days, 11% on code Yellow days, 18% on code Orange days, 17% on code Red days and 50% on code Purple days. Our preliminary results suggest that smoke from wildfires and prescribed burns has a substantial effect on regional air quality and accounts for a large percentage of days with unhealthy AQI levels.

Aditya A. Pophale¹, M. Talat Odman¹, Yongtao Hu¹, Michael E. Chang² and Armistead G. Russell¹

¹School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA, 30332

²Brook Byers Institute of Sustainable Systems, Georgia Institute of Technology, Atlanta, GA, 30332

Wild fires are expected to increase as a result of the changing climate; this will magnify pollutant emissions and degrade air quality. One potential solution is to increase the use of prescribed burning to prevent wildfires, but this can also increase emissions and negatively impact air quality. Dynamic forest and air quality management is a paradigm that considers the burning needs together with the consequences of burn emissions and maximizes the burn capacity while minimizing the impacts on air quality. Through the permitting system already in place in several states, burns can be restricted on poor air quality days and encouraged when meteorological conditions are more favorable.

With dynamic air quality management in mind, we have upgraded the HiRes air quality forecasting system, operating since 2006 in Southeastern USA. In addition to updating the models to the most recent versions, WRF3.6 and CMAQ5.0.2, we added the capability of forecasting the impacts of emission sources to the system. The sensitivities to power plant, traffic, and prescribed burn emissions are calculated by the DDM-3D feature of CMAQ and provided along with the air quality forecast for dynamic management purposes. The forecasting of prescribed burn impacts includes forecasting of burns to determine the baseline emissions required by the DDM-3D method.

In this presentation, the newly developed, weather- and need-based prescribed burn forecasting capability and the burn emission prediction component of the upgraded system, HiRes-2, will be described in detail. Results from the initial application to the 2015 burning season will be presented. The ongoing integration of the source-impact forecasts with the prescribed burning operation in Georgia, USA will be discussed. Ecological and air quality goals will be reviewed and dynamic management opportunities will be illustrated.

Uma Shankar¹, Jeffrey P. Prestemon², Don McKenzie³, E. Natasha Stavros⁴, Kevin Talgo¹, Aijun Xiu¹, Bok Haeng Baek¹, Mohammad Omary¹, Dongmei Yang¹

¹ The University of North Carolina-Institute for the Environment

² U.S. Forest Service Southern Research Station

³ U.S. Forest Service Pacific Northwest Research Station

⁴ NASA Jet Propulsion Laboratory

The U.S. Forest Service (USFS) and other federal agencies are federally mandated under the provisions the Resources Planning Act of 1974 to generate assessment reports on a ten-year cycle, describing projected conditions of forests and rangeland in the next 50 years. To support the current RPA assessment, the USFS Southern Research Station analyzed fire weather parameters downscaled from a suite of nine climate model projections from 2010-2060, along with projections of other factors known to influence fire activity, towards better understanding the climate drivers of disturbance in these renewable resources in the Southeastern U.S. These analyses were used to develop statistical models of county-level annual areas burned (AAB) over the Southeast, which take into account not only the meteorological drivers, but also the biophysical, socioeconomic, and land-use variables that have explained historical spatiotemporal variations of wildfire. To facilitate air quality impact assessments, the projected AAB were gridded over a Southeastern U.S. modeling domain at a spatial resolution of 12-km x 12-km at 5-year intervals from 2015-2060. They were used in a stochastic wildfire ignition model in five selected years, current and future, to constrain estimates of daily burned areas, which were in turn input to the BlueSky fire emissions model to create a point wildfire emissions inventory for the Southeast in each selected year. After progressive improvements in the fire inventory development, the impacts of the wildfire emissions on ambient concentrations of ozone and particulate species were examined using the Community Multiscale Air Quality model. Results show that differences from current National Emissions Inventory estimates in the projected emissions of criteria pollutants and precursors by mid-century are due to the combined influence of climate and socioeconomic variables on wildfires in the Southeast.

Rizzieri Pedruzzi¹, Taciana T. de A. Albuquerque^1,2, Barron Henderson³, Nikolle Aravanis³, Igor Baptista Araujo¹, Erick Sperandio¹, Neyval C. Reis jr. ¹, Davidson M. Moreira¹.

¹Dept. of Environmental Engineering - Federal University of Esperito Santo - UFES - BRAZIL

²Dept. of Sanitary and Environmental Engineering, Federal University of Minas Gerais - Brazil

³Dept. of Environmental Engineering Sciences - University of Florida - USA

rizzieri.pedruzzi@gmail.com, taciana@desa.ufmg.br .

The Region of Great Vitoria (RGV) - Brazil, although is an area with unique features and their own emissions, suffer with pollutants coming from others urban centers like Sao Paulo, Rio de Janeiro and Belo Horizonte, and emissions from facilities that are located on north and south of the RGV like mineral industries, metallurgical industries, cellulose facilities, thermoelectric, ports, etc., and additionally, there are intercontinental transport of pollutants. Besides that, the goal of this study is understand the influence of boundary conditions over RGV on response of the Community Multiscale Air Quality (CMAQ) model, with three different strategies to supply boundary conditions for the model. Alternative methods consist in providing BC from (1)
xed, time-independent, concentration pro
les, (2) concentrations predicted in a CMAQ mother domain (1km, 1h resolution) and (3) concentration values from the GEOS-Chem chemical-transport global model (2_2.5 , 3h resolution). High resolution (1 km) simulated concentrations of the main pollutants (NO, O₃, PM₁₀ and PM) will be compared through a comprehensive statistical analysis including observational data from 8 monitoring stations all over the RGV. Meteorological fields will be modelled using the Weather Research and Forecasting model WRFv3.6.1 for the 10-day period (from august 2th to august 11th), using Final Operational Global Analysis (FNL) data as initial and boundary condition with 1 x 1 horizontal resolution. It was used four nested domains with 27-km grid resolution, 9-km, 3-km and the inner domain with 120 x 120 km with 1-km grid resolution, centered on 20.251 S, 40.285 , and only the 1-km domain was aligned with SMOKE and CMAQ modeling simulations, which cover the entire Region of Great Vitoria in Brazil's Southeast. The SMOKEv3.6 emissions model was applied to build spatially and temporally resolved emissions using as input data the local emission inventory released in 2010 by the Esperito Santo State Environmental Protection Agency (Instituto Estadual de Meio Ambiente e Recursos H dricosm-IEMA). The CMAQ and SMOKE domains consist of 61 x 79 grid cells with 1 km horizontal spacing and 20 vertical layers. Aerosol processes and aqueous chemistry in CMAQv.5.0.1 (AERO6) were used, as well as the Carbon Bond 05 gas phase mechanism.

Igor Baptista de Araujo¹, Taciana T. de A. Albuquerque^1,2, Rizzieri Pedruzzi¹, Erick Sperandio¹, Neyval C. Reis jr. ¹, Davidson M. Moreira¹.

¹Dept. of Environmental Engineering - Federal University of Esperito Santo - UFES - BRAZIL

²Dept. of Sanitary and Environmental Engineering, Federal University of Minas Gerais - Brazil

rogibap@gmail.com, taciana@desa.ufmg.br .

The air quality of a region is the result of complex interactions involving the emission of air pollutants from stationary and mobile sources, local and remote, natural and anthropogenic, which together with the weather and topography of the region determine concentration of pollutants. Thus, it is crucial understand the emission inventory of pollutants aiming effective management of air quality in a region. The present work conducted the spatial and temporal modeling of the emissions of traffic routes in the metropolitan region of Vitoria, through the SMOKE emissions model, using data from the 2010 emission inventory, using the new emission factors and composition of the current vehicle fleet, proposed by the Ministry of Environment of Brazil, which had sources georeferenced and temporally allocated, emissions chemically speciated and subsequently included in the SMOKE model. The result showed the emission scenarios the main areas of emission of gases and particles, corroborating with that described in tabulated inventory and demonstrating the viability of the model as part of a powerful global tool for environmental management, and future possibility of using the EPA's model Motor Vehicle Emission Simulator (MOVES) due to national data available for model customization for the Brazilian reality, for improvement of emission inventory.

Alexis Zubrow, Alison Eyth, Brian Timin, Norm Possiel, Kirk Baker, Heather Simon, Sharon Phillips, Chris Misenis, Jim Kelly, and Pat Dolwick

EPA has updated its temporal allocation of on-road mobile source emissions by time of day, using measured traffic count data, and tested the effects of these emissions chages in the CAMx model for a 2011 basecase. The expectation was that improved temporal allocation of on-road mobile emissions may yield improvements to base case model performance. In particular, it was anticipated that the revisions may help resolve a previously identified issue with the 2011 CAMx modeling where NOx concentrations were most strongly overestimated in the morning and evening prime commuting periods. This in turn, led to diurnal patterns in ozone concentrations, especially at urban sites. Initial results suggest that the reallocation of emissions had only a marginal impacts on diurnal patterns of model bias.

James Keehn and Donald Hagen

Emissions from the commercial aviation sector are coming under increasing scrutiny despite advances in gas turbine engine efficiency and design. In order to assist with further reducing the effects of particulate matter (PM) emissions from these engines on local air quality, the Center of Excellence for Aerospace Particulate Matter Reduction Research at the Missouri University of Science and Technology has been conducting experimental studies to measure these emissions. These studies include experiments focused on the use of alternative aviation fuels sourced from renewable and non-renewable feedstocks. As these studies show reductions in PM emissions in terms of source activity levels, there is a need to model the resulting PM concentrations. This is especially important near airports where employees and nearby residents may be exposed to high levels of gas turbine engine emissions.

For this study, the Emissions and Dispersion Modeling System (EDMS) was chosen to generate an emissions inventory for the Hartsfield-Jackson Atlanta International Airport for a one-month period. EDMS is designed to pass source activity, weather data, and emissions data to the AMS/EPA Regulatory Model (AERMOD) in order to perform dispersion modeling of the emissions from airport-related sources. Dispersion modeling is ideal because its use is most appropriate when simulating levels of stable pollutants. Primary PM is the focus of the aforementioned measurement studies and is relatively stable over the short spatial and temporal scales which dispersion models excel at simulating. Because of the increased spatial resolution available, one result of interest is the dependence of PM concentrations on the movement of aircraft through the airport. It is common to generate emissions inventories for airports using a simplified representation of the gates and nearby taxiways when feeding the data into lower-resolution chemical transport models. This project includes scenarios based on both airport terminals and individual gates in order to examine whether PM concentrations are impacted by using a more accurate airport layout.

Measurement of PM emissions from aviation gas turbine engines most often takes place at the engine exit plane. Although the difficulty of measuring at the engine exit plane is increased, it limits the measured PM to its nonvolatile component. This simplifies the theoretical picture and allows for greater comparability between studies as certain influences such as the ambient weather conditions can be ruled out. However, the volatile PM components which are not included in such measurements are important when considering the air quality impacts of jet engine PM. This study attempts to account for total PM emissions in two different ways, in order to examine the impact of methodology for emissions inventory generation on the dispersion modeling results. The first method uses engine exit plane emissions of nonvolatile PM and then estimates the volatile components according to the First Order Approximation. The second method makes use of total PM measurements taken 145m from the engine exit plane during the Alternative Aviation Fuels EXperiment (AAFEX) II campaign. This data can be applied directly to the emissions inventory with fewer assumptions or approximations. The difference in modeling results using these two methods is a focus of the current project.

Chris Kite, Texas Commission on Environmental Quality

The SPECIATE Database maintained by the Environmental Protection Agency (EPA) contains multiple profiles for the volatile organic compounds (VOC) emitted through exhaust and evaporation from on-road vehicles. For the purposes of speciating on-road VOC for ozone modeling purposes, selecting the best available profiles can be a challenge for emissions processing staff that are not experienced with advanced photochemistry. After the initial release of the Motor Vehicle Emission Simulator (MOVES) model from EPA, the Texas Commission on Environmental Quality (TCEQ) updated the on-road VOC speciation to match the various MOVES emission processes of start exhaust, running exhaust, evaporative permeation, evaporative fuel vapor venting, and evaporative fuel leaks. For each emission process (e.g., running exhaust), a set of profiles was selected for review. The same on-road VOC emissions input file was separately processed with each profile into the 22 different species associated with the Carbon Bond 6 (CB6) chemical mechanism. The emissions by CB6 species were weighted according to their maximum incremental reactivity (MIR) values relative to the most reactive xylene (XYL) category. The weighted results were summed to obtain XYL-equivalent indexed totals per profile for comparison purposes. This approach enabled the TCEQ to quality assure and select newer speciation profiles of interest while confidently discontinuing the use of older profiles.

Muge Kafadar^a, Luca Pozzoli^a, Tayfun Kindap^a, Alper Unal^a

^aIstanbul technical University, Eurasia Institute of Earth Sciences, Deparment of Climate and Marine Science, Istanbul, Turkey

Istanbul which is the 19th most populated city in the world faces high air pollution levels in recent years. Among all the sources, transportation related air pollution has a critical role on human exposure as they are released at locations and levels where human activity is the highest. In a study conducted by Raaschou-Nielsen et al. on over 310,000 cohort members, it is found that an increase in road traffic of 4,000 vehicle-km per day within 100m of the residence has a Hazard Ratio of 1.09 (suggesting there is significant contribution to lung cancer).

In this study, our aim is to to accurately determine the impact of transport emissions on air quality for Istanbul via Models3 modeling system. For this purpose, The COPERT (COmputer Programme to calculate Emissions from Road Transport) model which is supported by European Environment Agency (EEA) is used to calculate emissions from the transportation sector. COPERT model uses an average speed based methodology with adjustment factors to estimate transport emissions. This paper will provide the reference case emission estimates for the city of Istanbul along with CMAQ results. The main contribution of the study is going to be on the effect of adjustment factors (such as fleet distribution of average speed on different roadways) on emissions and ambient concentrations as they are modeled with CMAQ model.

Raaschou-Nielsen, O., et al. (2013). Air pollution and lung cancer incidence in 17 European cohorts: prospective analyses from the European Study of Cohorts for Air Pollution Effects (ESCAPE). The Lancet Oncology.

Chi-tsan Wang¹, Yuzhi Chen¹, John Johansson², Johan Mellqvist², Dennis Mcnally³, William Vizuete^*1

1.Department of Environmental Science and Engineering, UNC Chapel Hill, U.S.

2.Department of Earth and Space Sciences, Chalmers University of Technology, Gothenburg, Sweden

3.Alpine Geophysics LLC

The Federal Aviation Administration's (FAA) Office of Environment and Energy (AEE) recently released the Aviation Environmental Design Tool (AEDT) Version 2b. AEDT 2b's release completes the replacement of multiple legacy tools, including the Integrated Noise Model (INM - single airport noise), Emissions and Dispersion Modeling System (EDMS - single airport emissions), Model for Assessing Global Exposure to the Noise of Transport Aircraft (MAGENTA - global noise), System for Assessing Aviation's Global Emissions (SAGE - global emissions) and AEDT 2a (regional noise). It is now the single required model for environmental compliance of airport actions, air traffic airspace and procedure actions, as well as policymaking support. AEDT is a software system that models aircraft performance in space and time to estimate fuel consumption, emissions, noise, and air quality consequences. AEDT is a comprehensive tool that provides information to stakeholders on each of these specific environmental impacts. AEDT is designed to model individual studies ranging in scope from a single flight at an airport to scenarios at the regional, national, and global levels. AEDT leverages geographic information system (GIS) and relational database technology to achieve this scalability and offers rich opportunities for exploring and presenting results. AEDT has already been used to support a number of domestic and international policymaking scenarios and will now make those capabilities available to all stakeholders.

Jeff Lundgren, David Chadder, Christian Reuten, Martin Gauthier, Carol McClellan, Golnoosh Bizhani, Ahammad Ali, Trevor Cavanaugh (RWDI) Margaret Mears (Kinder Morgan)

The Lower Fraser Valley (LFV) of British Columbia (BC) in Canada is home to two million people including the City of Vancouver and surrounding communities. The region experiences occasional episodes of degraded air quality during the summer months. Several large proposed and approved industrial projects will potentially increase emissions of NOx and VOC in the region. One such project is the proposed expansion of the Kinder Morgan Corporation (KMC) Trans Mountain Pipeline.

High resolution photochemical modelling with CMAQ was used to assess the potential cumulative effects of the KMC pipeline along with several other large projects and an estimation of future marine shipping emissions on air quality in the region. The study was conducted using WRF/SMOKE/CMAQ, configured with nested grids of 36-km, 12-km, 4-km, and 1-km resolution. Modelling was applied for the four historical meteorological episodes identified by Steyn et al. (2013) as capturing the meso-scale circulation patterns associated with high ozone concentrations in the LFV.

Baseline emissions were established from the most recent US (2011) and Canadian (2010) emissions inventories, with shipping, pipeline, and other project emissions forecasted to the year 2030. Project specific emissions were implemented within the 1-km resolution grid only. Detailed shipping emissions were also included in the 4-km domain.

Model results were investigated in terms of the resulting ground-level ozone and fine particulate matter concentrations and potential influences on visibility. The analysis also forms a potential approach to incorporate photochemical modelling into the regulatory air quality framework within the Province of BC.

References

D. G. Steyn, B. Ainslie, C. Reuten & P. L. Jackson (2013) A Retrospective Analysis of Ozone Formation in the Lower Fraser Valley, British Columbia, Canada. Part I: Dynamical Model Evaluation. Atmosphere-Ocean, Volume 51, Issue 2, 2013.

David Nunes, San Joaquin Valley Air Pollution Control District

In an effort to provide historical air quality exposure information to the residents of the San Joaquin Valley Air Pollution Control District in California, techniques were developed to estimate daily maximum 8-hr average ozone and 24-hr average PM2.5 for the past 20 years at a resolution of 4 km across the entire 22,600 square miles (58, 581 square kilometers) of the district. These daily estimates were then summarized into statistics of interest to residents and placed into a web-based tool for retrieval. Using this tool, residents can enter an address to compare their historical neighborhood air quality to county and basin summaries.

To essentially grid 20 years of air quality observations, we explored existing techniques, found them lacking and therefore develop new techniques to achieve our desired goals.

Air quality pollutants, even so called "regional" pollutants, can be highly variable in space and time. Monitoring networks are typically designed to measure air quality in highly populated areas or at assumed peak locations. Often these networks are inadequate when it comes to informing the public about air quality in their city or neighborhood. Spatial interpolation and regional models are the tools often used to estimate values between monitors. Although spatial interpolation techniques are generally quick, they may be inadequate for monitoring networks not designed to capture gradients associated with local emission, meteorology or complex terrain. Regional models do better at capturing variation between monitors and beyond the monitoring network. However, regional models are complex, computationally expensive and require bias correction to match monitor observations.

The San Joaquin Valley, is one such place where size (22,600 square miles), a generally linear deployment of monitors along the main transportation corridor (US 99), mountains, and heterogeneous emissions conspire to make spatial interpolation and regional modeling challenging. To address these issues, multiple linear regression equations were developed for unmonitored locations (neighborhoods) of the Valley using CMAQ model output as input. Using one year of CMAQ model output, equations were developed for grid cells without monitoring sites using grid cells with monitoring sites as the independent variables. R was utilized for this task using the leaps() function to evaluate all possible equations for each grid cell up with up to 7 variables. Bayesian Information Criteria (BIC) was used for model selection.

Using regression equations developed from the CMAQ model output, 8-hr average ozone observations from 1992 to 2014 and 24-hr average PM2.5 observations from 2004 to 2014 were used to estimate daily air quality at 3,800 neighborhoods (grid cells). These daily estimates are then used to create monthly and annual statistics for each neighborhood in the Valley as well as air quality maps. For years when both parameters exist, AQI was also calculated.

Although challenges remain (e.g., how to handle local impacts not observed or modeled) results to date have been encouraging. Observations at monitors excluded from the development process have been reproduced with a Pearson correlation coefficients generally greater than 0.95.

Fatema Parvez, Kristina Wagstrom

Near road emissions of different pollutants, adversely affect both human health and the environment. Vehicular emissions are one of the primary sources of air pollution in cities and lead to elevated morbidity and mortality rates in individuals living near roadways. The complex interactions between meteorology, traffic volume, road structures and regional air pollution makes assessment of human exposure to vehicular emissions difficult. In this study, we compare the temporal variation in the dispersion of near roadway particulate matter in an urban area.

We employ a steady state gaussian plume dispersion model, R-LINE, to simulate near road concentrations in Hartford, CT. R-LINE simulates the dispersion from line sources by numerically integrating the point source emissions along multiple road configurations.

We explore the seasonal and diurnal variability of vehicular emission dispersion under a variety of meteorological conditions. Variation in meteorological parameters such as wind speed, wind direction, atmospheric stability can play an important role in roadway dispersion. We estimate the monthly averaged pollutant concentration in different seasons and compare the impact of changes in meteorology and seasonal conditions on pollutant concentration. We also evaluate the diurnal variation of roadway dispersion. This approach leads to improved understanding of roadway pollution dispersion and how it varies with meteorology and different time periods.

Heather Simon, Benjamin Wells, Kirk R. Baker, Bryan Hubbell

Office of Air Quality Planning and Standards, U.S. EPA

In designing plans to reduce ozone levels, air quality planners must target reductions in NOx and VOC precursors. It is well known that ozone response to changes in NOx and VOC emissions can be nonlinear and depends on local atmospheric chemistry and meteorological conditions. In this analysis, we examine ozone changes in response to large NOx reductions in 3 urban areas: Philadelphia, Atlanta, and Chicago. We apply a method using Higher Order Decoupled Direct Method (HDDM) sensitivities from a 2007 CMAQ simulation to determine resulting spatial and temporal patterns of ozone responses. We see different types of responses in these 3 cities, with Chicago being the most NOx-saturated of the three and Atlanta being the most NOx-limited. Large shifts in spatial patterns of ozone within each of these cities are predicted. In addition, the model predicts seasonal shifts in mean ozone under scenarios of low NOx emissions. We explore the implications for population exposure and for designing future health studies.

Erick Giovani Sperandio Nascimento^a, Davidson Martins Moreira^a, Gilberto Fisch^b, Taciana Toledo de Almeida Albuquerque^a,c

^a Federal University of Esperito Santo

Av. Fernando Ferrari, s/n
29060-970 - Vitoria/ES - Brazil

Tel. (Fax): (27) 4009-2677

erick@lcad.inf.ufes.br, davidson@pq.cnpq.br

^b Institute of Aeronautics and Space

Pra a Marechal Eduardo Gomes, 50

12228-904 - Sao Jose dos Campos/SP - Brazil

Tel. (Fax): (12) 3941-5635

gilbertofischgf@iae.cta.br

^cFederal University of Minas Gerais

Department of Sanitary and Environmental Engineering, School of Engineering

Av. Ant nio Carlos, 6.627
CEP 31.270.901 - Belo Horizonte (MG) - Brazil

taciana@desa.ufmg.br

Rocket exhaust clouds are composed by hazardous pollutants, e.g. alumina, carbon monoxide and dioxide, and hydrogen chloride, which are generated during the burning of rocket engines. In the case of vehicle launching, huge and hot clouds are generated near the ground level and are composed by the buoyant exhaust products, which will rise, expand, stabilize, entrain the ambient air, and they will start to be dispersed according to the atmospheric conditions. This process takes a couple of minutes to occur, and generally human receptors located in populated areas nearby the launching center may be exposed to high levels of concentrations within a few to tens of minutes, up to less than one hour. Also, these pollutants may be carried farther due atmospheric dispersion, and chemically interact with other atmospheric compounds, forming new pollutants, reaching and impacting other populated areas located in farther distances. The launch centers around the globe, like spaceports, need to operationally assess the impact of rocket launchings events in the environment, requiring to evaluate both short and long range impacts prior to launchings through meteorological and air quality modeling. In general, however, air quality models do not account for calculating peak and average concentration for a short time scale, i.e. ranging from minutes to one hour. In addition, there is the fact that modeling rocket exhaust clouds formed due rocket/vehicle launching is quite a unique air quality problem. For this purpose, we chose to use a modern air quality model which targets this problem, named Modelo Simulador da Dispers o de Efluentes (MSDEF). For long range assessment, we chose the Community Multi-scale Air Quality (CMAQ) modeling system, since it represents the state-of-the-art in regional and chemical transport air quality modeling, and due to its capability to deal with chlorine gases - which is a considerable part of rocket exhaust clouds. In order to couple both models, the MSDEF code have been rewritten using the I/O API library, making it possible for MSDEF to generate the initial conditions as input to CMAQ model. Thus, it forms the basis for a hybrid, modern and multidisciplinary system which, in conjunction with the WRF model - which addresses for meteorological modeling - can be operationally used in different missions at the Alcantara Launching Center (the Brazilian gate to the space), as planning activities and environmental assessments, pre-and post-launching forecasts of the environmental effects of rocket operations.

Changbum CHO, Yun-Kyu LIM, Kyu Rang KIM, and Baek-Jo KIM

Alexander Cohan*1, Gary Wilson2, Greg Yarwood2
1LADCO
2Ramboll Environ

CAMx Ozone/PM Source Apportionment Technology (O/PSAT) currently utilizes a digital map of the modeling grid that defines how input emissions files will be tracked by region. Each grid cell is defined to wholly exist within a particular source region. A new approach is developed that allows CAMx to allocate fractional portions of grid cells to multiple source regions. This will improve characterization of source-receptor relations for emission sources that cover multiple tracking regions. This is an important distinction because state and county borders do not often match modeling grids. CAMx OSAT results are contrasted with and without fractional source regions using a 2011 model baseline.

Fahey, K.M., Pye, H.O.T., Luecken, D.J., and K.R. Baker

Clouds and fogs significantly impact the amount, composition, and spatial distribution of gas and particulate atmospheric species, not least of which through the chemistry that occurs in cloud droplets. Atmospheric sulfate is an important component of fine aerosol mass and in an environment where clouds or fogs are present, aqueous phase production of SO₄ can dominate over gas phase production (Seigneur and Saxena, 1988). More recent studies have suggested that the aqueous phase of clouds and wet aerosols may be an important medium for the production of secondary organic aerosol (Ervens et al., 2014; Ervens et al., 2011; McNeill, 2015).

Due in large part to computational constraints, historically, only a simple description of aqueous phase chemistry has been implemented in many chemical transport models. Aqueous phase chemistry in CMAQ, for example, is based on a simple sulfur oxidation scheme from RADM (Walcek and Taylor, 1986) with few updates to the mechanism in recent years (Carlton et al., 2008). Cloud and fog chemistry and redistribution between gas/aerosol/aqueous phases may have important impacts for a myriad of species, and as computational capabilities have expanded, so should the investigation and treatment of more complete aqueous chemistry mechanisms in regional modeling frameworks.

