Quantification of NO2 and SO2 emissions from the Houston Ship Channel and Texas City industrial areas during the 2006 Texas Air Quality Study

Authors


Abstract

[1] In August–September 2006, as part of the Second Texas Air Quality Study, NO2 and SO2 emissions from the Houston Ship Channel and Texas City industrial areas were quantified using mobile mini-differential optical absorption spectroscopy instruments. The measured NO2 emissions from the Houston Ship Channel and Texas City industrial areas were 2542 and 452 kg h−1, respectively, yielding NOx emissions 70% and 43%, respectively, above the reported inventory values. Quantified SO2 emissions from the Houston Ship Channel area were 1749 kg h−1 and were found to be 34% above the values reported in the inventory. Short-term variability of NO2 and SO2 emissions was found at the Houston Ship Channel. On 31 August 2006, a plume was detected at the HSC during three consecutive measurements, yielding a HCHO flux of 481 kg h−1. This event has been mainly attributed to photochemical production.

1. Introduction

[2] Air pollution is a problem in many cities of the world, and the Houston-Galveston, Texas, area (HGA) is no exception, with one of the most severe ozone (O3) problems in the United States. The process of tropospheric O3 production by photochemical oxidation of industrial hydrocarbons involving NO and NO2 as catalysts has been described in the literature [Crutzen, 1979]. This process is highly dependent on the concentration and chemical composition of hydrocarbons [Grover and Bradford, 2001], as well as on NOx [Jacob et al., 1996; Kasibhatla et al., 1991; Martin et al., 2002; Murphy et al., 1993; Penner et al., 1991]. Both NOx and VOCs are a product of combustion processes, from either stationary or mobile sources, and have biogenic sources as well [Conley et al., 2005; Finlayson-Pitts and Pitts, 2000]. In the HGA, emissions of these compounds are associated with petrochemical industrial facilities as well as traffic [Berkowitz et al., 2005; Buzcu-Guven and Fraser, 2008; De Gouw et al., 2009; Gilman et al., 2009; Jobson et al., 2004; Leuchner and Rappenglück, 2010; Smith et al., 2007; Xie and Berkowitz, 2007].

[3] The Texas Air Quality Study 2000 (TexAQS 2000) demonstrated that industrial complexes in the HGA are large emitters of NOx and hydrocarbons, which originate spatially localized plumes rich in O3 and hydrocarbon oxidation products (i.e., formaldehyde) [Daum et al., 2003, 2004; Kleinman et al., 2002; Ryerson et al., 2003]. Detailed studies have demonstrated that the high O3 production rates in the HGA are linearly dependent on the ratio of total nonmethane hydrocarbons to NOx [Berkowitz et al., 2005]. In the HGA, the production tends to be more rapid and efficient than in other urban areas [Kleinman et al., 2002; Lei et al., 2004]. In addition a detailed particulate growth study in the HGA concluded that SO2-rich plumes were associated with particle formation and substantial particle volume growth [Brock et al., 2003].

[4] Formaldehyde has a very short lifetime (generally 2–4 h) in the presence of sunlight, which is determined mainly by photolysis and reaction with OH radicals [Garcia et al., 2006; Wert et al., 2003]. It can be directly emitted by incomplete combustion or industrial processes or be secondarily formed by oxidation of VOCs [Finlayson-Pitts and Pitts, 2000]. Studies have shown that oxidation of ethene, propene, and isoprene tends to efficiently produce HCHO [Dodge, 1990; Goldan et al., 2000; Lee et al., 1998; Wert et al., 2003]. During TexAQS 2000, Wert et al. [2003] concluded that there was no evidence for strong primary emissions of HCHO when compared to the secondary emissions produced in several Houston area plumes.

