Journal of Geophysical Research: Atmospheres

Measurements of volatile organic compounds during the 2006 TexAQS/GoMACCS campaign: Industrial influences, regional characteristics, and diurnal dependencies of the OH reactivity

Authors

  • Jessica B. Gilman,

    1. Chemical Sciences Division, NOAA Earth System Research Laboratory, Boulder, Colorado, USA
    2. Cooperative Institute for Research in the Environmental Sciences, University of Colorado, Boulder, Colorado, USA
    Search for more papers by this author
  • William C. Kuster,

    1. Chemical Sciences Division, NOAA Earth System Research Laboratory, Boulder, Colorado, USA
    Search for more papers by this author
  • Paul D. Goldan,

    1. Chemical Sciences Division, NOAA Earth System Research Laboratory, Boulder, Colorado, USA
    2. Cooperative Institute for Research in the Environmental Sciences, University of Colorado, Boulder, Colorado, USA
    Search for more papers by this author
  • Scott C. Herndon,

    1. Aerodyne Research, Inc., Billerica, Massachusetts, USA
    Search for more papers by this author
  • Mark S. Zahniser,

    1. Aerodyne Research, Inc., Billerica, Massachusetts, USA
    Search for more papers by this author
  • Sara C. Tucker,

    1. Chemical Sciences Division, NOAA Earth System Research Laboratory, Boulder, Colorado, USA
    2. Cooperative Institute for Research in the Environmental Sciences, University of Colorado, Boulder, Colorado, USA
    Search for more papers by this author
  • W. Alan Brewer,

    1. Chemical Sciences Division, NOAA Earth System Research Laboratory, Boulder, Colorado, USA
    Search for more papers by this author
  • Brian M. Lerner,

    1. Chemical Sciences Division, NOAA Earth System Research Laboratory, Boulder, Colorado, USA
    2. Cooperative Institute for Research in the Environmental Sciences, University of Colorado, Boulder, Colorado, USA
    Search for more papers by this author
  • Eric J. Williams,

    1. Chemical Sciences Division, NOAA Earth System Research Laboratory, Boulder, Colorado, USA
    2. Cooperative Institute for Research in the Environmental Sciences, University of Colorado, Boulder, Colorado, USA
    Search for more papers by this author
  • Robert A. Harley,

    1. Department of Civil and Environmental Engineering, University of California, Berkeley, California, USA
    Search for more papers by this author
  • Fred C. Fehsenfeld,

    1. Chemical Sciences Division, NOAA Earth System Research Laboratory, Boulder, Colorado, USA
    2. Cooperative Institute for Research in the Environmental Sciences, University of Colorado, Boulder, Colorado, USA
    Search for more papers by this author
  • Carsten Warneke,

    1. Chemical Sciences Division, NOAA Earth System Research Laboratory, Boulder, Colorado, USA
    2. Cooperative Institute for Research in the Environmental Sciences, University of Colorado, Boulder, Colorado, USA
    Search for more papers by this author
  • Joost A. de Gouw

    1. Chemical Sciences Division, NOAA Earth System Research Laboratory, Boulder, Colorado, USA
    2. Cooperative Institute for Research in the Environmental Sciences, University of Colorado, Boulder, Colorado, USA
    Search for more papers by this author

Abstract

[1] An extensive set of volatile organic compounds (VOCs) and other gas phase species were measured in situ aboard the NOAA R/V Ronald H. Brown as the ship sailed in the Gulf of Mexico and the Houston and Galveston Bay (HGB) area as part of the Texas Air Quality (TexAQS)/Gulf of Mexico Atmospheric Composition and Climate Study (GoMACCS) conducted from July–September 2006. The magnitudes of the reactivities of CH4, CO, VOCs, and NO2 with the hydroxyl radical, OH, were determined in order to quantify the contributions of these compounds to potential ozone formation. The average total OH reactivity (ROH,TOTAL) increased from 1.01 s−1 in the central gulf to 10.1 s−1 in the HGB area as a result of the substantial increase in the contribution from VOCs and NO2. The increase in the measured concentrations of reactive VOCs in the HGB area compared to the central gulf was explained by the impact of industrial emissions, the regional distribution of VOCs, and the effects of local meteorology. By compensating for the effects of boundary layer mixing, the diurnal profiles of the OH reactivity were used to characterize the source signatures and relative magnitudes of biogenic, anthropogenic (urban + industrial), and oxygenated VOCs as a function of the time of day. The source of reactive oxygenated VOCs (e.g., formaldehyde) was determined to be almost entirely from secondary production. The secondary formation of oxygenated VOCs, in addition to the continued emissions of reactive anthropogenic VOCs, served to sustain elevated levels of OH reactivity throughout the time of peak ozone production.

1. Introduction

[2] Volatile organic compounds (VOCs) are pervasive throughout the lower atmosphere. From the most pristine to polluted environments, VOCs influence the oxidative capacity of the atmosphere via photolysis and reactions with ozone (O3), hydroxyl radicals (OH), nitrate radicals (NO3), and halogen radicals (Cl and Br). OH is formed primarily from the photolysis of ozone and has the ability to react with both saturated and unsaturated hydrocarbons on timescales ranging from minutes to several days. The products of these reactions result in the formation of alkyl peroxy radicals (ROO) which efficiently convert NO to NO2. Subsequent photolysis of NO2 ultimately results in the photochemical production of tropospheric ozone; however, formation of ROO is often the rate-limiting step. The initial rate of peroxy radical formation of various gas phase compounds (CO, CH4, and VOCs) can be compared by their relative OH reactivities, which are determined by the product of their concentration and the OH rate coefficient of that compound. By comparing the contributions of CO, CH4, VOCs, and NO2 to the total OH reactivity, one can determine the relative importance of these compounds to potential ozone formation.

[3] In 2000, an intensive field campaign was conducted in order to investigate the sources and atmospheric processes responsible for the formation and distribution of ozone in the greater Houston, Texas, metropolitan area [Banta et al., 2005; Berkowitz et al., 2004, 2005; Daum et al., 2003, 2004; Jobson et al., 2004; Karl et al., 2003; McGaughey et al., 2004; Roberts et al., 2003; Rosen et al., 2004; Ryerson et al., 2003; Wert et al., 2003]. Houston is unique in the fact that it is one of the largest cities in the U.S. and is home to a significant portion of the U.S.'s base petrochemical manufacturing capacity. The colocation of large point sources of VOCs and NOx (NO + NO2) from industrial activities in addition to urban sources (e.g., on-road emissions) can result in high ozone concentrations in Houston that often exceed the National Ambient Air Quality Standard (NAAQS). Ozone formation in Houston can often be much faster (on timescales less than one hour) [Ryerson et al., 2003] compared to other cities [Banta et al., 1998; Daum et al., 2000]. The Texas Air Quality (TexAQS) 2000 study showed that (1) plumes of ethene and propene released from petrochemical facilities could rapidly and efficiently produce large quantities of ozone downwind of Houston [Berkowitz et al., 2004; Daum et al., 2003, 2004; Kleinman et al., 2005; Ryerson et al., 2003] and (2) the emissions of these reactive VOCs were severely underestimated in the emission inventories used in Houston [Ryerson et al., 2003].

[4] The 2006 TexAQS/Gulf of Mexico Atmospheric Composition and Climate Study (GoMACCS) was conducted in order to revisit some of the questions intrinsic to the 2000 campaign with improved instrumentation and new measurement platforms such as the NOAA research vessel Ronald H. Brown (R/V Brown). This study utilizes an extensive set of VOC and other gas phase measurements in addition to detailed meteorological measurements conducted aboard the R/V Brown as it sailed throughout the central Gulf of Mexico, along Texas's gulf coast, and into the interior waters adjoining Houston, Texas. The goal of this study is to determine the major factors contributing to the OH reactivity and therefore possible ozone production in the Houston area. In order to put the measurements obtained during the TexAQS 2006 campaign into perspective, we first compare the VOCs in the greater Houston area to those in the Gulf of Mexico, other U.S. cities, and the previous TexAQS 2000 campaign. Then we evaluate the influence of industrial emissions, the regional distribution, and the effects of meteorology on the magnitude, variability, and composition of the OH reactivity of VOCs. Last, the diurnal profiles of the OH reactivity are examined in order to better understand the relative contributions of biogenic, anthropogenic (urban and industrial), and oxygenated VOCs in the photochemical production of ozone in Houston as a function of time of day as the timing of emissions of VOCs could have a large effect on the amount of ozone potentially formed [Nam et al., 2006].