Here we employ the Kinetic PreProcessor (KPP) to investigate the impacts of expanding CMAQ's baseline aqueous chemical mechanism. KPP has been used previously to generate Fortran90 code to simultaneously treat kinetic mass transfer between the gas-aqueous phases, ionic dissociation/association, wet deposition, scavenging of interstitial aerosol, and chemical kinetics (Fahey et al., 2013; Baek et al., 2011). We expand the baseline mechanism to include the aqueous phase formation of SOA in cloud water from isoprene epoxydiols and MPAN, as well as examine the impacts of implementing more rigorous mechanisms for inorganic and organic aqueous phase chemistry (e.g., Pandis and Seinfeld, 1989; Fahey and Pandis, 2001). We will examine the baseline and expanded model results for summer and winter periods and include comparisons between modeled and observational data.

Jiandong Wang^1,2, Shuxiao Wang^1,2*, Jonathan E. Pleim^3*, Francis S. Binkowski⁴, David C. Wong³, Jia Xing³, Bin Zhao^1,2, Rohit Mathur³, Jiming Hao^1,2

1 School of Environment, and State Key Joint Laboratory of Environment Simulation and Pollution Control, Tsinghua University, Beijing 100084, China

2 State Environmental Protection Key Laboratory of Sources and Control of Air Pollution Complex, Beijing 100084, China

3 Atmospheric Modeling and Analysis Division, National Exposure Research Laboratory, US Environmental Protection Agency, Research Triangle Park, NC, USA

4 Gillings School of Global Public Health, University of North Carolina, Chapel Hill, NC, USA

The mixing state of black carbon (BC), which refers to the degree of BC mixed with other aerosol species, significantly impacts on the absorption and radiative forcing of BC. In traditional models, BC is simply treated as either completely externally/internally mixed, or with a fixed external/internal ratio. To date, only a few models explicitly resolved the mixing state of BC particles by simulating the relative aging process. Riemer et al. (2009) reported a stochastic particle-resolved model PartMC-MOSAIC, which resolves the composition of each particle, hence, tracks the mixing state of particles. The aerosol model MADRID-BC (Oshima et al., 2009) simulated the mixing state of particles with two-dimensional aerosol sections and is applied in WRF-CHEM model (Matsui et al., 2013). However, these two methods are computationally expensive and cannot be directly used in modal models. Therefore, we developed a new module in the two-way coupled WRF-CMAQ modeling system. Fresh BC and aged BC (with coating) were represented as separated species. Similar treatment was also applied for other particles including sulfate, nitrate, ammonia, and organic matter. The freshly-emitted BC was considered as externally mixed. It would transfer to aged BC by condensation, coagulation and photo-oxidation processes. Parameters, such as the detailed reaction rate refers to smoke chamber experiments in recent references. Core-shell Mie theory was employed to calculate the scattering and absorption cross section. Two-way WRF-CMAQ model with BC mixing state module will be used to simulate a campaign conducted in Beijing, March, 2015. Species composition of single particle from Soot Particle Aerosol Mass Spectrometer (SP-AMS) and single particle albedo (SSA) from photo-acoustic spectroscopy will be used to validate the modeling results. The improvement of model performance are expected and will be further discussed.

References

Matsui, H., Koike, M., Kondo, Y., Moteki, N., Fast, J.D., and Zaveri, R.A. (2013), Development and validation of a black carbon mixing state resolved three-dimensional model: Aging processes and radiative impact, Journal of Geophysical Research-Atmospheres, 118(5), 2304-2326, doi:10.1029/2012jd018446.

Oshima, N., Koike, M., Zhang, Y., Kondo, Y., Moteki, N., Takegawa, N., and Miyazaki, Y. (2009), Aging of black carbon in outflow from anthropogenic sources using a mixing state resolved model: Model development and evaluation, Journal of Geophysical Research-Atmospheres, 114, doi:10.1029/2008jd010680.

Riemer, N., West, M., Zaveri, R.A., and Easter, R.C. (2009), Simulating the evolution of soot mixing state with a particle-resolved aerosol model, Journal of Geophysical Research-Atmospheres, 114, doi:10.1029/2008jd011073.

Golam Sarwar, Brett Gantt, Daiwen Kang, Donna Schwede, Rohit Mathur, USEPA

Reiner Schlitzer, Alfred Wegener Institut f r Polar- und Meeresforschung

Surface layer resistance plays an important role in determining ozone deposition velocity over seawater. Recent studies suggest that surface layer resistance over sea-water is influenced by wind-speed and chemical interaction at the air-water interface. Here, we investigate the effect of chemical interaction of iodide, dimethyl sulfide (DMS) and dissolved organic carbon (DOC) in seawater with atmospheric ozone on ozone deposition by using a resistance scheme that accounts for the effects of wind-speed and chemical interaction. Four different model simulations are performed using the hemispheric Community Multiscale Air Quality model for summer months in 2006. The first simulation is completed by including the surface layer resistance due only to wind-speed. The second, third, and forth simulations are completed by including the surface layer resistance due to wind-speed and the chemical effects of iodide, DMS, and DOC in seawater. Oceanic iodide levels are estimated using the sea surface temperature, DMS levels are obtained from the oceanic climatological DMS database, and DOC levels are obtained from an ocean model based on coupled physical/biogeochemical processes. Preliminary results suggest that each chemical interaction enhances ozone deposition velocity and decreases ozone over marine environments. Compared to the median ozone deposition velocity of 0.007 cm/sec from the wind-speed-based surface layer resistance, the iodide/ozone reaction enhances the median deposition velocity over marine regions by 0.02 cm/sec, the DMS/ozone reaction by 0.002 cm/sec, and the DOC/ozone reaction by 0.02 cm/sec. Consequently the iodide/ozone reaction decreases the median ozone by 0.61 ppbv, the DMS/ozone reaction by 0.06 ppbv, and the DOC/ozone reaction by 0.68 ppbv. Thus, the effects of the iodide/ozone and DOC/ozone reactions on ozone deposition are similar while the effect of the DMS/ozone reaction is much smaller. However, the magnitudes of these impacts have substantial spatial variation. The paper contains a discussion of the spatial impacts on ozone deposition velocity and atmospheric ozone and a comparison with observed data.

Yun-Kyu Lim¹⁾, Kyu Rang Kim¹⁾, Changbum Cho¹⁾, Mijin Kim¹⁾, Ho-seong Choi¹⁾, Mae Ja Han¹⁾, Inbo Oh²⁾, Baek-Jo Kim¹⁾

¹⁾ Applied Meteorology Research Division, National Institue of Meterological Research, Korea Meteorological Administraton, Jeju-do 697-845, Korea

²⁾ Environmental Health Center, University of Ulsan College of Medicine, Ulsan 682-714, Korea

Pollen is closely related to heath issues such as allergenic rhinitis and asthma as well ad intensifying atopic syndrome. Information on current and future spatio-temporal distribution of allergenic pollen is needed to address such issues. Pollen emission is one of the most important parts in the dispersal modeling system. Areal emission factor was determined from gridded areal fraction of oak trees, which was produced by the analysis of the tree type maps (1:5000) obtained from the Korea Forest Service. Daily total pollen production was estimated by a robust multiple regresion model of weather conditions and pollen concentration. Hourly emission factor was determined from wind speed and friction factor, daily total pollen production, and hourly emission factor. Forecast data from the UM LDAPS (Unified Model Local Data Assimilation and Prediction System) was utilized as input. For the verification of the model, daily observed pollen concentration from 12 sites in Korea during the pollen season of 2014. Although the model showed a tendency of over-estimation in terms of the seasonal and daily mean concentrations, overall concentration was similar to the observation. Comparison at the hourly output showed distinctive delay of the peak hours by the model at the 'Pocheon' site. It was speculated that the constant release of hourly number of pollen in the modeling framework caused the delay.

Heather A. Holmes, Olabosipo Osibanjo, S. Marcela Lori
a-Salazar

Atmospheric Sciences Program, University of Nevada, Reno, Reno, NV, USA

Stable atmospheric boundary layers are associated with a decrease in boundary layer height and inhibited mixing, both of which lead to an increase in ground level air pollution concentrations. This phenomenon is prevalent throughout the intermountain west where the complex, mountainous terrain and meteorology are favorable to cold air setting on the valley floor causing temperature inversions, also known as cold air pools (CAPs). Diurnal CAPs form in the early evening and erode in the morning when surface heating mixes out the stable nocturnal boundary layer. During winter, persistent CAPs can form that do not erode the nocturnal boundary layer; therefore the temperature inversion is present throughout the day. These persistent CAPs can last for days or even weeks, leading to an accumulation of harmful pollutants in the atmosphere.

Atmospheric models have a difficult time capturing the strength and evolution of CAPs, typically over-predicting turbulent mixing which leads to an under-prediction in pollutant concentrations. This is because the empirical data used to develop the turbulence parameterizations in numerical weather prediction models are based on data collected from experiments in flat, idealized terrain. Updated parameterizations are required to model the atmospheric flows in regions with elevation changes to capture the complex atmospheric physics. The first step to achieve this is to collect atmospheric turbulence data in regions with complex terrain during time periods with stable atmospheric conditions. Due to the complicated nature of the flow phenomena, multiple experiments are needed to incrementally increase the complexity of the atmospheric processes being investigated (e.g., hilly versus mountainous terrain, diurnal versus persistent CAPs).

The objective of this work is to calculate surface fluxes using observational data collected during one week in September 2014 from a monitoring site in Echo, Oregon. The site is located in the Columbia Basin with hilly terrain, irrigated farmland, and over 100 wind turbines. The 10m tower was placed in a small valley depression to isolate nighttime cold air pools. This investigation will present observations of heat, momentum, and carbon dioxide fluxes from data collected at a sampling frequency of 10Hz at 4 heights. Hourly momentum, latent heat, and sensible heat fluxes from the Weather Research and Forecasting (WRF) model will be evaluated using the observations. Different land surface models (LSM) and planetary boundary layer (PBL) parameterizations will impact the WRF surface flux results. Therefore, the evaluation will include multiple WRF simulations varying the PBL and LSM models. This analysis will use observations to assess the uncertainty in surface fluxes from WRF using different PBL and LSM parameterizations.

Eunhye Kim¹, Hyuncheol Kim^2,3, Byeong-Uk Kim⁴, Junghoon Cho⁵ and Soontae Kim¹

¹Ajou University, Dept.of Environmental Engineering, Suwon, Korea

²NOAA/Air Reasources Laboratory, College Park, MD

³UMD/Cooperative Institute for Climate and Satellites, College park, MD

⁴Georgia Environmental Protection Division, Atlanta, GA

⁵National Institute of Meteorological Research, Asian Dust Reseach Division, Jeju, Korea

We investigate impacts of meteorology downscaling in particulate matter (PM) forecast over East Asia. Considering rapid economic grow and anthropogenic emissions over region, and the geographic location Korea, meteorological conditions and regional transport patterns are crucial in understanding air quality issues in East Asia. Multiple global meteorology simulations have been utilized to initiate regional meteorology simulations, and their impacts on regional meteorology simulations, and their impacts on regional air quality simulations are studied. Weather Research and Forecasting (WRF) modeling system is used to simulate regional meteorology using initializations with (1) Global Forecast System (GFS) from National Centers for Environmental Prediction (NCEP), (2) Final Global Analysis (FNL) from NCEP, (3) United Model (UM) from Korea Meteorology Association (KMA), and (4) ERA Interim Reanalysis from European Centre for Medium-Range Weather Forecasts (ECMWF). Sparse Matrix Operating Kernel Emissions (SMOKE) - Community Multi-scale Air Quality (CMAQ) modeling framework is also used to simulate PM10 concentrations, and the brute force method is used to quantify their contributions from domestic and foreign emission sources. A 2014 February episode was chosen; a seriously high surface PM level lasted more than a week under a stationary anticyclonic pressure system during Feb. 20 ~ 28, 2014. A strong stagnant wind condition is a key to understand and to reproduce the surface PM concentrations. All simulation cases well reproduced general synoptic patterns. Each simulation, however, differs in detailed wind fields - intensity and flow directions, resulting in considerable changes in hourly PM₁₀ concentration (up to 100 micrograms per cubic meters) and diurnal variations, especially in nighttime. Finally, PM forecast showed a considerable improvement by ensemble of all 4 simulations.

Robert Nedbor-Gross¹,Barron H. Henderson.¹, Jorge E. Pachon ²

¹ University of Florida, Department of Environmental Engineering Sciences

² Universidad de la Salle, School of Engineering

Meteorology plays a key role in air quality modeling, and this work evaluates our ability to produce high quality meteorological datasets for Bogota Colombia. Bogota is a megacity at high elevation with complex topography that is often associated with poor model performance (Jimenez and Dudhia, 2012). Weather Research and Forecasting (WRF) modeling in Bogota adequately simulates most variables, but shows high error for wind direction. Model performance is generally sensitive to resolution, surface characterization, and physics parameterizations (Colle et al, 2002, Case et al., 2008, Ruiz et al., 2010). Previous research suggest that, for complex terrain, altering resolution and physics parameterizations has little effect in areas of complex terrain (Santos-Alamillos et la., 2013). This study focuses on updating model surface characteristics to improve model performance in Bogota.

The base WRF simulation was run for two periods of 25 days in 2012. Each period was composed of 5, 5.75-day overlapping segments with 27km, 9km, 3km and 1km domains and 2-way nesting. The 1km x 1km domain is centered over Bogota and designed to include a locally derived emission inventory. We used 5 physics parameterizations with and without spectral grid nudging. All domains and options showed reasonable performance for temperature, humidity, and windspeed - none showed passable mean absolute gross error for wind direction.

We test model performance as a function of 4 surface characteristics by substituting higher quality data with our 2 best performing physics parameterizations. Default WRF inputs have lower quality outside the United States (e.g., GTOPO30). We substitute the GTOPO30 elevation with the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER); the USGS soil classification database with the Harmonized World Soil Database; and the USGS land use database with the Moderate-resolution Imaging Spectroradiometer. Further, we tested the model sensitivity to initial soil moisture by spinning-up soil moisture with each combination of options. Results from 160 combinations are analyzed to identify key inputs.

Simulation performance with the highest quality inputs reveals persistent errors. Preliminary results show minor improvements for various combinations of inputs. This underscores the need for continued research into modeling options for complex topography.

Jose Luis Rodr guez Castilla.¹, Luis Carlos Angulo Argote.²

Gloria M. Restrepo³, Roberto E. Rojano⁴.

Cesar University (UPC), - Investigation group GEAB^1,2.

Antioquia University, investigation group PFA ^1,2,3,4

jolrodriguezca@gmail.com¹ ; lcangulo@unicesar.edu.co²

Cesar, Colombia.

The Climate of the Colombia country is strongly influenced by two air streams having highest running speeds between 900 and 1000 hPa, these are the Jet Streams of Choc and the Caribbean Low Level Jet. Although these streams move in opposite directions, their meet cause a series of atmospheric phenomena that determine the weather of much of this country.

In the present study are implemented the climate reanalysis NCEP/NCAR data for to know the climate of the stream low level jet of Central America (Caribbean Low Level Jet - CLLJ) and the jet stream of Choco (Western Colombian Jet - WCJ) in the period 1981-2010; taking into account the climatic variables that reveal the annual evolution of both jets. This job evaluated the variation of the factors that influence in the intensity of the CLLJ and the WCJ over the Valley of Cesar River.

Actually on the River Cesar Valley six opencast coal mining projects are operating, which are impacting greatly the air quality of this region, for this reason is very important to know as the weather phenomena are influencing in the dispersion of particulate matter (PM₁₀) that is emitted for these mining projects.

Rui Zhang¹, Daniel S. Cohan¹, Andrew White², Richard McNider² and Arastoo Pour Biazar²

¹Department of Civil and Environmental Engineering, Rice University, Houston, TX

²The National Space Science Technology Center, University of Alabama in Huntsville, Huntsville, AL

Biogenic volatile organic compounds (BVOCs) are in many locations the dominant summertime source of reactive hydrocarbons, which play a critical role in regional ozone and particulate matter (PM) formation. Reducing uncertainties in BVOC emission is a high priority issue for State Implementation Plan (SIP) modeling. BVOC estimates depend on the amount of radiation reaching the canopy (i.e. photosynthetically active radiation (PAR)) and temperature. One of the major uncertainty sources of current BVOC emission model (e.g. MEGAN) is the inaccurate model cloud prediction, which can be corrected by replacing with space-borne observations. The University of Alabama in Huntsville (UAH) developed a new PAR algorithm based on satellite cloud albedo and insolation retrievals from high resolution GOES imagers centered at 0.65 m visible channel. The UAH PAR/insolation retrieval products were first evaluated against several ground radiation monitors over the US and Texas (such as SURFRAD, SCAN and TCEQ broadband radiation monitoring network) during August 2006. The satellite retrieval showed much stronger correlations than a meteorological model (WRF) with observations in terms of temporal variations as well as spatial patterns. The new UAH PAR products were then cross-referenced with the discontinued University of Maryland (UMD) satellite PAR products and showed comparable statistic performance but much finer spatial texture due to the high resolution (4km).

The impact of BVOC emission estimates using satellite cloud and radiation information on air quality modeling was further quantified by a WRF-MEGAN-CMAQ simulation platform over two nesting domains covering the continental US (12km) and Texas (4km). The test simulation period coincides with the DISCOVER-AQ Houston campaign in September 2013. Three sets of PAR inputs were designed to drive MEGAN, namely PAR from a control WRF run, PAR from a WRF run with cloud assimilation, and PAR from GOES retrievals. Using satellite PAR, the modeled isoprene and monoterpene emission rate over southern US during that simulation period were on average 10-20% less than that derived from the control WRF model. The reduction occurs because the satellite data corrects the PAR over-prediction bias of WRF, which tends to under-predict cloud cover due to its inability to resolve subgrid clouds and because it misses some thin clouds. The impact of PAR inputs on ozone predictions depends on the local NO_x/VOC ratio. Over the VOC-limited region, the satellite PAR tends to shift predictions of ground-level ozone by 5-8%.

Due to a rapid decline in NOx emissions and uncertainty of regional emissions inventories, the uncertainty of emission inventory needs to be comprehensively evaluated. To end this, we assimilate tropospheric NO2 columns observed by Ozone Monitoring Instrument (OMI) into the Community Multi-scale Air Quality Model (CMAQ) on a 4x4 km2 resolution in September 2013. In this work, a Bayesian inversion of tropospheric OMI NO2 is conducted to update four different sectors of National Emission Inventory (NEI) 2011 (i.e., area, biogenic, mobile and point sources). The adjusted NEI-2011 demonstrates a general reduction in the anthropogenic sectors (i.e., area, mobile and point), while biogenic emission increases. A new run with the adjusted NOx emissions is evaluated with independent aircraft and Texas Commission on Environmental Quality (TCEQ)'s Continuous Ambient Monitoring Stations (CAMS) observations. The CMAQ simulation with the adjusted NOx emission reduces the mean error and RMSE. The RMSE and bias between the aircraft NOx measurements and simulated ones are 2.4 and 6.0ppbv for the default NEI2011 and 1.9 and 4.1ppbv for the adjusted NEI-2011. Regarding the comparison to surface observations, the mean bias and RMSE between CAMS and simulated NOx concentrations decrease respectively 0.8 and 0.9 ppbv from the default NEI-2011 to the updated one.

Authors: Josephine Bates, Audrey Flak, Howard Chang, Heather Holmes, David Lavoue, Mitchel Klein, Matthew Strickland, Lyndsey Darrow, James Mulholland, Armistead Russell

Improving spatial resolution of air quality models is critical for accurate population exposure estimates used in advanced epidemiologic studies. This presentation will compare two data fusion methods on downscaling from coarser resolutions (e.g., CMAQ results with 12-km spatial resolution, or satellite observations with 4-10 km resolution) to a 250-meter resolution. The first method utilizes a Bayesian statistical modeling approach to downscale gridded chemical transport model simulations. The model uses ground level observations from monitoring networks, CMAQ simulations, link based mobile emissions or R-LINE dispersion model results, and meteorology predictors. CMAQ acts as a spatial and temporal predictor while the emissions and R-LINE results act as spatial predictors. The second data fusion method downscales CMAQ data using fine resolution R-LINE results. CMAQ concentration fields (12-km resolution) are first adjusted using observations from monitors, spatially smoothed using bilinear interpolation, and then linearly combined with traffic pollution results from R-LINE (250-m resolution). Daily, spatially resolved (250m) PM_2.5 concentrations for the Atlanta area will be presented from 2005 to 2007. Each method is applicable to other pollutants and other coarse spatial fields with varying grid resolutions developed from CMAQ and other methods. The model results are evaluated using data withholding and monitoring data from an independent dataset. These downscaling approaches that incorporate observations developed in this work will improve prediction performance of chemical transport models while providing finer spatial resolutions of air quality exposure fields.

Daniel Tong, Menghua Wang, Hang Lei, Li Pan, Pius Lee, Brett Gantt, Sarwar Golam, Jeff McQueen, and Ivanka Stajner

As anthropogenic emissions are decreased and tighter ozone standards are being considered, marine emissions and chemistry become increasingly important for assessing ozone background and coastal air quality. In conjunction with recent efforts of implementing marine chemistry in CMAQ, we are developing a global dataset of marine emissions using satellite observations and global meteorological prediction. Isoprene emitted from marine phytoplankton has been linked to important atmospheric chemistry and climate processes, although to what extent it perturbs the Earth's atmosphere is still under debate. Prior studies have developed emission algorithms to estimate global marine isoprene emissions, but these data were generated to support specific short-term studies. Supported by the NOAA Joint Polar Satellite System Proving Ground program, we have developed a high-resolution (4km) global dataset of isoprene emission from the oceans at monthly and hourly temporal resolutions that can be used to support both air quality and climate modeling. Using a revised emission algorithm originally proposed by Gantt et al. (2009), we have developed a marine isoprene emission algorithm based on the ocean color data retrieved from two sensors: the MODIS abroad Aqua and the VIIRS abroad Soumi-National Polar-orbiting Partnership (SNPP). The emission algorithm is driven either by NCEP's Global Forecast System (GFS) meteorology or by user-supplied meteorological fields. Results show that both MODIS and VIIRS products can capture seasonal and spatial variability of global oceanic isoprene emission, which is controlled by a myriad of biological and environmental variables including chlorophyll-a concentration, phytoplankton functional types, seawater light attenuation rate, solar radiation, wind speed, and sea surface temperature. The isoprene products have been compared against in-situ measurements collected at various open-sea and near-shore sites around the globe. Comparisons show that the retrieved emission flux is generally within the range of these measurements. The data will be made freely available to the modeling community, including CMAQ users, with appropriate data interpolation tools.

Tropospheric ozone is a known respiratory irritant, and it is important to understand the sources of ozone to address the reduction of human exposure. Recent studies have shown that policies aimed at reducing ozone exceedances of the NAAQS have been successful. However, while high ozone concentrations have decreased, annual median levels have remained unchanged. Understanding the complex relationships between many emission sources and ozone concentrations will be important as new regulations are required. In this study, the Community Multiscale Air Quality (CMAQ) model is used to estimate the sensitivity of ozone concentrations to emissions from major sources, including on-road and non-road sources, power plants, chemical and oil industries, solvent usage, and biogenic emissions. Ozone sensitivities are estimated at 4-km resolution over the southeastern U.S. using the decoupled direct method (DDM). Optimized source impact adjustment factors are generated using a novel hybrid source apportionment model that takes into account observed concentrations, as well as model and measurement uncertainty. Adjustment factors are applied to ozone sensitivities for improved estimates. Results are evaluated by comparing with observations, as well as comparing with results from similar studies. Additional benefits from this study include increasing our understanding of source-specific ozone responses and providing fine-scale response estimates for use in health studies.

Beata Czader, Daniel Cohan, Nancy Sanchez, Frank Tittel, and Robert Griffin

Natural gas has the potential to reduce greenhouse gas and air pollutant emissions by substituting for higher-emitting fossil fuels such as coal and oil. However, potential reductions could be offset by leaks of methane, the primary constituent of natural gas and a potent greenhouse gas that also contributes to background levels of ozone pollution. The rate of leaks from urban infrastructure is one of the most uncertain components in estimating overall emission rates from the natural gas life cycle, and has not been quantified previously in the greater Houston region. This has motivated our research team at Rice to undertake a measurements and modeling campaign to quantify these rates.

Modeling efforts will first develop a spatial distribution of expected methane and ethane emissions from different sources for the Houston region. Ethane is considered because it is co-emitted with methane from natural gas sources but not from landfills, wastewater treatment plants, and natural sources of methane (e.g., wetlands). Natural gas processing facilities and petrochemical plants report their emissions to the US EPA, and the routes of major transmission pipelines are available in geographic information systems. Because detailed characteristics of local natural gas distribution infrastructure are unavailable, we will use data on neighborhood age, housing density, and other features to map the expected spatial distribution of methane emissions from the local gas distribution system in the Houston area.

The Community Multiscale Air Quality (CMAQ) model will then be used to simulate the expected distribution of methane and ethane in the Houston region. CMAQ does not currently simulate variations in methane levels. Thus, we will modify CMAQ to simulate methane explicitly, expanding the model's capabilities for future studies to better characterize this important greenhouse gas and precursor of ozone pollution. Comparisons of modeled and measured levels of the gases will help identify potential discrepancies between the emission inventory and actual emission rates.

Nutthida Kitwiroon and Sean Beevers

The CMAQ-Urban is a one way coupled model of CMAQ and ADMS-Roads, developed as part of the UK research council project, Traffic Pollution and Health in London.

To address the need to account for those people who travel UK wide, the CMAQ-Urban model has been extended from a London model (Beevers et al.,2012) and now makes pollution predictons at UK national scale. The mode estimates the exposure of individuals by providing hourly concentrations of NO_x,NO₂,O₃, PM₁₀ and PM_2.5 at 20x20m grid resolution. Using CMAQ-Urban at country scale requires the model code to be parallelized and this will be discussed, as will accounting for double counting of road traffic emissions, solved using CMAQ's DDM.

The model outputs have been evaluated against the measurements from the UK national and London air quality monitoring networks, focusing on the results in major UK cities, such as, London, Manchester, Birmingham, Edinburgh and Glasgow. Th results in this study demonstrate CMAQ-Urban's capability in quantifying the dynamic exposure of populations across the UK.

Michael D. Moran, Qiong Zheng, Junhua Zhang, Radenko Pavlovic, and David Niemi

GEM-MACH, an on-line chemical transport model embedded within Environment Canada's multi-scale operational weather forecast model GEM, has served as Environment Canada's operational regional air quality forecast model since November 2009. The current forecast system is run twice daily to produce 48-hour forecasts of hourly O₃, PM_2.5, and NO₂ fields on a North American grid with 10-km horizontal grid spacing and 80 vertical levels. One known weakness of the GEM-MACH forecasts has been a tendency for episodic overpredictions of NO₂ and PM_2.5 concentrations over some Canadian cities, especially in the wintertime. Such overpredictions may occur for a number of reasons, but one obvious candidate is the overallocation to these urban areas of NO₂ and primary PM_2.5 gridded model emissions. The spatial allocation of emissions is a particular challenge when processing Canadian emissions inventories because emissions are usually reported by province as compared to the county-level reporting used in U.S. inventories.

