Final Report: Assessing anthropogenic impact on secondary pollutant formation in the South Eastern US via airborne formaldehyde measurements

EPA Grant Number: R835406
Title: Assessing anthropogenic impact on secondary pollutant formation in the South Eastern US via airborne formaldehyde measurements
Investigators: Keutsch, Frank N
Institution: University of Wisconsin - Madison
EPA Project Officer: Hunt, Sherri
Project Period: April 1, 2013 through March 31, 2015 (Extended to March 31, 2016)
Project Amount: $194,183
RFA: Anthropogenic Influences on Organic Aerosol Formation and Regional Climate Implications (2012) RFA Text |  Recipients Lists
Research Category: Air Quality and Air Toxics , Global Climate Change , Climate Change , Air


The primary objective of the research was to improve our mechanistic understanding of the anthropogenic influence on processes transforming primary emissions into secondary pollutants, especially ozone and secondary organic aerosol (SOA), and to constrain the temporal and spatial scales of this anthropogenic impact in the southeastern United States. Specifically, airborne measurements of formaldehyde, in conjunction with a full suite of chemical and meteorological observations, were used to evaluate the impact of anthropogenic volatile organic carbon (VOC) and nitrogen oxide (NOx) emissions on ozone and SOA production efficiency at the urban-rural interface. Analysis focused on quantifying the degree of fragmentation versus functionalization, and the related extent of chain terminating reactions during the photochemical lifecycle of biogenic and anthropogenic VOCs. These factors determine the relative ozone production rates and mass of reactive carbon in the gas and condensed phases. We evaluated how synergistic and competitive mechanisms interact to influence SOA loadings and optical properties. Formaldehyde measurements also enable improved isoprene emission models that then can be used as a reference for isoprene inferred from formaldehyde via satellite retrievals, which are critical to regional models of air quality. The objectives of the research did not change from the original application and there has been no change in key personnel.

Summary/Accomplishments (Outputs/Outcomes):

The work under this grant was extremely successful in pursuing the objective of investigating anthropogenic influence on secondary pollutant formation in the southeastern United States, as evidenced by the large number of presentations and publications disseminating the results associated with this grant. Below we discuss each output/outcome with respect to achieving the specific objectives of the grant.

1. We successfully measured formaldehyde on the NOAA P-3 as part of the Southeast Nexus (SENEX) component of the Southeast Atmosphere Study (SAS) during which the instrument performed extremely well for all flights. Data were promptly submitted to the official data archive and are being used by our research team as well as numerous collaborators. A primary goal was sampling across a wide range of anthropogenic influence and Figure 1 shows that this was achieved (see also Warneke, et al. Atmos. Meas. Techn. 2016).

Atmospheric oxidation of VOCs is directly coupled to the formation of secondary pollutants such as ozone and secondary organic aerosol. These oxidation processes also form formaldehyde and allow the use of formaldehyde as a downstream test of how well VOC oxidation is represented in models. The different P-3 flight patterns (Figure 1) provide the opportunity to conduct a number of different studies, such as investigation of the chemistry within plumes and contrasting urban and rural regimes of secondary pollutant formation.

Figure 1. Covariation of isoprene, NO, and HCHO mixing ratios in the summertime southeast United States. Data are limited to daytime boundary-layer observations. Histograms show the corresponding NO and isoprene distributions. A large variety of regions was sampled as well as VOC / NO regimes spanning a wide range of anthropogenic influence (shown here NOx).

2. Ratio of glyoxal to formaldehyde (RGF) as metric of anthropogenic influence (Kaiser, et al., Atmos. Chem. Phys. 2015;15:7571-7583)

The yield of formaldehyde (HCHO) and glyoxal (CHOCHO) from oxidation of VOCs depends on precursor VOC structure and the concentration of NOx (NOx = NO + NO2). Previous work has proposed that the ratio of CHOCHO to HCHO (RGF) can be used as an indicator of precursor VOC speciation, and absolute concentrations of the CHOCHO and HCHO as indicators of NOx. Because this metric is measurable by satellite, it is potentially useful on a global scale; however, absolute values and trends in RGF have differed between satellite and ground-based observations. To investigate potential causes of previous discrepancies and the usefulness of this ratio, we analyzed measurements of CHOCHO and HCHO over the southeastern United States (SE US) from the 2013 SENEX (Southeast Nexus) flight campaign (Figure 2), and compared these measurements with OMI (Ozone Monitoring Instrument) satellite retrievals.

