Final Report: Bayesian and Adjoint Inverse Model Analysis of PM Sources in the United States Using Observations from Surface, Aircraft, and Satellite Platforms

EPA Grant Number: R832158
Title: Bayesian and Adjoint Inverse Model Analysis of PM Sources in the United States Using Observations from Surface, Aircraft, and Satellite Platforms
Investigators: Jacob, Daniel J. , Seinfeld, John
Institution: Harvard University , California Institute of Technology
EPA Project Officer: Chung, Serena
Project Period: January 1, 2005 through December 31, 2007 (Extended to December 31, 2008)
Project Amount: $450,000
RFA: Source Apportionment of Particulate Matter (2004) RFA Text |  Recipients Lists
Research Category: Particulate Matter , Air Quality and Air Toxics , Air

Objective:

Our objective in this project was to better quantify and understand particulate matter (PM) sources in the United States on a continental scale through combined forward, Bayesian and adjoint inverse model analyses of observations from surface sites, aircraft, and satellites. A central component of our analysis was the GEOS-Chem global chemical transport model, which provides a continental and intercontinental perspective on sources and also allows the interpretation of aerosol column data from satellites.

Summary/Accomplishments (Outputs/Outcomes):

The project has produced eight publications in the peer-reviewed literature acknowledging support from the EPA-STAR Program. We summarize our findings below with shortened abstracts from the individual publications. Full papers are included as chapters in the body of the report. Support from the grant contributed to the education of Harvard Ph.D. students Colette Heald (now asst. prof. at Colorado State U. ), Duncan Fairlie (now research scientist at NASA/LaRC), and Easan Drury (Ph.D. expected 2009), Caltech Ph.D. student Daven Henze (now asst. prof. at U. Colorado), and Harvard postdoc Rokjin Park (now asst. prof. at Seoul National U.)

2.1 Concentrations and sources of organic aerosols in the free troposphere over North America, by C.L. Heald, D.J. Jacob, S. Turquety, R.C. Hudman, R.J. Weber, A.P. Sullivan, R.E. Peltier, E.L. Atlas, J.A. de Gouw, C. Warneke, J.S. Holloway, A. Neuman, F.M. Flocke, and J. H. Seinfeld, J. Geophys. Res., 111, D23S47, 2006.

Aircraft measurements of water-soluble organic carbon (WSOC) aerosol over the northeastern US during summer 2004 (ITCT-2K4 campaign) were simulated with GEOS-Chem to test our understanding of the sources of organic carbon (OC) aerosol in the free troposphere and the boundary layer (BL). WSOC aerosol concentrations observed from the aircraft in the BL are consistent with OC aerosol observed at the IMPROVE surface network. The model is too low by 30%, which we attribute to missing secondary organic aerosol (SOA) formation processes. We conclude that primary anthropogenic emission and biogenic SOA formation contribute OC aerosol sources of comparable magnitude over the northeastern US in summer.

2.2 Global secondary organic aerosol from isoprene oxidation, by D.K. Henze and J.H. Seinfeld, Geophys. Res. Lett., 33, L09812, 2006.

We find that inclusion of isoprene as a source of SOA in the GEOS-Chem global model increases the global burden of SOA from all sources by more than a factor of two. The isoprene source substantially increases SOA concentrations in the free troposphere, because isoprene oxidation products have much greater concentrations at higher altitudes than other biogenic SOA precursors, highlighting the importance of semi-volatile organics for SOA formation. This additional source of SOA enhances production of SOA from other parent hydrocarbons by 17%.

2.3 Global modeling of secondary organic aerosol formation from aromatic hydrocarbons: high vs. low-yield pathways, by D.K. Henze, J.H. Seinfeld, N.L. Ng, J.H. Kroll, T.-M. Fu, D.J. Jacob, and C.L. Heald, Atmos. Chem. Phys., 8, 2405-2421, 2007.

Formation of SOA from the aromatic species toluene, xylene, and, for the first time, benzene, was added to a global chemical transport model (GEOS-Chem). A simple mechanism was presented that accounts for competition between low and high-yield pathways of SOA formation, wherein secondary gas-phase products react further with either nitric oxide (NO) or hydroperoxy radical (HO­2) to yield semi- or non-volatile products, respectively. Benzene is estimated to be the most important aromatic species with regards to global formation of SOA, with a total production nearly equal that of toluene and xylene combined. Global production of SOA from aromatic sources via the mechanisms identified here is estimated at 3.5 Tg/yr, resulting in a global burden of 0.08 Tg, twice as large as previous estimates.