A number of possible improvements to some of the spatial surrogate fields used to allocate Canadian provincial emissions to the GEM-MACH grid have been tested for important sectors such as on-road mobile sources and residential wood combustion (RWC). One approach was to replace standard population- and dwelling-density spatial surrogates with capped-density surrogates for some sectors based on the argument that some emissions activities such as local traffic, RWC, residential meat cooking, and lawn-mower and snow-blower use are unlikely to scale linearly with population or dwelling density in dense urban cores. In addition, new provincial RWC surrogate fields were constructed based on wood-consumption data from the Statistics Canada Households and the Environment Survey and other factors. The improvements in GEM-MACH performance that resulted from the use of new emissions files generated using these new Canadian spatial surrogates will be shown.

Shih Ying Chang^1,2, Saravanan Arunachalam¹, Marc Serre², Vlad Isakov³

¹Institute for the Environment, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA

²Department of Environmental Sciences and Engineering, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA

³Office of Research and Development, U.S. Environmental Protection Agency

Emission from onroad vehicles is a major contributor of air pollution-related premature death. Previous studies have estimated that onroad emissions in the U.S. cause 29,000 to 53,000 ozone and PM_2.5-related premature deaths. In these studies, air quality chemical transport models (CTM) were used to provide ambient concentration estimates. These grid-based models were usually run at a relatively coarse spatial resolution (i.e. 36 km * 36 km or 12 km * 12 km) that fails to fully capture the concentration hotspots at the proximity of emission sources, and thus high-risk areas were not captured. Further, studies have shown that people living close to major roads have higher risk to develop respiratory diseases than those living several hundred meters away. To capture this sharp gradient and improve characterization of the exposure and risk from traffic-related air pollutants, fine-resolution modeling is required. Fine-resolution modeling with CTM, however, is resource intensive because of the increased grid number and is not viable for large-scale applications. In this study, we will use a line source dispersion model, R-LINE, to provide PM_2.5 concentration estimates in the North Carolina Piedmont region at the Census block level. While R-LINE will be used to provide estimates of primary PM_2.5 due to onroad emissions, secondary PM_2.5 will be estimated with a two-step approach combining CMAQ and space-time ordinary kriging (STOK). In the first step, brute-force CMAQ simulations will be performed with and without onroad emissions to estimate the fraction of secondary PM_2.5 from onroad source to total PM_2.5 at a coarse resolution. In the second step, the estimated fraction from Step 1 will be applied to the observed PM_2.5 from surrounding monitoring sites to generate the "soft data" for STOK to estimate Census block-level secondary PM_2.5 from onroad source. Concentration-response functions from the literature will be used to estimate premature death based on predicted ambient concentration. The resultant premature death estimate will be compared to the estimate using CTM model to evaluate the potential underestimates by using only coarse-resolution grid-based modeling.

Youhua Tang^1,2 (youhua.tang@noaa.gov), Li Pan^1,2 (Li.Pan@noaa.gov), Pius Lee¹ (pius.lee@noaa.gov), Jeffery T. McQueen⁴ (jeff.mcqueen@noaa.gov), Jianping Huang^4,5 (jianping.huang@noaa.gov), Daniel Tong^1,2,3 (daniel.tong@noaa.gov), Hyun-Cheol Kim^1,2 (hyun.kim@noaa.gov), Min Huang^1,3 (min.huang@noaa.gov), Dale Allen⁶ (allen@atmos.umd.edu) and Ken Pickering⁷ (Kenneth.E.Pickering@nasa.gov)

1. NOAA Air Resources Laboratory, 5830 University Research Court, College Park, MD 20740

2. Cooperative Institute for Climate and Satellites, University of Maryland, College Park, MD 20740

3. Center for Spatial Information Science and Systems, George Mason University, Fairfax, VA 22030

4. NCEP Environmental Modeling Centers, 5830 University Research Court, College Park, MD 20740

5. I.M Systems Group Inc., Rockville, MD 20852

6. Department of Atmospheric and Oceanic Science, University of Maryland, College Park, MD 20742

7. NASA Goddard Space Flight Center, Greenbelt, MD 20771

Lightning NO_x could be a significant NOx emission source. The U.S. EPA Community Multi-scale Air Quality (CMAQ) Model version 5.0.2 has a lightning NOx scheme that determines the flash rate using a convective precipitation-based method and an assumed value of moles of NOx produced per flash. The flash rates can be scaled using monthly average National Lightning Detection Network (NLDN) data. We also tested other schemes for estimating the flash rate including Wong et al. (2013) and Fehr et al. (2004). The direct usage of NLDN data by Wang et al (1998) was also tested asa reference case as it does not rely on meteorological model results. The results of Beirle et al. (2006) as well as Pickering et al. (2015) are tested for lightning NOx production rates. Their corresponding CMAQ results are compared to various observations, including DISCOVER-AQ aircraft data. Besides the uncertainty of NOx production rate per lightning flash, the uncertainties in modeled precipitation rate and NLDN data also affect the lightning NOx estimation. Our discussion includes the impact of lightning NOx on surface ozone and PM2.5.

Adel F. Hanna1, Jason Ching1 and Joseph P. Pinto2

1Institute for the Environment, University of North Carolina, Chapel Hill, NC; 2National Center for Environmental Assessment, US Environmental Protection Agency, RTP, NC

Heat waves are a major source of mortality and morbidity, particularly in urban areas, and are expected to create additional risks as they become more severe in the future. Global and regional scale models have been widely used to characterize future climate conditions. However, there is considerable spatial variation in structural (building heights, aspect ratios etc.) and surface cover properties that contribute to the heat load within cities and is not captured by regional scale models. The concept of Local Climate Zones (LCZs) was developed by Stewart and Oke (2012) to characterize the urban heat island effect in terms of differences in LCZs. Our study simulated current (2003) and future (2050) climate seasonal scenarios for 120 days (May to August) using the Weather Research Forecasting (WRF) Model (at 12-km horizontal resolution) nested in the Community Climate System Model (CCSM) as a base case. We used LCZs to further downscale the mesoscale model simulations to assess intra-urban differences in heat stress for six cities in the central and eastern United States (Chicago, IL; Boston, MA; New York, NY; Washington, DC; Atlanta, GA; and Miami, FL). We used the wet bulb globe temperature (WBGT) to characterize heat stress, as a number of organizations around the world use it as a basic measure of heat stress (e.g., ISO 7243 heat stress standard). ISO 7243 is a heat stress index with thresholds relating directly to levels of physical activity. Following this basic investigation using very generic inputs to characterize LCZs, we hope to eventually be able to incorporate more detailed information from data sets such as WUDAPT (World Urban Database and Access Portal Tools).

Disclaimer: The views expressed are those of the authors and do not necessarily represent the views or policies of the U.S. Environmental Protection Agency.

Zachariah Adelman1, Mohammad Omary1, Dongmei Yang1, Ralph Morris2, Tom Moore3
1University of North Carolina Institute for the Environment, 2Ramboll Environ, 3Western States Air Resources Council (WESTAR)

The U.S. EPA released a new version of the Motor Vehicle Emission Simulator (MOVES) in October 2014 for estimating the emissions from on-road mobile sources. MOVES2014 is a major upgrade from the previous version of the model (MOVES2010b) and includes several updates to the database, algorithms, and source types simulated by the software. We conducted a model sensitivity experiment under the Western Air Quality Study to evaluate the incremental change in simulated air quality by replacing MOVES2010b emissions with MOVES2014. We used the CAMx regional chemistry transport model to simulate January and July 2011 on a 36/12/4-km nested modeling domain that focuses on the U.S. intermountain West. This presentation will present details and comparisons of the emissions and simulated air quality from the MOVES201b and MOVES2014 CAMx simulations. The presentation will include a model performance evaluation comparison between the two simulations for criteria pollutants at both urban and rural monitoring networks. We will include an analysis of the changes in model performance across different spatial and temporal scales.

Yongtao Hu¹, M. Talat Odman¹, Michael E. Chang² , Armistead G. Russell¹ and Hope Moorer³

¹School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA, 30332

²Brook Byers Institute of Sustainable Systems, Georgia Institute of Technology, Atlanta, GA, 30332

³Georgia Ports Authority, Savannah, GA, 31402

The Garden City Terminal (GCT) is a major inland container port on the Savannah River operated by the Georgia Ports Authority. The annual total container trade at GCT more than tripled between 2000 and 2014. With this rapid increase of ocean going carrier calls and container movements, coupled with plans to further expand capacity at GCT, emissions related to vessel, truck and rail transport and cargo handling equipment, and their impact on air quality has become a concern. Beginning in 2007, various efforts have been implemented to reduce emissions rates at the terminal. We chose one week in July in each of 2002 and 2010 to assess the local and regional air quality impacts of GCT and to evaluate the effectiveness of the emissions mitigation efforts. Detailed emissions inventories of the vessels and port activities at GCT were developed based on the records of pilot vessels, cargo handling equipment operations, and truck and rail movements at the terminal for the selected episodes. Emissions within the inventory were allocated by hour (e.g., by labor shift rules or by schedule records when available) and into 1-km grid cells by location within the GCT complex (e.g., at the river channel, berths, terminal yard area, or gates). All changes in emissions between 2002 and 2010 due to growth in activity, mandated regulations, and voluntary changes in equipment and operations were considered. Fine scale modeling was then conducted using the WRF-SMOKE-CMAQ system with nested grids that start from a 36-km resolution over the United States and decrease to a 1-km resolution over the Savannah metropolitan area. Both the 2002 and 2010 July episodic simulations were evaluated against ground-based air quality measurements. The CMAQ-DDM-3D technique was also used to assess the impact of GCT emissions on O₃ and PM_2.5 concentrations. Additional model simulations assessed the effects of the changes in emissions on air quality under different meteorology conditions. Finally, the changes in air quality impacts of GCT from 2002 to 2010 were evaluated relative to the increase in cargo handling.

Metin Baykara^a, Luca Pozzoli^a, Tayfun Kindap^a, Alper nal^a

^aIstanbul technical University, Eurasia Institute of Earth Sciences, Deparment of Climate and Marine Science, Istanbul, Turkey

Biogenic emissions are among the most significant sources that contribute to the formation of ozone (O₃) and secondary organic aerosols (SOA) in Europe. However, considerable uncertainties still exist in biogenic emissions estimation methodologies. In this study, the impact of biogenic emissions on ozone and SOA for Europe were investigated using two different models: the Model of Emissions of Gases and Aerosols from Nature (MEGAN) and the Nucleus for European Modeling of the Ocean (NEMO). The meteorological inputs for this work were obtained from The Weather Research and Forecasting (WRF) model. Community Multi-scale Air Quality (CMAQ) chemistry-transport model was used to determine the variability of Ozone and SOA levels for two episodes (i.e., January and July 2008).

Analysis of the results showed that ozone produced due to biogenic precursors using MEGAN has an average of 3 ppb in Central and Northern Europe but in certain Southern areas, such as Balkans and Iberian Peninsula, it is around 5 ppb. Similar spatial distribution was found in NEMO results, however, especially for Mediterranean region NEMO showed up to 5 ppb difference in monthly mean of ozone in July. SOA from terpene oxidation values are higher for NEMO as compared to MEGAN with a factor of 10 in Mediterranean region. This paper will present the differences between two models and discuss the reasons of these differences.

Saravanan Arunachalam

coming soon.

Alison Eyth, Rich Mason, Alexis Zubrow

EPA has developed an updated emissions modeling platform for 2011 based on Version 2 of the 2011 National Emissions Inventory (NEI) including projections to future years such as 2017 and 2025. The projected inventories incorporate inputs from multiple sources: improvements to projection approaches developed by EPA, information submitted by states and other interested parties received as part of EPA's comment process for the 2018 emissions modeling platform, and improvements to EPA models such as the Motor Vehicle Emissions Simulator (MOVES) and the Integrated Planning Model (IPM), including improved input data for those models. EPA spent a substantial amount of time refining the projection approaches for electric generating units (EGUs), oil and gas sources, residential wood combustion sources, other stationary sources, and onroad mobile sources. Updates were also made to approaches for projecting emissions in Mexico, the upstream impacts of mobile source rules, and fugitive dust sources. The resulting projected inventories will be used for multiple regulatory analyses including transport modeling and development of regulatory impact assessments for national ambient air quality standards. The improved methods and the impacts of updating the projections on the emissions will be discussed.

Robert Nedbor-Gross¹, Barron H. Henderson.¹, Jorge E. Pachon. ², Maria P. Perez Pen ²

¹ University of Florida, Department of Environmental Engineering Sciences

² Universidad de la Salle, School of Engineering

Health effects from particulate matter (PM) have been increasing with continuous economic growth and have been previously estimated from interpolated observations (Secretar a Distrital de Ambiente, Informe Anual de Calidad del Aire de Bogota, 2011). Studies have shown that observed PM can be insufficient to characterize health outcomes and recommend models as a source for better coverage (Bell and Ebisu, 2012). At the Universities of Florida and de La Salle, an air quality modeling system was specifically designed to assess pollution and proposed regulations in Bogota, Colombia. The modeling platform combines the Weather Research and Forecasting Model (WRF), local emissions, and the Community Multi-scale Air Quality model (CMAQ). This platform is the first of its kind to be implemented in the megacity of Bogota, Colombia. Major emission sources that were included are commercial, industrial, motor vehicle, and road dust. This presentation focuses on improving key emission source estimates.

Road dust is the major contributor to local PM emissions and has a high degree of uncertainty. Road dust emissions were developed by extrapolating EPA AP-42 methods to Bogota roads and road conditions. Using these methods, road dust makes up 90% of particulate emissions by mass, and produces a large high bias compared to measured PM₁₀ (http://ambientebogota.gov.co). This study constrains dust emissions with ground monitor observations. We assume that observations are accurate and that dust emissions have the highest uncertainty, which is consistent with our emission database development. We then minimize simulated mean fractional error (MFE) to produce dust emission scaling factors. Improving the estimate of dust emissions will produce more accurate representation of health effects by improving exposure fields.

Using improved more accurate exposure fields; we quantify health effects in present day and potential futures. The dust emission scaling factors from the present day are applied to future year emission inventories, which include economic growth and current regulations. In addition, we simulate futures with a series of regulations proposed by the local authority. Using these simulations, we quantify the health benefits from proposed regulations. Our estimates are useful in evaluating the efficacy of future regulations.

Kristen E Brown, Daven K Henze, Jana B Milford

We use MARKAL to model different policy scenarios in which fees are applied to emissions related to generation and use of energy. MARKAL, which solves for the least cost way to satisfy demand for energy and calculates the associated emissions, is run with a modified version of the EPA US 9 region v1.1 database. The fees are based on literature values of damages for climate and health impacts and are applied to upstream and combustion emissions related to electricity generation, industrial energy use, transportation energy use, residential energy use, and commercial energy use. These fees lead to reductions in emissions that vary based on the fees applied. The resulting emissions in 2045 are used in conjunction with CMAQ to determine the change in air quality due to different emissions reduction scenarios. The air quality in fee scenarios is compared to a future scenario without fees to determine the effect of the policies. By going beyond the energy-system modeling done by MARKAL, we can further evaluate how air quality may respond to such emissions scenarios, exploring differences in the distribution of ozone and particulates in urban and rural areas. We can also explore the occurance of any unexpected dis-benefits in air quality, where emissions are reduced but air quality in some locations is worsened.

Clare Flynn

Kenneth Pickering

Christopher Loughner

Andy Weinheimer

Glenn Diskin

Richard Clark

Gina Mazzuca

Russell R. Dickerson

The first deployment of the NASA Earth Venture -1 DISCOVER-AQ (Deriving Information on Surface conditions from Column and Vertically Resolved Observations Relevant to Air Quality) project was conducted during July 2011 in the Baltimore-Washington region. The P-3B aircraft provided in situ vertical profiles of meteorological quantities, trace gases, and aerosols over six Maryland Department of the Environment (MDE) air quality monitoring sites over fourteen flight days. Additionally, two sites launched ozonesondes and tethersondes during the campaign, supplementing the P-3B profiles. A major goal of DISCOVER-AQ is to understand the processes, such as vertical mixing, linking column abundances to surface concentrations for O₃ and NO₂. This information is vital for better use of satellite data for air quality applications. The in situ P-3B and sonde observations, as well as model simulations performed with the coupled WRF/CMAQ model, will be used to investigate the impact of vertical mixing on O₃, NO₂, and CO profiles as well as the column-surface relationship. Simulations using six commonly used PBL schemes within the coupled WRF/CMAQ model system will be run and the results will be compared to observations, to determine which scheme (if any) best captures the PBL depth, the observed state of boundary layer mixing and trace gas profiles during July 2011, as well as the observed column-surface relationship for O₃ and NO₂. Reasons for differences among PBL schemes will also be explored.

Shannon L. Capps, Rene Nsanzineza, Matthew D. Turner, Daven K. Henze, Shunliu Zhao, Matthew G. Russell, Amir Hakami, and Jana B. Milford

Natural gas and oil exploration and production processes emit gases that contribute to tropospheric ozone formation, which negatively impacts human health and public welfare. Attaining the ozone National Ambient Air Quality Standard (NAAQS) has been challenging for some urban regions, including certain cities adjacent to oil and gas development. For these locations, understanding the relative contribution of emissions from oil and gas activities is important to evaluating emissions control strategies. In this investigation, we elucidate the influences of NO_x and volatile organic compound (VOC) emissions from this sector. Furthermore, we compare and contrast these influences between urban regions with differing background concentrations and oil and gas-related activity.

Specifically, we use the adjoint of CMAQ to calculate the sensitivity of urban ozone concentrations and NAAQS exceedances and of ozone-related health risks to emissions of precursor gases. The adjoint efficiently determines these relationships for each emitted species and sector. We contrast the relative influences of VOCs and NO_x on ozone in urban centers in the Colorado Front Range, the Northeastern corridor, and eastern Texas, which have experienced recent acceleration of development of oil and natural gas resources nearby, with those in the southern California area, which has managed adjacent oil and gas production for a longer period of time. We extend the analysis by projecting the influence of revised emissions estimates on ozone impacts.

Sandra Rayne, Heather Holmes, Barbara Zielinska and Alan Gertler

The Lake Tahoe Basin is located on the border of California and Nevada northeast of the Central Valley. Despite its pristine beauty and water clarity, the Lake Tahoe Basin is facing problems related to air pollution inlcuding peak ozone concentrations that approach or slightly exceed multiple ambient air quality standards. The meteorology in this region is unusually complex due to the Sierra Nevada Mountains and other topographical features. Thermally driven wind systems are a common phenomenon found in mountainous regions throughout the world. These wind systems, along with the structure of the atmospheric boundary layer are key for understanding the distribution and transport of atmospheric pollutants in complex terrain. It is important in air pollution modeling to correctly represent the planetary boundary layer (PBL) and different PBL parameterization schemes within the Weather Research and Forecast (WRF) model have different assumptions determining the transport of mass, moisture and energy. The performance of different PBL schemes varies depending on the meteorological conditions, thus emphasizing the importance of meteorological inputs when developing air quality control strategies. Although several studies have examined the sensitivity of WRF model predictions to PBL schemes, such a study has not been done over the complex terrain of the Lake Tahoe Basin. This study is needed to determine the meteorological model parameterizations to drive air quality simulations in this geographical domain. During the period of July 21-26 2012, a field study was conducted in the Lake Tahoe Basin designed to characterize the precursors and pathways of secondary pollutant formation, including ozone and secondary organic aerosols (SOA). This analysis looks at three WRF PBL schemes (Yonsei University Scheme, Mellor-Yamada Janic Scheme and the Asymmetric Convective Model version 2 Scheme) and evaluates their performance for this area. Based on the model evaluation results, one PBL scheme in WRF is used to investigate the thermally forced wind circulation in the region. The WRF results are then compared to the ozone observations from the field study to better understand the impact of meteorology on the transport of pollutants in this area.

Fernando Garcia-Menendez¹, Rebecca K. Saari², Erwan Monier¹ and Noelle E. Selin²

1)Joint Program on the Science and Policy of Global Change; Massachusetts Institute of Technology

2) Engineering Systems Division; Massachusetts Institute of Technology

As part of EPA's Climate Change Impacts and Risk Analysis (CIRA) project, we evaluate the effects of climate change and climate policy on U.S. air quality and health using integrated economic, climate, and air pollution projections. Our modeling framework is based on the MIT Integrated Global System Model coupled to a global atmospheric chemistry model (CAM-Chem) and health/economic impacts model (BenMAP). Using an ensemble simulation of 21^st century climate change we assess the "climate penalty" on O₃ and PM_2.5 concentrations under different policy scenarios and compare the value of air quality-related health benefits derived from slowing climate change to mitigation costs. In addition, we explore the influence of important sources of uncertainty in climate modeling on air quality projections. Our simulations suggest that climate change, exclusive of variations in pollutant emissions, can significantly impact air quality across the U.S. and increase associated health effects. Greenhouse gas mitigation efforts can substantially reduce these impacts and climate-specific air pollution health benefits alone can offset a significant fraction of policy costs. Uncertainties related to emissions, climate system response, and natural variability strongly propagate to air quality projections. However, our results reveal that the impacts of climate change on air pollution should not be overlooked in climate policy analyses.

Barry Baker¹ , Rick Saylor¹, Pius Lee²

¹NOAA Air Resources Laboratory, Atmospheric Turbulence and Diffusion Division, Oak Ridge, TN 37830

²NOAA Air Resources Laboratory, College Park, MD

The bulk Richardson number is commonly used in observational and modeling studies to determine atmospheric stability. A constant critical bulk Richardson number (Ri_bc) is frequently assumed in current methods to predict the level at which turbulence transitions from the boundary layer to the free-atmosphere defined the height of the boundary layer. Recently, a dynamic Ri_bc formulation was developed using large eddy simulations and wind tunnel data and has been found to be applicable to field measurements. In the current study, the dynamic Ri_bc algorithm was implemented into the Weather Research and Forecasting (WRF) model. Results show that the dynamic Ri_bc produces a higher boundary layer height than when using a static Ri_bc value and that nocturnal low-level wind speeds are not sensitive to a change in the nocturnal boundary layer height. Simulations using meteorological input data from the dynamic Ri_bc formulation were then done with the Comprehensive Air Quality Model with Extensions (CAMx) over the TexAQS II domain and over a mid-Atlantic domain to investigate the contribution of nocturnal boundary layer dynamics on daytime ozone concentrations. Model results for ozone were found to be dependent on the predominant chemical regime. Predictions of mean 8-hr ozone concentrations from CAMx were not improved in the Texas NOx-limited regime., but in the mid-Atlantic VOC-limited domain, predicted median 8-hr ozone concentrations during a 22-day simulation improved by 10% compared to simulations using a static Ri_bc.

Julia Gamas, Ph.D.

U.S. EPA Office of Air Quality Planning and Standards

Dan Loughlin, Ph.D., and Rebecca Dodder, Ph.D.

U.S. EPA Office of Research and Development

This presentation will provide an update on the development and evaluation of four Air Quality Futures (AQF) scenarios. These scenarios represent widely different assumptions regarding the evolution of the U.S. energy system over the next 40 years. The primary differences between the scenarios relate to the availability of advanced technologies that may have lower environmental impacts and society's willingness to adopt less polluting consumption patterns when faced with future environmental challenges. We briefly describe how the four qualitative scenarios were incorporated into the MARKet Allocation (MARKAL) by adjusting hurdle rates and other values so that the model was free to choose technologies and fuels consistent with the implications of the scenarios, in response to the application of additional policies or other assumptions. In this application, NOx reduction targets are applied in each scenario. The resulting technology and fuel choices for each are examined, as are ancillary benefits, such as reductions of other criteria and greenhouse gas emissions. We evaluate how features of each scenario may have influenced the technology and fuel choices, and the implications for future air quality management.

Marjan Soltanzadeh, Robyn Chatwin-Davies, Amanda Pappin, Amir Hakami

Department of Civil and Environmental Engineering, Carleton University, Ottawa

Various studies have investigated the Greenhouse Gas (GHG) reduction co-benefits, or ancillary air quality benefits, associated with reduction of GHG (in particular, CO₂) emissions. The studies have all been scenario-based, i.e., they evaluate air quality co-benefits from implementing an ensemble of measures that are prescribed in a climate policy or emission scenario. Scenario-based evaluations have been used to gain insight on sector-specific co-benefits from GHG reduction. However, since scenario-based analysis considers emissions changes in specific sectors and across all locations, it cannot distinguish between benefits accrued from GHG reductions among sources in different locations. Source-specificity (i.e. location dependence) can be an important factor for co-benefits, and can play an important role as a policy metric in addition to the control cost. The motivation for our study is a) to quantify how reduction of criteria co-pollutants can have ancillary benefits for public health, b) to determine how these co-benefits vary spatially and by sector, and c) to compare their spatial distribution with that of abatement costs or the social cost of carbon.

In this study we, for the first time, apply the adjoint of CMAQ to quantify the health benefits associated with emission reduction of criteria pollutants (NO_X, CO, PM) in different sectors of on-road, non-road (diesel and gasoline), and power plants, on a location-by-location basis across the US. These health benefits are then converted to CO₂ emission reduction co-benefits by accounting for source-specific emission rates of criteria pollutants in comparison to CO₂. The initial focus in this study is on the co-benefits due to the reduced NO_X as an ozone precursor and the subsequent impact on long-term mortality. Emission data for NO_X and CO₂ mobile data are taken from the 2011 National Emission Inventory (NEI 2011) at the county level, and then aggregated up to 36-km resolution. CO₂ emission data for point sources are taken from Fossil Fuel Data Assimilation System (FDDAS 2010). Our preliminary results show that the monetized health impacts (based on O₃-related short-term mortality) associated with reductions in 1 ton of CO₂ emissions ranges from 20-150 $/ton CO₂ for mobile on-road sector, 50-900 $/ ton CO₂ for mobile non-road and 1-100 $/ton CO₂ for point sources. These values are comparable to those found previously in scenario-based studies (Nemet et al., 2010), and yet are significant in comparison with the price of carbon in various markets. In general, co-benefits show a great deal of spatial variability across different emission locations. Consequences of such spatial variability in devising control policy options that effectively address both climate and air quality objectives will be discussed.

Nemet, G. F., Holloway, T. & Maier, P. Implications of incorporating air-quality co-benefits into climate change policymaking. Environ. Res. Lett. 5,014007 (2010).