Figure 2. Daytime flight tracks colored by HCHO, CHOHO, and RGF. Power plant markers are scaled by NOx emissions.

Our high time-resolution flight measurements show that high RGF is associated with monoterpene emissions, low RGF is associated with isoprene oxidation, and emissions associated with oil and gas production can lead to small-scale variation in regional RGF (Figures 3-5).

Figure 3. The relationship of CHOCHO and HCHO for each flight, gridded to OMI satellite resolution. Flights with extreme values of  inlude those to the Haynesville shale (10 and 25 June) and the Ozarks (26 June).

Figure 4. Flight tracks for 10 June (a, c) and 25 June (b, d) over the Haynesville shale, colored by RGF and the measured monoterpene mixing ratio. The southeast corner highlights high RGF in a region with high monoterpene concentrations. The blue circle (b) indicates the location of high RGF discussed further in the text. Figure 4 shows meteorological and trace gas measurements acquired at this location. National parks are shown in green, and the Kisatchie National Forest is labeled in (a).

Figure 5. Flight track for 26 June over the Fayetteville shale, the independence power plant, and the Ozarks, colored by the specified trace gas mixing ratio and RGF. The blue arrow highlights the region of elevated monoterpene mixing ratios. National forests are shown in green.

During the summertime in the southeast United States, RGF is not a reliable diagnostic of anthropogenic VOC emissions, as HCHO and CHOCHO production are dominated by isoprene oxidation. However, this also reflects the fact that anthropogenic VOC oxidation is of low significance compared to isoprene oxidation with respect to ozone production and overall oxidation rate. Our results show that the new CHOCHO retrieval algorithm reduces the previous disagreement between satellite and in situ RGF observations. As the absolute values and trends in RGF observed during SENEX are largely reproduced by OMI observations, we conclude that satellite-based observations of RGF can be used alongside knowledge of land use as a global diagnostic of dominant hydrocarbon speciation.

Overall, the flight-based measurements presented here show that RGF can be indicative of VOC speciation. High RGF (> 3%) is observed consistently in areas with high monoterpene emissions, and low RGF (< 2.5%) is associated with strong isoprene emissions. The previously observed fast and short (2–5 min) increase in RGF in Di-Gangi, et al. (2012) may have been a result of extremely fresh emissions (e.g., diesel trucks emit at a rate of CHOCHO / HCHO = 9.4%; Schauer, et al., 1999), and not indicative of larger-scale changes in dominant VOC speciation. Emissions associated with oil and gas production areas can cause RGF to deviate from the values observed over their background levels (Figures 3-5). However, the absolute value of RGF in such regions likely is dependent on background BVOC (biogenic volatile organic compounds) emissions, speciation of AVOCs (anthropogenic volatile compounds), and any direct OVOC (oxygenated volatile organic compounds) emissions.

Compared to previous literature, absolute values of flight-based RGF are in better agreement with satellite observations using the new CHOCHO retrieval algorithms. While time resolution plays a large role in direct comparisons of point-based measurements and satellite retrievals, the trend of high RGF over areas with monoterpenes and low RGF over areas with isoprene is broadly in agreement for the two platforms. With these trends validated by ground measurements, RGF based on satellite retrievals may be useful as a diagnostic of BVOC emissions. As these retrievals become available at higher time and spatial resolutions, RGF can be used to help identify the speciation of VOCs leading to secondary pollutant formation on a regional scale.