[5] Worsening air quality in the HGA has increased the concern of the general public and policymakers regarding environmental problems generated by outdoor pollution and its relationship with human health [Brody et al., 2004; Conley et al., 2005]. Because O3 threshold mixing ratios established by the National Ambient Air Quality Standard are frequently exceeded, the HGA has been designated an O3 nonattainment region, and it is therefore important to identify and quantify O3 precursors and reduce their emissions [Allen et al., 2004; Byun et al., 2007; Grover and Bradford, 2001; Kleinman et al., 2002]. With the objective to provide detailed accurate emission information, a 5-week field campaign in the HGA was performed during the summer of 2006. NO2, SO2, and HCHO differential vertical columns were quantified in the Houston Ship Channel (HSC) and Texas City (TC) industrial areas by traversing the plume of these industrial complexes using a mini-differential optical absorption spectroscopy (DOAS) instrument installed on a car. The DOAS measurements of NO2 and SO2 presented here were conducted in parallel with airborne and mobile solar IR measurements to quantify VOC emissions [De Gouw et al., 2009]. The latter study shows large discrepancies between measured and reported emissions. It was therefore of great interest to investigate whether discrepancies exist also for NO2 and SO2 emissions. The objective of this paper is to determine the consistency of our measurements with available emission inventories of the HGA.

2. Method and Data Analysis

[6] DOAS is a technique that allows the remote detection of trace gases. It is based on the absorption of electromagnetic radiation by matter and is widely used for continuously quantifying atmospheric gases [Finlayson-Pitts and Pitts, 2000; Platt, 1994; Platt et al., 1979; Platt and Stutz, 2008]. DOAS measurements can be performed using a wide variety of experimental setups. In this field campaign, passive DOAS using scattered sunlight by air molecules and particles has been used in zenith scattered light configuration. Zenith scattered sunlight measurements offer a large variety of applications, are one of the earliest applications of passive DOAS, and have been used to study stratospheric chemistry and the radiative transport in clouds [Platt and Stutz, 2008].

2.1. Mobile Mini-Differential Optical Absorption Spectroscopy

[7] A mobile miniDOAS instrument, described in detail by Galle et al. [2002], was used to perform the measurements presented in this paper. MiniDOAS instruments have been widely adopted by the scientific community to quantify emissions from urban [Johansson et al., 2008, 2009], industrial [McGonigle et al., 2004; Rivera et al., 2009], and volcanic [Bobrowski et al., 2003; Bobrowski and Platt, 2007; Galle et al., 2002; Mori et al., 2006] sources. Our system utilizes an 8 mrad field of view telescope to collect scattered UV light. The telescope has a quartz lens and, only for SO2 measurement purposes, a Hoya U-330 filter was used to reduce stray light, blocking visible light with wavelength higher than 360 nm. The telescope is then coupled to a 600 μm quartz fiber optic which transfers light into an Ocean Optics S2000 spectrometer with spectral resolution of approximately 0.6 nm and spectral range of 280–420 nm (SO2 and HCHO) and 336–480 nm (NO2). The spectrometer and a GPS were connected to a laptop computer controlled by custom-built software, Mobile DOAS [Johansson and Zhang, 2004], which acquired and preliminarily evaluated spectra in real time.

2.2. Measurement Method

[8] The mobile miniDOAS instrument was mounted on a car and aligned toward zenith, and spectra were recorded encircling the sources of pollution. Differential vertical columns of the species of interest were measured upwind and downwind of the sources in order to derive inflow and outflow, respectively, with the flux being the difference between them. For the Texas City industrial area, measurements were conducted traversing the downwind plume. Typically measurements were done 100 m to about 10 km (only the case of the HSC industrial area) downwind of the sources since these are distributed over large areas with no accessibility except for public roads. Every measurement series started with the recording of a background reference spectrum, considered to be the reference for all the following spectra recorded during the measurement series. Ideally the reference spectrum is expected not to include any concentration above ambient of the trace species of interest; however, in urban and industrial areas this is difficult to achieve, and therefore our measurement in this case will produce the difference in vertical columns between the background reference spectrum and all measured spectra across the plume for every measurement series. All vertical columns derived from spectra collected after the reference spectrum are relative to the latter and will be further referred to as differential vertical columns.

[9] It is important to notice that we have assumed that the light path is vertical and no air mass factor has been applied to the differential vertical columns, which could bring uncertainties to the results of our measurements. The measurements presented in this study were conducted during days of clear blue sky without presence of clouds. In order to give insight for high aerosol loads which could lead to enhanced light paths through multiple scattering, the oxygen dimer (O4) was included into the DOAS fit. All measurements conducted during the field campaign that showed indication of enhanced light paths during the traverses were disregarded. More details about this procedure are given by Johansson et al. [2008]. In section 2.5, the uncertainty related to light path extension inside the plume (multiple scattering) as well as to scattering of photons into the instrument's field of view by particles beneath the plume (light dilution) is further discussed.