Hyun Cheol Kim^1,2, Pius Lee¹, Fong Ngan^1,2, Youhua Tang^1,2, Hye Lim Yoo^1,2, Li Pan^1,2 and Ivanka Stajner³

¹ NOAA/Air Resources Laboratory, College Park, MD

² UMD/Cooperative Institute for Climate and Satellites, College Park, MD

³ NOAA/National Weather Service, Silver Spring, MD

Surface ozone variability with respect to cloud coverage is evaluated in a regional air-quality forecast model using satellite-observed cloud fraction (CF) information and observations from a surface air-quality monitoring system. We compared CF and daily maximum ozone from the National Oceanic and Atmospheric Administration's National Air Quality Forecast Capability (NOAA NAQFC) with CFs from the Moderate Resolution Imaging Spectroradiometer (MODIS) and the U.S. Environmental Protection Agency's AirNow surface ozone measurements during May to October, 2014. We found that observed surface ozone shows a clear (negative) correlation with the MODIS CFs, showing around 1 ppb decrease for 10% MODIS CF change over the Contiguous United States, while the correlation of modeled surface ozone with the model CFs is much weaker, showing only -0.5 ppb per 10% NAQFC CF change. Further, daytime CF differences between MODIS and NAQFC are correlated with modeled surface-ozone biases between AirNow and NAQFC, showing -1.05 ppb per 10% CF change, implying that spatial- and temporal- misplacement of the modeled cloud field might have biased surface-level ozone in the model. Current NAQFC cloud fraction seems to be too low compared to MODIS retrievals (mean NAQFC CF = 0.38 and mean MODIS CF = 0.55), which could be contributing up to 35% of surface-ozone bias in the current NAQFC system.

Timothy Vinciguerra, Emily Bull, Timothy Canty, Hao He, Eric Zalewsky, Michael Woodman, Sheryl Ehrman, Russell Dickerson

Using EPA 2018 version 1 emissions (2011 base year) and CMAQ v5.0.2, a series of sensitivity tests were performed to evaluate the significance of NOx emission reductions from electricity generating units (EGUs) in several states in the eastern United States. EGU NOx emissions were adjusted to match the lowest/best NOx rates observed during the ozone seasons of 2005-2012 (Scenario 3A) and the highest/worst rates during the same time period (Scenario 3B). Scenario 3C adjusted rates to match the observed rates from 2011, and Scenario A4 assumed additional reductions to Maryland units. Finally, Scenario 3D assumed NOx reduction if SCR controls were applied to currently uncontrolled units. This analysis outlines a local strategy for attainment in Maryland, but in anticipation of a decrease of the ozone standard, also provides profound benefits for upwind states where most of the regional EGU NOx originates.

Tammy M. Thompson,¹ Michael G. Barna,² Bret A. Schichtel², and Kristi A. Gebhart²

¹Cooperative Institute for Research in the Atmosphere, Colorado State University, Fort Collins, CO 80523-1375, USA

²National Park Service, Air Resources Division, Lakewood, CO, USA

Excess wet and dry deposition of nitrogen-containing compounds is a concern in many areas throughout the US. In some high alpine lakes in Rocky Mountain (RMNP) and Grand Teton (GTNP) national parks, this increased deposition is causing biogeochemical changes. Source apportionment in chemical transport models can be used to identify sources of excess nitrogen deposition and guide the development of effective control strategies. However, accurate source apportionment depends on accurate model inputs, including meteorological representation. In the Rocky Mountains, mountainous terrain creates unique challenges for meteorological modeling. For example, model evaluation of a nitrogen source apportionment study in RMNP suggests that ammonia sources in eastern Colorado are likely under-represented due to the difficulty of capturing upslope air flow on the eastern side of the Continental Divide with meteorological models. A comparison of modeled versus measured wind direction at a RMNP meteorological stationshows that during 16% of the total year-long source apportionment modeling episode, the wind in the model is westerly, while observed winds are easterly. In this study, the Weather Research and Forecasting-Advanced Research WRF model (WRF-ARM) is used with the Comprehensive Air quality Model with extensions (CAMx) and the same 2009 emissions inventory and emissions pre-processing to test the sensitivity of chemical transport modeling source apportionment results to meteorological modeling. The meteorological modeling options selected for this sensitivity study include those commonly used for US federal and state regional chemical transport modeling efforts as well as model options selected based on demonstrated ability to forecast upslope flow events on the Colorado Front Range. The output from WRF sensitivity runs are evaluated versus measured data then used as inputs to re-run source apportionment modeling for nitrogen in RMNP. Source apportionment results are compared and reported for all scenarios. Results also have implications for general model performance in challenging terrains.

Human-socioeconomic activities, energy use, and technology changes have important impacts on air quality and climate science and policy. A fully-coupled earth system model provides a powerful tool to link energy, emissions, air quality, and climate change for integrated science and policy applications. We have developed an Integrated Technology-Driven Earth System Model (ITDEaSM) to link technology options, energy uses, and policy choices and resultant emissions with Earth system processes. The ITDEaSM is developed base on (1) a Technology-Driven Model (TDM) that generates technology splits to project emissions, (2) the Community Earth System Model (CESM) and the Weather Research and Forecasting Model with Chemistry (WRF/Chem) with advanced chemistry, aerosol, and cloud treatments that simulate air quality and climate over global and regional scales, respectively, and (3) a water quality model and an ecosystem model that use climate and deposition predictions from WRF/Chem to simulate the impact of climate change on water quality and forest ecosystem. Our objectives are to improve model representations of the feedbacks among climate change, air/water quality, and ecosystem and perform decadal simulations at global through urban scales to identify technology choices for co-benefits of climate/Earth system mitigation. In this work, we will present our results from decadal simulations of CESM and WRF/Chem. Driven by the Representative Concentration Pathways (RCP) scenarios and TDM projected emission scenarios, decadal simulations using WRF/Chem with boundary conditions from CESM are performed over continental U.S. for current (2001-2010) and future (2046-2055) decades. A comprehensive evaluation is performed for current decade using available observations from surface networks and satellites. Simulations with emission projections for two RCP scenarios (4.5 and 8.5) and one TDM emission scenario (A1B) are intercompared to examine the impacts of technology/energy choices on future climate and air quality. The policy implications of our results in supporting climate mitigation and earth system management will be also discussed.

Pius Lee ^1*, Robert Atlas², Youhua Tang^1,3,

Li Pan^1,3, Hyuncheol Kim^1,3, Daniel Tong^1,3,4, Sean Casey⁵

1. Air Resource Laboratory (ARL), NOAA, College Park, MD

2. Atlantic Oceanographic and Meteorological Laboratory, Miami, FL

3. Cooperative Institutes for Satellite and Climate, University of Maryland, College Park. MD

4. Center for Spatial Information Science and Systems, George Mason University, Fairfax, VA

5. Center for Satellite Application and Research, NESDIS/STAR, College Park, MD

Recent and the scheduled soon-to-happen launches of NPP and JPSS satellites reinforce the emphasis of NOAA warning and forecasting operational systems tap heavily on the satellite technology to improve its understanding of the current and future states of the earth systems. Air quality is an important part of this integral system. Observation System Simulation Experiment (OSSE) is a powerful tool. It pinpoints and can be applied specifically to rank and quantify the benefit of future potential instrumentation on those satellites that can benefit air quality forecast and analysis. The new instrumentation gives additional data but their accompanying retrieval algorithms and uncertainty estimates are also important input. In this study we attempt to quantify the performance gain of a regional state-of-the-science chemical data assimilation and air quality forecasting system when incremental sets of measurements are acquired into the Global Forecasting System (GFS) of the National Centers for Environmental prediction (NCEP). There are several sets of new instrumentations and algorithms that are being considered for the NOAA Geostationary Operational Environmental Observing Satellite R-series (GOES-R). One of them is the Atmospheric Infrared Sounder (AIRS) system. This study illustrates how much such hyperspectral data can benefit the forecasting performance of a regional air quality forecasting model. AIRS with its hyperspectral sensors helps to constrain moisture and wind fields in the troposphere. It has reasonably good sensitivity in the lower troposphere and thus provides considerable benefit to constrain those meteorological fields and subsequently it is hopeful that it can also benefit the performance of an air quality model that is driven by those observation constraint meteorological fields. The air quality model may better capture air pollutant concentration in those lower layers. Further options in the OSSE will give a more holistic metrics on how one can recommend future instrumentations that can optimally benefit regional air quality modeling.

Shaohui Zhang[¹]*, Ernst Worrell¹, Wina Crijns-Graus¹

¹Copernicus Institute of Sustainable Development, Utrecht University, Heidelberglaan 2, 3584 CS Utrecht, The Netherlands

Several studies have shown that actions to reduce the combustion of fossil fuels often decrease GHG emissions as well as air pollutants, bring multiple benefits for improvement of energy efficiency, climate change, and air quality associated with human health benefits. Therefore, air quality and health co-benefits can provide strong additional motivation for improving energy efficiency. In China, the cement industry is the second largest energy consumer and key emitter of CO₂ and air pollutants. It accounts for 7% of total energy consumption in China and 15% of CO₂, 21% of PM, 4% SO₂ and 10% of NOx of total emissions, respectively.

In this study, an integrated approach that comprises of a number of different methods and tools within the same platform (i.e. provincial energy conservation supply curves (ECSC), Greenhouse Gas and Air Pollution Interactions and Synergies (GAINS) model, geographical information system (GIS), TM5, and Health Impact Assessment (HIA)) is developed and used to assess the potential of energy savings and emission mitigation of air pollutants, as well as the environmental and health impacts of pollution arising from China's cement industry at the provincial level during the period 2011-2030. The results show significant heterogeneity across provinces in terms of potential of energy saving as well as emission mitigation of CO₂ and air pollutants (i.e. PM, SO₂, and NOx) in the next two decades. In addition, the current commercially available energy efficiency measures would decrease 25% of SO₂, 20% of NOx, and 5% of PM_2.5 meaning that 0.017 0 of premature deaths might be avoided (adults e 30 ages), compared to the baseline scenario. Therefore, It is more cost effective for policy makers to consider both air quality and health impacts together when planning and implementing energy policy than to pay attention at each issue separately.

Marina Astitha1* , S.T. Rao2 , Jaemo Yang1 , Huiying Luo1

1Department of Civil & Environmental Engineering, University of Connecticut, Storrs-Mansfield, CT, USA,

2Department of Marine, Earth & Atmospheric Sciences, North Carolina State University, Raleigh, NC, USA

*astitha@engr.uconn.edu

The reducible and irreducible uncertainties in uncoupled and fully coupled meteorologyatmospheric chemistry models have been studied extensively in the past two decades. Current practices in reducing model uncertainties involve data assimilation, ensemble modeling and bias reduction techniques (Hogrefe et al. 2000; Gilliam et al. 2012 among others). Nevertheless, the inherent uncertainty cannot be eliminated since our inability to properly characterize the initial state of the atmosphere will prevent the accurate prediction of its future state. This work focuses on the assessment of inherent uncertainties in atmospheric and air quality modeling systems by evaluating the impact of different initial conditions on weather parameters and their effect on tropospheric ozone and particulate matter concentrations. The modeling systems applied for this work are the coupled meteorology-chemistry modeling system RAMS/ICLAMS (Cotton et al. 2003; Solomos et al. 2011) and the Comprehensive Air Quality Model with Extensions (CAMx, Environ 2015). The simulations were performed during the summer of 2006 using various initialization fields from global model analyses outputs. The uncertainty is quantified for weather variables that affect air pollutant transport and dispersion and atmospheric pollutant concentrations for continental and Northeastern US. The results indicate that the most impacted atmospheric fields are precipitation, cloud cover, ventilation coefficient, and sea salt loading. The predicted daily maximum ozone concentration has shown substantial variability ranging from 10% to 40% (standard deviation 10-15ppb) which is consistent with the results of Gilliam et al. (2012) using a different methodological approach. The main goal of this work is to provide confidence limits in the modeling predictions that focus on regulatory applications regardless of the selected modeling system.

Alvarez A.R., Kumar N., Knipping E., Shaw S.

New air quality standards for ozone will likely impose a heavy burden on utilities and other industrial sources to further reduce emissions. Research has shown that emissions from low level sources such as motor vehicles may be more efficient at producing ozone than those from the elevated sources like power plants. Control strategies have typically targeted power plants and other industrial sources. Controls to motor vehicles have also been imposed through the mandatory installation of catalytic converters and fuel re-formulation and improvements in fuel efficiency. Consequently, newer cars are less polluting, more fuel-efficient and burn fuel more cleanly than previous cars. However, more air quality benefits may be achieved by reducing emissions from older vehicles that are still circulating in our roads in significant numbers. Newer cars may also add to the problem due to poor maintenance practices that may deteriorate over time the proper working of the catalytic converter and make the car run less efficiently, burning more fuel. The fact is that according to some estimates 90 percent of emissions from vehicles come from 10 percent of the fleet.

Faced with further emissions reductions, utilities and other heavy polluters could gain pollution credits by purchasing the older or more polluting vehicles from consumers and destroying those vehicles so "bad emitters" are eliminated from the fleet and encouraging owners to purchase newer, more efficient vehicles. Currently, some states have such buy-back programs but most face limited budgets. There is no mechanism for air quality emissions trading that different sources could make use of to improve air quality in a more cost-effective manner.

The main research question addressed in this paper is to quantify the tradeoffs in overall air quality when emissions from a power plant are offset with similar level of emissions from retiring older or more polluting vehicles and compare the costs of control of emissions from a power plant versus the cost of buying back those vehicles.

The CAMx and CMAQ models were used to answer this question. Preliminary results indicate that retiring around 100,000 old vehicles may be an effective way to reduce emissions and ozone levels at target areas, compared to the installation of SCR controls running at 90% efficiency at one of the most polluting power plant units located in the neighborhood. Installation of such controls can be prohibitely expensive, depending on plant design and just running existing SCR controls at full efficiency during the ozone season can run into millions of dollars in costs.

Chih-Yuan Chang and Kuo-Jen Liao*

Department of Environmental Engineering, Texas A&M University-Kingsville

Over the past two decades, air quality has significantly improved in the United States (U.S.) due to controls of air pollutant emissions from anthropogenic sources (e.g., power plants, mobile vehicles, etc.). However, unconventional energy production (e.g., shale oil and gas) flourished in recent years can be a new source of air pollutant emissions in the U.S. The U.S. Environmental Protection Agency (EPA) updates the national emission inventory (NEI) roughly every three years, and NEI alone is inadequate for estimating emissions from shale oil and gas-related activities because of the fast growth in unconventional energy production. In this study, we used an integrated bottom-up (i.e., NEI) and top-down (i.e., satellite remote sensing) approach to estimate air pollutant emissions over major shale plays from 2011-2014 in the U.S. For the bottom-up approach, we investigate 2010 U.S. air pollutant emissions modeled using Sparse Matrix Operator Kernel Emission (SMOKE). For the top-down approach, we use vertical column densities (VCDs) of NO₂ from the Ozone Monitoring Instrument (OMI). The 2010 air pollutant emissions estimated using the bottom-up approach is used as baseline emission for estimating NO_x emissions from 2011-2014.

The correlation between OMI-retrieved NO₂ column densities and baseline NO_x emissions estimated using SMOKE with the EPA NEI are investigated for the 2010 summer episode (i.e., June, July and August). The correlation between OMI-retrieved NO₂ VCDs and bottom-up NOx emissions are then applied to time-series of OMI-retrieved NO₂ VCDs to estimate NO_x emissions from U.S. shale plays from 2011 to 2014. The 2011 NOx emissions estimated using the integrated approach will also be compared against NO_x emission estimated using the EPA 2011 NEI which includes emissions from oil and gas sectors. This approach can provide near real-time estimation of total NOx emissions from key shale plays in the U.S. and help understand the contribution of shale oil and gas-related NO_x emissions to formation of ozone and particulate matters in down-wind areas of the shale plays.

Kaili Stevens1, Valerie Garcia2, Chris Nolte2, Tanya Spero2, and Jim Crooks2

1ORISE, Oak Ridge National Laboratory, Tennessee
2U.S. Environmental Protection Agency, Research Triangle Park, North Carolina

Severe weather events, such as flooding, drought, forest fires, and dust storms can have a serious impact on human health. Dust storm events are not well predicted in the United States, however they are expected to become more frequent as global climate warms through the 21st century. Understanding what causes dust storms will facilitate the prediction of vulnerability to dust storm events within the context of a changing climate. In turn, this will inform adaptation and mitigation strategies to help protect human health, and prevent crop and property damage. Using data on dust storms in the United States from 1992 to 2010 collected by the U.S. National Weather Service, and output from the Weather Research and Forecasting Model containing meteorological variables for the affected portion of the country, we will predict vulnerability to dust storms under future climate scenarios. More specifically, a logit regression model will be used to investigate the association between these meteorological variables and the presence or absence of a dust storm event. The regression model will be applied to generate probability maps of dust storm events in the United States under a changing climate scenario. A comparison will also be performed of the dust product simulated by the Community Multiscale Air Quality (CMAQ) model. The results of this project will be extended to determine the effects of dust storms on population health.

Chaopeng Hong1,2, Khairunnisa Yahya1, Yang Zhang1, and Qiang Zhang2

1Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, NC 27695, U.S.A.
2 Ministry of Education Key Laboratory for Earth System Modeling, Center for Earth System Science, Tsinghua University, Beijing, China

Real-time air quality forecasting (RT-AQF) over multiple years allows for comprehensive model evaluation and examination of changes in air quality over a long-term period. The online-coupled Weather Research and Forecasting model with Chemistry with the Model of Aerosol Dynamics, Reaction, Ionization, and Dissolution (referred to as WRF/Chem-MADRID) has been applied for RT-AQF for seven O3 seasons (May to September) and six winters (December to February) during 2009-2015 to forecast ozone (O3) and fine particles (PM2.5) over the southeastern U.S. In this work, a comprehensive model evaluation of forecasted air quality and weather is performed using observations from available surface networks and satellites in terms of spatial distribution, temporal variation, discrete and categorical performance statistics. Changes in air quality and meteorology during 2009- 2015 are analyzed along with their correlations to assess the model's ability in reproducing the observed changes.
WRF/Chem-MADRID shows consistently good skills for O3 and PM2.5 forecast in terms of both categorical and discrete statistics during 2009-2015. It performs well in both O3 and winter seasons with most NMBs within 15% for O3 and within 10% for PM2.5 against observations from AirNow. Larger biases are found for secondary PM2.5 species against observations from other networks. These biases are due possibly to uncertainties in meteorology, emissions, and boundary conditions, as well as limitations in model treatments for several processes such as aerosol thermodynamics and chemistry, and secondary organic aerosol formation. While observed seasonal-mean maximum 1-hr and 8-hr O3 concentrations from AirNow remain nearly constant since 2009, observed seasonal-mean PM2.5 concentrations show stronger inter-annual variabilities than O3, partly reflecting the effects of emission reductions since 2009. Forecasted seasonal-mean O3 levels also remain nearly constant during O3 and winter seasons 2009-2015 despite somewhat inconsistent inter-annual variabilities with the observations. Forecasted seasonal-mean PM2.5 levels reproduce well their observed increasing trend from 2009 to 2011 and their observed declining trend from 2011 to 2014 during O3 seasons and the nearly constant observations during winter. Inter-annual variabilities of forecasted O3 and PM2.5 concentrations are mainly attributed to changes in meteorological variables and the subsequent changes in meteorology-dependent biogenic emissions. The results from this work will provide valuable information to further enhance the forecasting skills of the RT-AQF model through improving the model inputs and treatments.

Dan Loughlin and Chris Nolte, U.S. EPA

Steve Smith, Pacific Northwest National Laboratory

Jason West, Univ. of North Carolina at Chapel Hill

The Global Climate Assessment Model (GCAM) is a global integrated assessment model used for exploring future scenarios and examining strategies that address air pollution, climate change, and energy goals. GCAM includes technology-rich representations of the energy, transportation, buildings, and agriculture sectors, which are linked to representations of the economy, climate and land use systems. For various scenarios, GCAM produces outputs such as estimates of technology adoption, fuel use, and climate and air pollutants. GCAM has been used in a variety of high-profile applications, including the production of the Representative Concentration Pathway (RCP) 4.5 W/m2 scenario. Development of GCAM continues, and a new version called GCAM-USA now represents the U.S. at the state-level resolution within the larger global framework. We describe GCAM-USA and highlight how the model is being updated to more fully account for U.S. air quality regulations. We also demonstrate the application of the model to examine the state-level climate and air pollutant emissions associated with alternative assumptions about policy regimes. We conclude by discussing ongoing efforts to evaluate how GCAM-USA can be integrated into a decision support framework, with the goal of facilitating state and federal air, climate and energy analyses.

Dan Loughlin, Ph.D.

U.S. EPA Office of Research and Development

Kathy Kaufman, Brian Keaveny, and Alex Macpherson, Ph.D.

U.S. EPA Office of Air Quality Planning and Standards

A marginal abatement cost curve (MACC) traces out the relationship between the quantity of pollution abated and the marginal cost of abating each additional unit. In the context of air quality management, MACCs typically are approximated by sorting end-of-pipe controls by their respective cost effectiveness. Alternative measures, such as renewable electricity, energy efficiency, and fuel switching (RE/EE/FS), are not considered as it is difficult to quantify their abatement potential. In this presentation, we demonstrate the use of an energy system model to develop regional and sectoral MACCs for nitrogen oxides (NOx) that incorporate both end-of-pipe controls and these alternative measures. The resulting MACCs may be incorporated into other modeling tools, such as Integrated Assessment Models, and may be of use in developing regional emission control strategies.

Min Huang^1,2, Li Pan^1,3, Pius Lee¹, Daniel Tong^1,2,3, Youhua Tang^1,3, Ivanka Stajner⁴, Jeff McQueen⁵, Ariel Stein¹, Julian Wang¹

¹NOAA/OAR/ARL, NOAA Center for Weather and Climate Prediction, College Park, MD 20740, USA

²Center for Spatial Information Science and Systems, George Mason University, Fairfax VA 22030, USA

³Cooperative Institute for Climate and Satellites, University of Maryland, College Park, MD 20740, USA

⁴NOAA/NWS/STI, Silver Spring, MD 20910, USA

⁵NOAA/NWS/NCEP/EMC, NOAA Center for Weather and Climate Prediction, College Park, MD 20740, USA

Josephine T. Bates, Rodney J. Weber, Joseph Abrams, Vishal Verma, Ting Fang, Cesunica Ivey, Mitchel Klein, Matthew J. Strickland, Stefanie E. Sarnat, Howard H. Chang, James A. Mulholland, Paige E. Tolbert, Armistead G. Russell

It is hypothesized that inhalation of specific sources of PM2.5 can catalytically generate reactive oxygen species (ROS) in excess of the body's antioxidant capacity, which leads to oxidative stress. Field studies measuring ROS-generation potential of ambient water-soluble PM2.5 in Atlanta from June 2012 - June 2013 are used with source apportionment techniques to investigate emissions sources of ROS-generating PM2.5 and to develop a long-term estimate of daily ROS levels for use in an acute epidemiologic study. A dithiothreitol (DTT) assay is used to assess the ROS-generation potential of water-soluble PM2.5. The Chemical Mass Balance Method with ensemble-averaged source impact profiles is used for PM2.5 source apportionment in Atlanta, GA, and source apportionment results from CMAQ-DDM are used for all other DTT measurement sites (Yorkville, GA, Centerville, AL, and Birmingham, AL). Sources of PM2.5 are related to DTT activity using least squares linear regression modeling. Results for Atlanta show that light-duty gasoline vehicles (LDGV) exhibit the highest intrinsic DTT activity, followed by biomass burning (BURN) and heavy-duty diesel vehicles (HDDV) (0.11, 0.069, and 0.052 nmol min-1 g-1source, respectively). BURN contributes the largest fraction to total DTT activity, followed by LDGV and HDDV (45%, 20% and 14%, respectively). High correlations between DTT activity and LDGV and BURN (R = 0.54 and R = 0.61, respectively) are consistent with these results. Biogenicly-derived secondary organic carbon and secondary sulfate species are not found to be significantly DTT active. The regression models developed for each site are applied to PM2.5 source impacts from 1998-2010 to estimate historical DTT activity of ambient PM2.5 for use in health studies. Epidemiologic analyses in the 5-county Atlanta area find significant associations between estimated DTT activity and emergency department visits related to congestive heart failure (CHF) and asthma/wheezing attacks. In two-pollutant models with PM2.5 and DTT activity, DTT activity continues to be significantly associated with asthma/wheeze and CHF while PM2.5 does not, even though DTT activity is likely not estimated as well as PM2.5 is measured. These results support the hypothesis that PM2.5 health effects are, in part, due to oxidative stress and suggest that DTT activity is a useful indicator of aerosol-health impacts that may help explain observed associations between PM2.5 mass and health. 

Jonathan Steffens, Sue Kimbrough, Gayle Hagler, Timothy Barzyk, Vlad Isakov, Ryan Brown, Alan Powell

EPA's Geospatial Monitoring of Air Pollution (GMAP) vehicle - an all-electric vehicle measuring real-time concentrations of particulate and gaseous pollutants - was utilized to map air pollution trends near the Port of Charleston in South Carolina. High-resolution monitoring was performed along driving routes near several port terminals and rail yard facilities, recording geospatial coordinates and measurements of pollutants including black carbon, size-resolved particle count ranging from ultrafine to coarse (6 nm to 20 um), carbon monoxide, carbon dioxide, and nitrogen dioxide. Additionally, a portable meteorological station was used to characterize local meteorology. Port activity data was provided by the Port Authority of Charleston and includes counts of ships and trucks, and port service operations such as cranes and forklifts during the sampling time periods. Measurements are supplimented with modeling performed with AERMOD and RLINE in order to characterize the impact of the various terminals at the Port of Charleston on local air quality. Specifically, the data are used to determine the magnitude of the increase in local, near-port pollutant concentrations as well as the spatial extent to which concentration is elevated above background. These effects are studied in relation to a number of potentially significant factors such as 1) source emissions as characterized by port activity data, 2) time of day, 3) type of pollutant, and 4) local meteorological characteristics.

Minah Bae¹, Hyun Cheol Kim^2,3 and Soontae Kim¹

¹Ajou University, Dept. of Environmental Engineering, Suwon, Korea.

²NOAA/Air Resources Laboratory, College Park, MD

³UMD/Cooperative Institute for Climate an Satellites, College Park, MD

Weekly variations of surface nitrogen oxides (NOx) and Ozone in the Seoul Metropolitan Area (SMA), Korea, during 2001-2014 are investigated using surface measurements, satellite observations and regional air quality model. In many urban areas, mobile emissions of NOx are lower on weekends due to less traffic on the roads, which often causes a higher surface ozone level during weekends, especially for NOx-saturated chemical condition. This so-called weekend-effect has an important implication to the emission regulation policy-making on balancing regulations for NOx emission and other precursors (e.g. volatile organic compounds) to maximize the efficiency of regulation to public air quality improvement. Surface measurements from National Institute of Environmental Research (NIER, Korea) and NO₂ vertical column density remote sensing from Ozone Monitoring Instrument (OMI) and Global Ozone Monitoring Experiment 2 (GOME-2) are analyzed to produce weekly signals. We further utilize Moderate Resolution Imaging Spectroradiometer (MODIS) cloud fraction information to exclude unexpected interruptions from meteorological variations since they have significant impacts on tropospheric ozone photochemistry. Preliminary results show considerable changes in weekly cycles in ozone and its precursors with enhanced weekend-to-weekday differences in recent years compared to early 2000, suggesting rapid changes in chemical environment in the SMA including the changes of life style (e.g. Five-day workweek policy and eventual traffic patterns). We suggest that better understanding of chemical environment and its yearly change can be crucial on the planning of efficient emission regulation policy.