2. Ship-Based Measurements

2.1. Measurement Locations

[5] The entire cruise track is shown in Figure 1a. The R/V Brown left Charleston, South Carolina, on 27 July 2006 heading south along the eastern coast of the United States and around the tip of Florida. From 30 July to 1 August the ship sailed directly across the Gulf of Mexico heading toward the primary study area near Houston, Texas. During this time, the R/V Brown was the furthest away from any land-based pollution sources for the entirety of the cruise. A total of 96 VOC samples were collected on the transit across the Gulf of Mexico (Figure 1a). The data collected in this region is hereafter referred to as the “central gulf.”

Figure 1.

Cruise track of the R/V Brown for TexAQS/GoMACCS. (a) Data points collected in the central gulf region. (b) Detail of the dashed box in Figure 1a depicting the data points in the coastal offshore and HGB areas.

[6] From 1 August to 11 September, the R/V Brown remained in the primary study area covering the northwestern coast of the Gulf of Mexico extending from Corpus Christi, Texas, to Beaumont, Texas. Throughout this time, various transits were made along the Texas coast as shown in Figure 1b. The data labeled as “coastal offshore” is composed of 283 VOC samples that were collected approximately 20 km from the shoreline. The most intense area of study included a total of 596 VOC samples collected inland of Galveston, Texas. This region is defined as the “Houston and Galveston Bay” (HGB) area and is highlighted in Figure 1b. Within this region, the ship made multiple transects of Galveston Bay (192 VOC samples) and the Houston Ship Channel (HSC; 104 VOC samples) which ends 8 km east of downtown Houston making it well within the urban corridor. The ship spent the most time sampling at a container terminal, Barbours Cut (BC; 300 VOC samples), located 6 km east of the municipal airport in La Porte, Texas, where a similar set of measurements were collected during the 2000 Texas Air Quality Study conducted in August and September [Jobson et al., 2004; Karl et al., 2003].

2.2. Gas Phase Measurements

[7] All of the inlets for the gas phase measurements were located at the top of an 8 m tall scaffold erected near the bow of the R/V Brown so that the inlets were approximately 20 m above the waterline and well forward of any sources of contamination from the ship itself when pointed into the wind.

[8] A custom-built gas chromatographic (GC) system capable of analyzing two samples in parallel was used to separate and identify a large number of VOCs in situ. A detailed description of the sample inlet configuration and of the instrument itself is described by Goldan et al. [2004], but a brief overview is provided here. The GC sample inlet consisted of a 15 m Teflon line (6 mm O.D.). A sample pump on the GC pulled in ambient air at a rate of ∼7 std L min−1. From this sample stream, two separate parallel samples were cryogenically collected for 5 min at a flow rate of 70 sccm resulting in a sample size of 350 mL at STP for each channel. The two sample channels have slightly different configurations designed to eliminate O3, CO2, and H2O before trapping the sample. A sample trap temperature of −165°C is used to collect the lightest VOCs measured (C2–C5 alkanes, C2–C4 alkenes, and ethyne) for subsequent analysis on an 18 m Al2O3/KCl PLOT column (Chrompak) with a flame ionization detector (FID). All other VOCs (C5–C10 alkanes, C5–C9 alkenes, C6–C9 aromatics, C1–C5 alcohols, C2–C7 aldehydes and ketones, C1–C5 alkyl nitrates, acetonitrile, dimethylsulfide, several monoterpenes and halocarbons) were collected with a slightly warmer sample trap at −145°C for separation on a semipolar 20 m DB-624 capillary column (J&W Scientific) and detection by an electron impact linear quadrupole mass spectrometer (Agilent 5973). The GC-FID/MS system is totally automated and repeats the entire sampling and analysis process every 30 min. The detection limit, precision, and accuracy varied by compound. Detection limits ranged from 0.5 to 1.0 pptv (parts per trillion by volume) except for the C2–C3 hydrocarbons which had a detection limits of approximately 10 pptv. The precision of the VOC mixing ratios was determined from the standard deviation in multiple measurements from the same standard or from ambient air samples for compounds showing little or not temporal variation. This was typically ±1/2 the respective detection limits stated above +1% of the reading. The GC was calibrated using numerous multicomponent gas standards. Calibration accuracy was about 15% for most compounds, with the exception of oxygenates which had a calibration accuracy of ∼25%.

[9] Formaldehyde was measured using tunable infrared laser differential absorption spectroscopy (TILDAS) [Zahniser et al., 1995]. During TexAQS 2006, the quantum cascade laser (QCL) was operated over a 1.2 cm−1 window centered at 1765 cm−1. The acquired spectra were analyzed using the measured rotation-vibration transitions of HCHO and HCOOH. Long temporal averages revealed no other significant interfering absorbers in this wavelength region. The inlet and overall performance of the instrument were characterized using a permeation source of HCHO every hour. The overall accuracy (7%) of the retrieved HCHO mixing ratio is based on the line strength determination [Herndon et al., 2005] and the measurement of pressure and temperature in the multipass cell. The 1 s root mean square was routinely less than 220 pptv. Allan Variance analysis [Allan, 1966] of the time series of the measurement of zero gas and stable air masses in the MBL typically reveal minimums of 50–25 pptv at 30–90 s. We conservatively attribute a 3 sigma detection limit for the instrument during this deployment to be 75 pptv in 100 s. Several different tests were conducted in order to rule out any potential artifacts from the inlet itself. These tests included (1) varying the flow rate through the inlet to see whether increased transit time resulted in systematically different formaldehyde concentrations and (2) deliberately spiking the inlet with varying amounts of ozone to look for production of formaldehyde inside the inlet. These tests did not reveal measureable artifacts in formaldehyde.

[10] Carbon monoxide (CO) was measured via a modified AeroLaser GmbH (Garmisch-Partenkirchen, Germany) AL5002 Ultra-Fast CO analyzer, a commercially available vacuum-UV resonance fluorescence instrument [Gerbig et al., 1999]. Mixing ratios were reported for ambient air by correcting for the removed water vapor using the water vapor mixing ratio measured at the inlet via Vaisala RH probe. The water mixing ratio was typically 30 ppth (3%) during the campaign, and the correction was always less than 4%. CO data were collected at 1 Hz and averaged to 1 min; the total uncertainty is estimated at 3%, with a limit of detection of 1.5 ppbv.

[11] Nitric oxide (NO) was detected via ozone-induced chemiluminescence; nitrogen dioxide (NO2) was measured via photolysis followed by chemiluminescent detection [Gao et al., 1994; Ryerson et al., 2000]. NO2 is partially converted to NO (typical conversion fraction range = 0.35–0.60), and the sum of NO and converted NO2 (NOx) is detected as NO. Uncertainty (at ±1σ) for NO during TexAQS was 3.8% at 1 min resolution, with a LOD of 0.010 ppbv. Uncertainty for NO2 cannot be easily stated as a single value, since it is determined by the relative mixing ratios of NO and NO2, along with inherent instrumental and environmental parameters at NO2/NO > 3, NO2 uncertainty was ∼13% and at NO2/NO = 1, NO2 uncertainty was ∼17%. LOD for NO2 was 0.06 ppbv in conditions where NO was below LOD. Ozone (O3) was measured via UV absorbance by a commercial instrument (Thermo Environmental Instruments, Inc., Model 49c).

[12] The gas phase measurements of HCHO, CO, NO2, and O3 had faster time responses (sub-Hertz sample rates) compared to the GC-FID/MS sample acquisition time of 5 min. In order to integrate the various data sets, the HCHO, CO, NO2, and O3 mixing ratios have been averaged over a 5 min window coincident with the GC sampling.

2.3. Meteorological Measurements

[13] Atmospheric boundary layer wind, aerosol backscatter, and turbulence profiles were measured using NOAA's high-resolution Doppler lidar (HRDL) [Grund et al., 2001]. HRDL is a coherent Doppler lidar used to make velocity and return signal strength measurements with 30 m range resolution along a line of sight at 2 Hz update rates. Velocity measurements acquired during motion stabilized conical scans are processed into high vertical resolution profiles of average horizontal wind speed and direction, while zenith staring data provide profiles of vertical velocity variance with 30 m resolution. These profiles were updated every 15 min along the ship track and were used to determine the boundary layer depth, referred to here as the mixing height (MH) which determines how high near-surface emitted pollutants are mixed. During TexAQS 2006, mixing heights were determined using the HRDL wind, wind shear, and velocity variance profiles [Tucker et al., 2008] by finding the altitude at which the strength of surface-connected turbulence dropped below a given threshold.