[10] Stratospheric NO2 variations are expected to be small in the short time period (few minutes to 1 h) our measurements take when compared with tropospheric NO2 variations present in the polluted plumes we are quantifying. In addition, using a background reference spectrum for each measurement series eliminates the stratospheric NO2 part and has less influence of instrumental instabilities, as already pointed out by Sinreich et al. [2005].

[11] GPS data were recorded providing time and position before and after each spectrum was collected. Fluxes from traverses were calculated by multiplying differential total vertical columns by the distance traversed perpendicular to the wind direction and by the wind speed. Wind information was obtained from GPS sondes regularly launched during the field campaign (see section 2.4).

2.3. Spectral Evaluation

[12] SO2, HCHO, and NO2 were retrieved from the recorded spectra using the WinDOAS software (version 2.1) (Van Roozendael and Fayt, 2001) in the wavelength regions of 310–325, 324–347, and 420–460 nm, respectively. This procedure has been described in detail elsewhere [Galle et al., 2002; Platt and Stutz, 2008]. The cross sections used in the DOAS fit (SO2 [Bogumil et al., 2003], HCHO [Meller and Moortgat, 2000], and NO2 [Vandaele et al., 1998]) were collected from the literature and degraded to our instrument's resolution. In addition, the cross sections of O3 [Voigt et al., 2001], O4 [Hermans et al., 1999], and H2O [Coheur et al., 2002; Fally et al., 2003] were included in the fitting procedure as well. To compensate for the filling up of the Fraunhofer lines in the solar spectrum by inelastic scattering, the so-called Ring effect [Platt and Stutz, 2008], a synthetic Ring spectrum calculated from a high-resolved solar spectrum (R. L. Kurucz, I. Furenlid, J. Brault, and L. Testerman, 1984, Solar flux atlas from 296 to 1300 nm) (http://kurucz.harvard.edu/sun/fluxatlas/) using the DOASIS software [Kraus, 2003] was also included in the fit. Details of every fitting interval and cross sections used are shown in Table 1. Examples of DOAS fits for NO2, HCHO, and SO2 are shown in Figure 1. The number of spectra collected for the SO2 and NO2 measurements was in all cases automatically adapted to give a time resolution of approximately 5 s per measurement point. However, for HCHO evaluation, spectra were coadded in order to increase the signal-to-noise ratio, resulting in a time resolution of 86 s per measurement point.

Figure 1.

Examples of differential optical absorption spectroscopy (DOAS) fit in different wavelength regions. Fitted cross sections of (a) NO2 in the wavelength region of 420–460 nm, (b) HCHO in the wavelength region of 324–347 nm, and (c) SO2 in the wavelength region of 310–325 nm.

Table 1. Fitting Intervals and Molecules Included in the Differential Optical Absorption Spectroscopy Fit
Fitting Interval Wavelength Range (nm)Species
310–325SO2, O3, NO2, HCHO, Ring
324–347HCHO, O3, NO2, SO2, O4, Ring
420–460NO2, O3, O4, H2O, Ring

2.4. Wind Data

[13] Wind speed and wind direction data were obtained from wind fields recorded by GPS balloon sondes (RS80-15N-V2D-A-GPS) obtained from Environmental Science Corporation, Boulder, Colorado. GPS sondes were launched in between our measurements and nearby the measurement paths used. Typically, three or four balloon sondes were launched every day; the rise speed of the balloons was typically 5 m s−1. In the HSC the soundings were carried out close to the Lynchburg Ferry crossing (Figure 2), and in Texas City the soundings were launched westward of the town (Figure 3). When balloon sondes were not available, wind data were used from a radar profiler operated by the Texas Commission of Environmental Quality within the NOAA profiler network at the La Porte Airport south of the HSC.

Figure 2.