T. Clune, NASA Goddard Space Flight Center, MC 610.8, Greenbelt, MD 20771, Megan R. Damon, NASA Goddard Space Flight Center and Science Systems and Applications, Inc., MC 606, Greenbelt, MD 20771 and, George Delic(*), HiPERiSM Consulting, LLC, P.O. Box 569, Chapel Hill, NC 27514.

This is a report on performance enhancements for the Global Modeling Initiative (GMI) chemistry-transport model (CTM) that add new levels of parallelism and replace the legacy algorithm in the Gear sparse matrix gas-phase chemistry solver. The role of GMI modeling is to study and predict the chemical interaction, atmospheric transport, and deposition of airborne pollutants on global scales. The GMI model implements multiple science modules that describe various physical processes such as advection, diffusion, photolysis, aqueous chemistry and cloud dynamics, gas-phase chemistry, etc. The gas phase chemistry solver used in the CTM is one of the most computationally intensive modules. Gas-phase solvers implement algorithms for integration of a stiff system of ordinary differential equations (ODE), with sparse Jacobians, to describe production and loss of chemical species in reaction mechanisms. In this report the Gear algorithm is used for the system of ODEs. To improve performance a new version makes two important changes in the standard methodology. The first change replaces the sparse matrix solver used for chemical species concentrations. The second modification integrates the new solver into the transit over grid cells so that separate cells in the grid domain are distributed to different threads. This report describes how the sparse solver (FSparse) replaces the legacy JSparse solver method based on the work of Jacobson and Turco [1]. The FSparse solver is a Fortran implementation of Gaussian elimination procedures from the CSparse library of Davis [2]. As a result, the revised Gear solver adds both instruction level parallelism and thread level parallelism to the existing distributed memory (message passing) level in the GMI. In response to direction from the Advanced Software Technology Group at NASA GSFC this modification of the GMI model code has been tested on AMD and Intel commodity processors, as well as multiple Intel Phi Many Integrated Core (MIC) devices. This overview of methodology and results will be at a level suitable for modeling scientists interested in capabilities of new computer architectures.

(*) Work effort performed as Subcontractor/Consultant under a Task Order issued by the National Aeronautics and Space Administration, Goddard Space Flight Center.

[1] M. Jacobson and R.P. Turco (1994), Atmos. Environ. 28, 273-284

[2] T.A. Davis, Direct Methods for Sparse Linear Systems, SIAM, Philadelphia, 2006.

CiaoKai Liang, Raquel A. Silva, J. Jason West, HTAP modelers

Numerous modeling studies indicate that emissions from one continent influence air quality over others. Reducing air pollutant emissions from one continent can therefore benefit air quality and health on multiple continents. Several studies have suggested that using output from a multimodel ensemble leads to robust estimates of air pollutant concentrations and allows characterization of uncertainty due to intermodel differences. In this study, we estimate the impacts of the intercontinental transport of ozone (O₃) and fine particulate matter (PM_2.5) on human mortality by using an ensemble of global chemical transport models coordinated by the Task Force on Hemispheric Transport of Air Pollution (TF HTAP). We use simulations of 20% reductions of all anthropogenic emissions from the North America, Europe, South Asia, East Asia, Middle East, and Russia/Belarus/Ukraine regions to calculate their impact on premature mortality within that region and elsewhere in the world. To better understand the impact of potential control strategies, we also use three major sector-based perturbations: Power and Industry, Ground Transport, and Residential. Following previous studies, human premature mortality resulting from each perturbation scenario is calculated using a health impact function based on a log-linear model for O₃ and an integrated exposure response model for PM_2.5 to estimate relative risk. The exposed population (adults aged 25 and over) is obtained from the LandScan 2011 Global Population Dataset. Baseline mortality rates for chronic respiratory disease, ischemic heart disease, cerebrovascular disease, chronic obstructive pulmonary disease, and lung cancer are estimated from the GBD 2010 country-level mortality dataset for population aged 25 and over. We also conduct several sensitivity analyses by considering different parameters, the shape of the exposure response model, and a low/high concentration threshold. Our results allow us to identify the contributions of emissions from different source regions and sectors to air pollutant concentrations and to premature human mortality in each region.

Yu Morino¹, Kayo Ueda², Akinori Takami¹, and Tatsuya Nagashima¹

(¹National Institute for Environmental Studies, ²Kyoto University)

Chemical transport models are a useful tool to evaluate health effects and source contributions of PM_2.5 in the atmosphere. However, model results generally include large uncertainties, because of problems with their input data (e.g., meteorological, boundary, and emissions data), the parameterization of each process, and missing science elements. Thus, evaluations and improvements of model performance are critical for model applications. Recently, we evaluated model performance of PM_2.5 chemical compounds using simultaneous measurement data over Japan. It was found that concentrations of organic aerosol were better reproduced by a model with an AERO6-VBS module than that with an AERO6 module. In addition, concentrations of aerosol nitrate were better reproduced by a model with dry-deposition velocities of nitric acid and ammonia enhanced by a factor of five, as was done in a previous study.

In this study, we evaluated sensitivities of simulated PM_2.5 health effects and source contributions to these model setups. As compared to the standard simulation, simulated excess mortality due to organic aerosol was higher by a factor of two and excess mortality due to nitrate aerosol was lower by a factor of three in the improved simulation. Differences in PM_2.5 source contributions to the model setups are also discussed.

Quang-Hung Le

Cumulus convection plays a central role in most of the interactions and the representation of cumulus convection, generally called cumulus parameterization, has almost always been at the core of our efforts to numerically model the atmosphere. The new Weather Research and Forecasting model (WRF), like other numerical models, can make use of several cumulus parameterization schemes (CPS). This study focuses on how different CPSs in WRF model simulate Tropical Cyclone (TC) convection. A simulation of Typhoon Fanapi (2010) is used to better understand the physical processes of interaction in TCs. This TC is recorded creating a high precipitation number in the south of Taiwan. Because clouds and their associated physical processes strongly influencing the couplings between the atmosphere and oceans (or ground) through modifications of radiation and planetary boundary layer (PBL) processes, a further investigate by applying new data set of land use (LU) and roughness length (Zo) show that they can improve the location and intensity of convection in this particular case.

Jian Sun, Joshua S. Fu, Kan Huang, Yang Gao

We propose to use Environmental Benefits Mapping and Analysis Program (BenMAP-CE) to evaluate the PM_2.5- and ozone-related mortality at present (2000s) and in the future (2050s) over the continental United States. Atmospheric chemical fields are simulated by WRF/CMAQ (horizontal resolution: 12km 12km), applying the dynamical downscaling technique from global climate-chemistry model under the Representative Concentration Pathways scenario (RCP 8.5). Future air quality results predict that the annual mean PM_2.5 concentration will decrease nationwide, especially in the eastern US and west coast. However, the ozone concentration is projected to decrease in the Eastern US but increase in the Western US. Future mortality is assessed under two scenarios (1) holding future population and baseline incidence rate at the present level and (2) using the projected baseline incidence rate and population in 2050. For PM_2.5, the entire continental US presents a decreasing trend of PM_2.5-related mortality by the 2050s in Scenario (1), primarily resulting from the emissions reduction. While in Scenario (2), almost half of the continental states show a rising tendency of PM_2.5-related mortality, due to the dominant influence of population growth. In particular, the highest PM_2.5-related deaths and the biggest discrepancy between present and future PM_2.5-related deaths will both occur in California in 2050s. For the ozone-related premature mortality, the simulation shows nation-wide rising tendency in 2050s under both two scenarios, mainly due to the projected increase of ozone concentration and population.

Yasar Burak Oztaner^a, Luca Pozzoli^a, Tayfun Kindap^a, Amir Hakami^b, Alper Unal^a

^aIstanbul technical University, Eurasia Institute of Earth Sciences, Deparment of Climate and Marine Science, Istanbul, Turkey

^bCarleton University, Department of Civil and Enviromental Engineering, Ottowa, Ontario, Canada

Surface ozone is a secondary air pollutant that is also a short-term climate forcer. Long-term exposure to high concentration of ozone causes significant damage to plant and crops. A UNECE study showed that ozone pollution reduced annual yield by 3 -16 % with an impact of 14-26 billion dollars (UNECE, 2010). As the ozone formation in the atmosphere relies on meteorological and chemical conditions, the spatial and temporal variation is significant. Air quality modeling is one of the most widely used methods to understand this variability. Adjoint sensitivity is one of the most efficient methods that aims to answer how emissions from specific source location and times can contribute to various ozone metrics such as human health, crop yield, etc.

The objective of this study is to determine sensitivity of crop yield loss to reduction in ozone precursor emissions. For this purpose, we used CMAQ-Adjoint to evaluate the ozone exposure on crop yield over Europe for the growing season (i.e., May-Aug) in 2014. We investigated the relative impacts of emissions to wheat production using two ozone metrics: W126, which is cumulative exposure index over a period of months; and AOT40, which is accumulated dose of ozone over a threshold of 40 ppb. This is the first time where adjoint method is used over Europe to estimate crop yield loss sensitivity. This paper will present the findings of different ozone metrics and discuss the adjoint sensitivity analysis results.

Yuqiang Zhang¹, J. Jason West¹

¹ Environmental Sciences and Engineering Department, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599

Global anthropogenic emissions of ozone (O₃) precursors (methane (CH₄), nitrogen oxides (NO_x), non-methane volatile organic compounds (NMVOCs), and carbon monoxide (CO)) have increased since 1980. These increases are greatest in developing countries, such as China and India, while emissions have generally decreased in North America and Europe. Consequently, emissions have shifted southwards. Modeling studies have shown that the tropospheric O₃ burden and resulting radiative forcing are more sensitive to emission changes in the tropics and Southern Hemisphere than in other regions. Here we investigate the influence of global emission changes from 1980 to 2010 on the tropospheric ozone burden and surface ozone, using the CAM-Chem global chemical transport model. We separate the influence of the change in the spatial distributions of global emissions from that of the change in the emission magnitude. The influence of the change in global CH₄ concentration is also investigated.

Results show that the global tropospheric O₃ burden has increased by an estimated 28.12 Tg from 1980 to 2010, neglecting possible effects of climate change, with the largest increases over 30 ° S-30 ° N (17.93 Tg). Of the total tropospheric O₃ burden change, the influence of the change in the spatial distributions of global anthropogenic emissions contributes 16.39 Tg, nearly double the effect of the change in emission magnitude (8.59 Tg), and also greater than the effect of the global CH₄ concentration change (7.48 Tg) . The three-month O₃ season maximum daily 8-hr average surface O₃ has decreased over the U.S. and Europe, and increased over East and South Asia from 1980 to 2010, mainly because of the changes in the spatial distributions of emissions . For the annual zonal average O₃, significant increases from 1980 to 2010 are modeled in the upper troposphere (500 to 150 hPa) between 15° N and 40° N. Upper tropospheric O₃ also increases over the U.S. and Europe, despite large emissions reductions in these regions. These upper tropospheric changes result from the change in the spatial distributions of emissions, with little from contributions from the changes in emissions magnitude or methane, and reflect the intercontinental transport of O₃ and its precursors. We conclude that the changing spatial distribution of emissions has been the most important influence on global ozone, more important even than the change in the global emission magnitude.

Doug Boyer, Weining Zhao

Most photochemical model performance evalutions are static and many programs that produce the graphical output are difficult to adjust. A unique, interactive, and intuitive web-based model performance evaluation tool is presented using Google Maps with CAMx photochemical modeling and surface observation overlays. Animations of the hourly model results and surface observations along with radar and infrared satellite layers will be shown that help characterize model errors. Using the tool, observed surface ozone increases in proximity to passing thunderstorms will also be presented.

A WRF-SMOKE-CMAQ modeling system was used to study Volatile Organic Compound (VOC) emissions and their impact on surface VOC and ozone concentrations in southeast Texas during September 2013. The model was evaluated against the ground-level Automated Gas Chromatograph (Auto-GC) measurement data from the Texas Commission on Environmental Quality (TCEQ). The comparisons indicated that the model over-predicted benzene, ethylene, toluene and xylene, while under-predicting isoprene and ethane. The VOC emissions were adjusted using the ratios of simulated/observed surface concentrations. This involved enhancing emissions for isoprene and ethane and reducing emissions for ethylene, benzene, toluene and xylene. The modifications improved model performance of each VOC species substantially, while the change in ozone was marginal. The variability of the National Aeronautics and Space Administration (NASA)'s Ozone Monitoring Instrument (OMI) formaldehyde columns over Southeast Texas matched better with simulated formaldehyde columns from CMAQ with the adjusted emissions during the month. A detailed process analysis indicated that for the VOC species, emission was the main positive contribution throughout all day, while the daytime key negative contribution came from vertical diffusion due to active mixing. During nighttime, horizontal and vertical advection were the main processes regulating species concentrations. The magnitude of the contribution of each process changed correspondingly as the emissions were adjusted. For ozone, a significant consumption of ozone by chemical reaction during most times of the days suggested that the Houston industrial region might be treated as a NOx saturate (i.e. VOC-sensitive) region. Adjusting VOC emissions produced a certain amount of ozone during late morning to early afternoon, but this increase was compensated by change in magnitude of those of meteorological processes, which could be one possible explanation for the marginal simulated ozone change. The impact of increased VOC emissions on ozone formation depended on the background distribution of NOx emission and concentrations of OH and HO2 radicals. Adjusting VOC amounts to the observed concentrations shifted the ozone hotspots outside the industrial/urban region and enhanced the peaked ozone in the outflow region with consistent southerly/southeasterly winds during the afternoon time (1-5pm). This study helps in the understanding of these processes which are critical to constrain high peak ozone values in the outflow regions.

A. Chtcherbakov, R. Bloxam, L. Huang, S. Wong, Y. Hall

Ontario Ministry of Environment and Climate Change

Analysis of spatial distribution of ozone and PM2.5 based only on observed data or modelled results could be a challenge. Ozone and PM2.5 observed data is available from Ontario's network of 40 ambient air monitoring stations across the province (National Air Pollution Surveillance Program (NAPS)) and Environment Canada's network of 33 Canadian Air and Precipitation Monitoring Network (CAPMoN) sites (4 of them are in Ontario). A majority of the stations are located in more populated cities in southern Ontario resulting in gaps in monitoring data and thus a lack of reliable data for interpolation, especially in the northern part of the province. Modelled concentrations are available for the entire province, but modelling results are often biased due to a number of model uncertainties and assumptions.

This study describes the approach of merging modelled and observed data in a spatial analysis using a krigging technique, which takes advantage of both observed data and modelled results. Different krigging capabilities were used (i.e. ellipse of influence, variogram models, exclusion areas with a lack of observed data, application of observed-modelled ratios and/or differences for krigging, spline interpolation, etc.). This approach was applied to several metrics relating to air quality in Ontario (i.e. annual PM2.5 concentration, 4-th highest 8-hour daily maximum ozone concentration, high end of ozone concentration distribution such as the averaged 98-th percentile and the averaged ozone concentration above certain threshold values (i.e. 70 and/or 65 ppb)). Maps showing spatial distribution of these metrics across the province were created.

This merging technique has been implemented on observational datasets from the monitoring networks for the entire year of 2010, coupled with corresponding model outputs from Community Multi-scale Air Quality (CMAQ) modelling system runs with relatively coarse resolution (36 km). Results are discussed and improvements in spatial pattern of the air quality metrics over Ontario are shown.

Garnet Erdakos¹, Kenneth Craig¹, Jennifer DeWinter¹, Stephen Reid¹

¹Sonoma Technology, Inc., Petaluma, CA

The MetDat meteorological database system was developed in collaboration with the U.S. Environmental Protection Agency (EPA) and includes a synthesis of raw and computed meteorological parameters from surface and upper air observations, model reanalyses, and trajectory calculations.

The MetDat system consists of streamlined computer scripts and programs for acquisition and processing of historical observed and predicted meteorological data to produce a database of more than 100 meteorological parameters. Input data for MetDat include National Climatic Data Center (NCDC) Integrated Surface Hourly (ISH) and Integrated Global Radiosonde Archive (IGRA) quality-controlled surface and upper-air meteorological observation data sets, and the regional, long-term (1979 to present), dynamically consistent, high-resolution, three-dimensional National Center for Environmental Prediction (NCEP) North American Regional Reanalysis (NAYY) data set. Using extraction algorithms, parameterizations, and the Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model, MetDat produces observed and derived meteorological parameter outputs for either short- (1, 3, 4, 6, and 12-hour) or long-term (daily and monthly) time periods at selected monitoring sites. This process can be executed from a UNIX/LINUX environment command line for any historical data period.

EPA has used MetDat to develop statistical models for adjusting air quality trends to account for year-to-year meteorological variability, and the system has recently been made available to state and local air quality agencies through an AirNow Tech data portal to support analyses such as the development of statistical regression equations for ozone forecasting (https://www.airnowtech.org/metdat/metdathome.cfm).

Brian Eder, Rob Gilliam, George Pouliot

AMAD/NERL/US EPA

Historically, the U.S. EPA has evaluated retrospective, often annual length, simulations of CMAQ, summarizing the performance using monthly or seasonal statistical summaries. While informative, such an approach often masks finer scale temporal (i.e. diurnal to weekly) and spatial (meso to synoptic) variability that greatly impacts the atmosphere and hence air quality. In order to ensure CMAQ's state-of-the-science status as well as its ability to address emerging Agency needs, it is crucial that newer evaluation approaches are utilized that will allow for a more rapid testing and hence more efficient evolution of the modeling system's science. Accordingly, the AMAD began running CMAQ continuously and in near real-time in 2013, allowing for immediate and ongoing analysis at finer spatial and temporal scales, thereby facilitating model evaluation of PM_2.5 mass and O₃ concentration and hence improve model performance. Following the protocol established when the Division was involved with the development and implementation of the National Air Quality Forecast Capability (NAQFC), results from the simulations have been immediately examined and discussed by Division scientists in bi-weekly meetings while antecedent meteorological and air quality conditions remain familiar. Advantages of evaluating CAMQ in near real-time are numerous and have led to identification of deficiencies related to: 1) boundary conditions, 2) episodic emission events, both natural (wildfires, Saharan dust) and anthropogenic (residential wood burning); 3) evolution of the nocturnal boundary layer and 4) unrealistic lake temperatures, among others.

Lucas Henneman, Cong Liu, Cesunnica Ivey, Yongtao Hu, Armistead Russell

Current implementations of first-principles regional air quality models include a feature that directly calculates the sensitivity of various output parameters to input parameters across detailed spatial and temporal scales. For example, CMAQ-DDM estimates sensitivities of various pollutants to emission sources using many of the same terms used in the continuity equations, making it an efficient estimator. However, the estimation of sensitivities is subject to the same propagation of uncertainties as other model outputs, including biases in emissions and meteorological inputs, numerical error, and misspecification of atmospheric processes. Statistical models that are trained using combinations of observed and modeled inputs provide an evaluation tool for sensitivities estimated by firs principles models. While statistical models are subject to their own sources of bias and uncertainty, benefits include the ability to use observed data and their comparatively less intense computational effort. This work will compare results from CMAQ-DDM and statistical models that link emissions near Atlanta, Georgia with observed concentrations in city center. Differences between the models on various scales (temporal, observed concentration, emissions) will be traced to aspects of each model that may lead to biases. Outcomes have implications in the interpretability of results of both models.

Syuichi Itahashi¹

Hiroshi Hayami¹

Itsushi Uno²

1 Environmental Science Research Laboratory, Central Research Institute of Electric Power Industry, Abiko, Chiba, Japan

2 Research Institute for Applied Mechanics, Kyushu University, Kasuga, Fukuoka, Japan

In this presentation, we introduce the investigated results of the emission source contributions for tropospheric ozone (O₃) over East Asia. Source contributions have been comprehensively evaluated with the HDDM (higher-order decoupled direct method) of sensitivity analysis and the OSAT (ozone source apportionment technology) of mass balance analysis in CAMx (Comprehensive Air Quality Model with eXtensions) model. The response of O₃ to emissions reductions at various levels in mainland China, Korea, and Japan were examined. The performance of HDDM and OSAT were compared with results calculated by the traditional approach of BFM (brute force method) where one model parameter are varied one-at-a-time. Abilities were assessed at three receptor sites in Japan where experienced severe pollution event in May 2009. For emissions from China, HDDM assessed O₃ response with only up to 3 ppbv bias (a relative error of 4.5%) even in 50% reduction, but failed to assess more extreme reduction. OSAT was reasonably available at 100% reduction, with a -4 ppbv (-7%) bias, but less accurate at moderate ranges of reduction (50-70%). For emissions from Korea and Japan, HDDM remarkably captured the nonlinear response at all receptor sites and at all reduction levels to within 1% in all but one case; however, OSAT had increasing bias with increasing reduction of emissions. One possible reason for this is that OSAT does not account for NO titration; to address for this, a term of potential ozone (PO; O₃ and NO₂ together) was introduced. Application of PO instead of O₃ improved performance of OSAT, especially for emissions reductions from Korea and Japan.

Modeling and evaluating trace metal PM2.5 concentration estimates is challenging and has inherent uncertainties, such as determining source contribution estimates and comparing results with observed concentrations. In this work, a novel data assimilation approach is used to generate new PM2.5 source profiles for 20 source categories. The approach takes into account measurement concentrations and uncertainties, as well as source fingerprint uncertainties. After generating new profile estimates, new results are incorporated into the SMOKE modeling system, and new emissions are calculated. New emissions are used as inputs for a one year CMAQ simulation at 36-km resolution over continental U.S. CMAQ simulations are evaluated by comparison to observed trace metal concentrations. Improving modeled trace metal concentrations strengthens source-tracer analyses and provides more reliable data for air toxicity studies related to human health.

Prakash Karamchandani¹, Biswanath Chowdhury², Bart Brashers³, Lynsey Parker¹, Aaron Kaulfus⁴, Eladio Knipping⁵

¹Ramboll Environ, 773 San Marin Drive, Suite 2115, Novato, CA 94998

²Sage - an Xator Company, 15 Roszel Rd. Suite 102, Princeton, NJ 08540

³Ramboll Environ, 19020 33^rd Ave. W., Suite 310, Lynnwood, WA 98036

⁴Southern Company, 600 North 18th Street, Birmingham, AL 35291

⁵Electric Power Research Institute, 1325 G Street NW Suite 1080, Washington, DC 20005

SCICHEM 3.0 is the latest improved version of SCICHEM, a state-of-the-science non-steady-state puff model with complete chemistry treatment, including gas-phase chemistry, aerosol chemistry, and cloud chemistry modules that are based on modules used in CMAQ. The model represents a continuous plume by a series of Gaussian puffs governed by ordinary differential equations and puff transport is determined by atmospheric turbulence in addition to the mean winds and the terms representing puff interactions handle nonlinearities introduced by turbulent effects and concentration variations. SCICHEM 3.0 was developed to determine the short-range and long-range impacts of individual sources and source complexes on downwind concentrations of primary and secondary pollutants, including ozone and fine particulate matter (PM_2.5). This paper describes recent improvements to the model and presents results from extensive testing of the model as well as model performance evaluation results. The testing includes annual simulations over different regions of the United States and a variety of emission source types. The evaluations are conducted using both recent and historical plume measurement databases.

Chris Kite and Jim Smith, Texas Commission on Environmental Quality

On December 3, 2014, the Environmental Protection Agency (EPA) released "Draft Modeling Guidance for Demonstrating Attainment of Air Quality Goals for Ozone, PM2.5, and Regional Haze." From 2012 through June 2015, the Texas Commission on Environmental Quality (TCEQ) was preparing an attainment demonstration for the Dallas-Fort Worth (DFW) ozone nonattainment area, one of only two large urban areas classified as Moderate under the 2008 ozone National Ambient Air Quality Standard (NAAQS) of 75 parts per billion averaged over eight hours. This attainment demonstration modeled a 2018 future year, and provided an ideal opportunity to compare projected future design values calculated under the "all days" test from the corresponding 2007 guidance with those calculated under the "top ten days" test from the 2014 draft guidance. This paper compares calculation methodologies between versions of the guidance, and also explores employing model performance measures to filter the days included in the top ten days test. When performing anthropogenic precursor culpability assessment (APCA) analyses for attainment years, the TCEQ typically averages the APCA output for the episode days included in the relative response factor (YYF) test, and indexes the results for each monitor with the future ozone design value. When the top ten days YYF test is used instead of the all days test, preliminary APCA analyses are showing a relatively higher contribution of ozone formation from local anthropogenic emission source categories versus regional ones.

Yuanwei MA, Dezhong WANG

School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai, China

Achieving information of source terms is an import task in emergency response during a nuclear accident. Affected by large errors of atmospheric model, the reverse modeling of atmospheric dispersion faces great challenges. The traditional scaling method adopted in Fukushima nuclear accident is affected by deviation of the observed and predicted values. To cope with the deviations, an assimilation model based on Ensemble Kalman filter (EnKF) was established with consideration of model errors and observation errors. It was validated by 7 groups of tracer experiments. The result shows that the EnKF model could estimate the release rate of source term successfully. This scheme was applied to emergency response system of Qinshan nuclear power plant.

Christopher Marciano, Kiran Alapaty, Jerry Herwehe, Gary Lackmann, Wei Wang

Heavy precipitation events are important drivers of the water cycle and have significant socioeconomic impacts due to the potential for flooding. Air and water quality are also impacted by heavy precipitation through the removal of hygroscopic pollutants from the atmosphere and nutrient loading to water reservoirs. Given their far-reaching importance, it is a great benefit to the society to increase the accuracy of forecasting of heavy precipitation events. Previous work has demonstrated that employing a new scale-dependent Kain-Fritsch cumulus parameterization scheme improves the representation of precipitation intensity and distribution in high-resolution, mesoscale numerical weather prediction models. This new multi-scale Kain-Fritsch (MSKF) scheme introduces several new modifications in order to improve model performance at higher resolutions including cloud-radiative interactions with the convective parameterization, a dynamic adjustment timescale, and scale-dependent stabilization and entrainment methods.

Performance of the MSKF scheme for heavy precipitation events is assessed here using the Weather Research and Forecasting (WRF) model. Events of interest are objectively chosen based on event type, duration and intensity using a pre-existing database of heavy precipitation events. In order to test the sensitivity of the MSKF scheme to horizontal resolution, simulations are performed with horizontal grid spacings of 36, 12, 4 and 1-km. Performance of the MSKF at each horizontal grid spacing is assessed by comparing the simulations with Multisensor Precipitation Estimates (MPE) for precipitation and quality controlled local climate data (QCLCD) for surface parameters such as temperature. Similar analysis will be performed for simulations of the same events implementing explicit convection at the 4 and 1-km grid spacings. Comparison of results using explicit convection and convection parameterized with the MSKF scheme will yield insight on the added benefit of the MSKF for heavy precipitation forecasting at varying horizontal resolutions.