3. Results

3.1. Overview of VOC Measurements

[14] Median, average, and maximum mixing ratios for 83 VOCs, CO, and NO2 for both the central gulf and Houston and Galveston Bay (HGB) regions are compiled in Table 1. While the compounds have been organized primarily by chemical classes, certain compounds were also grouped on the basis of their sources. For example, the “terpenes and oxidation products” and “other alkenes/monomers” could be folded into the general category of “alkenes,” but they were kept separate on the basis of the relative contributions from biogenic, urban (i.e., automobile exhaust) and industrial sources which will be discussed later.

Table 1. Listing of Observed Mixing Ratios, OH Reaction Rate Coefficients, and OH Reactivities for the Central Gulf and Houston and Galveston Bay Areasa
CompoundCentral GulfHouston and Galveston BayOH Reactivities in HGB
MedianAverageMaximumMedianAverageMaximumkOHbReferencecAverageMaximum
Alkanes
Ethane0.3240.3913.9294.4078.353197.010.2510.051.20
Propane0.0410.1814.0582.7136.666347.491.110.189.32
n-Butane0.0170.1301.5890.9554.177467.632.410.2427.15
i-Butane0.0080.0410.5400.8584.377182.072.110.239.50
n-Pentane0.0070.0440.4330.3372.227183.103.810.2117.12
i-Pentane0.0060.0390.3670.7013.608183.103.610.3216.22
n-Hexane0.0010.0100.1150.1930.85680.855.210.1110.34
2,2-Dimethylbutane0.0040.0290.0774.842.210.000.27
3-Methylpentane0.0050.0500.1240.42533.215.210.054.25
Sum 2,3-Dimethylbutane and 2-Methylpentane0.0010.0070.0710.2050.73360.535.6est.0.108.34
Methylcyclopentane0.0020.0190.0940.35831.378.620.086.64
Cyclohexane0.0010.0140.0640.31622.567.010.053.87
n-Heptane0.0010.0030.0290.0520.21329.306.810.044.87
Methylcyclohexane0.0010.0210.0470.21323.629.610.055.60
n-Octane0.0050.0150.0889.368.110.021.87
n-Decane0.0020.0020.0050.0100.0623.4311.010.020.93
 
Alkenes
Ethene0.0600.0700.1790.7962.69254.548.510.5611.43
Propene0.0410.0500.1600.2231.53655.5826.310.9935.96
2-Methylpropene0.0150.0230.1560.0620.27911.3751.410.3514.37
1-Butene0.0120.0280.24823.5431.410.1918.18
cis-2-Butene0.0060.0050.08610.7556.410.1214.92
trans-2-Butene0.0080.0060.12314.5864.010.1922.95
2-methyl-1-Butene0.0340.0540.1275.7961.010.198.68
3-methyl-1-Butene0.0030.0090.3131.810.010.24
2-methyl-2-Butene0.0130.0230.2286.910.050.48
1-Pentene0.0130.0502.6431.410.042.04
cis-2-Pentene0.0070.0414.2465.010.076.78
trans-2-Pentene0.0130.0768.3667.010.1313.78
cis-2-Hexene0.0314.3362.030.056.61
trans-2-Hexene0.0212.5360.240.033.74
trans-3-Hexene0.0131.7862.030.022.71
Ethyne (Acetylene)0.0400.0420.0800.3010.4735.790.9050.010.13
 
Terpenes and Oxidation Products
Isoprene0.0560.1401.5810010.353.89
alpha-Pinene0.0130.0240.2552.310.030.32
beta-Pinene0.0070.0130.2174.310.020.39
Limonene0.0030.0110.9516410.043.82
Methacrolein (MACR)0.0060.0080.0330.0240.0520.5529.010.040.39
Methyl vinyl ketone (MVK)0.0030.0040.0180.0290.0810.7320.010.040.36
 
Other Alkenes/Monomers
1,3-Butadiene0.0160.0822.8966.610.134.73
Styrene0.0040.0413.2758.060.064.67
1-Octene0.0060.0805.9136.270.075.26
Propylene oxide0.0100.0110.0270.0240.20721.590.5080.000.27
Methyl acrylate0.05518.1625.190.0311.21
Methyl methacrylate0.0121.1426.190.010.73
Vinyl acetate0.0060.36876.1349.030.4491.77
Acrylonitrile (2-propenenitrile)0.0150.26417.208.530.063.60
 
Aromatics
Benzene0.0060.0070.0140.1930.41811.931.210.010.36
Toluene0.0010.0020.0120.1610.49615.865.610.072.20
Sum m&p-Xylenes0.0020.0020.0120.0520.21710.9819.0est.0.105.13
o-Xylene0.0010.0010.0050.0230.0873.8713.610.031.30
Ethylbenzene0.0010.0010.0050.0260.0823.077.010.010.53
1,2,3-Trimethylbenzene0.0010.0030.0211.2332.710.020.99
1,2,4-Trimethylbenzene0.0030.0140.0773.8032.510.063.04
1,3,5-Trimethylbenzene0.0010.0030.0191.1056.710.031.54
1-Ethyl-2-methylbenzene0.0010.0030.0191.0411.910.010.30
Sum 1-Ethyl-3&4-mebenzene0.0030.0110.0835.6615.6est.0.032.17
n-Propylbenzene0.0030.0150.835.810.0020.12
i-Propylbenzene0.0020.0181.076.310.0030.17
 
Oxygenates
Formaldehyde* (HCHO)0.5210.5400.8491.6042.44319.189.410.564.42
Acetaldehyde0.1110.1460.3720.5800.7795.0915.010.291.88
Propanal0.0270.0360.0940.1660.2282.0320.010.111.00
Acetone1.2921.3982.6242.7413.54543.050.1710.010.18
2-Butanone (MEK)0.0410.0530.1540.1950.43022.901.210.010.69
Methanol0.6240.6611.0222.5814.764144.230.910.113.34
Ethanol0.0330.0360.0860.5820.86626.633.210.072.10
i-Propanol0.0290.0310.0720.1750.2987.065.110.040.89
Methytertbutylether (MTBE)0.0200.12513.752.910.0090.99
Methyl Formate0.0480.0530.240.1810<0.0010.001
 
Others (N, S, Cl, and F)
Acetonitrile0.1080.1080.1320.1530.1907.070.0211<0.0010.004
Methylnitrate0.0020.0020.0030.0030.0060.100.0210<0.001
Ethylnitrate0.0020.0030.0060.040.181<0.001<0.001
n-Propylnitrate0.0050.0060.0130.180.581<0.0010.003
n-Butylnitrate0.021.61<0.001<0.001
Dimethylsulfide (DMS)0.0290.0310.0640.0200.0280.405.412<0.0010.053
Chloroacetone0.0030.0050.0070.080.4013<0.001<0.001
Chlorobenzene0.0111.720.6014<0.0010.025
Trichloroethylene0.0110.0120.0210.0070.0170.362.211<0.0010.019
Tetrachloroethylene0.0020.0070.0130.220.1711<0.001<0.001
CHCl30.0080.0080.0090.0200.0250.190.1011<0.001<0.001
CCl40.0930.0930.0980.0930.0950.194e-41500
CFCl30.2940.2930.3070.2660.2570.404e-61500
CFCl2-CF2Cl0.0840.0840.0880.0800.0800.113e-41600
 
Non-VOCs
CO*8180841241497440.2150.883.66
NO2*0.0800.1481.5365.9088.53553.9258.7151.8211.54
CH4 (global average)1775  1775  6e-310.27 

[15] In the central gulf, the most abundant VOCs based on median mixing ratios were acetone (1.292 ppbv), methanol (0.624), formaldehyde (0.521), ethane (0.324), acetaldehyde (0.111), and acetonitrile (0.108). The relative abundance of oxygenated VOCs and other long-lived compounds, such as ethane and acetonitrile, indicate that the majority of the air sampled in the central gulf was well aged [de Gouw et al., 2005; Goldan et al., 2004]. While formaldehyde is expected to be a relatively abundant VOC in aged air masses within the marine boundary layer, formaldehyde mixing ratios in the central gulf were higher than expected on the basis of first principles and comparisons to other studies [Snow et al., 2007]. While measurement artifacts cannot be ruled out completely, the inlet tests that were performed in this relatively clean environment did not indicate that formaldehyde was artificially formed in the inlet from oxidants such as ozone. It should be noted that the central gulf region was not completely devoid of “fresh” air masses containing reactive hydrocarbons. There were two episodes that contained appreciable amounts of alkanes (>1 ppbv), one relatively prolonged (31 July 2006 at 1400–2200 UTC) and the other only lasting 1.5 h (1 August 2006 at 1500–1630 UTC). The maximum mixing ratios for all C2–C6 alkanes in the central gulf region (as listed in Table 1) occurred in one of these two air masses. The increase in C2–C6 alkanes associated with these two episodes did not correlate with chemical tracers such as CO, SO2, NO, or NO2. A correlated increase in any combination of these compounds could point to recent emissions from the exhaust of the R/V Brown itself or other ships nearby. While oil and natural gas drilling platforms are common in the Gulf of Mexico, they are located along the continental shelf and are not found in the central gulf. A biological source cannot be ruled out but is highly unlikely. While the origin of these alkanes remains unknown, the infrequent nature of these events indicates that the central gulf was only modestly influenced by anthropogenic emissions. Therefore, the median mixing ratios of the VOCs measured in the central gulf (excluding biogenic VOCs) are used to represent the lowest concentrations of VOCs that the coastal urban area of Houston could encounter with strong onshore winds from the Gulf of Mexico.