2.4 Fire and biofuel contributions to annual mean aerosol mass concentrations in the United States, by R.J. Park, D.J. Jacob, and J.A. Logan, Atmos. Environ., 41, 7389-7400, 2007.

We estimated the contributions from biomass burning (summer wildfires, other fires, residential biofuel, and industrial biofuel) to seasonal and annual aerosol concentrations in the United States . Our approach was to use total carbonaceous (TC) and non-soil potassium (ns-K) aerosol mass concentrations for 2001–2004 from the nationwide IMPROVE network of surface sites, together with satellite fire data. We find that summer wildfires largely drive the observed interannual variability of TC aerosol concentrations in the United States . The summer wildfire contributions to annual TC aerosol concentrations for 2001–2004 are 0.26 μg C m−3 in the west and 0.14 μg C m−3 in the east; Canadian fires are a major contributor in the east. Non-summer wildfires and prescribed burns contribute on an annual mean basis 0.27 and 0.31 μg C m−3 in the west and the east, highest in the southeast because of prescribed burning. Residential biofuel is a large contributor in the northeast with annual mean concentration of up to 2.2 μg C m−3 in Maine . Industrial biofuel (mainly paper and pulp mills) contributes up to 0.3 μg C m−3 in the southeast. Total annual mean fine aerosol concentrations from biomass burning average 1.2 and 1.6 μg m−3 in the west and east, respectively, contributing about 50% of observed annual mean TC concentrations in both regions and accounting for 30% (west) and 20% (east) of total observed fine aerosol concentrations. Our analysis supports bottom-up source estimates for the contiguous US of 0.7–0.9 Tg C yr−1 from open fires (climatological) and 0.4 Tg C yr−1 from biofuel use.

2.5 The impact of transpacific transport on mineral dust in the United States, by T.D. Fairlie, D.J. Jacob, and R.J. Park , Atmos. Environ., 41, 1251-1266, 2007.

We used GEOS-Chem to estimate the impact of transpacific transport of mineral dust on aerosol concentrations in North America during 2001. We implemented two different dust mobilization schemes in the model (GOCART and DEAD) and find that the best simulation of North American surface observations is achieved by combining the topographic source used in GOCART with the entrainment scheme used in DEAD. This combination restricts dust emissions to year-round arid areas but includes a significant wind threshold for dust mobilization. The model reproduces the timing and distribution of Asian dust outbreaks in North America during April–May. Beyond these outbreaks we find persistent Asian fine dust (averaging 1.2 μg m−3) in surface air over the western US in spring, with much weaker influence (0.25 μg m−3) in summer and fall. Asian influence over the eastern US is 30–50% lower. We find that transpacific sources accounted for 41% of the worst dust days in the western US in 2001.

2.6 Improved algorithm for MODIS satellite retrievals of aerosol optical depths over western North America, by E.E. Drury,  D.J. Jacob, J. Wang, R.J.D. Spurr and K. Chance, J. Geophys. Res., 113, D16204, 2008.

Inference of surface PM concentrations from satellite observations of aerosol optical depth (AOD) can be compromised by inconsistent assumptions of aerosol optical properties and errors in surface reflectance estimates. We developed an improved AOD retrieval algorithm for the MODIS satellite instrument using locally derived surface reflectances and GEOS-Chem aerosol optical properties for the 0.47, 0.65, and 2.13 mm MODIS channels. Comparison with coincident ground-based (AERONET) AOD observations at 16 sites in the western and central US in summer 2004 shows poor correlation in the daily data but the correlation improves as averaging time increases. Averaging over the available coincident observations (n = 11–44 days) results in strong correlations (R0.47mm = 0.90, R0.65mm = 0.67) and a 19% low bias, representing considerable improvement over the operational MODIS AOD products in this region.

2.7 Development of the adjoint of GEOS-Chem, by D.K. Henze, A. Hakami, and J.H. Seinfeld, Atmos. Chem. Phys, 7, 413-433, 2007.

We developed the adjoint of GEOS-Chem for the purpose of this project, focusing on the chemical and thermodynamic relationships between sulfate–ammonium–nitrate aerosols and their gas-phase precursors. The adjoint model was constructed from a combination of manually and automatically derived discrete adjoint algorithms and numerical solutions to continuous adjoint equations. Explicit inclusion of the processes that govern secondary formation of inorganic aerosol was shown to afford efficient calculation of model sensitivities such as the dependence of sulfate and nitrate aerosol concentrations on emissions of SOx, NO­x, and NH3. The accuracy of the adjoint model was extensively verified by comparing adjoint to finite difference sensitivities, which are shown to agree within acceptable tolerances. The potential for inverse modeling using the adjoint of GEOS-Chem was assessed in a data assimilation framework using simulated observations, demonstrating the feasibility of exploiting gas- and aerosol-phase measurements for optimizing emission inventories of aerosol precursors.