Chandrasekar Radhakrishnan (chanradh@in.ibm.com)

Rashmi Mittal ( rashmitt@in.ibm.com)

Thomas George (thomasgeorge@in.ibm.com)

IBM Research, New Delhi, India

Li Pan ^1,2, Hyun Cheol Kim ^1,2, Pius Lee ¹, YouHua Tang^1,2, Daniel Tong ^1,2, Ivanka Stajner³, and Weiwei Chen^1,4,

¹ NOAA/Air Resources Laboratory, College Park, MD

² UMD/Cooperative Institute for Climate and Satellites, College Park, MD

³ NOAA/NWS Office of Science and Technology Integration, Silver Spring, MD

⁴ Northeast Institutes of Geography and Agroecology, Chinese Academy of Sciences, Changchun, China

In order to quantitatively describe wildfire emissions and their contributions to air quality, HMS-BlueSky-SMOKE, a wildfire emissions calculation algorithm by USDA Forest Service, was used in CMAQ simulations during SENEX field experiment, which includes Hazard Mapping System (HMS) - wildfire detecting algorithm, BlueSky-wildfire emission estimation algorithm, and SMOKE-wildfire plume rise algorithm. Due to the high uncertainties of wildfire emissions, the first step of quantitatively evaluating fire emissions focused on fire signal capture capacity of this system in the fire events. In this study, we conducted two CMAQ scenario tests with and without fire emissions during the time period of SENEX field experiments. The differences between the two model simulations demonstrate the impact of fires in the former CMAQ model simulation. At same time, the observations from various sources, which include ground observations (Interagency Monitoring of Protected Visual Environments (IMPROVE)), satellite retrievals (Automated Smoke Detection and Tracking Algorithm (ASDTA) smoke AOD) and aircraft measurements (SENEX campaign), are evaluated using several criteria to determine which observations are affected by fire impacts. Fire signals detected in observations and fire signals simulated in the model are directly compared to each other to evaluate this wildfire detection in observations and wildfire simulation in CMAQ. Agreements and discrepancies between observations and model simulations will be highlighted in case studies.

George Pouliot

Atmospheric Modeling Division, National Exposure Research Laboratory, Environmental Protection Agency, Research Triangle Park, NC 27711

pouliot.george@epa.gov

Ana Rappold

Environmental Public Health Division, National Health and Environmental Effects Research Laboratory, Environmental Protection Agency, Research Triangle Park, NC 27711

rappold.ana@epa.gov

Jeanette Reyes

Environmental Public Health Division, National Health and Environmental Effects Research Laboratory, National Exposure Research Laboratory, Environmental Protection Agency, Research Triangle Park, NC 27711

reyes.jeanette@epa.gov

Kristen Foley

Atmospheric Modeling Division, National Exposure Research Laboratory, Environmental Protection Agency, Research Triangle Park, NC 27711

foley.kristen@epa.gov

Tom Pierce

Atmospheric Modeling Division, National Exposure Research Laboratory, Environmental Protection Agency, Research Triangle Park, NC 27711

pierce.tom@epa.gov

Biomass burning has been identified as an important contributor to the degradation of air quality because of its impact on ozone and particulate matter. Two components of the biomass burning inventory, wildfires and prescribed fires are routinely estimated in the national emissions inventory. However, there is a large amount of uncertainty in the development of these emission inventory sectors. We have completed a 5 year set of CMAQ model simulations (2008-2012) in which we have simulated regional air quality with and without the wildfire and prescribed fire inventory. This paper examines the regional impact of wildfires and prescribed fires and total PM2.5 burden on the atmosphere. We will examine CMAQ model performance over regions with significant PM2.5 contribution from prescribed fires and wildfires. We will also examine the PM speciation including EC and OC at standard monitoring locations where fires were detected and compare with modeling results.

Jeanette Reyes, MS

PhD Student

University of North Carolina at Chapel Hill

reyesjm@live.unc.edu

Marc Serre, PhD

Associate Professor

University of North Carolina at Chapel Hill

marc_serre@unc.edu

Will Vizuete, PhD

Associate Professor

University of North Carolina at Chapel Hill

airquality@unc.edu

The EPA employs a vast monitoring network to measure ambient PM2.5 concentrations across the United States with one of its goals being to quantify exposure within the population. However, there are several areas of the country with sparse monitoring spatially and temporally. One means to fill in these monitoring gaps is to use PM2.5 modeled estimates from chemical transport models specifically the Community Multi-scale Air Quality (CMAQ) model. CMAQ is able to provide complete spatial coverage but is subject to systematic bias and inherent variability due to model uncertainty. Due to the deterministic nature of CMAQ, often these uncertainties are not quantified. Much effort is employed to quantify the efficacy of these models through different metrics of model performance. Currently evaluation is specific to only locations with observed data. Multiyear studies across the United States are challenging because the bias and model performance of CMAQ are not uniform over such large space/time domains. Bias changes regionally and temporally. Because of the complex mix of species that constitute PM2.5, CMAQ bias is also a function of increasing PM2.5 concentration. To address this issue we introduce a model performance evaluation for PM2.5 CMAQ that is regionalized and non-linear. This model performance evaluation leads to a bias quantification for each CMAQ grid. This leads to areas and time periods of bias being better qualified. The regionalized bias correction approach is non-linear and is therefore more flexible at characterizing model performance than approaches that rely on linearity assumptions and assume homoscedasticity of CMAQ predictions errors. Total error is then divided in terms of being either systematic or random. This allows modelers to have a better understanding of where areas of model improvement are best allocated.

Shantha Daniel and Jim Smith, Texas Commission on Environmental Quality

Ou Nopmongcol, Ramboll Environ

Numerous comparisons between the two photochemical modeling systems, Community Model for Air Quality (CMAQ) and the Comprehensive Air Quality Model with Extensions (CAMx) have been performed over the years. Because of the varying input requirements for these modeling systems, many, if not most of these comparisons have involved meteorological, emissions, and/or boundary conditions that are not consistent, making direct comparisons between the chemical transport models impossible. Such comparisons have been especially problematic when using Texas' emissions inventory, since the Texas Commission on Environmental Quality (TCEQ) uses a highly-customized and enhanced version of the Emissions Processing System, version 3 (EPS-3), while most users of CMAQ process their emissions through the Sparse Matrix Operator Kernel Emissions (SMOKE) system. This fundamental difference in emissions processing has also led many modeling groups to rely on emissions that can readily be processed through SMOKE, such as the 2011 National Emissions Inventory (NEI), rather than the more highly-resolved (spatially, temporally, and chemically) emissions used by the TCEQ. In 2014 the TCEQ contracted with Ramboll ENVIRON to develop software, CAMx2CMAQ, which converts CAMx model-ready emissions and boundary conditions directly into CMAQ input format. This paper describes the software, presents the results of comparisons of CMAQ and CAMx for Texas using the CMAx2CMAQ converter, and discusses additional issues which need to be addressed to better align the inputs to the two modeling systems.

Tammy M. Thompson,¹ Michael G. Barna,² and Bret A. Schichtel²

¹Cooperative Institute for Research in the Atmosphere, Colorado State University, Fort Collins, CO 80523-1375, USA

²National Park Service, Air Resources Division, Lakewood, CO, USA

Human activity, including fossil fuel combustion and agriculture has greatly increased the amount of reactive nitrogen (RN) in the atmosphere and its subsequent deposition to land. Increases in deposition of RN compounds can adversely affect sensitive ecosystems and is a growing problem in many natural areas. Grand Teton National Park is particularly sensitive to increased RN deposition and there is evidence that accumulating RN may have crossed sensitive critical load thresholds. However, additional information is needed to better understand the total RN deposited in the park and the sources responsible for this RN. To address these information gaps, the National Park Service in conjunction with Colorado State University researchers and assistance from the Forest Service conducted the Grand Teton Reactive Nitrogen Deposition Study (GrandTReNDS) involving spatially and temporally detailed measurements of RN during spring/summer 2011. In this work it was found that during summer months at the high elevation site Grand Targhee, 62% of the nitrogen deposition was due to reduced nitrogen, about equally split between dry and wet deposition, oxidized nitrogen accounted for 27% and the remaining was wet deposited organic nitrogen. Subsequent monitoring of ammonia gas indicates that 2011 ammonia concentrations were atypically low and deposition of reduced N is likely higher in other years. An important next step to GrandTReNDS is the use of chemical transport models CTMs) to estimate source contributions to RN in the park. Given the large contribution of reduced nitrogen species to total nitrogen deposition in the park, understanding and properly characterizing ammonia in CTMs is critical to estimating the total nitrogen deposition. As part of the modeling effort, the influence of ammonia deposition parameterization will be explored. This will include a comparison of CAMx model uni-directional model to CMAQ bi-direction and uni-directional model simulations and evaluation of each model's performance compared to GrandTReNDS and other datasets.

Yadong Xu, William Vizuete, Marc L. Serre

Bayesian Maximum Entropy Integration of Ozone Observations and Model Predictions at Multiple Time Scales

Yadong Xu, William Vizuete, Marc L. Serre

University of North Carolina, Chapel Hill, NC, USA

Abstract:

Objective: We developed and compared the predictive capacity of two upscaling methods: USM1 (data aggregation from hourly to daily followed by BME spatiotemporal interpolation) and USM2 (perform BME spatiotemporal interpolation on hourly ozone followed by data aggregation to daily) for ambient ozone exposure estimations.

Data: The observed hourly concentrations were downloaded from the Air Quality System (AQS) maintained by the U.S. Environmental Protection Agency (EPA) and Chemical Transport Model (CTM) hourly data obtained from the Comprehensive Air Quality Model with Extensions (CAMx) annual simulation of year 2005 with 36x36km grid cell resolution for the contiguous United States and 12x12km grid cell resolution for the eastern United States.

Methods: The AQS hourly data was paired with CTM modeled hourly ozone concentrations on space/time. Then Localized bias-corrected CTM data were constructed as the soft data through a moving spatiotemporal window-air quality model performance (MSW-AMP) approach. A transformation of these data was used, which consisted in removing from the data an offset obtained using an exponential kernel smoothing of the data. The exponential kernel smoothing was set so that the offset captured the spatial variability of the data over intermediate spatial distances and intermediate time scales. A 3-term exponential/exponential/cosine space/time covariance model was used to characterize the space/time autocorrelation in the offset removed data. The BME method was then used to estimate hourly O3 at un-sampled locations using the offset removed daily observations treated as hard data and localized bias-corrected hourly CTM data treated as soft data. We perform cross-validations at multiple time scales by USM1 and USM2 to assess the estimation accuracy across space and time.

Results: Compared with the kriging methods of classical linear geostatistics, the integration of CTM soft information by the BME method can effectively increase the estimation accuracy for both hourly and daily ozone concentrations. The spatiotemporal distributions of estimation errors from USM1 and USM2 were similar, but the implementation of USM1 is associated with much lower computation burden.

Air quality and climate modeling with the Representative Concentration Pathways (RCP) scenarios provides assessments on the effects of emissions on air quality and its trends as well as risks from climate change impacts that are needed for air pollution and climate mitigation policy-making. The use of atmospheric models with advanced physics and chemistry treatments can reduce uncertainties associated with such assessments thus potentially improve the effectiveness of the policies made based on more accurate impact assessments. Most global and regional climate modeling systems applied for past climate studies, however, do not include detailed chemistry and aerosol treatments as well as various chemistry-climate feedbacks that are necessary for accurate assessments.
In this work, an online-coupled downscaling model system with advanced chemistry and aerosol treatments is applied to the continental U.S. to simulate current (2001-2010) and future (2046-2055) atmosphere under two different RCP scenarios: RCP4.5 and RCP8.5. The downscaling model systems consist of two models: the modified Community Earth System Model (CESM) at and the modified online-coupled Weather, Research and Forecasting model with Chemistry (WRF/Chem) v3.6.1 which is applied at a horizontal resolution of 36-km to downscale CESM simulations at 0.9 1.25 . The main objectives are to evaluate the decadal performance of WRF/Chem in terms of meteorology, air quality, and aerosol-cloud-radiation feedbacks; quantify and contrast the changes in predictions between current and future RCP scenarios; and assess the advantages of downscaling with WRF/Chem. Preliminary evaluation for the current decadal simulations shows that WRF/Chem performs relatively well for meteorological and radiation variables (e.g., a slight underprediction of -0.5 C for temperature at 2-m), however, it gives a large wet bias for precipitation and large underpredictions for cloud variables including cloud liquid water path, cloud droplet number concentration and cloud optical thickness. While the model has good performance for both hourly O3 and daily PM2.5 at all types of sites, different trends are found for chemical performance against different surface networks. Ozone tends to be underpredicted against the CASTNET network but overpredicted against AQS, while PM2.5 concentrations tend to be overpredicted against IMPROVE but underpredicted against STN. Compared to the RCP4.5 scenario, the RCP8.5 scenario generally projects a future atmosphere with higher temperatures, stronger downward shortwave radiation over land, which in turn drive more drastic changes in latent heat fluxes and precipitation. While RCP4.5 and RCP8.5 scenarios project similar levels of decreases in PM2.5, they predict different changes in O3 with decreases by RCP4.5 but increases by RCP8.5. The major findings from this work will provide insight into the impacts of different climate policy scenarios on future U.S. air quality, climate, and the feedbacks of chemical components to cloud, radiation, and climate.

Xinxin Zhai 1, Jim Mulholland 1, Yongtao Hu 1, Armistead Russell 1, Byeong Kim 2, Yunhee Kim 2, David D'Onofrio 3

1. Georgia Institute of Technology, 2. Georgia Environmental Protection Division, 3. Atlanta Regional Commission

Mobile source impacts are a major source of primary PM2.5. To help understand the spatial and temporal trends of the mobile source impacts, we applied both measurement-based models and emission-based models at different scales. We estimated mobile source impacts in 2011 using three methods: 1.the measurement-based Chemical Mass Balance (CMB) model; 2. the Research-Line dispersion model (RLINE); and 3.the Integrated Mobile Source Indicator model (IMSI) based on CMAQ. CMB results are available at three sites in the Atlanta metropolitan area: South DeKalb (urban), Jefferson Street (urban), and Yorkville (rural). The RLINE estimated mobile source impacts at 200 m resolution for PM2.5, CO, and NOx are based on emissions of 40641 actual links with the impact on primary PM2.5, CO, and NOx. The IMSI estimated mobile source impacts were developed at a 4km resolution using concentration fields from the chemical transportation model CMAQ after fusing the results with observations. We found that although the annual spatial pattern of the RLINE and IMSI results correspond well, there was substantial variation in the daily spatial patterns. RLINE results contain high spatial gradients, which were rescaled using regression analysis with measurement-calibrated IMSI at 4km resolution. The temporal correlations at 4-km resolution grid cells are good between the RLINE and IMSI results. This work demonstrates the limitations in RLINE model results, and suggests a correction is necessary for overestimated spatial gradients.

Xinxin Zhai 1, Poornima Sampath 1, Jim Mulholland 1, Yongtao Hu 1, Armistead Russell 1, Byeong-Uk Kim 2, Yunhee Kim 2, David D'Onofrio 3

1. Georgia Institute of Technology, 2. Georgia Environmental Protection Division, 3. Atlanta Regional Commission

The Research-Line (RLINE) is a newly developed dispersion modeling tool for estimating the impacts of line source emissions on air quality, (e.g., on-road gasoline vehicles and diesel vehicles) at very fine spatial scales. However, dispersion models can overestimate the spatial gradients of concentrations. Here we compare RLINE annual estimates of PM2.5 for the year 2011 at 200m resolution with three other datasets: measured PM2.5 concentrations at nine locations in Atlanta, measurement-based source impacts of PM2.5 by CMB at three sites in Atlanta, and mobile source impacts by Integrated Mobile Source Indicators (IMSI) based on CMAQ output at 4km resolution. The comparison of the two spatially resolved methods, IMSI and RLINE, shows good correlation, with Pearson R about 0.91, indicating that RLINE corresponds well with the spatial gradients from the Community Multiscale Air Quality Modeling (CMAQ version 4.6 with updated SOA module). The correlation of CMB estimated mobile source impacts with RLINE is good but has very high slope, suggesting that RLINE simulates sharper spatial gradients than reality. These findings consistently suggest that RLINE results captured the spatial variations, but were biased high near the sources in terms of impacts. To rescale the RLINE results to measurement levels, we developed four regression models of the annual averages at the three sites to rescale the RLINE estimates using the annual averages of RLINE results in 200m resolution based on emissions of 40641 actual links and the CMB-derived mobile source impacts, as well as, separately, the observed PM2.5 from nine impacts. Our comparison indicates that RLINE estimates are eight times higher than measurement-derived source impacts in Atlanta area, but is consistent with chemical transport model spatial predictions.

Daniel Cohan and Beata Czader

New techniques have been introduced recetly by US EPA and others for using the high-order decoupled direct method (HDDM) to predict the amounts of emission reductions needed to meet more stringent ambient ozone standards. These approaches enable more accurate extrapolation of first- and second-order sensitivity relationships to the amounts of emission reductions likely to be needed to meet challenging targets. However, the impacts of these peak-targeted controls on off-peak ozone and on particulate matter throughout the year have been less thoroughly examined.

Here, we will report a modeling study that assesses the level of emission reductions to meet the new standard relative to the emission levels reported in the 2011 National Emissions Inventory. Different potential emission reduction strategies and the heterogeneity of their spatial and temporal impacts on ozone will be evaluated. Ozone response to emissions under different meteorological conditions and different levels of biogenic emissions will be analyzed. In particular, we will show how strategies for reducing peak-period ozone affect ozone at other times of day and on days with low or moderate ozone levels.

The study will also analyze how alternate ozone approaches would affect each component of particulate matter. In particular, we will characterize the responses of particulate matter by species, region, and season to reductions in NOx and VOC emissions.

K. Wyat Appel, Sergey Napelenok, Christian Hogrefe, George Pouliot, Brian Eder, Kristen M. Foley and Shawn J. Roselle

The Community Multiscale Air Quality model is a state-of-the-science air quality model that simulates the emission, transport and fate of numerous air pollutants, including ozone and particulate matter. The Atmospheric Modeling and Analysis Division (AMAD) of the U.S. Environmental Protection Agency develops the CMAQ model and releases a new major version of the model every several years (non-major releases can occur more often), with the last major release of the model occurring in 2012. Currently, the latest version of the CMAQ model (v5.0.2) can be downloaded from the Community Modeling and Analysis System website (CMAS; https://www.cmascenter.org/). The next major release of the CMAQ model (v5.1) is scheduled for fall 2015. The AMAD will perform two annual CMAQ model simulations, one with the current publically available version of the model (CMAQv5.0.2) and the with the new version of the model (CMAQv5.1). In additional, several sensitivity runs testing the response of the new model to changes in NOX, SOX, and VOC emission changes will also be performed. The results of each model simulation will then be compared to observations and the performance of the model simulations assessed. This work will summarize the performance of each model simulation and how the performance of the new model compares to the current model.

Amanda Pappin, Morteza S. Mesbah, Amir Hakami

The health benefits of emission controls depend on atmospheric conditions conducive to secondary pollutant formation along the transport pathway from sources to receptors, and hence on emission quantities themselves. Such dependence would extend to the magnitude and shape of the concentration-response function (CRF) drawn from epidemiological studies. Recently, a non-linear and concave CRF has been suggested for mortality and long-term PM_2.5 exposure in North America. This shape of the CRF indicates a heightened sensitivity of populations at lower concentrations, implying that the benefits of reducing emissions would increase as the atmosphere becomes less polluted.

We investigate how atmospheric chemistry and the shape of the CRF influence estimates of the benefits-per-ton of NO_x and VOC emission control. We do so using adjoint sensitivity analyses in CMAQ for various forms of CRFs for chronic exposure mortality, and for various emission control scenarios for North America. We find that even with a linear CRF for O₃, benefits-per-ton increase substantially and without exception as the atmosphere becomes less polluted, due entirely to the role of atmospheric chemistry. At baseline, benefits-per-ton of emitted NO_x far exceed current abatement costs (>$75,000/ton for sources in the eastern U.S./California). Benefits-per-ton increase from there, by as much or more than 3 times with large, North American-wide emission reductions. We argue that for PM_2.5, compounding benefits-per-ton would hold true due to nonlinearity in secondary PM formation from emitted NO_x and other precursor species. A nonlinear CRF for PM_2.5 (and potentially NO₂ based on recent epidemiological evidence) would only further support this idea. We argue that unlike the traditional view of benefits-per-ton based on a linear CRF, which dictates a focus on emitters within urban areas and immediately upwind, a concave CRF also draws attention to emissions located in moderately populous or rural areas.

Our findings indicate unforeseen and long-term benefits of emissions control, suggesting that current emission controls make future abatement efforts more valuable. Assessment of recent emission trends in the U.S. indicates that we are currently at an important point on the abatement trajectory, where the benefits of abatement have increased in the past, and will do so considerably in the near future. This constitutes a new and significant deviation from a widely accepted paradigm of convexity, or diminishing returns, in air pollution economics. Consideration of the compounding nature of emission control benefits can cast abatement policies in a self-propelling, and thus an increasingly rewarding, light in the long-term. Conversely, disregarding the compounding nature of benefits could result in significant underestimation of the societal benefits of air pollution control policies.

Changhan Bae¹, Soontae Kim¹, Hyun Cheol Kim ^2,3 and Byeong-Uk Kim⁴

¹Ajou University, Dept. of Environmental Engineering, Suwon, Korea

² NOAA/Air Resources Laboratory, College Park, MD

³ UMD/Cooperative Institute for Climate and Satellites, College Park, MD

⁴ Georgia Environmental Protection Division, Atlanta, GAd

Sensitivity of customized emission inventory based on the Comprehensive Air quality Model with extensions (CAMx) Particulate matter Source Apportionment Technology (PSAT) is evaluated using operational particulate matter (PM) forecast system over northeast Asia. Among various uncertainties that current air quality forecast systems have, the lack of information on the latest emission inventory is one of the biggest problems to improve the performance of PM forecast system. To reduce emission inventory uncertainties, we attempted to enhance our base inventory using source contribution analysis results obtained with PSAT. The Integrated Multidimensional Air Quality System for Korea (IMAQS/K) is a regional PM forecast system based on the Weather Research and Forecasting (WRF)-Sparse Matrix Operator Kernel Emissions (SMOKE)-Community Multi-scale Air Quality (CMAQ) modeling framework with nested domains; Northeast Asia (27-km), South Korea (9-km), and over the Seoul Metropolitan Area (SMA, 3-km). For the base run, emission inventories from Intercontinental Chemical Transport Experiment - Phase B (INTEX-B) 2006 and Clean Air Policy Support System (CAPSS) 2007 are used for Asia and for South Korea, respectively. High PM episode during February 24 - 27, 2014 is selected. In the period average comparisons, the base run showed underestimations of PM_2.5 (~35 ug/m³, 30%) and sulfate (~14 ug/m³, 55%) and overestimation of nitrate (~12 ug/m³, 195% ) over Bulkwang super site from the AirKorea surface observation network. As the CAMx/PSAT analysis suggests that the most of sulfate concentration is contributed from sources in China, we conducted a sensitivity test by increasing Chinese SO_x emissions 3 times and decreasing Korean NOx emission by 50%. The test run with adjusted emission showed a considerable improvement in the performance of PM2.5 forecast: R² increased from 0.5309 to 0.5549, RMSE is reduced from 33.51 to 26.21, and IOA also increased from 0.71 to 0.83. We will discuss details about what we observed and our future plan for further improvement.

Gail Tonnesen, EPA Region 8; Pat Brewer, National Park Service; Tom Moore, WESTAR/WRAP

The Clean Air Act sets a national goal of remedying existing and preventing future visibility impairment in Class I national parks and wilderness areas due to manmade emissions. The Regional Haze Rule requires states to submit plans every 10 years that implement emission control strategies to demonstrate reasonable progress in improving visibility in Class I areas. To define effective emission controls, states need to understand the contributions at each Class I area from: 1) U.S. anthropogenic source categories and regions, 2) natural sources such as wildfire and windblown dust, and 3) transported international emissions. This study applied the 2008 WestJumpAQMS emissions inventories and the Comprehensive Air quality Model with extensions (CAMx version 5.41) with the Particle Source Apportionment Tool (PSAT) to evaluate source contributions to particle composition and visibility impairment. A 12-km modeling domain over 17 western states was nested within a 36-km continental U.S. modeling domain. Boundary conditions for the continental U.S. domain were extracted from the MOZART global model. Anthropogenic emissions were based on the 2008 National Emissions Inventory with updated regional oil and gas emissions, biogenic emissions estimated using the MEGAN model version 2.10, and fire emissions based on the Joint Fire Sciences Program Deterministic & Empirical Assessment of Smoke's Contribution to Ozone (DEASCO3) study . Model results are available daily for 365 days in 2008. The IMPROVE monitoring network collects 24-hr particle samples every 3 days, roughly 122 days per year, at 70 Class I areas in the western U.S. Samples are analyzed for particle chemical species and the visibility impact determined using IMPROVE light extinction calculation protocols. Daily IMPROVE monitoring data for 2008 were compared to CAMx PSAT results of daily particle species composition and visibility impact for every 12-km grid cell with an IMPROVE monitor. PSAT apportioned emissions from states, anthropogenic source sectors (point, mobile, oil and gas, other area, prescribed fire, agricultural fire, dust) plus natural sources (wildfire, windblown dust, lightning) and international emissions. Example results will be presented and contrasted for individual Class I areas exhibiting high influences from controllable U.S. anthropogenic emissions, wildfire, and international transport. These analyses will assist states in identifying effective emissions control strategies for Regional Haze state implementation plans due in 2018.

Andrew Beddows, King's College London

Elevated pollution levels over the UK during the summer of 2006 have been the focus of many modelling studies, and the sensitivity of modelled ozone concentrations to various inputs, such as boundary conditions and NO_x and VOC emissions, have been evaluated. Such simple sensitivity studies, where model inputs are varied one at a time, fail to capture the complex nature of interactions between inputs, and do not reveal how the apparent sensitivity to one particular input is affected by uncertainty in other inputs.

For the first time a full global sensitivity analysis - that is, one in which all model inputs are perturbed simultaneously over their full ranges of uncertainty - has been performed for the July 2006 ozone pollution episode over the UK. This has previously been intractable because of the many thousands of model runs required by such an analysis, but has now been made possible with the use of Gaussian process emulation. This model emulation technique has enabled quick to run surrogates for CMAQ to be produced with a high degree of accuracy so that they can be used with confidence in global sensitivity analysis methods.