[16] In the HGB area, ethane (4.407 ppbv), acetone (2.741), propane (2.713), methanol (2.581), formaldehyde (1.604), n-butane (0.955), iso-butane (0.858), and ethene (0.796) were the most abundant VOCs measured on the basis of median values. This represents an approximately 10–100 fold increase in the median mixing ratios for the C2–C5 alkanes and 2–17 fold increase of the oxygenated VOCs compared to the central gulf region. The influence of large point sources of hydrocarbons is clearly identifiable in the extremely high maximum mixing ratios (>100 ppbv) for the C2–C5 alkanes and methanol. Other species with very high maximum mixing ratios (>50 ppbv) include the reactive compounds ethene, propene, and vinyl acetate, and n-hexane.

[17] The impact of industrial VOC sources in Houston can be assessed by comparing VOC mixing ratios to other urban settings. Data from 28 U.S. cities [Baker et al., 2008] and from the 2002 New England Air Quality Study (NEAQS 2002) [Goldan et al., 2004], which includes data collected along the New England coast near Boston and New York Harbors, are used to represent “typical” urban areas that lack the large industrial component unique to the HGB area. Ethyne (acetylene) and benzene are commonly used anthropogenic tracers because they are relatively unreactive compounds that are emitted in most urban areas by automobiles [Fortin et al., 2005; Harley et al., 2001; Lough et al., 2005; Whitby and Altwicker, 1978]. As shown in Figure 2, benzene and ethyne are well correlated for both data sets (R2 = 0.75 for 28 U.S. Cities and R2 = 0.92 for NEAQS 2002) owing to their similar sources and long chemical lifetimes [Fortin et al., 2005; Jobson et al., 1998; Parrish et al., 1998]. In contrast, the lack of correlation between benzene and ethyne in HGB (R2 = 0.14) indicates large and frequent emissions of benzene in addition to those from strictly urban sources. These large benzene sources are attributed to industrial emissions including chemical plants and petroleum refineries. Figure 2 also shows that there are a few samples (<3%) when ethyne mixing ratios are much greater than that of benzene indicating an infrequent but direct source of acetylene from industrial activities in the HGB area. This is in accordance with the findings from the TexAQS 2000 study in La Porte, Texas, which identified an industrial plume near Houston that contained mixing ratios of ethyne significantly higher than what could be attributed solely to vehicle exhaust [Jobson et al., 2004].

Figure 2.

Scatterplot of benzene versus ethyne in the Houston and Galveston Bay area. Data from 28 United States cities [Baker et al., 2008] and NEAQS 2002 [Goldan et al., 2004] have been included for comparison.

[18] The VOC measurements in HGB during the TexAQS 2006 campaign were compared to those previously collected in La Porte, Texas, during the TexAQS 2000 campaign [Jobson et al., 2004]. It should be noted that the HGB 2006 data set (1) contains fewer samples (575 data points) compared to La Porte (1191 data points) and (2) is composed of a greater geographic region. The range and variability in measured mixing ratios of ethyne, ethene, and propene in HGB 2006 and La Porte 2000 are illustrated in the mixing ratio histograms (Figure 3). The distribution of ethyne mixing ratios for HGB 2006 was reasonably lognormal compared to the bimodal distribution seen in La Porte 2000 (Figure 3a). This suggests that ethyne mixing ratios for HGB 2006 were not as heavily influenced by industrial emissions as compared to La Porte 2000. For ethene (Figure 3b) and propene (Figure 3c), the general shape of the mixing ratio distributions for both HGB 2006 and La Porte 2000 were skewed toward higher mixing ratios indicating that industrial emissions of these two compounds had a comparable influence on the variability of the mixing ratios for both data sets. One feature that is common to all three compounds is that the modes of the mixing ratio distributions (Figures 3a3c) have shifted to lower values for the HGB 2006 data set. This is also evident in the median mixing ratios of these compounds. For La Porte 2000, the median mixing ratios of ethyne, ethene, and propene were 0.41, 1.83, and 0.45 ppbv [Jobson et al., 2004], respectively, compared to HGB 2006 where the median values decreased to 0.301, 0.796, and 0.223 ppbv (Table 2), respectively. Comparisons of the aircraft measurements during TexAQS 2000 and 2006 also indicate that the mixing ratios of compounds such as ethene were lower in the 2006 study (Rapid Science Synthesis Report to the Texas Commission on Environmental Quality (TCEQ), accessed January 2009 at http://www.tceq.state.tx.us/assets/public/implementation/air/texaqs/doc/rsst_final_report.pdf). While these simple comparisons suggest that ambient concentrations of ethene and propene have decreased throughout the greater Houston area compared to the initial TexAQS 2000 study, a more detailed analysis is required in order to accurately account for intrinsic differences in the meteorological conditions, instrumentation, and geographic regions of the two data sets.

Figure 3.

Histograms of the mixing ratios of (a) ethyne, (b) ethene, and (c) propene for HGB 2006. Data from the TexAQS 2000 campaign at a ground site in La Porte, Texas, [Jobson et al., 2004] have been added for comparison.

Table 2. Observed OH Reactivities
 Central Gulf of MexicoCoastal OffshoreHouston and Galveston Bay
Median (s−1)Median Percent of ROH,TOTALAverage (s−1)Average Percent of ROH,TOTALMedian (s−1)Median Percent of ROH,TOTALAverage (s−1)Average Percent of ROH,TOTALMedian (s−1)Median Percent of ROH,TOTALAverage (s−1)Average Percent of ROH,TOTAL
  • a

    ROH,CH4 was determined by assuming a global average methane mixing ratio of 1775 ppbv [Intergovernmental Panel on Climate Change, 2007] for all data sets.

  • b

    ROH,TOTAL (s−1) is the total OH reactivity as determined from the sum of OH reactivities of CH4, CO, VOCs, and NO2. The minimum ROH,TOTAL (s−1) of each data set is as follows: 0.75 for central Gulf of Mexico and coastal offshore and 0.83 for Houston and Galveston Bay. The maximum ROH,TOTAL (s−1) of each data set is as follows: 1.41 for central Gulf of Mexico, 16.0 for coastal offshore, and 200 for Houston and Galveston Bay. Number of points is 77 for central Gulf of Mexico, 180 for coastal offshore, and 575 for Houston and Galveston Bay.

  • c

    The standard deviation (σ), which shows the spread of the data where 68% of the calculated ROH,TOTAL's lie within 1σ of the average, is 0.13 for central Gulf of Mexico, 2.76 for coastal offshore, and 16.8 for Houston and Galveston Bay. The overall uncertainties in the calculated ROH,TOTAL's are as follows: 10% for central Gulf of Mexico, 10% for coastal offshore, and 12% for Houston and Galveston Bay. The overall uncertainties were estimated using the reported uncertainties of the reaction rate coefficients and of the measured/assumed gas-phase mixing ratios.