2.8 Inverse modeling and mapping US air quality influences of inorganic PM2.5 precursor emissions using the adjoint of GEOS-Chem, by D.K. Henze, J.H. Seinfeld, and D.T. Shindell, Atmos. Chem. Phys., Discuss, 8, 15,031-15,099, 2008.

Influences of specific sources of inorganic PM2.5 on peak and ambient aerosol concentrations in the US were evaluated using a combination of inverse modeling and sensitivity analysis. First, sulfate and nitrate aerosol measurements from the IMPROVE network assimilated using the 4D-Var method into the GEOS-Chem CTM in order to constrain emissions estimates in four separate month-long inversions (one per season). Of the precursor emissions, these observations primarily constrain ammonia (NH3). While the net result is a decrease in estimated US NH3 emissions relative to the original inventory, there is considerable variability in adjustments made to NH3 emissions in different locations, seasons and source sectors, such as focused decreases in the midwest during July, broad decreases throughout the US in January, increases in eastern coastal areas in April, and an effective redistribution of emissions from natural to anthropogenic sources. Implementing these constrained emissions, the adjoint model is applied to quantify the influences of emissions on representative PM2.5 air quality metrics within the US . The resulting sensitivity maps display a wide range of spatial, sectoral and seasonal variability in the susceptibility of the air quality metrics to absolute emissions changes and the effectiveness of incremental emissions controls of specific source sectors. NH3 emissions near sources of sulfur oxides (SOx) are estimated to most influence peak inorganic PM2.5 levels in the East; thus, the most effective controls of NH3 emissions are often disjoint from locations of peak NH3 emissions. Controls of emissions from industrial sectors of SOx and NOx are estimated to be more effective than surface emissions, and changes to NH3 emissions in regions dominated by natural sources are disproportionately more effective than regions dominated by anthropogenic sources. NOx controls are most effective in northern states in October; in January, SOx controls may be counterproductive.

Conclusions:

Results from this project have led to significant improvements in understanding and quantifying PM sources in the US. We developed new parameterizations for organic aerosol and mineral dust sources in models, and showed that these led to improvements in the ability of the GEOS-Chem model to fit observations. We quantified the contributions to PM2.5 from open fires and biofuel use and showed that these were important on an annual basis. We developed a new algorithm to enable the exploitation of aerosol data from the MODIS satellite sensor to constrain PM2.5 concentrations and sources in the US. Finally, we developed an adjoint of the GEOS-Chem CTM and applied it to examine the observational constraints from speciated PM2.5 concentrations in the US on the sources of PM precursors and their regional distributions.