The previously discussed analysis demonstrates that RGF reflects the dominant VOCs, often isoprene in the southeastern United States, and this makes small-scale influence, e.g., of anthropogenic VOCs, hard to observe. However, the absolute concentration of formaldehyde and glyoxal represent a stronger tracer for anthropogenic influence via NOx within one dominant VOC domain. Figure 6 shows multiple intersections of an urban plume with high NOx within an isoprene dominated airmass.

Figure 6. Measurements acquired on the 12 June flight. Shaded regions indicate high anthropogenic influence. While the measurements alter between AVOC/high NOx and BVOC/low NOx regimes, little change is seen in RGF. The maximum values of NOx and (MVK + MACR) / isoprene fall above the limits shown here.

Whereas the RGF values (inset g) remain unchanged, as isoprene continues to be the dominant VOC, the absolute concentrations are clearly increased as a result of anthropogenic NOx. This likely results from increased oxidant concentrations, resulting in faster VOC oxidation chemistry, consistent with the lower isoprene concentrations and higher oxidation product to isoprene ratios, which reflects the extent of oxidation of the VOCs in the airmass. This is an important result within the context of secondary pollutant formation as the oxidation processes forming formaldehyde are directly coupled to the formation of secondary pollutants.

3. Analysis of anthropogenic influence over the photochemical lifecycle of biogenic isoprene (Wolfe, et al. Atmos. Chem. Phys. 2016;16:2597-2610, doi:10.5194/acp-16-2597-2016).

The chemical link between isoprene and formaldehyde (HCHO) is a strong, nonlinear function of anthropogenic influence via NOx (i.e., NO + NO2) and provides direct information on the radical chain propagation versus terminating that is central to VOC oxidation. Therefore, it also is a benchmark for overall photochemical mechanism performance with regard to VOC oxidation. This relationship also is a linchpin for top-down isoprene emission inventory verification from orbital HCHO column observations.

Using a comprehensive suite of airborne in situ observations over the southeast United States, we quantified HCHO production across the urban-rural spectrum (Figure 7).

Figure 7. NOx modulates the relationship between observed HCHO and calculated initial isoprene mixing ratios. Symbols denote all 1 Hz data. Dashed lines illustrate representative major-axis fits of NOx-grouped subsets at mean NOx values of 170, 380, and 810 pptv (see text for details of fitting procedure). The slope (b) and intercept (c) of these fits are the prompt HCHO yield and background HCHO mixing ratio, respectively. Error bars in (b) and (c) are 3σ fitting uncertainties.

Analysis of isoprene and its major first-generation oxidation products allowed us to define both a “prompt” yield of HCHO (molecules of HCHO produced per molecule of freshly emitted isoprene) and the background HCHO mixing ratio (from oxidation of longer-lived hydrocarbons potentially including isoprene oxidation products). Over the range of observed NOx values (roughly 0.1–2 ppbv), the prompt yield increased by a factor of 3 (from 0.3 to 0.9 ppbv ppbv−1), while background HCHO increased by a factor of 2 (from 1.6 to 3.3 ppbv). Therefore, we see a strong anthropogenic influence on HCHO formation and VOC oxidation (Figure 8).

Figure 8. Comparison of observed and model-derived relationships between HCHO and initial isoprene versus NOx. Slopes (a) and intercepts (b) are calculated as described in the text. The observed values (blue line with shading) are the same as those shown in Figure 7b–c. Symbols represent fit results for the global AM3 model (red circles) and the 0-D UWCM box model (black diamonds). Error bars denote 3σ fitting uncertainties.

We applied the same method to evaluate the performance of both a global chemical transport model (AM3) and a measurement-constrained 0-D steady-state box model. Both models reproduced the NOx dependence of the prompt HCHO yield, illustrating that models with updated isoprene oxidation mechanisms can adequately capture the link between HCHO and recent isoprene emissions. On the other hand, both models underestimated background HCHO mixing ratios, suggesting missing HCHO precursors, inadequate representation of later-generation isoprene degradation and/or underestimated hydroxyl radical concentrations. Detailed process rates from the box model simulation demonstrate a three-fold increase in HCHO production across the range of observed NOx values, driven by a 100% increase in OH and a 40% increase in branching of organic peroxy radical reactions to produce HCHO. Figure 9 compares the total HCHO and peroxy (RO2) radical production rate as a function of NOx. The latter is an important quantity as it is directly related to the ozone production in VOC limited regimes and as it represents the total processing rate of organic carbon, which also is related to SOA formation.