CH4 (ROH,CH4)a0.27280.27270.27170.27100.2750.273
CO (ROH,CO)0.40410.39380.68420.61220.62120.737
VOCs (ROH,VOC)0.28290.32320.44421.44533.02587.2572
NO2 (ROH,NO2)0.0220.03230.22140.42151.26251.8218
ROH,TOTALb0.97 1.01c 1.6 2.74c 5.2 10.1c 

3.2. OH Reactivity

[19] The relative contribution of VOCs to potential ozone formation is investigated here by comparing the various sinks of the hydroxyl radical (OH). The actual amount of ozone eventually produced is dependent upon the particular oxidation mechanism and NOx concentrations [Carter, 1994]. The reaction sequence initiated by OH has several important consequences including (1) the regeneration of OH and (2) the secondary formation of reactive and photolabile VOCs such as aldehydes. Both of these act to perpetuate a chain reaction sequence which is ultimately terminated by the reaction of OH and NO2 in polluted atmospheres. The product of this reaction, HNO3, results in the permanent removal of both reactive oxidants from the atmosphere. The role of VOCs in determining the balance between perpetuation (OH + CH4, CO and VOCs) and termination (OH + NO2) of ozone-forming reaction sequences is determined by the relative contribution to the total OH reactivity (ROH,TOTAL, s−1), defined as sum of the OH reactivities of CO, CH4, all reported VOCs, and NO2 as shown by the following equation:

equation image

The OH reaction rate coefficients of the various species (kOH, cm3 molec−1 s−1) at 298 K and 1013 mbar are listed in Table 1 in addition to the average and maximum values of the OH reactivities for all the reactants investigated in HGB. There were no measurements of methane so a global average abundance of 1774.6 ± 1.2 ppbv [Intergovernmental Panel on Climate Control, 2007] was used. ROH,VOC was determined using all the VOCs reported in Table 1. While the list is extensive, the calculation of ROH,VOC does not include all of the VOCs monitored by the GC-FID/MS or any VOCs that were undetectable. Therefore, the values of ROH,VOC presented here correspond to minimum values, but these values are expected to adequately represent the ambient OH reactivity.

[20] The median and average OH reactivities for the central gulf, coastal offshore region, and HGB are shown in Table 2. In the central gulf, CO is the dominant OH sink followed by VOCs and CH4. NO2 is negligible by comparison. This is not the case for the HGB area where VOCs are the principal contributors to ROH,TOTAL and NO2 represents a significant portion. Although methane is considered to be the same concentration in all locations, the average contribution of methane as an OH sink decreased from approximately 27% in the gulf to only 3% in HGB. In an area with a significant number of liquid and natural gas refineries, the ambient concentration of methane in HGB is expected to be greater than the global average of 1774.6 ppbv. Even if the methane concentration were twice that of the global average–which is highly unlikely–the contribution of methane to ROH,TOTAL would still be smaller than that of CO in HGB. While the average CO mixing ratio increased from 80 ppbv in the gulf to 149 ppbv in HGB (Table 1), the relative contribution of CO as a potential OH sink decreased from 38% to only 7% in HGB.

[21] The distributions of ROH, NO2, ROH,VOC, and ROH,TOTAL for HGB are shown in Figure 4. While the distribution of ROH,NO2 (Figure 4a) is heavily skewed, both the range and median values of ROH, NO2 in HGB are comparable to La Porte 2000 (median ROH,NO2 = 1.4 s−1) [Jobson et al., 2004]. The distribution of ROH,VOC (Figure 4b) is relatively symmetric and spans 3 orders of magnitude. The median ROH,VOC for HGB (3.02 s−1) is comparable to both the mode of the VOC+CO distribution shown for La Porte 2000 [Jobson et al., 2004] and for aircraft measurements during TexAQS 2000 [Kleinman et al., 2005]; however, the fraction of samples with ROH,VOC > 10 s−1 appears to be greater for La Porte 2000 compared to the HGB 2006 data set. This is in accordance with the observed decrease in the ethene and propene mixing ratios for HGB 2006 compared to La Porte 2000. While the distribution ROH,TOTAL (Figure 4c) for HGB is reasonably lognormal, there is a tail that extends toward higher ROH,TOTAL values indicating a significant number of air masses containing high concentrations of reactive compounds. The frequency distribution shown in Figure 4d clearly shows that the composition of these highly reactive air masses (ROH,TOTAL > 10s−1) is dominated almost entirely by VOCs. The presence of these concentrated VOC plumes in HGB (1) results in a sizable increase of the average ROH,TOTAL compared to the median and (2) will have a direct impact on the local air quality and should therefore remain evident throughout the ensuing analyses. Since extreme values are not conveyed by medians, average values of the OH reactivity will be the focus of subsequent discussions. The average ROH,TOTAL in HGB was 10.1 ± 12% s−1, where 12% is the average uncertainty in ROH,TOTAL as determined from the accumulation of the individual reaction rate coefficient uncertainties and the stated uncertainties in the measured gas phase concentrations. The variability of ROH,TOTAL in HGB, as evidenced by the large standard deviation, far outweighs the uncertainties in ROH,TOTAL which are good to within 12% of the calculated values.

Figure 4.

Histograms of (a) ROH,NO2, (b) ROH,VOC, (c) ROH,TOTAL, and (d) cumulative frequency distributions of ROH,CH4, ROH,CO, ROH,NO2, ROH,VOC, and ROH,TOTAL in the Houston and Galveston Bay area.

3.2.1. Effects of Industrial Plumes on ROH,VOC

[22] As shown in Figure 4d, ROH,VOC is the largest contributor to ROH,TOTAL at values > 10 s−1. This was a direct result of sampling individual plumes of reactive VOCs from industrial sources in close proximity to the sampling location. The impact of these industrial plumes on the variability of the magnitude ROH,VOC and the VOC composition is evidenced in the time series shown in Figure 5. The maximum ROH,VOC encountered during the entire campaign was 200 s−1 (Figure 4; 7 September 2008 at 1100 UTC) while the R/V Brown was stationed within the Houston Ship Channel. Not surprisingly, extremes in ROH,VOC were most often encountered when sampling close to the highly industrialized ship channel and nearby Barbours Cut.

Figure 5.

A time series in the Houston and Galveston Bay area of (a) ROH,VOC, ozone mixing ratios, and sunlight and (b) the fractional contribution of each VOC group to ROH,VOC.

[23] In general there were two types of industrial emissions encountered during this study: those composed of a range of species in a similar chemical class or those composed of one single dominant species. The sample with the highest reactivity in Figure 5 (7 September 2008 1100 at UTC) was composed of a mixture of C2–C5 alkenes with a combined mixing ratio of 190 ppbv. The preponderance of a range of alkenes is indicative of recent emissions from petrochemical production facilities that line the Houston Ship Channel [Jobson et al., 2004; Ryerson et al., 2003]. Plumes dominated by a single compound, such as one composed of 76 ppbv of vinyl acetate (Table 1) encountered on 15 August in Barbours Cut (time series not shown), may be from an isolated event, such as venting or flaring, at a chemical production facility [Kim et al., 2005]. There are several known emitters of vinyl acetate in the Houston area, including a chemical plant within 4 km of the sampling site at Barbours Cut (2004 TCEQ Annual Point Source Emission Inventory) although none of the local chemical plants reported any emission “events” for vinyl acetate during the entire campaign (TCEQ Air Emission Event Report Database, accessed September 2008 at http://www11.tcequation state.tx.us/oce/eer/index.cfm). Several studies aimed at characterizing industrial emissions have shown that the magnitudes of emissions are routinely variable over time and often include episodic releases that are just under reportable quantities [Murphy and Allen, 2005; Webster et al., 2007]. Documentation of the variability in both the magnitude and composition of industrial emissions is one crucial piece of information required for the reconciliation of emission inventories with field measurements.

3.2.2. Effects of the Regional Distribution of VOCs on ROH,VOC

[24] The regional distribution of VOCs is determined by the continuous interplay of their sources, sinks, transport, mixing and photochemical aging. The influence of these processes on the magnitude of ROH,VOC and the evolution of the VOC composition will be examined here. Diurnal profiles of ROH,VOC and ozone for the central gulf, coastal offshore, and HGB areas are shown in Figure 6. Also included are pie charts that represent the average contribution of each VOC class to ROH,VOC in the designated regions. The diurnal profiles of ROH,VOC were determined from the averages and medians of all samples collected at the same time of day as governed by the GC-MS sampling time. Plotting ROH,VOC on a diurnal timescale is useful since the oxidation of VOCs by OH and the potential photochemical production of ozone are both diurnally dependent. While the profiles of the average and median ROH,VOC show that the diurnal trends are similar to both metrics; the impact of industrial plumes of VOCs are more clearly conveyed by the averages which will be the focus of the following discussions.

Figure 6.

(top) Average and median diurnal profiles of ROH,VOC and average diurnal profile of ozone mixing ratios in (a) the central gulf, (b) coastal offshore, and (c) the Houston and Galveston Bay areas. (bottom) The average contribution of each VOC group to ROH,VOC in each designated region.