Using an efficient screening method a total of 224 input variables covering reaction rates, boundary conditions, emissions and deposition velocity were perturbed to produce a shortlist of 31 which had a significant effect on output concentrations of ozone and NO₂. Model surrogates were then constructed which emulate the changes in those outputs to changes in the 31 shortlisted inputs and used with the Fourier amplitude sensitivity test (FAST). This sensitivity test perturbs all of the inputs at once to induce a variance in the model output which is then decomposed into contributions from each of the inputs. 21 day time series of model sensitivities have been produced for various locations around the UK and reveal a complex spatio-temporal pattern of the sensitivity of ozone and NO₂ concentrations to those inputs.

YangjunWang¹, Shuxiao Wang ²*, Yongtao Hu⁴,Jiandong Wang²,Bin Zhao², Jiankun Jiang², Mei Zheng³, ArmisteadG. Russell⁴ , Jiming Hao²

Extremely heavy haze episodes occurred in Beijing in January 2013, which are described as strong intensity, long duration and extensive coverage in eastern China. During that period, there was a record-breaking high concentrations of fine particulate matter (PM_2.5) which exceeded 700 g m^-3. In order to alleviate the extremely haze pollution level and furthermore avoid the taking place of haze episodes in Beijing, China, understanding the origin of fine particulate matter is crucial to proposing effective strategies for government. In this study we applied the Particulate Matter Source Apportionment Technology (PSAT) in CAMx (Comprehensive Air Quality Model with Extensions) to quantify the impacts of different emission regions on the concentrations of PM_2.5 at the urban center of Beijing during heavy haze episodes on January 03-23 2013. The emission regions considered here are thirteen cities in Jing-Jin-Ji region , part of Henan and Shandong provinces, and regions outside finest simulation domain (i.e. long-range transport). The results indicate that the average local (Beijing) contribution reached 82.70%, while the average contribution of long range transportation was 6.80% during January 03-23 2013. In addition, the average contributions from Tianjin city and Hebei province were 1.47% and 6.06%, respectively. Local contribution dominated the hourly concentrations of PM_2.5 almost at all the time during January 2013. Especially for the extremely high concentration of PM_2.5, the local contributed much more, as a result of the meteorological conditions with highly unfavourable for pollutant dispersion occurring simultaneously. Therefore, controlling local emissions will be the most important step for Beijing government to mitigate the extremely heavy haze pollution. Nevertheless, the concentration of PM_2.5 at the urban center of Beijing sharply reduced but still was too high to meet the National Air Quality Standards around the noon of January 13 with the help of meteorological conditions with moderate dispersion ability, and it was dominated by the contribution from surrounding cities and long-range transport. During that period, the contributions of Hebei province, Tianjin city, Shandong province and long range transport were 23.50%, 3.46%, 9.30% and 18.04%, respectively. Moreover, Among the eleven cities in Henan province, the biggest contributor is Tangshan (6.90%), followed by Shijiazhuang (3.22%), Cangzhou (3.01%) and Baoding (2.81%). In short, Controlling local emissions should be given priority for Beijing government in order to alleviate the extremely high concentration of PM2.5 effectively, and the emission controlling in its surrounding cities, especially cities in Jing-Jin-Ji region, as well as long-range cities, is essential to improve Beijing air quality, and furthermore to meet the National Ambient Air Quality Standards. This study can provide valuable scientific insights into the extremely heavy haze over Beijing for the future policymaking.

Sophie Cousineau, Annie Duhamel, Sylvain Mnard, Rodrigo Munoz-Alpizar, Nedka Pentcheva, Jacinthe Racine, Mourad Sassi, Mehrez Samaali, Calin Zaganescu

Air Quality Modelling Applications Section, Meteorological Service of Canada, Environment Canada, Montreal, Qubec, Canada

The Air Quality Modelling Applications Section (AQMAS) of Environment Canada (EC) updated its policy modelling platform from base year 2006 to base year 2010. The motivation behind this transition is to take into account the latest emission inventory information and model updates upon which sound advice can be given to policy management. The latest data available, at the beginning of the transition process, include 2010 North-American emission inventories, 2010 meteorology piloting files, and the latest tools such as the meteorological (GEM) and chemical transport (AURAMS) models. The development of this updated modelling platform encompasses the meteorology generation and interpolation, the emissions inventories and processing tools, post-processing of the modelling outputs, preparation of inputs for health and environmental benefits valuation models as well as the performance verification. Being able to assess the air quality for a current period is mandatory when managing future emissions regulatory purposes. This presentation will first provide an overview of the policy modelling platform, followed by 2010 base case evaluation results. Finally, a brief insight of the next generation of Environment Canada's policy platform which is based on its Canadian operational online air quality forecast model: GEM-MACH, will be given.

Jared H. Bowden

Adrienne Wootten

Ryan Boyles

Adam Terando

Located in the Northern Caribbean Sea, Puerto Rico is in a region dominated by the easterly trade winds. This combined with the topography leads to dramatic changes in rainfall over short distances over the island. The coarse projections available for the Caribbean cannot represent the precipitation regime over the island of Puerto Rico. Therefore, there is a need for high resolution projections of climate change for Puerto Rico to assist conservation and adaptation decision making. The purpose of this study is to test the ability of the Weather Research and Forecasting Model (WRF) as a regional climate model to simulate the precipitation within Puerto Rico by downscaling NCEP-DOE Reanalysis. Eight annual simulations were performed using different combinations of cumulus parameterizations and interior grid nudging techniques to assess the sensitivity on the island precipitation. The set of simulations include a comparison of activating the cumulus parameterization in the 2-km innermost domain versus relying on the explicit microphysics. The results show that while the WRF precipitation is dryer than the observations in Puerto Rico, activating the cumulus parameterization in the inner domain improves the timing, intensity, and placement of rainfall compared to the explicit microphysics. The results also show that using interior grid nudging techniques in the outer domains improves the placement and intensity of rainfall in the inner domain.

Sulfate aerosol forms from the gas- and aqueous-phase oxidation of sulfur dioxide and is an important component of atmospheric aerosols. Dimethylsulfide (DMS) present in sea-water can be emitted into the atmosphere which can then react with atmospheric oxidants to produce sulfur dioxide leading to sulfate formation. The current Community Multiscale Air Quality (CMAQ) model, however, does not include DMS emissions and its atmospheric chemistry. In this study, we implement a DMS emission scheme and its atmospheric chemistry into the CMAQ model. DMS emissions are calculated based on oceanic climatological DMS concentrations and the total resistance to gas-transfer at the air/sea interface. Our CMAQ-predicted DMS fluxes are similar to previous estimates reported in the literature. The updated atmospheric chemistry in CMAQ includes gas-phase oxidation of DMS by hydroxyl radical, nitrate, chlorine radical, chlorine monoxide, iodine monoxide, and bromine monoxide. Model simulations without and with the DMS chemistry over the Northern Hemisphere show enhanced concentrations of sulfur dioxide over marine environments. Preliminary results suggest that the inclusion of DMS chemistry enhances sulfate aerosol over marine environments and coastal areas. However, the magnitude of its impact has substantial spatial and temporal variation. The paper will detail the spatial impact of DMS on sulfate concentration and compare model predictions with observed data.

Melanie Follette-Cook (Melanie.Cook@nasa.gov)

Christopher Loughner (Christopher.P.Loughner@nasa.gov)

Kenneth Pickering (Kenneth.E.Pickering@nasa.gov)

Rob Gilliam (Gilliam.Robert@epa.gov)

Jim MacKay (Jim.MacKay@tceq.texas.gov)

During the month of September, 2013 the Deriving Information on Surface conditions from Column and Vertically Resolved Observations Relevant to Air Quality (DISCOVER-AQ) campaign used two aircraft in tandem with an extensive suite of ground-based in-situ and remote sensing instruments to measure trace gases and aerosols over the Houston area. Ten flight days and over 200 profiles of trace gases and aerosols were completed over the course of the month. The variety of meteorological flight conditions, air quality conditions, consistent flight patterns, and large sample make the DISCOVER-AQ dataset ideal for model evaluation. Several high-resolution WRF simulations were performed to test different initial and boundary conditions, spatial resolutions, nudging options, and simulation methods (e.g. re-initialization vs. iterative methods). Our evaluation of the WRF and CMAQ simulations using in-situ aircraft and ground based DISCOVER-AQ data indicate several potential areas for model improvement, as well as highlight the need for certain diagnostics to better evaluate data vs. model discrepancies.

Alexander Macpherson^a, Heather Simon^a, David Misenheimer^a, Charles Fulcher^a, Bryan Hubbell^a, Robin Langdon^a

^a Office of Air Quality Planning and Standards, Office of Air and Radiation, United States Environmental Protection Agency, 109 T.W. Alexander Drive, Research Triangle Park, NC 27711 USA.

The United States Environmental Protection Agency (EPA) is currently reviewing the National Ambient Air Quality Standard (NAAQS) for ozone. States with areas designated as nonattainment with the standards are required to develop State Implementation Plans (SIPs) to demonstrate how pollution levels will be reduced to meet the standard. Historically, many states have developed SIPs independently. However, for ozone, some states have at times recognized the important role of regional transport (for example the Ozone Transport Commission which addresses ozone air quality in the Northeast) and have developed regional agreements to control ozone-forming emissions. These types of regional air quality management approaches have the potential for improved pollution control efficiency if states collaboratively determine the least-cost controls within or across regions. We present a Mixed Integer Programming model for devising least cost control strategies that recognize the potential for interstate transport of ozone. While linear programming models have been used to assess regional ozone control strategies, this model applies the framework nationally to identify efficiencies from reducing regional transport. Air quality is characterized by a source-receptor matrix estimating the impact of regional NO_X and local VOC emissions reductions on ozone concentrations at monitors. In select areas where large NO_X reductions are anticipated, the source-receptor approach also approximates the non-linearity of ozone response to emissions reductions. Least cost control strategies are determined by choosing across (potentially multiple) technically-feasible control technologies for each controllable emissions source. This tool allows user-defined policy constraints about which ozone precursors and emissions locations to consider in identifying the least cost attainment strategy. A case study is presented using information from a series of emissions sensitivity air quality model simulations along with current emissions control supply information. The model holds promise for evaluating alternative scenarios, testing the role of pollutant transport in compliance strategies, and identifying monitors exerting disproportionate influence on attainment strategies. The case study is a proof of concept but is limited by the geographic and source specificity of the source-receptor matrix. As additional air quality simulations are performed, more refined information about the response of ozone to emissions reductions in specific locations and from specific source types will improve the accuracy of the model.

Disclaimer: This work represents the views of the authors and does not reflect official Environmental Protection Agency policy.

S. Galmarini and E. Solazzo

European Commission,

Joint Research Centre, Institute for Environment and Sustainability,

Air and Climate Unit, Ispra (Italy)

The evaluation of air quality models has been for several years based on mainly the operational and diagnostic modes of the four identified by Dennis et al. (2010). Operational evaluation assesses how far the model is from the measurements and the diagnostic one tries to identify the specific process that could be cause for a possible discrepancies. Neither of the two is exhaustive, since operational evaluation does not provide explanation on why a model is wrong and if correct neither allows identifying whether the latter is causal or systematic. Diagnostic evaluation being based on Ockham principles strips the contribution elements leaving only the essential ones but does not allow determining whether it is the specific process or module the cause of the error or more the interaction among multiple components of the model. The approach we propose is more holistic than the diagnostic one but more specific than the operational one proving therefore clear indications on the scales that are carrying and propagating the error and therefore giving clear indications on the parts of the model or input data that are sources of errors. The results presented are based on a detailed ananlysis of AQMEII 1,2 and 3 datasets thus constituting a solid basis of analysis form the statistical point of view with a wide range of variables, models, simulated years and conditions.

Ivanka Stajner (1), Jeff McQueen(2), Pius Lee(3), Ariel Stein(3), Jinaping Huang (2,5), Li Pan (3,6), Daniel Tong (3,6), Ho-Chun Huang (2,5), Perry Shafran (2,5) Jerry Gorline (4) ,Phil Dickerson(7), Sikchya Upadhayay (1,8)

1. NOAA NWS/OSTI
2. NOAA NWS/NCEP
3. NOAA ARL
4. NOAA NWS/OSTI/MDL
5. IMSG
6. CICS, University of Maryland
7. EPA
8. Syneren Technologies

NOAA provides operational predictions of ozone and wildfire smoke for the United States (U.S.) and predictions of airborne dust and is available at http://airquality.weather.gov/. For ozone predictions and testing of fine particulate matter (PM2.5) predictions, NOAA National Centers for Environmental Prediction (NCEP) operational North American Mesoscale (NAM) weather predictions are combined with the Community Multiscale Air Quality (CMAQ) model. For smoke and dust predictions the Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model is used. Verification of ozone predictions relies on AIRNow compilation of observations from surface monitors. Verification of smoke and dust predictions uses satellite retrievals of smoke and dust.

Recent updates to predictions of operational ozone and developmental PM2.5 have focused on using partially updated emissions from NEI 2011. Recent testing of CMAQ modeling system includes updated emissions, contributions from global dust predictions through lateral boundary conditions, and additional vertical layers. Evaluation shows ozone prediction accuracy is maintained and improved. Seasonal biases in PM2.5, while still present, have been decreasing with model and emission updates. A bias correction procedure is in testing to further reduce biases in the developmental PM2.5 predictions. The bias correction procedure improves average diurnal cycle and further evaluation is focusing on day-to-day variability.

Jianping, Huang^1,2*, Jeff McQueen², Perry Shafran^1,2, Jerry Gorline³, Ho-chun Huang^1,2, Jun Wang^1,2, Irina V. Djalalova⁴, Dave Allured⁴, James Wilczak⁵, Pius Lee⁶, Li Pan^6,7, Daniel Tong^6,7, Youhua Tang^6,7, Sikchya Upadhayay^{8, 9}, Geoff DiMego², and Ivanka Stajner⁹

¹I.M. Systems Group Inc., Rockville, MD

²National Oceanic and Atmospheric Administration (NOAA), National Centers for Environmental Prediction (NCEP), College Park, MD

³NOAA, Meteorological Development Laboratory (MDL), Silver Spring, MD

⁴Cooperative Institute for Research in the Environmental Sciences (CIRES), University of Colorado, Boulder, CO

⁵NOAA, Earth Systems Research Laboratory (ESRL), Boulder, CO

⁶ NOAA Air Resources Laboratory, Silver Spring, MD

⁷ University of Maryland, College Park, MD

⁸Syneren Technologies Corporation, Arlington, VA 22201

⁹NOAA, Office of Science and Technology Integration, Silver Spring, MD

Persistent seasonal forecast biases are seen in predictions of surface particulate matters with diameter less than 2.5 mm (PM_2.5) produced by the NOAA National Air Quality Forecasting Capability (NAQFC). The NAQFC is the off-line coupling forecasting system, consisting of the NOAA NCEP regional operational weather forecasting model, the Non-hydrostatic Multi-scale Model on the Arakawa staggered B-grid (NMMB) and the EPA Community Multiscale Air Quality (CMAQ) model. It is critical to quantify and reduce the biases before these PM_2,5 predictions can be transitioned into operations. In this study, two advanced bias correction approaches, the analog (AN) and the Kalman-filtering combined with historical forecast analogs (KFAN) are employed to reduce the NAQFC PM_2.5 prediction errors. Currently the NMMB-CMAQ system is run four times per day but provides 48-h forecasts for the 06z and 12z cycle by NOAA/NWS/NCEP at 12-km horizontal resolution with 35 vertical levels, and provides operational products of surface ozone and developmental products of PM_2.5 nationwide. The KFAN approach calculates the hourly CMAQ PM_2.5 forecast biases at each observational site, searches an analog case from historical records based on key meteorological parameters during the training period, adds the identified forecast biases to the nearby grids, and finally spreads bias correction to the gridded CMAQ predictions. This study is focused on the CONUS domain and the study periods cover July 2014, January 2015, and April-August 2015. The PM_2.5 predictions with and without bias correction are verified against AIRNow observational data. We compare the performance of KFAN and AN in different regions and seasons, and quantify the impact of training periods on the bias correction approaches for operational use. Furthermore, we present detailed analyses of cases associated with large dust and wildfire smoke concentrations when the application of bias correction is more challenging.

Shaocai Yu¹, Pengfei Li¹, Renchang Yan¹, Qingyu Zhang¹, Liqiang Wang¹, Si Wang¹, Bixin Chen¹, Weiping Liu¹, Yang Zhang², David Wong³, Kiran Alapaty³, Jon Pleim³ and Rohit Mathur³

¹ Research Center for Air Pollution and Health, College of Environmental and Natural Resources, Zhejiang University, Hangzhou, Zhejiang 310058, P.R. China.

² Air Quality Forecasting Lab, Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, NC 27695, USA

³Atmospheric Modeling and Analysis Division, National Exposure Research Laboratory, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711, USA

Over the past three decades, China, especially in the megacity areas, has suffered from air pollution and heavy haze because of its decades-long burst of economic growth and rapidly expanding clout as an industrial giant. The Beijing-Tianjin-Hebei, Yangtze River Delta, and Pearl River Delta regions are three key areas with heavy haze pollution in China. In this study, PM_2.5 concentrations in Hangzhou metropolitan city are forecasted with the newly-developed two-way coupled WRF-CMAQ and WRF-Chem model systems. While WRF/Chem includes indirect aerosol effects on grid-scale clouds, such treatments have been implemented in the newly-developed two-way coupled WRF-CMAQ model by including parameterizations for both cloud droplet and ice number concentrations calculated from the CMAQ-predicted aerosol particles. The real-time forecasts of PM_2.5 are carried out over East Asia at a 36-km grid resolution, eastern China at a 12-km grid resolution, and Hangzhou at a 4-km grid resolution. Evaluations of model performance on PM_2.5, PM₁₀, O₃, SO₂, NO₂, CO, air quality index (AQI), and aerosol optical depth (AOD) are performed by comparing to observations from satellites such as MODIS and surface monitoring networks over eastern China.

Christian Hogrefe¹, Jia Xing¹, Johannes Flemming², George Pouliot¹, Shawn Roselle¹, and Rohit Mathur¹

¹ Atmospheric Modeling and Analysis Division, U.S. EPA

² European Centre for Medium-Range Weather Forecasts, Reading, U.K

Global and hemispheric chemistry-transport models are often used to prepare chemical boundary conditions for regional scale air quality simulations. This process as well as the attribution of errors in the regional scale simulations can be complicated by differences in the representation of atmospheric chemistry between the modeling systems. To avoid such complications, it is desirable to employ a consistent modeling framework spanning all scales of interest. In the context of performing CMAQ simulations over the U.S., such a framework requires performing corresponding simulations over the entire Northern Hemisphere to represent background concentrations and intercontinental transport. In this study, we apply the hemispheric version of CMAQ5.0.2 to simulate air quality over the Northern Hemisphere for the year 2010. The emission inputs for this simulation were prepared for the modeling activities under the task force on Hemispheric Transport of Air Pollution (HTAP) using a combination of regional and global emission inventories. We present a comparison of these hemispheric CMAQ simulation against observations collected for the Air Quality Model Evaluation International Initiative (AQMEII), including ground-based observations over North America and Europe, ozonesonde measurements, and MOZAIC aircraft profiles. In addition, the simulations are also compared against 2010 global model simulations using the same emission inventory and performed as part of the HTAP activity, including the Composition - Integrated Forecast System (C-IFS) model developed by the European Center for Medium Range Weather Forecasting (ECMWF) and used as source of chemical boundary conditions for the regional scale simulations performed under AQMEII. The results are discussed in terms of their implications for using the different global and hemispheric simulations as boundary conditions for regional scale simulations.

James T. Kelly¹, Kirk R. Baker¹, Christopher G. Nolte², William C. Keene³, and Alexander A. P. Pszenny⁴

¹Office of Air Quality Planning and Standards, U.S. EPA, Research Triangle Park, NC, USA

²Office of Research and Development, U. S. EPA, Research Triangle Park, NC, USA

³Department of Environmental Sciences, University of Virginia, Charlottesville, Virginia, USA

⁴Institute for the Study of Earth, Oceans and Space, University of New Hampshire, Durham, NH, USA

Mathematical modeling of atmospheric aerosol processes is useful for air quality planning related to National Ambient Air Quality Standards for particulate matter, regional haze in Federal class I areas, and deposition to sensitive ecosystems. Model evaluations to support air quality modeling assessments are often limited by a lack of comprehensive datasets. As part of the Nitrogen, Aerosol Composition, and Halogens on a Tall Tower (NACHTT) campaign, the phase partitioning of HNO₃, NH₃, and HCl with size-resolved particles and related meteorological conditions were measured continuously during February and March 2011 in Colorado northeast of Denver. Here we evaluate model predictions of gas-particle partitioning of HNO₃, NH₃, and HCl as well as predictions of size-resolved SO₄^2-, NH₄⁺, NO₃^-, Cl^-, Na⁺, Ca²⁺, Mg²⁺, and K⁺ concentrations using the NACHTT dataset. Modeling is based on the Community Multiscale Air Quality (CMAQ) model for a domain covering the continental U.S. with 12 km horizontal resolution. In addition to the base simulation, several sensitivity simulations are conducted to explore the influence of emissions on model predictions. Source apportionment modeling is also performed to further characterize model predictions at the NACHTT site. Our evaluation demonstrates that the national modeling platform can simulate the distributions of observed quantities over the study period with reasonable confidence. Differences between predictions and observations and areas for future work are identified and will be discussed in the presentation.

Kan Huang and Joshua Fu

Department of Civil and Environmental Engineering, The University of Tennessee, Knoxville, Tennessee, USA

Black carbon plays a unique role in the Arctic climate system due to its multiple effects. It causes Arctic warming by directly absorbing sunlight from space and by darkening the surface albedo of snow and ice, which indirectly leads to further warming and melting, thus inducing an Arctic amplification effect.

BC depositions over the Arctic are more sensitive to regions in close proximity. In this study, we reconstruct BC emissions for Russian Federation, which is the country that occupies the largest area in the Arctic Circle. Local Russia information such as activity data, emission factors and other emission source data are used. In 2010, total anthropogenic BC emission of Russia is estimated to be around 254 Gg. Gas flaring, a commonly ignored black carbon source, contributes a dominant 43.9% of Russia's total anthropogenic BC emissions. Other sectors, i.e., residential, transportation, industry, and power plants, contribute 22.0%, 17.8%, 11.5%, and 4.8%, respectively.

BC simulations were conducted using the hemispheric version of CMAQ with polar projection. Emission inputs are from a global emissions database EDGAR (Emissions Database for Global Atmospheric Research)-HTAPv2 (Hemispheric Transport of Air Pollution) and EDGAR-HTAPv2 with its Russian part replaced by the newly developed Russian BC emissions, respectively. The simulations using the new Russian BC emission inventory could improve 46 - 61% of the Absorption Aerosol Optical Depth (AAOD) measured at the AERONET sites in Russia throughout the whole year as compared to that using the default HTAPv2 emissions. At the four air monitoring sites (Zeppelin, Barrow, Alert, and Tiksi) in the Arctic Circle, surface BC simulations are improved the most during the Arctic haze periods (October - March). Emission perturbation studies show that Russia's BC emissions contribute over 50% of the surface BC concentrations over the Arctic during the cold seasons.

This study demonstrates the good capability of H-CMAQ in simulating the transport of BC particles to the Arctic and suggests that the impact of Russian emissions on the Arctic haze has likely been underestimated, which is one of the causes that previous modeling works struggled in reproducing the BC levels in the Arctic region.

Allison Michaelis, Kiran Alapaty, Jerry Herwehe

The Model for Prediction Across Scales (MPAS) is a new generation variable grid resolution global model with promising features suitable for weather prediction as well as air quality modeling. In light of persistent reductions in NAAQS for ozone, models such as MPAS offer a natural solution to account for the global transport of air pollution that contributes to the background concentrations at local scales. Unfortunately, almost all global models suffer from the lack of scale aware physics and chemistry, constraining the accuracy of meteorological and air quality predictions. Of these processes, convective clouds play a prominent role on the atmospheric dynamics, thermodynamics, and chemical state of the atmosphere. To introduce multi-scale physics in such variable resolution models, we introduce the Multi-scale Kain Fritsch (MSKF) scheme to the MPAS.

The objective of this research project is to test the performance of the MSKF convective parameterization scheme (Alapaty et al. 2012; Zheng et al. 2015) using the MPAS-A numerical model. MPAS-A is a non-hydrostatic, global atmosphere model that makes use of centroid Voronoi tessellations to create variable resolution horizontal meshes. This technique allows for a gradual expansion from a localized region of mesh refinement to a larger grid spacing on the rest of the globe, rather than a harsh discontinuity between domains with different grid lengths. The current Kain-Fritsch convective parameterization scheme available in MPAS-A was designed for optimization at grid lengths of 25-km. Therefore, precipitation biases tend to arise when this scheme is implemented at higher resolutions. To combat these biases, the MSKF scheme introduces several modifications. Sub-grid scale cloud-radiation interactions, a dynamic adjustment timescale, and scale dependent entrainment effects are few among other modifications that will allow the CP scheme to scale appropriately with varying resolutions.

To study the performance of the MSKF scheme, a season long global simulation is performed starting from May 15, 2006 for 3.5 months. A global horizontal grid varying from 15-km over the continental United States to 60-km over the remainder of the globe was used. Results from the simulation using the MSKF scheme are compared to a simulation using the default KF scheme as well as to observational datasets, such as the Tropical Rainfall Measuring Mission (TRMM) for precipitation an quality controlled local climate data (QCLCD) for surface parameters. Evaluation results of this research will shed light on the performance of the MSKF in a global setting.

Dan Goldberg, Kyle Hosley, Tim Vinciguerra, Tim Canty, Chris Loughner, Ross Salawitch, and Russell Dickerson

We simulate ozone over the eastern United States (66 - 94 W longitude) using the Comprehensive Air-Quality Model with Extensions (CAMx) v6.10 to determine the attribution of ozone to individual source regions and the model domain boundary. Boundary condition ozone (BC_O3) - defined as the summation of ozone transported into and formed from precursors that are transported into the modeling domain - represents 38.8% of the total surface ozone in Maryland during the July 2011 mean. Even during the worst air quality days, BC_O3 in Maryland represents greater than 25 ppbv of the total surface ozone. In future years, the portion attributed to the boundary will become a larger percentage as anthropogenic nitrogen oxide (NO_x) and volatile organic compound (VOCs) emissions in our domain are predicted to decrease. We also demonstrate that BC_O3 in Maryland has increased in an absolute sense from 26.0 to 27.2 parts per billion by volume (ppbv) between July 2002 and July 2018, despite identical initialization of meteorology, land-use patterns, and trace gases at the boundary in each simulation. We show that two main sinks for ozone are decreasing in magnitude: (1) peroxy radical (HO₂) concentrations are declining in non-urban, non-industrial regions and (2) decreases in NO_x concentrations are causing less nighttime removal via N₂O₅ hydrolysis. Reductions in anthropogenic NO_x and VOCs have yielded the unintended consequence of increasing the lifetime of ozone.