Figure 9. NOx dependence of chemical properties related to HCHO production, extracted from the UWCM simulation of SENEX observations. (a) Production rates for HCHO (blue) and total RO2 (orange). All quantities are averaged over NOx using 10 bins with equal numbers of points. Solid lines show the mean and shading is 1σ variability.

4. Airborne measurements of the atmospheric emissions from a fuel ethanol refinery (de Gouw, et al. J. Geophys. Res., 2015;120, doi:10.1002/ 2015JD023138).

Joost de Gouw (NOAA, CIRES and UC Boulder) used our formaldehyde measurements as part of an analysis of ethanol emissions from fuel ethanol refineries.

Ethanol made from corn now constitutes approximately 10% of the fuel used in gasoline vehicles in the United States. The ethanol is produced in more than 200 fuel ethanol refineries across the nation. DeGouw used airborne measurements, including formaldehyde, from the SENEX campaign downwind from Decatur, Illinois, where the third largest fuel ethanol refinery in the United States is located to estimate emissions of reactive VOCs, primarily ethanol (Figure 10). The estimated emissions were compared with the total point source emissions in Decatur according to the 2011 National Emissions Inventory (NEI-2011), in which the fuel ethanol refinery represents 68.0% of sulfur dioxide (SO2), 50.5% of nitrogen oxides (NOx =  NO + NO2), 67.2% of volatile organic compounds (VOCs), and 95.9% of ethanol emissions. Emissions of SO2 and NOx from Decatur agreed with NEI-2011, but emissions of several VOCs were underestimated by factors of 5 (total VOCs) to 30 (ethanol) (Figure 11). By combining the NEI-2011 with fuel ethanol production numbers from the Renewable Fuels Association, deGouw calculated emission intensities, defined as the emissions per ethanol mass produced. Emission intensities of SO2 and NOx are higher for plants that use coal as an energy source, including the refinery in Decatur. By comparing with fuel-based emission factors, we find that fuel ethanol refineries have lower NOx, similar VOC, and higher SO2 emissions than from the use of this fuel in vehicles. The VOC emissions from refining could be higher than from vehicles, if the underestimated emissions in NEI-2011 downwind from Decatur extend to other fuel ethanol refineries. Finally, chemical transformations of the emissions from Decatur were observed, including formation of new particles, nitric acid, peroxyacyl nitrates, aldehydes, ozone, and sulfate aerosol.

Figure 10. Estimated fluxes of (a) sulfur, (b) nitrogen, (c) organic carbon species, and (d) ozone as a function of distance and transport time from the ADM plant in Decatur. The uncertainties in the estimated fluxes are ±50%.
Figure 11. Emission intensities of SO2, NOx, VOCs, and ethanol from fuel ethanol refineries derived from combining the 2011 National Emissions Inventory and ethanol production numbers reported by the Renewable Fuels Association. Note that the x axes are on logarithmic scales. Average emission intensities (total emissions divided by total production) are shown by the solid blue lines. Plants that use coal are indicated in black, all others in red. Emission intensities for the ADM plant calculated from the NEI-2011 and estimated from the measurements downwind from Decatur here are shown by the arrows. For comparison, fuel-based emission factors from gasoline vehicles are shown by the dashed blue lines.

Joost de Gouw, one of our SENEX collaborators at NOAA, Boulder, published this important finding in the Journal of Geophysical Research (see publications for this grant).