[25] In the central gulf (Figure 6a), ROH,VOC is relatively low (average = 0.32 s−1) and there are only minor diurnal variations. Similarly, the diurnal profile of ozone is nearly constant with an average ozone mixing ratio of 20 ppbv. Oxygenated VOCs are the dominant contributor to ROH,VOC in this region. Formaldehyde alone accounts for 33% of the average ROH,VOC followed by acetaldehyde at 18%, with OH reactivities of 0.106 s−1 and 0.057 s−1, respectively. This is indicative of a well aged air mass where most of the primary reactive hydrocarbons have been oxidized. Alkenes, primarily ethene and propene, contribute to ROH,VOC in this clean environment possibly because of their globally dispersed oceanic sources [Lewis et al., 2001; Ratte et al., 1998; Riemer et al., 2000]. The ocean is also a significant source of dimethylsulfide (included in the “Others” grouping) which is fairly reactive with OH because of the twin methyl groups; however, its contribution to ROH,VOC is minor. FLEXPART analyses [Stohl et al., 1998] (data not shown) indicate that air sampled throughout the central gulf had come from the Atlantic Ocean after several days of transport time, most of which was over open water with minor exposure to continental emissions. While it is important to note that air in the central gulf did not originate from Texas, this type of air mass serves to characterize the inflow of background air into Texas under southerly conditions.

[26] The coastal offshore region serves as an intermediate case between the central gulf and HGB. Samples were collected approximately 20 km offshore and were composed of those from the central gulf (southerly winds) and those from continental sources (northerly winds). This intermediary region is characterized by an overall increase in ROH,VOC (average = 1.44 s−1) and a more pronounced diurnal profile compared to that in the central gulf (Figure 6b); however, this increase is surpassed by that of ozone mixing ratios (average ozone = 58 ppbv) which are almost 3 times that of the central gulf. We have shown that the central gulf is not the source of reactive VOCs or high levels of ozone, so the increase in these compounds is directly dependent upon transport from coastal sources. ROH,VOC is still dominated by oxygenated VOCs as seen in the central gulf; however, the increase in the alkenes and alkanes is indicative of the influence of more recent emissions of reactive VOCs. Thus, the regional distribution of VOCs just off the coast of Texas represents a mixture of the processed air masses associated with the central gulf and the predominantly fresh emissions originating in HGB.

[27] The strong temporal changes in the diurnal profiles of ROH,VOC and ozone in the HGB area clearly distinguish this region from the others as shown in Figure 6c. The maximum value in the average diurnal profile of ROH,VOC increased from 5 s−1 in the coastal offshore region to 26 s−1 in HGB. This is due in part to the additional increase in the contribution of reactive alkanes and alkenes along with significant increases in the contribution from biogenics, monomers, and aromatics. As noted previously, the C2–C4 alkanes are the most abundant hydrocarbons; however, ethene (average = 0.60 s−1) and propene (average = 0.98 s−1) are the dominant contributors to ROH,VOC because of their high OH reactivities. Formaldehyde (average = 0.48 s−1) is the third largest contributor to ROH,VOC, but the overall contribution of all the oxygenates is smaller because of the cumulative increase in all other VOCs. As shown in Figure 6c, ozone mixing ratios increase in conjunction with solar radiation whereby both reach a daily maximum near solar noon (∼1800 UTC) indicating the dependence of ozone on secondary photochemical formation. Overnight, ozone mixing ratios begin to decrease because of frequently observed titration by NOx. The diurnal profiles of ROH,VOC are in direct opposition to that of ozone. During the evening hours (approximately 0000–1200 UTC), ROH,VOC continually increases reaching a maximum daily average of 26 s−1 shortly before daybreak. Upon sunrise, ROH,VOC quickly decreases before stabilizing at a relatively low average of 3–4 s−1. The factors that contribute to the diurnal trends seen for the average ROH,VOC profile will be further explored in the following sections.

3.2.3. Effects of Mixing Heights on ROH,VOC

[28] Mixing heights are expected to influence measured mixing ratios of VOCs and in turn ROH,VOC. Mixing height (i.e., boundary layer depth) is defined here as the height of the layer of the atmosphere that is in turbulent connection with the surface of the Earth. For a given surface area, mixing height will be directly proportional to the volume of air that a VOC is mixed into on timescales typically shorter than its chemical lifetime (approximately 10–30 min). Figure 7a shows the mixing heights in HGB which were determined using wind turbulence profiles from the high-resolution Doppler lidar aboard the R/V Brown [Tucker et al., 2008]. The average diurnal profile of ROH,VOC in HGB (Figures 6c and 7b), which is characterized by increasing values overnight followed by a sharp decrease upon sunrise, is partly governed by the diurnal trend in the mixing height (Figure 7a). VOCs that are emitted during the afternoon when mixing heights are high will be mixed into a larger volume, effectively diluting them. The result is lower measured mixing ratios during the day compared to the same emissions at night when the mixing heights are typically much lower. The inverse relationship between mixing heights and measured mixing ratios should be expected to hold true for all VOCs that are emitted or formed near the surface that are not well mixed throughout the lower troposphere.

Figure 7.

(a) Mixing heights determined from the high-resolution Doppler lidar in the Houston and Galveston Bay area. The average diurnal profile of the mixing height and a constant mixing height of 500 m has been added as a guide for the eye. (b) The average diurnal profiles of ROH,VOC, NormROH,VOC, and sunlight in the HGB area.

[29] To account for this strong dependence of the VOC mixing ratios on mixing heights we can normalize the data to a constant mixing height of 500 m. This height was chosen because (1) it is the approximate average of the day and night mixing heights and (2) it corresponds to the relatively constant mixing height over the gulf. Because mixing heights do not affect the reaction rate coefficients, scaling the mixing ratio will have the same effect as normalizing ROH,VOC as shown by the following equation:

equation image

where NormROH,VOC is the normalized VOC OH reactivity (s−1), MH is the mixing height (m) at the time each VOC sample was collected, and (500 m)−1 is the scalar used to account for changes in mixing height. The diurnal profiles of ROH,VOC and NormROH,VOC plotted in Figure 7b were determined from the averages of all samples collected at the same time each day. Even though the low nocturnal mixing heights emphasize the magnitude of ROH,VOC, Figure 7b shows that the average diurnal profile of NormROH,VOC gradually increases over the course of the night and remains elevated throughout the daytime hours. Normalizing ROH,VOC to a constant mixing height helps to isolate one physical parameter that is shown to have a significant effect on the VOC mixing ratios. For this reason, the average diurnal profile of the NormROH,VOC will be used throughout the remainder of this study in order to better identify the relative magnitudes and contributions of various sources of VOCs as a function of the time of day.

3.3. Contributions of VOC Sources to the Diurnal Profile of NormROH,VOC

3.3.1. Biogenic VOCs

[30] A detailed investigation of the average diurnal profiles of ROH,VOC and NormROH,VOC for the biogenic VOCs is presented here in order to (1) illustrate the validity of normalizing the VOC data to a constant mixing height and (2) to determine the sources and relative contribution of biogenic VOCs to NormROH,VOC as a function of time of day. Biogenic VOCs are a good case study because their emissions are highly dependent upon sunlight and temperature, both of which have strong diurnal cycles. Figure 8a shows the biogenic diurnal profile of ROH,VOC as well as the diurnal profiles of the mixing height (MH) and solar radiation. The contributions from isoprene, its oxidation products methyl vinyl ketone (MVK) and methacrolein (MACR), and the monoterpenes (alpha-pinene, beta-pinene, and limonene) are stacked so that the height of each bar represents the average value of ROH,VOC for all biogenic VOCs in HGB at that time. The ratio of the fractional contribution of each biogenic group to the diurnal profile is shown in Figure 8b.

Figure 8.

(a) Average diurnal profiles of mixing height, sunlight, and ROH,VOC for all biogenic VOCs for the HGB area. (b) The fractional contribution of isoprene, the monoterpenes (alpha- and beta-pinene and limonene), and methyl vinyl ketone (MVK) and methacrolein (MACR) to ROH,VOC. (c) The average diurnal profile of NormROH,VOC for the biogenics. (d) A comparison of the average diurnal profile of NormROH,VOC for all VOCs (biogenic + anthropogenic + oxygenated VOCs) and the contribution solely from biogenic VOCs.