Measurement performed on 31 August 2006 starting at 1037 LT at the Houston Ship Channel. Shown are the spatial distributions of (a) NO2, (c) SO2, and (e) HCHO along the Houston Ship Channel quantified during the traverse, the measured differential vertical columns of (b) NO2, (d) SO2, and (f) HCHO, (g) the spatial distribution of HCHO measured by the Piper Aztec aircraft, and (h) the differential vertical column of O4 measured during the traverse. The black square indicates the starting point of the traverse. The site where balloon sondes were launched (close to the Lynchburg Ferry crossing) is indicated by a black triangle. In gray scale, the known NOx, SO2, and HCHO sources from the EI in the HSC are shown. The black arrow at the right upper corner of Figures 2a, 2c, and 2e indicates the wind direction obtained from wind fields recorded by a GPS balloon sonde launched at 0953 LT.

Figure 3.

Measurement performed on 20 September 2006 starting at 1634 LT at Texas City. (a) The spatial distribution of NO2 along the measurement path quantified during the traverse. (b) Measured differential vertical columns of NO2 along the traverse. The black square indicates the starting point of the traverse. The site where balloon sondes were launched is indicated by a black triangle. In gray scale, the known NOx sources from the EI in Texas City are shown. The black arrow at the right upper corner of Figure 3a indicates the wind direction obtained from wind fields recorded by a GPS balloon sonde launched at 1428 LT.

[14] The measurements presented in this study were conducted in sunny conditions. Under these conditions the industrial emission plumes mix rather quickly in the vertical, giving a homogeneous distribution of the pollutants versus height through the mixing layer even several kilometers downwind. In addition to the atmospheric mixing, the plumes from process industries exhibit an initial lift since they are usually hotter than the surrounding air. This assumption of rapid mixing agrees with Doppler lidar measurements conducted from the ship RV Ronald H. Brown by NOAA during the TexAQS 2006 [Tucker et al., 2009]. From these measurements information about mixing height and the vertical mixing of the atmosphere could be obtained showing typical daytime vertical mixing speeds of ± (0.5–1.5) m s−1.

[15] The measurements presented in this study were typically conducted at a plume transport time of 200–500 s downwind of the industries, which, according to the discussion above, means that sufficient time has elapsed for the emission plume to mix at heights of up to several hundred meters above the ground. For this reason we have used the average wind over the lower 500 m of the atmosphere as a proxy for the mass-weighted wind.

2.5. Emission Error Estimation

[16] Several types of errors can affect the calculation of emissions using a miniDOAS instrument. A spectroscopic error is related to errors in the absorption cross sections, temperature changes in the spectrometer, fitting errors, and spectral interferences. The spectroscopic error in this study reflects the error in calculated emissions due to the DOAS retrieval of the differential vertical column and was calculated as 10%, 14%, and 40% for NO2, HCHO, and SO2 respectively.

[17] A scattering error is expected due to light path extension inside the plume (multiple scattering) as well as to scattering of photons into the instrument's field of view by particles beneath the plume (light dilution). The scattering error strongly depends on meteorological conditions (i.e., cloud coverage, fog, and rain) [Millán, 1980; Moffat and Millán, 1971] and changes as a function of the distance of the instrument from the plume and the wavelength region used in the DOAS retrieval, becoming more significant at shorter wavelengths for the case of SO2 [Mori et al., 2006]. Because the measurements presented in this study were conducted during days of clear blue sky without presence of clouds and care was taken to disregard measurements with clear indication of enhanced light paths during the traverses (see section 2.2), the scattering error was assumed to be 10% for this study.

[18] The main uncertainty, however, is due to assumptions about the vertical distribution of the plume, which will lead to errors in attributing the true wind speed for each molecule in the plume. Because we are not able to calculate the plume height with our measurements, we have assumed that the emission plume from an industry mixes rather quickly vertically, due to convection from solar heating of the ground, with a speed of about 0.5–1.5 m s−1 in addition to initial plume lifting due to high temperature. Further discussion about the wind data used in this study can be found in section 2.4 and the paper of Mellqvist et al. [2010]. Uncertainty in the wind speed and wind direction from GPS balloon sondes, used to calculate emissions presented in this study, is about 30% [Mellqvist et al., 2010]. Our estimated uncertainty in the wind field accounts for possible measurement errors and systematic spatial wind differences, in addition to errors in estimated plume height and therefore errors in wind speed and wind direction used to calculate fluxes, as well as errors due to changes in wind speed and wind direction during the period of time a measurement is being conducted. Further details about expected errors using this technique are given by Johansson et al. [2008] and Rivera et al. [2009].