5. Ongoing Work

There are numerous additional studies ongoing resulting from this grant and the most important within the context of the grant's objectives are summarized here. Additional work has been submitted that directly relates to SOA formation, in this case via glyoxal. This work by our collaborators Jingqiu Mao and Jingyi Li (GFDL Princeton) is close to publication (replies to reviews in JGR submitted) and focuses on our formaldehyde measurements as well as glyoxal measurements during SENEX to investigate the ability of current model mechanisms to accurately predict SOA formed from glyoxal in the southeast United States, one of the main SOA formation pathways from isoprene. Additional work led by Joos de Gouw at NOAA is close to being submitted for publication. This work focuses on our formaldehyde as well as isoprene measurements during SENEX to evaluate anthropogenic influence on oxidative capacity in the southeast United States, a critical aspect of anthropogenic influence on secondary pollutant production. Margaret Marvin, from our direct science team, is conducting an additional analysis comparing the link between isoprene and formaldehyde in different model mechanisms. Lastly, Jennifer Kaiser, also a member of our direct science team, is conducting a study using the SENEX formaldehyde and other measurements that builds on her and Glenn Wolfe’s work to advance the use of formaldehyde satellite retrieval for obtaining isoprene emissions.


Successful formaldehyde measurements were obtained and data submitted to archive. We have successfully evaluated the capability of using RGF as an indicator of anthropogenic influence on secondary pollutant formation. In the southeast United States, VOC oxidation is dominated by isoprene so that RGF shows little variability (Kaiser, et al. ACP 2015). However, the absolute concentration of formaldehyde reflects the rate of oxidative processing as a function of anthropogenic NOx (Wolfe, et al., 2016 and de Gouw, et al., in preparation). The results provide direct information on the details of oxidative VOC processing, such as the relative importance of chemical pathways recycling or terminating radicals and fragmenting or functionalizing reactive carbon. The work under this grant also has resulted in publication of findings on anthropogenic influence via fuel ethanol refineries. Our observations and scientific results have been integrated into regional models (Li, et al., in revision JGR; Marvin, et al., in preparation) and will provide an improved understanding of anthropogenic influence on secondary pollutant formation.