[31] The diurnal profile of ROH,VOC for the biogenics (Figure 8a) closely matches that seen for isoprene mixing ratios measured at La Porte in 2000 [Jobson et al., 2004; Roberts et al., 2003] suggesting a large contribution of isoprene to the biogenic ROH,VOC. The diurnal profile of the biogenic ROH,VOC increases along with sunlight as expected for the light- and temperature-dependent emissions of isoprene. The biogenic ROH,VOC reaches an average daily diurnal maximum at 1400 UTC (0900 LT) which is the result of emissions of isoprene into a relatively shallow boundary layer (MH < 500m). As the mixing height continues to increase throughout the daytime, there is a sharp decline in ROH,VOC with a local minimum at solar noon. This minimum is attributed to increased vertical mixing and photoxidation of the biogenics. Photoxidation of isoprene is evidenced by the increasing contribution of MVK and MACR during the daytime (Figure 8b). A slight increase in the diurnal profile of ROH,VOC is seen late in the local daytime (∼2100 to 0300 UTC) in response to decreasing mixing heights and slower photochemical removal while biogenic emissions remain strong. Monoterpenes have the largest relative contribution to the biogenic OH reactivity at night as a result of their continued emissions while those of isoprene are minimal. The chemical losses of monoterpenes can be strongly affected by NO3 during the night when OH concentrations are considered to be zero [Warneke et al., 2004]; however, in HGB, NO3 is not an efficient sink of biogenics (and alkenes in general) because of large primary emissions of NO resulting in generally low NO3 and N2O5 mixing ratios observed at night [Osthoff et al., 2008].

[32] The biogenic diurnal profile of NormROH,VOC (Figure 8c) shows a distinctive diurnal pattern that increases with sunlight, but reaches a plateau rather than a daily maximum as per the ozone profile. This plateau is attributed to photochemical oxidation as vertical mixing has been effectively compensated for. Interestingly, a “spike” becomes magnified in the average diurnal profile of NormROH,VOC at 0330 UTC. This spike is attributed to a plume of limonene that occurred on 7 September 2006 (shown in Figure 5). Within this plume, limonene was measured at 0.95 ppbv, which represents an 86-fold increase over the average mixing ratio of only 11 pptv (Table 1). This is one clear indication that industrial sources of monoterpenes exist in HGB. Limonene is used as an additive in several materials including solvents and cleaners to give citrus scent while isoprene is used in the production of natural rubber and is often polymerized with styrene to make SIS (styrene-isoprene-styrene) plastic. Jobson et al. [2004] reported a prolonged and totally unnatural spike of 25 ppbv of isoprene that was attributed to an industrial source in the HGB area. Isoprene has also been attributed to vehicle exhaust [McGaughey et al., 2004] thereby partially categorizing it as urban emissions as well. While industrial and urban sources of biogenics have been noted, the diurnal profile strongly suggests that most of the biogenic VOCs are indeed from natural sources. Overall, the biogenics in HGB are a minor contributor to ROH,VOC at only 7% on average (Figure 6c); however, the diurnal profile of NormROH,VOC for the biogenic VOCs shows that the largest contribution to the OH reactivity as a function of time occurs during the late afternoon (Figure 8d). During this time, biogenic VOCs can account for up to 20% of the VOC reactivity. Emissions of isoprene in the afternoon hours could contribute to ozone formation depending upon the ambient levels of NOx [Lee and Wang, 2006; Li et al., 2007; Wiedinmyer et al., 2001]. During the evening hours, the biogenics have a negligible contribution to the average diurnal profile of NormROH,VOC.

3.3.2. Anthropogenic VOCs

[33] In HGB, anthropogenic activities are the predominant source of alkanes, alkenes, monomers, and aromatics. Any natural sources of these compounds will be dwarfed by the large variety and immense scale of anthropogenic sources of these compounds in HGB. In this study, the total reactivity of all anthropogenic VOCs represents the sum of the contributions from urban and industrial sources; however, air masses containing extreme values of ROH,VOC are attributed exclusively to industrial emissions because an urban source would be highly unlikely.

[34] The diurnal profile of NormROH,VOC for the alkanes (Figure 9a) displays a fairly broad distribution throughout the day overlaid by sharp spikes. These sharp spikes are the result of discrete plumes of reactive alkanes composed predominantly of iso-butane, n-butane, iso-pentane, and n-pentane. Overall, the C4–C5 alkanes were the dominant contributors (58% on average) to the total reactivity of alkanes. Large point sources of these compounds may include evaporation during refining processes, leaks from storage tanks and certain industrial flares [Blake and Rowland, 1995; Harley et al., 1992; Strosher, 2000; Viswanath, 1994]. The remaining contribution to the alkane OH reactivity is split between ethane and propane (C2–C3; 14%) and the heavier C6–C10 (28%) alkanes. Industrial sources of ethane and propane are associated primarily with natural gas facilities [Harley et al., 1992]. Plumes of ethane and propane generally do not cause spikes in the diurnal profile because they are less reactive than the C4–C5 alkanes; however, the occurrence of these plumes is evidenced by mixing ratios of up to 197 ppb of ethane and 350 ppb of propane (Table 1). Because ethane and propane are relatively unreactive, they are expected to accumulate in the region and add to the elevated background of VOCs in HGB. Higher molecular weight C6–C10 alkanes are more reactive, but these compounds occurred in low enough concentrations that they only accounted for 28% of the total alkane reactivity with hexane being the dominant single contributor of this subgroup. These alkanes are commonly used in industrial solvents, paints, and inks. The largest urban sources of alkanes include gasoline engines which tend to emit midweight alkanes ranging from C5–C9 [Schauer et al., 2002; Watson et al., 2001] while diesel engine emissions skew toward higher carbon numbers C6–C10 [Lowenthal et al., 1994; Schauer et al., 1999].

Figure 9.

Average diurnal profiles of sunlight and NormROH,VOC for (a) alkanes, (b) alkenes, (c) other alkenes/monomers, and (d) aromatics for the HGB area. (e) A comparison of the average diurnal profile of NormROH,VOC for all VOCs (biogenic + anthropogenic + oxygenated VOCs) and the contribution solely from anthropogenic VOCs.

[35] For the alkenes (Figure 9b), the diurnal profile is similar to that of the alkanes; however, the magnitude of NormROH,VOC has increased by approximately a factor of two. As mentioned previously (Figure 6c), alkenes are the dominant contributors to the average OH reactivity for HGB. Because the alkenes are so reactive, small increases in their concentrations can have a disproportionately large increase in the OH reactivity. Alkenes are associated with emissions from mobile sources [McGaughey et al., 2004; Schauer et al., 2002; Watson et al., 2001] as well as the petrochemical industry [Jobson et al., 2004; Ryerson et al., 2003]. Urban emissions would be expected to have a stronger diurnal pattern as a result of recurrent daily routines such as rushhour commutes in the morning and evenings. A prolonged increase in NormROH,VOC from 1230–1400 UTC (0730–0900 LT) coincides with the local rush hour traffic; however, weak correlations between ethene mixing ratios and other exhaust markers such as ethyne, NOx, and CO, indicate that industrial emissions are likely mixed with urban emissions during this time. The OH reactivity of the alkenes is dominated by ethene and propene which account for half of the reactivity of this chemical class (52% on average). The increase in the magnitude and variability in the concentrations of ethene and propene can be attributed to the large concentrations of petrochemical industrial facilities that are located throughout the Houston metropolitan area [Ryerson et al., 2003].

[36] The impact of industrial emissions, which are likely to be independent of the time of day, is evident in the highly variable diurnal profile for the other alkenes/monomer grouping shown in Figure 9c. This chemical class consists primarily of compounds used in the production of polymers and is primarily attributed to industrial emissions. While 1,3-butadiene is known to have an urban component [Jobson et al., 2004; McGaughey et al., 2004], it is included in the monomers category because of the abundance of industrial point sources in Houston. According to the U.S. Environmental Protection Agency's (EPA) toxic release inventory (TRI) program, there were approximately 1,251,000 lbs of 1,3-butadiene and 214,000 lbs of vinyl acetate released in Harris County (includes Houston, Texas) in 2006. Even with over a million pounds of 1,3-butadiene released in 2006, it was ranked only 10th for total on-/off-site releases for all industries. For the monomers, vinyl acetate and 1,3-butadiene were the largest contributors to the OH reactivity of this group; however, plumes of methyl acrylate, 1-octene, styrene, and acrylonitrile were also encountered. These compounds are used to make a variety of polymers including polyvinyl acetate (PVA), ABS plastic (Acrylonitrile-Butadiene-Styrene), and high-density polyethylene (HDPE) which contains 8–10% 1-octene as a conditioner. Monomers are of particular interest because (1) they are not commonly monitored because of their diminished sources outside of highly industrialized cities such as Houston [Grant et al., 2007; Jobson et al., 2004; Karl et al., 2003] and (2) they have been listed by the U.S. EPA as hazardous air pollutants/toxic air contaminants in the Clean Air Act. The diurnal profile of the monomers lacks the broad distribution seen for the alkanes and alkenes; therefore, the monomers are not expected to contribute much to the background VOC OH reactivity in HGB. However, documentation of these highly reactive compounds unique to the Houston area strengthens the understanding of photochemistry in the region and the connections between emission events of highly reactive compounds and high ozone days.