[19] Another uncertainty regarding the measurement method is the fact that the measurements are conducted using a mobile van and driving on roads exposed to traffic emissions. Because in this study it is not possible to separate industrial plumes from traffic emissions, we have assumed that emissions from traffic are negligible compared to emissions by industries and form part of the background emissions.

[20] The combined uncertainty for calculated emissions taking into consideration all errors and uncertainties is therefore 33%, 35%, and 51% for NO2, HCHO, and SO2, respectively.

2.6. Airborne Measurements

[21] A Piper Aztec (N6670Y) aircraft operated by the Baylor Institute for Air Science conducted several research tasks during the TexAQS 2006 field experiment. Aircraft-based measurements were focused on the collection of lower tropospheric air quality data from August through October 2006 in the Houston region.

[22] NO, NO2, HCHO, and reactive alkenes, among other parameters, were measured during the conducted flights. NO and NO2 data recorded during flights conducted on the same measurement paths as our measurements by the Piper Aztec aircraft were used in this study with the objective to gain knowledge of the NO2/NOx ratio. NO was measured using an Eco Physics CLD 77. NO2 mixing ratios were measured using the photolysis-chemiluminescence technique [Kley and McFarland, 1980], consisting of a Thermo Environmental 42C instrument and a Droplet Measurement Technologies BLC photolytic converter. Baseline corrections, from the periodic in-flight zeroing of the instruments, were applied automatically for both species. Corrections for both ambient water and pressure were also applied. The uncertainty of NO and NO2 was 0.4 and 0.2 ppbv, respectively [Alvarez et al., 2007].

3. Results and Discussion

[23] Measurements were performed between 30 August and 29 September 2006 aimed at quantifying emissions from different areas in the HGA. All measurements were compared with reported values in order to give insight into the consistency of our measurements with the available emission inventories of the area. An emissions inventory (EI) developed by the Texas Commission on Environmental Quality was used in this study (the 2006 Point Source Emissions Inventory) (http://www.tceq.state.tx.us/implementation/air/industei/psei/psei.html). It contains daily averaged emission data for approximately 3000 sources for the time period 15 August to 15 September 2006. During this period, some industrial facilities in the HGA reported their monitored hourly emissions of NOx, SO2, and VOCs, among other pollutants. In the EI used, these data have been merged with an annual database, providing daily emissions for sources not included in the hourly monitored data set. It is important to notice that daily emissions reported by the industries show very little variability (about 1%) during the days our measurements were compared with the inventory.

[24] Table 2 shows a comparison between the median of total fluxes measured in this field study and reported emissions in the inventory for HSC and TC sites. The median absolute deviation from the median (MAD) refers to the variability of daily quantified fluxes during the field campaign and not to uncertainty of the measurement itself (discussed in section 2.5). In addition, a detailed account of daily measurements carried out during the field campaign is presented in Table 3.

Table 2. Summary of Total Quantified Fluxes (Median) and Comparison With Emissions Inventorya
NO2NOx (kg h−1)SO2HCHO
Flux (kg h−1)No. Meas.Adjusted ValuesbEIFlux (kg h−1)No. Meas.EIFlux (kg h−1)No. Meas.EI
  • a

    The median absolute deviation from the median is shown in parentheses. Abbreviation is as follows: EI, emission inventory.

  • b

    The NOx fluxes have been adjusted from the original quantified total NO2 values according to the NO2/NOx ratio derived by aircraft measurements: 0.74 for Houston Ship Channel and 0.73 for Texas City. NOx fluxes have no further correction applied for rapid photochemical NOx loss.

Houston Ship Channel
2542 (214)93435 (289)20151749 (160)31308481 (21)314
 
Texas City
452 (115)11619 (158)433      
Table 3. NO2, SO2, and HCHO Quantified Fluxes During the Field Campaign
DateStart Time (LT)SpeciesWind Direction (°)Wind Speed (m s−1)Flux (kg h−1)
Houston Ship Channel
31 Aug 20060833NO2752.072756
 0946NO2752.072650
  HCHO  542
 1037NO2752.072540
  SO2  3493
  HCHO  481
 1153NO2852.041411
  SO2  1589
  HCHO  460
4 Sep 20061250NO2813.64946
  SO2  1749
19 Sep 20061127NO24710.784075
 1221 316.552817
 1512 384.262542
 1604 384.262402
 