Journal Articles on this Report : 10 Displayed | Download in RIS Format

Other project views: All 25 publications 10 publications in selected types All 10 journal articles
Type Citation Project Document Sources
Journal Article Drury E, Jacob DJ, Wang J, Spurr RJD, Chance K. Improved algorithm for MODIS satellite retrievals of aerosol optical depths over western North America. Journal of Geophysical Research 2008;113:D16204. R832158 (Final)
  • Full-text: Wiley-Full-text PDF
    Exit
  • Abstract: Wiley-Abstract & Full-text HTML
    Exit
  • Journal Article Fairlie TD, Jacob DJ, Park RJ. The impact of transpacific transport of mineral dust in the United States. Atmospheric Environment 2007;41(6):1251-1266. R832158 (2006)
    R832158 (2007)
    R832158 (Final)
  • Full-text: ScienceDirect-Full-text HTML
    Exit
  • Abstract: ScienceDirect-Abstract
    Exit
  • Other: ScienceDirect-Full-text PDF
    Exit
  • Journal Article Heald CL, Jacob DJ, Turquety S, Hudman RC, Weber RJ, Sullivan AP, Peltier RE, Atlas EL, de Gouw JA, Warneke C, Holloway JS, Neuman JA, Flocke FM, Seinfeld JH. Concentrations and sources of organic carbon aerosols in the free troposphere over North America. Journal of Geophysical Research 2006;111:D23S47. R832158 (2006)
    R832158 (2007)
    R832158 (Final)
  • Full-text: Wiley-Full-text PDF
    Exit
  • Abstract: Wiley-Abstract & Full-text HTML
    Exit
  • Journal Article Henze DK, Seinfeld JH. Global secondary organic aerosol from isoprene oxidation. Geophysical Research Letters 2006;33(9):L09812 (4 pp.). R832158 (2006)
    R832158 (2007)
    R832158 (Final)
    R831075 (Final)
  • Full-text: AGU-Full Text PDF
    Exit
  • Abstract: AGU-Abstract
    Exit
  • Journal Article Henze DK, Hakami A, Seinfeld JH. Development of the adjoint of GEOS-Chem. Atmospheric Chemistry and Physics 2007;7(9):2413-2433. R832158 (2006)
    R832158 (2007)
    R832158 (Final)
  • Full-text: ACP-Full-text PDF
    Exit
  • Abstract: ACP-Abstract HTML
    Exit
  • Journal Article Henze DK, Seinfeld JH, Ng NL, Kroll JH, Fu T-M, Jacob DJ, Heald CL. Global modeling of secondary organic aerosol formation from aromatic hydrocarbons:high-vs. low-yield pathways. Atmospheric Chemistry and Physics 2008;8(9):2405-2420. R832158 (2007)
    R832158 (Final)
  • Full-text: ACP-Full Text PDF
    Exit
  • Abstract: ACP-Abstract
    Exit
  • Journal Article Henze DK, Seinfeld JH, Shindell DT. Inverse modeling and mapping US air quality influences of inorganic PM2.5 precursor emissions using the adjoint of GEOS-Chem. Atmospheric Chemistry and Physical Discussions 2008;8(4):15031-15099. R832158 (Final)
  • Full-text: ACP Discussions-Full Text PDF
    Exit
  • Abstract: ACP Discussions-Abstract
    Exit
  • Journal Article Henze DK. Inverse modeling and mapping US air quality influences of inorganic PM2.5 precursor emissions using the adjoint of GEOS-Chem. Atmospheric Chemistry and Physics 2009;9(4):5877-5903. R832158 (Final)
  • Full-text: ACP-Full-text PDF
    Exit
  • Abstract: ACP-Abstract HTML
    Exit
  • Journal Article Park RJ, Jacob DJ, Logan JA. Fire and biofuel contributions to annual mean aerosol mass concentrations in the United States. Atmospheric Environment 2007;41(35):7389-7400. R832158 (2007)
    R832158 (Final)
  • Full-text: Science Direct-Full Text HTML
    Exit
  • Abstract: Science Direct-Abstract
    Exit
  • Other: Science Direct-Full Text PDF
    Exit
  • Journal Article van Donkelaar A, Martin RV, Park RJ, Heald CL, Fu T-M, Liao H, Guenther A. Model evidence for a significant source of secondary organic aerosol from isoprene. Atmospheric Environment 2007;41(6):1267-1274. R832158 (2005)
    R832158 (Final)
  • Full-text: Science Direct-Full Text HTML
    Exit
  • Abstract: Science Direct-Abstract
    Exit
  • Other: Science Direct-Full Text PDF
    Exit
  • Supplemental Keywords:

    Particulate matter, emissions, GEOS-Chem, inverse modeling, model adjoint,, RFA, Scientific Discipline, Air, Ecosystem Protection/Environmental Exposure & Risk, RESEARCH, particulate matter, Environmental Chemistry, Monitoring/Modeling, Monitoring, Environmental Monitoring, Ecological Risk Assessment, Environmental Engineering, remote sensing, particulate organic carbon, atmospheric dispersion models, atmospheric measurements, model-based analysis, chemical characteristics, emissions monitoring, environmental measurement, positive matrix factorization, airborne particulate matter, air quality models, adjoint inverse model analysis, air quality model, carbon particles, air sampling, diesel exhaust, particulate matter mass, mobile sources, PM2.5, aerosol optical depth data, aersol particles, modeling studies, aerosol analyzers, chemical speciation sampling, particle size measurement, carbonaceous particulate matter

    Relevant Websites:

    http://www.as.harvard.edu/ctm/Exit EPA DisclaimerResearch web site of Daniel Jacob
    http://www.che.caltech.edu/groups/jhs/research.shtmlExit EPA Disclaimer Research web site of John Seinfeld

     

    Progress and Final Reports:

    Original Abstract
  • 2005 Progress Report
  • 2006 Progress Report
  • 2007 Progress Report