[37] Of the anthropogenic VOCs, aromatics lack a strong diurnal trend (Figure 9d) and contribute the least to the OH reactivity, 5% on average (Figure 6c). Aromatics are most commonly associated with petroleum-based sources including on-road emissions and refineries [Karl et al., 2003]. The impact of large industrial sources of benzene in addition to those from urban sources was clearly documented by the lack of correlation between benzene and ethyne which is attributed solely to on-road emissions (Figure 2). Karl et al. [2003] attributed only 20% of the aromatic reactivity and 37% of the variance to tailpipe emissions at La Porte, Texas, while the remainder is ascribed to industrial emissions. Industrial sources of the C8–C9 aromatics in HGB are evident in maximum mixing ratios that are commonly several orders of magnitude larger than the median values (Table 1). While emissions of aromatics are not expected to contribute much to ozone formation, they can have a significant impact on regional air quality because of their known toxicities and possible contributions to secondary aerosol formation.

[38] The contribution of the anthropogenic VOCs to the average diurnal profile of NormROH,VOC for all VOCs is shown in Figure 9e. It is clear that the anthropogenic VOCs dominate the reactivity at night and are responsible for the increasing magnitude of NormROH,VOC just before sunrise. This pool of reactive anthropogenic VOCs would provide ample fuel for the photochemical production of ozone that typically commences with the sunrise and the morning rush hour commute. The accumulation of these anthropogenic VOCs results from their continued emissions and lack of chemical losses overnight. Alkenes and monomers are the only anthropogenic compounds that are reactive with NO3 and O3. As mentioned previously, NO3 concentrations were low and ozone reaction rates are slow enough making either of these chemical sinks minor in comparison to that of OH during the day. The elevated reactivity of anthropogenic VOCs during the daylight hours suggests continuous and robust emissions of these compounds in order to compensate for chemical losses due to reactions with OH.

3.3.3. Oxygenated VOCs

[39] Oxygenated VOCs can be directly emitted by natural, urban, and industrial sources. Additionally, oxygenated VOCs can be formed in the atmosphere as secondary products from hydrocarbon oxidation. Industry commonly uses these compounds as solvents and cleaners. Primary emissions of methanol from industrial sources have been documented in La Porte, Texas [Jobson et al., 2004], and are evidenced in this study by the exceptionally high maximum mixing ratio of methanol at 144 ppb (Table 1). MTBE has been used in the past as a fuel additive and other carbonyls, such as the C4–C6 aldehydes, have been linked to primary emissions from vehicle exhaust from both gasoline and diesel engines [Ban-Weiss et al., 2008; Jakober et al., 2008; Schauer et al., 2002]. Primary sources of formaldehyde include incomplete combustion of fossil fuels (i.e., urban) [Ban-Weiss et al., 2008; Sigsby et al., 1987], vegetative emissions (i.e., biogenic) [Holzinger et al., 2007; Seco et al., 2007], and in the production of various resins, plastics, adhesives, foams, and fertilizers (i.e., industrial); however, these primary emissions are small and are typically outweighed by the secondary photochemical production in urban areas [Li et al., 1994; Wert et al., 2003].

[40] The diurnal profile of NormROH,VOC for the oxygenated VOCs (Figure 10a) follows the solar radiation trace and mimics the diurnal profile for ozone (Figure 6c) indicating the importance of the secondary production of oxygenated VOCs on potential ozone formation in HGB. On average, formaldehyde and acetaldehyde account for 47% and 30%, respectively, of the oxygenate OH reactivity in HGB. The contribution of formaldehyde to NormROH,VOC in HGB and the central gulf is shown in Figure 10a. The daytime NormROH,VOC of formaldehyde increased from 0.1 s−1 in the central gulf to greater than 1.0 s−1 in HGB. This tenfold increase is a direct result of the increased amount of reactive precursors such as ethene and propene from anthropogenic sources in HGB [Wert et al., 2003]. The contribution of the oxygenated VOCs to the average diurnal profile of NormROH,VOC is shown in Figure 10b. Oxygenated VOCs account for approximately 40% of the total OH reactivity of VOCs during peak production hours equaling that seen for anthropogenic VOCs. The elevated daytime values of NormROH,VOC for the oxygenated VOCs are the result of the secondary formation of reactive aldehydes from a steady source of predominately anthropogenic alkenes in addition to a small contribution from biogenic VOCs. These results show that accurate representation of the reactive alkenes and their oxidized products, formaldehyde and acetaldehyde, will be critical to air quality models and the development of efficient ozone control strategies for Houston.

Figure 10.

(a) Average diurnal profiles of sunlight and NormROH,VOC for all oxygenated VOCs and the fractional contribution of formaldehyde (HCHO) in Houston and Galveston Bay. The average diurnal profile of NormROH,VOC for formaldehyde the central gulf has been added for comparison. (b) A comparison of the average diurnal profile of NormROH,VOC for all VOCs (biogenic + anthropogenic + oxygenated VOCs) and the contribution solely from oxygenated VOCs in the HGB area.

4. Summary

[41] A total of 81 VOCs and several other gas phase compounds were measured in situ aboard the NOAA R/V Brown as part of the 2006 TexAQS/GoMACCS field experiment. The role of VOCs in tropospheric photochemistry in the Gulf of Mexico and the Houston and Galveston Bay areas (HGB) was determined from the total OH reactivity (ROH,TOTAL) which provides an objective measure of the relative role of various species to potential ozone formation. In the central gulf, VOCs were only a minor contributor to ROH,TOTAL which was dominated by CO and methane. The VOC composition was dominated by oxygenated VOCs and longer-lived hydrocarbons resulting in a low VOC OH reactivity (ROH,VOC). In HGB, VOCs were the dominant contributor to ROH,TOTAL identifying them as a critical component in the photochemical production of ozone in the Houston area. The increase in ROH,VOC in the HGB area was a direct result of fresh emissions of reactive VOCs from anthropogenic sources which were heavily influenced by industrial emissions. The impact and variability of industrial sources were evidenced by (1) very high maximum mixing ratios (>50 ppbv) of a variety of VOCs, (2) a disparity between the correlation between benzene and acetylene for typical urban areas compared to the highly industrialized HGB, (3) plumes of VOCs with markedly high reactivities (>20 s−1) that were composed of a single chemical or chemical class, and (4) spikes in the diurnal profiles of the normalized OH reactivity (NormROH,VOC) for the alkanes, alkenes, monomers, and biogenic VOCs. Detailed documentation of industrial releases of VOCs will help reconcile the disparity between ambient measurements and emission inventories, which currently underestimate the emissions of highly reactive VOCs [Ryerson et al., 2003].

[42] The diurnal profile of NormROH,VOC provided insight into the sources and atmospheric chemistry of VOCs that potentially contribute to the photochemical production of ozone as a function of the time of day. On the basis of the NormROH,VOC diurnal profiles, (1) the biogenics were determined to be almost exclusively natural in origin even though there was direct evidence of at least some industrial emissions, (2) emissions of anthropogenic VOCs and the lack of chemical removal throughout the night resulted in the accumulation of these reactive compounds just before sunrise, and (3) the source of the most reactive oxygenated VOCs was found to be overwhelmingly from secondary production. Coincident with the peak ozone production hours, elevated values of NormROH,VOC were primarily sustained by continued emissions of anthropogenic VOCs and from the secondary formation of oxygenated VOCs from a steady source of alkenes from industrial and urban sources. High levels of reactive VOCs in the afternoon could contribute to ozone production by reacting with an increased level of free radicals produced from the photolysis of ozone and other VOCs. These results show that accurate representation of the reactive alkenes and their oxidized products, formaldehyde and acetaldehyde, will be critical to air quality models and the development of efficient ozone control strategies for Houston.

Acknowledgments

[43] The authors would like to gratefully acknowledge the crew of the NOAA R/V Brown, NOAA's Air Quality and Climate Research and Modeling Programs, and funding from the Texas Commission on Environmental Quality (TCEQ). J. Gilman would like to thank J. Roberts and D. Welsh-Bon of the NOAA/ESRL; H. Osthoff of the University of Calgary for helpful discussions; and D. Hamilton, D. Coffman, and T. S. Bates of NOAA/PMEL for their help and expertise provided throughout the campaign.

Ancillary