Lakshmi Pradeepa Vennam^1,2, Saravanan Arunachalam², Kevin Talgo², Mohammad Omary², Jia Xing³, Rohit Mathur³, and William Vizuete¹

¹ Department of Environmental Sciences and Engineering, University of North Carolina at Chapel Hill, Chapel Hill, NC

² Institute for Environment, University of North Carolina, Chapel Hill, NC

³ U.S. Environmental Protection Agency, RTP, NC

Intercontinental pollution attributable to aviation can be a key concern with increase in aviation activity. Aircraft emissions budgets in upper troposphere lower stratosphere (UTLS) region and their potential impacts on both upper troposphere and surface air quality are not well understood. Given the uncertainties in prior global studies due to coarse horizontal resolution (4x5 degree, 2x2.5 degree) and not considering lightning emissions (another altitude varying emission source for NO) to study the aviation impact in UTLS, we developed an enhanced modeling platform with highly resolved vertical resolution (44 layers upto 50 mb) to understand the impact of full-flight aircraft emissions (including both landing and takeoff as well as cruise mode) at both Continental US (36 km) and Hemispheric (108 km) scales. The regional domain over the U.S. was driven by the hemispheric-domain boundary conditions to prevent higher O₃ concentrations in upper model layers which is one of the main issues observed among global model downscaled regional boundary conditions. This approach also removes potential inconsistencies in the physical and chemical processes when downscaling from global scale models to regional-scale models, which have inherently different algorithms. We performed model evaluation with aircraft in-situ observational data from MOZAIC and INTEX campaigns to illustrate the improvements observed in the modeling approach. In order to understand the changes occurring in the UTLS chemistry we used process analysis to illustrate the aviation-attributable concentrations for key criteria pollutants such as O₃, NO₂ and PM_2.5. We will present results from this application, along with an assessment of concentrations along the isentropes to understand the transport of cruise altitude emissions at hemispheric scale to the surface layer.

Nina Randazzo, Daniel Tong, Pius Lee, Li Pan, Min Huang

Carbon Monoxide (CO) is a criteria air pollutant regulated by the National Ambient Air Quality Standards. CO is also an ozone precursor and so is becoming increasingly important as ozone standards are being tightened. The performance of CMAQ to predict CO concentration, however, is rarely examined. This work evaluates the CO prediction by the NOAA National Air Quality Forecast Capability (NAQFC) system during summertime in 2013, with emphases on both surface concentrations and vertical profiles during biomass burning (BB) season. Days are separated into high-BB influence and low-BB influence cases using the mixing ratios of co-located BB markers. For surface concentration, the model is evaluated against the EPA Air Quality System (AQS) ground observations. Sites are divided into urban, rural, and suburban sites using the AQS site classification, which are further checked against the ORNL Harmonized Global Land Use classification. Predictions aloft are compared to data from aircraft campaigns such as the Southeast Nexus (SENEX) campaign. Several days are chosen based on the availability of aircraft data and the representation of high-BB and low-BB days in the region. Ground concentration time series and high-altitude observations for each region are compared with model output for high-BB and low-BB days. The evaluation of the vertical profile of CMAQ CO predictions under different biomass burning conditions has important implications for the ability of satellites to detect CO plumes and for the ability of CMAQ to predict satellite detection. The evaluation of the ground concentration time series CMAQ predictions in conjunction with a vertical profile analysis allows a comparison between how CMAQ performs at satellite-detectible altitudes versus the surface. Our results provide in-depth evaluation of the accuracy of the CMAQ model to predict CO based on the amount of biomass burning in the area and on the relative importance of local emissions and transport, both at the surface and aloft.

K.M. Seltzer, C.G. Nolte, T.L. Spero, K.W. Appel, J. Xing

Prakash Karamchandani¹, Greg Yarwood¹, David Parrish², Thomas Ryerson³, Paul O. Wennberg⁴, Alex Teng⁴, John D. Crounse⁴

¹Ramboll Environ, 773 San Marin Drive, Suite 2115, Novato, CA 94998

²4555 13th St. #2E, Boulder, CO 80304

³NOAA ESRL Chemical Sciences Division, 325 Broadway, R/CSD7, Boulder, CO 80305

⁴California Institute of Technology, 1200 E. California Blvd., Pasadena, CA 91125

Industrial emissions of highly reactive volatile organic compounds (HRVOCs), namely ethene, propene, butenes, and 1,3-butadiene, can contribute to localized ozone (O₃) production in the Houston area. Quantifying the relative contributions of these alkenes to O₃ formation through ambient measurements is difficult. Atmospheric degradation of HRVOCs in the presence of nitrogen oxides (NOx) forms known yields of hydroxynitrates that are specific to parent alkenes. Aircraft measurements of C₂-C₄ hydroxynitrates and other species during the NASA Studies of Emissions and Atmospheric Composition, Clouds and Climate Coupling by Regional Surveys (SEAC⁴RS) project in 2013 provide a unique opportunity to link observed O₃ and HCHO enhancements to emissions of specific alkenes. A data analysis and plume modeling study was conducted to quantify these relationships using SEAC⁴RS measurements of plumes from the Houston Ship Channel. The data analysis component of the study identified 4 periods when plumes from the ship channel were sampled during SEAC⁴RS. These periods are simulated with SCICHEM 3.0, a state-of-the-science puff model with complete chemistry treatment. The base gas-phase chemistry mechanism in SCICHEM is updated to CB6r2 and a scheme for b-hydroxynitrate formation during alkene oxidation is added to the mechanism. The b-hydroxynitrate and ozone concentrations predicted using SCICHEM are compared with data from aircraft transects to evaluate whether the model chemistry is consistent with SEAC⁴RS observations. The results from the plume modeling are analyzed to compare modeled yields of O₃ and formaldehyde (HCHO) from parent alkenes with yields obtained from analysis of the measurements.

Chris Nolte, Tanya Spero, Jared Bowden

Projecting climate change scenarios to local scales is important for understanding, mitigating, and adapting to the effects of climate change on society and the environment. Many of the global climate models in the Intergovernmental Panel on Climate Change Fifth Assessment Report do not fully resolve regional-scale processes and therefore cannot capture regional-scale changes in temperatures and precipitation, particularly acute extreme events. We use the Weather Research and Forecasting model as a regional climate model (RCM) to dynamically downscale the NCAR/DOE
Community Earth System Model over North America for an 11-year period centered on 2000 as well as the 11-year period centered on 2030 for three different Representative Concentration Pathway (RCP) projections. These RCM scenarios are used as meteorological inputs for the Community Multiscale Air Quality modeling system to explore influences of regional climate change on air quality. We will present changes in climate and air quality under those RCP scenarios, including changes in temperature extremes and the frequencies of ozone and PM episodes.

Michael D. Moran, Junhua Zhang, Radenko Pavlovic, and Samuel Gilbert

GEM-MACH, which consists of an on-line chemical transport model embedded within GEM, Environment Canada's multi-scale operational weather forecast model, has been used in a limited-area configuration in Environment Canada's operational Regional Air Quality Deterministic Prediction System (RAQDPS) since November 2009. The RAQDPS is run twice daily to produce 48-hour forecasts of hourly O₃, NO₂, and PM_2.5 on a North American grid. Over the 5-year-plus period since 2009 there have been 11 upgrades of varying importance made to the RAQDPS. For example, in 2009 the model grid had 15-km horizontal grid spacing and 58 vertical levels extending from the surface to 0.1 hPa whereas the current grid has 10-km horizontal grid spacing and 80 vertical levels over the same depth. The input emissions files have also been updated several times.

It is now possible to examine the evolution of RAQDPS performance over a five-year period from January 2010 to December 2014. A set of quality-controlled near-real-time North American air quality measurements of hourly O₃, NO₂, and PM_2.5 has been used for this evaluation, and changes in the observing network and ambient levels over this period and the quality control steps that were employed will be described briefly. Broadly consistent monthly and diurnal variations are evident from year to year in five-year time series of such standard performance metrics as mean bias, correlation coefficient, and unbiased root mean square error (URMSE). The impact of model upgrades on model performance are also evident as time trends in some performance metrics, particularly URMSE. Analyses of the influence of geographic region, degree of urbanization, and forecast lead time on model skill will also be shown.

Jared H. Bowden

Kevin. D. Talgo

Tanya L. Spero

Christopher G. Nolte

Megan Mallard

projections for the Contiguous United States. The development of this ensemble includes model development and testing numerous downscaling techniques to improve the simulated regional climate. Research efforts include improving the precipitation via changes made to a cumulus parameterization scheme, use of interior grid nudging, and improvements in the representation of large lakes. The current regional climate ensemble includes downscaling of multiple GCMs (NASA-GISSE, CESM, GFDL-CM3), emission scenarios (RCP4.5, RCP6.0, and RCP8.5), and future decades (2030s-2050s).

Here we focus on projected changes in the Southeast US climate with an emphasis on meteorological parameters important for air quality during the summer season. One recently noted atmospheric circulation of concern is the North Atlantic Subtropical High (NASH) and the observed and projected westward shift as the climate warms. Our evaluation will focus on this shift and its impact on changes in ventilation and temperature extremes.

Kenneth Pickering, Lok Lamsal, Christopher Loughner, Scott Janz, Nick Krotkov, Andy Weinheimer, Alan Fried

Air mass factors (AMFs, which are the ratios of slant column abundances to vertical column amounts) are required to convert slant column observations of trace gases such as NO2 and HCHO from existing and future satellite instruments such as OMI and TEMPO to vertical column abundances. Similarly, slant column observations from the ACAM, GCAS and Geo-TASO instruments flown on NASA aircraft during the DISCOVER-AQ deployments require AMFs to obtain vertical columns of trace gases below the aircraft. The AMF is computed from a radiative transfer code which requires assumed vertical profiles of the trace gas mixing ratios as input. We have used profiles from CMAQ model simulations for the Maryland and Houston DISCOVER-AQ campaigns in the AMF calculations as part of retrieval sensitivity studies. AMFs from various CMAQ model configurations have been compared with those computed using NO2 and HCHO profiles measured in-situ by the NASA P-3B aircraft in spirals over surface air quality monitoring stations. Profile shapes from the model are altered through various combinations of changes of the boundary layer scheme used in the WRF model that drives CMAQ, of the emissions, and of model chemistry. The sensitivity of the AMFs to the resulting model profiles are computed in terms of the percent error with respect to the AMF computed from the observed profiles. Retrieval errors of up to 30% on a monthly mean basis occur due to profile shape errors with certain combinations of aerosol optical depth and surface albedo. Minimization of these errors with accurate CMAQ simulation of profile shapes will become especially important for the hourly observations from the geostationary TEMPO instrument due to fly by 2019.

Evan Couzo, Erwan Monier, Fernando Garcia-Menendez, Nick Hoffman, Minjoong Kim, Rokjin Park, and Noelle Selin

Climate penalty calculations do not typically consider the effects of non-GHG emissions. In this work, we estimate the US climate penalty (for ozone and particulate matter) as a function of different local (US) and global emissions scenarios, thus quantifying the avoided climate penalty achieved by non-GHG emissions reductions resulting from climate policy. Our global modeling framework is fully self-consistent and integrates global economic growth, emissions, and climate projections for two scenarios: a reference no-climate-policy case and a climate policy that stabilizes radiative forcing at 3.7 W/m2 by 2100. Pollutant concentrations are predicted using GEOS-Chem from 20-year simulations of present (1991-2010) and future (2041-2060 and 2091-2110) climate and emissions. Climate and meteorological fields are taken from the Community Atmosphere Model.

Chowdhury Nazmi¹- PhD Candidate, NOAA CREST, Department of EE, The City College of New York, NY 10031

Barry Gross- NOAA CREST, Professor, Department of EE, The City College of New York

Fred Moshary-NOAA CREST, Professor, Department of EE, The City College of New York

Shobha Kondragunta, NOAA STAR, College Park, MD

Particulate Matters, or PM, are small particles that are found in the air, including inorganic sulfates, nitrates as well as biomass products such as smoke. Fine Particulate Matter with diameters < 2.5 microns are called PM2.5 and they are believed to pose significant health risks. Because of their small size, they can be easily inhaled and lodge deeply into the lungs. Therefore, the Environmental Protection Agency (EPA) has developed strict 24 hour concentration guidelines for PM2.5. To address this, the EPA (as well as state government) has set up a network of PM2.5 monitoring stations. However, surface sampling can be quite expensive and the existing networks are very limited resulting in significant data gaps that can affect the ability to forecast PM2.5 over a 24 hour period. To address these data gaps, satellite observations of aerosols can be useful. Satellites provide column integrated Aerosol Optical depth information that can potentially be used to estimate PM2.5. In recent studies, it has been found that particle mass can be linearly related to optical scattering coefficient of the particles which implies that the total column integrated aerosol optical depth (AOD) measurements would also be connected to surface PM2.5 through a simple linear model. However, a wide range of factors such as aerosols variability, meteorology and the vertical structure of aerosols affect the PM2.5-AOD relationship. Clearly, one very important factor is the aerosol type. In particular, the sulfate/nitrate dominated local aerosols will have very different mass/optical properties than smoke. Studies also suggest that the relationship between PM2.5 and AOD does not work well in the presence of aloft plumes. Therefore, any tools that help quantitatively distinguish these categories would be a considerable benefit in reducing PM25 retrieval errors. The ability to identify and quantify smoke plumes is critical for better interpreting the linkage of passive satellite observations of aerosol optical depth (AOD) and surface aerosol concentration (PM2.5). A number of numerical models which combine meteorological transport and satellite observations has been developed which attempt to quantify plume vertical height and extent including the Navy's NAAPS model and NOAA's GOES ASDTA product. In this research proposal, we propose to study the performance of these models in plume forecast and whether they can be used to filter out contaminated cases which will result in a better PM2.5 to AOD relationship. Preliminary results show that multi-year GOES-ASDTA smoke product and NAAPS aerosol model can be used to get a useful smoke indicator that can effectively eliminate smoke contaminated cases improve the correlation and RMSE between PM2.5 and AOD. In this study multi-year MODIS AQUA/TEYYA aerosol optical depth and PM2.5 concentration from 20 stations in NY State were used. We also studied statistics of aloft smoke from CALIPSO, NAAPS and GOES ASDTA as well as the usefulness of NAAPS Aloft smoke product in filtering out contaminated cases. In addition to that, LIDAR imagery from CCNY LIDAR webpage were used to filter out aloft plume days and observe the effect of plumes on potential linear behavior between PM2.5 and AOD. We will further study the following topics: 1)Exploring the use of the SI as an extra input to modify existing PM25 estimators including multivariable regressions and NN. 2)We will also look at other smoke tools such as OMI's Absorbing Index and the VIIRS smoke mask and look for correlations and best combinations to use between all of these products.

Yuqiang Zhang¹, Jared Bowden², Zachariah Adelman^1,2, Vaishali Naik³, Larry W. Horowitz⁴ ,

Steven J. Smith⁵, J. Jason West¹

¹ Environmental Science and Engineering Department, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599

² Institute for the Environment, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599

³ UCAR/NOAA Geophysical Fluid Dynamics Laboratory, Princeton, NJ 08540

⁴ NOAA Geophysical Fluid Dynamics Laboratory, Princeton, NJ 08540

⁵ Joint Global Change Research Institute, Pacific Northwest National Laboratory, College Park, MD 20740

Policies to mitigate greenhouse gas (GHG) emissions will not only slow climate change, but can also have ancillary benefits of improved air quality, and consequently human health. Here we examine the co-benefits of global and regional GHG mitigation on U.S. air quality and human health in 2050 at fine resolution, using dynamical downscaling methods, based on a previous global co-benefits study (West et al., 2013). The co-benefits of reducing GHGs on U.S. air quality are quantified via two mechanisms: through reductions in co-emitted air pollutants from the same sources, and by slowing climate change and its influence on air quality. We use the WRF model to dynamically downscale a realization of climate change from the GFDL global model to the regional scale, the SMOKE model to directly process global anthropogenic emissions into the regional domain, and we provide boundary conditions from global simulations to the regional CMAQ model. BenMAP is used to calculate the avoided premature human mortality in the U.S. due to the improved air quality.

The total co-benefits of an aggressive global GHG mitigation strategy are estimated to be higher in the eastern U.S. (ranging from 0.6-0.9 g/m³) than the west (0-0.4 g/m³) for PM_2.5, with an average 0.47 g/m³ over U.S.; for O₃, the total co-benefits are fairly uniform at 2-5 ppb with U.S. average of 3.55 ppb. Comparing the two mechanisms of co-benefits, we find that reductions of co-emitted air pollutants have much greater influence on both PM_2.5 (96% of the total co-benefits) and O₃ (89% of the total) than the second co-benefits mechanism via slowing climate change, consistent with West et al. (2013). GHG mitigation from foreign countries contributes more to the U.S. O₃ reduction (76% of the total) than that from domestic GHG mitigation only (24%), highlighting the importance of global methane reductions and the intercontinental transport of air pollutants. For PM_2.5, the benefits of domestic GHG control are greater (74% of total). The avoided premature human mortality in the U.S. due to the air quality changes, including both PM_2.5 and O₃, will also be calculated and presented for this study. Since foreign contributions to co-benefits are comparable to domestic reductions, especially for O₃, previous studies that evaluate co-benefits locally or regionally may greatly underestimate the total co-benefits of global GHG reductions. We conclude that the U.S. can gain significantly greater domestic air quality co-benefits by engaging with other nations to control GHGs.

This paper presents a long-term (2005-2013) analysis of tropospheric NO2 and HCHO using the Ozone Monitoring Instrument (OMI) satellite and surface ozone using the Texas Commission on Environmental Quality (TCEQ)'s Continuous Ambient Monitoring Stations (CAMS) sites in large cities of Texas. Rather than focusing solely on seasonal changes, the authors consider other harmonic changes using Least Squares Harmonic Estimation (LS-HE), which reduces uncertainty in trends by 5-15%. Tropospheric NO2 observed from OMI demonstrated a downward trend in Austin (-0.06 0.02 1015molecules cm-2yr-1), Dallas (-0.21 0.04 1015molecules cm-2yr-1), Fort Worth (-0.16 0.04 1015molecules cm-2yr-1), Houston (-0.14 0.05 1015molecules cm-2yr-1), and San Antonio (-0.07 0.02 1015molecules cm-2yr-1). These decreasing trends could be attributed to efficient emission control policies. The OMI HCHO observations exhibited both negative and positive annual trends in Austin (+0.09 0.05 1015molecules cm-2yr-1), Dallas (-0.01 0.04 1015molecules cm-2yr-1), Fort Worth (-0.04 0.04 1015molecules cm-2yr-1), Houston (-0.06 0.05 1015molecules cm-2yr-1), and San Antonio (+0.01 0.03 1015molecules cm-2yr-1) mainly resulting from differences in decreases in anthropogenic emissions, the impact of climatic variations and biogenic emissions. The discrepancy between the annual trends of HCHO and NO2 provides strong evidence that all of the cities gradually became more NOx-sensitive. Summertime ozone surface mixing ratio trends are mainly governed by NO2 annual trends as well as climatic influences. We obtained the downward annual trend of the ozone mixing ratio from CAMS observations for Austin (-0.27 0.17 ppbv yr-1), Dallas (-0.41 0.18 ppbv yr-1), Fort Worth (-0.55 0.18 ppbv yr-1), Houston (-0.53 0.18 ppbv yr-1), and San Antonio (-0.03 0.16 ppbv yr-1). The results show that prevailing chemical conditions control the annual trends of surface ozone. While Austin and San Antonio exhibited similar declines in OMI NO2 levels, a more prevailing NOx-sensitive regime over Austin resulted in its greater sensitivity to NO2 changes. The dominant NOx-saturated regime over Dallas, however, led to its lower sensitivity to high annual downward trends of OMI NO2. To investigate the impact of reductions in ozone precursors on surface ozone, we employ the WRF-Chem regional model to simulate NO2, HCHO, and O3 levels in September 2013 in two scenarios using National Emissions Inventory 2005 (NEI-2005) and NEI-2011 anthropogenic emissions. The model describes a decrease in ozone levels from the first to second scenario in Dallas (3%) and Forth Worth (0.5%) but an increase in Austin, San Antonio, and Houston (0.7%, 0.8% and 7%, respectively), which may be attributed to uncertainty in VOC concentrations and overestimation of NOx in the NEI-2005 inventory that led to inaccurate HCHO/NO2 ratios. The results indicate that the emission uncertainties should be constrained to better inform emissions control policy based on air quality modeling.

Raquel A. Silva, J. Jason West, Jean-Fran ois Lamarque, Drew T. Shindell and the ACCMIP modelers

Ambient air pollution from ground-level ozone and fine particulate matter (PM_2.5) has an adverse impact on human health, particularly on cardiovascular and respiratory mortality and lung cancer. Future air pollutant concentrations will be driven by natural and anthropogenic emissions and by climate change due to its influence on air quality. To support the IPCC AR5, four Representative Concentration Pathway scenarios (RCPs) of future global greenhouse gas and air pollutant emissions were projected to 2100. The ACCMIP ensemble of chemistry climate models simulated future concentrations of ozone and PM_2.5 using anthropogenic and biomass burning emissions from the RCPs.

We use global distributions of the concentrations of ozone and PM_2.5 from the ACCMIP ensemble to quantify the human mortality impacts of future ambient air pollution, and separate the impact of future climate change from that of changing emissions. Premature mortality in 2030, 2050 and 2100 is estimated for each scenario and for each model using a health impact function based on changes in concentrations of ozone and PM_2.5 relative to 2000. We use future population and cause-specific baseline mortality rates as projected by the International Futures (IFs) integrated modeling system. Model output and IFs projections were processed and regridded to a common horizontal resolution (0.5 x0.5 ). We isolate the impact of climate change by comparing simulations with present emissions and climate versus present emissions but future climate.

Global PM_2.5 mortality generally decreases in the future in most RCP scenarios, consistent with the projected decrease in most air pollutant emissions. Ozone mortality increases in some scenarios/periods/regions, particularly in RCP8.5 in 2100, likely driven by the large increase in methane emissions and by the net effect of climate change in this scenario. Climate change has a significant contribution to future premature mortality (due to changing emissions and climate change), particularly in 2100: 40% of ozone-related deaths and over 16% of PM_2.5-related deaths. Mortality estimates differ among models, and the reported uncertainty accounts for uncertainty in relative risk associated with exposure and the spread of model results. Increases in exposed population and in baseline mortality rates of respiratory diseases magnify the impact on mortality of changes in air pollutants concentrations.

Brittany Phillips, Kyle Dawson, Taylor Royalty, Robert Reed, Markus Petters, and Nicholas Meskhidze

North Carolina State University

Aerosols play an important role in controlling the Earth's radiation balance, cloud formation and microphysical properties, biogeochemical cycling of nutrients, the chemistry of the troposphere, and have been increasingly recognized for their adverse effects on air quality and human health. Since aerosols arise from both natural sources and anthropogenic activities, it becomes increasingly important to accurately characterize the relative impacts of aerosols of different origin. The characterization of both emission and deposition fluxes of aerosols is important for understanding the life cycle of aerosols. Due to experimental challenges, the direct measurement of size- and composition-resolved emission fluxes are rare.

Here we present a first analysis from a newly constructed aerosol flux measurement system with the capability of retrieving hygroscopicity- and size-resolved particle fluxes. The instrument suite includes a 3-D ultrasonic anemometer, a scanning mobility particle sizer (SMPS), a hygroscopicity tandem differential mobility analyzer (HTDMA), a single particle soot photometer (SP2), and a cloud condensation nuclei (CCN) counter. Results are presented field measurements on marine aerosol production at the US Army Corps of Engineers' Field Research Facility in Duck, NC, conducted during spring 2015. Measurements include 200 nm sized aerosol growth factor distributions and sea spray particle flux measurements, total sub-micron sized aerosol concentration and hygroscopicity distributions, and size-resolved black carbon particle concentration and fluxes. Additional information including meteorological (wind speed and direction, and air temperature) and seawater (salinity, depth, pH, temperature, dissolved oxygen, and Chlorophyll a relative fluorescent units) parameters were also recorded. The presentation will discuss the results for sub-micron aerosol loadings, size distribution, and CCN properties under variable environmental conditions. These new datasets could reduce uncertainties in tropospheric aerosol budget and provide the critical data required for improved regional and global scale modeling.

This presentation will describe the inline coupling of CMAQ with an advanced, next-generation atmospheric modeling system capable of simulating from global down to meter scales. The modeling system was developed to study how changes in climate and landscape will affect regional air pollution, while at the same time simulating how local changes in emissions and land use can feedback and influence the entire global climate system. An unstructured grid is used to cover the entire earth for simulating global climate and long-range transport of pollutants. The mesh can seamlessly telescope down to much higher resolution in areas of interest for simulating local and urban air quality.

An evaluation of the modeling system over the U.S. will be shown, along with an analysis of long-range transport of pollutants on the global unstructured mesh. The global representation and transport of ozone will be shown, since it has an important background contribution due to the transport of stratospheric ozone into the troposphere from weather systems. Applications on how this modeling system can be used to study air quality and climate interactions will be discussed.

Karoline K. Johnson, Michael H. Bergin, Duke University

Armistead G. Russell, Georgia Institute of Technology

Gayle S.W. Hagler, US EPA

Low cost air quality sensors are becoming more prevalent and widely used in a variety of monitoring applications. Not only are these sensors lower cost but their smaller size, lower power requirements and lack of need for an external shelter allows them to be much more portable and easier to deploy. Although it should be noted that before sensors can be used to make measurements of any utility they must first be evaluated for precision, accuracy and detection limits to determine if the information that they supply is meaningful. At this time there are many sensors available for both particulate matter and gas-phase pollutants that do not adequately characterize pollutants at typical ambient concentrations. For this work a variety of well characterized pollutant sensors including particulate matter (PM), black carbon (BC), and nitrogen oxides (NOx) sensors were paired with a carbon dioxide (CO₂) sensor and placed at the roadside to monitor concentrations and estimate on-road emissions factors. These emissions factors are calculated by measuring the change in pollutant concentration over the change in CO₂ concentration and assuming a carbon fuel composition. Emissions factors can vary based on a variety of factors including driving patterns, fleet composition, and varying regulations. This makes it especially important to know emissions factors at a variety of locations since values measured at one site may not be applicable in other locations and will change over time as fleets and regulations change. Tests were run at the roadside in both Atlanta, GA and in Durham, NC.

The sensor approach we will describe could be used at roadside sites across the United States and even internationally to provide air quality data and emissions factors for a variety of locations. This could provide valuable information to policy makers, researchers, and citizens about these sites at costs orders of magnitude less than conventional methods ($100s-$1000s versus $10,000s-$100,000s for conventional measurements). Emissions factor packages like the ones used in this study have the potential to also be used to generate values for additional sources such as ports, airports, or smaller scale sources like trash burning piles or cook stoves. This data also has the potential to improve modeling efforts providing a denser array of air quality information and more specific emissions factors to be used in modeling efforts.