Journal Articles on this Report : 7 Displayed | Download in RIS Format

Other project views: All 24 publications 7 publications in selected types All 7 journal articles
Type Citation Project Document Sources
Journal Article de Gouw JA, McKeen SA, Aikin KC, Brock CA, Brown SS, Gilman JB, Graus M, Hanisco T, Holloway JS, Kaiser J, Keutsch FN, Lerner BM, Liao J, Markovic MZ, Middlebrook AM, Min K-E, Neuman JA, Nowak JB, Peischl J, Pollack IB, Roberts JM, Ryerson TB, Trainer M, Veres PR, Warneke C, Welti A, Wolfe GM. Airborne measurements of the atmospheric emissions from a fuel ethanol refinery. Journal of Geophysical Research: Atmospheres 2015;120(9):4385-4397. R835406 (2014)
R835406 (Final)
  • Full-text: Wiley Online-Full Text-HTML
  • Abstract: Wiley Online-Abstract
  • Other: Wiley Online-Full Text-PDF
  • Journal Article Kaiser J, Wolfe GM, Min KE, Brown SS, Miller CC, Jacob DJ, de Gouw JA, Graus M, Hanisco TF, Holloway J, Peischl J, Pollack IB, Ryerson TB, Warneke C, Washenfelder RA, Keutsch FN. Reassessing the ratio of glyoxal to formaldehyde as an indicator of hydrocarbon precursor speciation. Atmospheric Chemistry and Physics 2015;15(13):7571-7583. R835406 (2014)
    R835406 (Final)
  • Full-text: ACP-Full Text-PDF
  • Abstract: ACP-Abstract
  • Other: ResearchGate-Abstract & Full Text-PDF
  • Journal Article Kaiser J, Skog KM, Baumann K, Bertman SB, Brown SB, Brune WH, Crounse, JD, de Gouw, JA, Edgerton, ES, Feiner PA, Goldstein AH, Koss A, Misztal PK, Nguyen TB, Olson KF, St. Clair JM, Teng AP, Toma S, Wennberg PO, Wild RJ, Zhang L, Keutsch FN. Speciation of OH reactivity above the canopy of an isoprene-dominated forest. Atmospheric Chemistry and Physics 2016;16(14):9349-9359. R835406 (Final)
    R835407 (Final)
  • Full-text: ACP-Full Text PDF
  • Abstract: ACP-Abstract & Full Text HTML
  • Journal Article Li J, Mao J, Min K-E, Washenfelder RA, Brown SS, Kaiser J, Keutsch FN, Volkamer R, Wolfe GM, Hanisco TF, Pollack IB, Ryerson TB, Graus M, Gilman JB, Lerner BM, Warneke C, de Gouw JA, Middlebrook AM, Liao J, Welti A, Henderson BH, McNeill VF, Hall SR, Ullman K, Donner LJ, Paulot F, Horowitz LW. Observational constraints on glyoxal production from isoprene oxidation and its contribution to organic aerosol over the Southeast United States. Journal of Geophysical Research: Atmospheres 2016;121(16):9849-9861. R835406 (Final)
  • Abstract: AGU Publications-Abstract
  • Journal Article Millet DB, Baasandorj M, Farmer DK, Thornton JA, Baumann K, Brophy P, Chaliyakunnel S, de Gouw JA, Graus M, Hu L, Koss A, Lee BH, Lopez-Hilfiker FD, Neuman JA, Paulot F, Peischl J, Pollack IB, Ryerson TB, Warneke C, Williams BJ, Xu J. A large and ubiquitous source of atmospheric formic acid. Atmospheric Chemistry and Physics 2015;15(11):6283-6304. R835406 (Final)
    R835402 (2014)
    R835402 (Final)
  • Full-text: ACP-Full-text-PDF
  • Abstract: ACP-Abstract
  • Other: ResearchGate-Abstract
  • Journal Article Warneke C, Trainer M, de Gouw JA, Parrish DD, Fahey DW, Ravishankara AR, Middlebrook AM, Brock CA, Roberts JM, Brown SS, Neuman JA, Lerner BM, Lack D, Law D, Hubler G, Pollack I, Sjostedt S, Ryerson TB, Gilman JB, Liao J, Holloway J, Peischl J, Nowak JB, Aikin KC, Min K-E, Washenfelder RA, Graus MG, Richardson M, Markovic MZ, Wagner NL, Welti A, Veres PR, Edwards P, Schwarz JP, Gordon T, Dube WP, McKeen SA, Brioude J, Ahmadov R, Bougiatioti A, Lin JJ, Nenes A, Wolfe GM, Hanisco TF, Lee BH, Lopez-Hilfiker FD, Thornton JA, Keutsch FN, Kaiser J, Mao J, Hatch CD. Instrumentation and measurement strategy for the NOAA SENEX aircraft campaign as part of the Southeast Atmosphere Study 2013. Atmospheric Measurement Techniques 2016;9(7):3063-3093. R835406 (Final)
    R835410 (2013)
  • Full-text: AMT-Full Text PDF
  • Abstract: AMT-Abstract
  • Journal Article Wolfe GM, Kaiser J, Hanisco TF, Keutsch FN, de Gouw JA, Gilman JB, Graus M, Hatch CD, Holloway J, Horowitz LW, Lee BH, Lerner BM, Lopez-Hilifiker F, Mao J, Marvin MR, Peischl J, Pollack IB, Roberts JM, Ryerson TB, Thornton JA, Veres PR, Warneke C. Formaldehyde production from isoprene oxidation across NOx regimes. Atmospheric Chemistry and Physics 2016;16(4):2597-2610. R835406 (Final)
  • Full-text: ACP-Full Text PDF
  • Abstract: ACP-Abstract
  • Supplemental Keywords:

    formaldehyde, secondary pollutant formation, ozone, anthropogenic influence 

    Progress and Final Reports:

    Original Abstract
    2013 Progress Report
    2014 Progress Report