Texas City
2 Sep 20061340NO2991.51186
 1405 1052.23250
14 Sep 20060950NO21023.96452
 1118 1023.96449
20 Sep 20061012NO2676.16567
 1059 676.16634
 1310 774.9404
 1425 933.72375
 1506 933.72499
 1619 933.72644
 1634 933.72596

3.1. NOx Emissions

[25] Figure 2a shows the spatial distribution of differential NO2 columns (color coded) quantified during a measurement conducted at the HSC on 31 August 2006 starting at 1037 LT. The known emission sources from the EI inside the HSC are shown as gray dots. Figure 2b shows the differential NO2 columns quantified during the measurement. The traverse started at the northeast part of the HSC (black square in Figure 2a) and continued toward the west. Two main peaks were found during the traverse; the highest peak found at the southwest is thought to include most of the accumulated emissions of the HSC, while the second, smaller peak most likely represents emissions from the eastern part of the HSC. As discussed in section 2.2, the oxygen dimer (O4) was included into the DOAS fit in order to indicate the presence of high aerosol loading, which could lead to enhanced light paths through multiple scattering. Figure 2h shows the differential O4 columns quantified during the measurement where a stable behavior of the oxygen dimer during the traverse is evident, indicating no enhanced light paths during the traverse.

[26] From NO2 emissions data reported in Table 3, it can be seen that differences in NO2 emission rates were found at the HSC during the same day and among different days. On 31 August, NO2 quantified emissions dropped by half comparing a traverse conducted at 1153 LT to traverses conducted previously during the same day.

[27] Figure 3a shows the spatial distribution of differential NO2 columns (color coded) quantified during a measurement conducted at TC on 20 September 2006 starting at 1634 LT. The known NOx emission sources of the TC area are shown as gray dots. Figure 3b shows the differential NO2 columns quantified during the measurement. The traverse started at the north part of a highway located west of TC (black square in Figure 3a) and continued toward the southeast. A main peak was identified during the traverse, found at the western part of the TC industrial area. This peak is thought to represent one of the largest sources of NOx in the area located at the western part. The other, smaller peaks most likely represent emissions from the whole TC area in general. Some differences were found in NO2 quantified emissions at TC among different days (Table 3).

[28] The instrument used in this field campaign is capable of quantifying NO2 but not NOx, which is given in the inventories. Knowledge about the ratio NO2/NOx becomes important in order to assess the consistency of our measurements with the available inventories. Here we have used data from measurements conducted by the Piper Aztec aircraft flying above the same measurement paths we have used to quantify emissions from the HSC and TC industrial sites. Aircraft-based measurements indicate a NO2/NOx ratio of 0.74 for HSC and 0.73 for TC along the measurement paths we have used. Figure 4 shows recorded NO2 and NOx mixing ratios by the Piper Aztec aircraft while performing a vertical profile between 356 and 1265 m above mean sea level along the HSC on 31 August 2006 at 0800-0815 LT. As can be seen, most of the quantified NOx is in the form of NO2. For this field experiment, NOx average emissions have been estimated from the original quantified total NO2 values according to the NO2/NOx ratio derived by aircraft measurements: 0.74 for HSC and 0.73 for TC. This ratio was estimated for only 1 day for each industrial area; however, it was extrapolated to other measurement days, slightly adding uncertainty to our emission estimates of NOx.

Figure 4.

NO2 and NOx mixing ratios recorded by the Piper Aztec aircraft while performing a vertical profile between 356 and 1265 m above mean sea level along the Houston Ship Channel on 31 August 2006 at 0800–0815 LT.

[29] In summary, large differences in quantified NO2 emissions were found during and among different days of the field campaign. Our results show that NOx-derived emissions from the HSC and TC industrial areas are 70% and 43%, respectively, above the reported inventory values, with 33% uncertainty.

[30] A study conducted by Williams et al. [2009] during the TexAQS 2006 reveals that NOx emissions from commercial marine shipping in the HGA are significant and present large variability (30%–40% relative standard deviation), which could explain NOx emissions measured in excess in the HSC and probably to a lesser extent in TC, as well as the measured short-term variability in emissions found at the HSC during the field campaign.

3.2. SO2 Emissions

[31] During the measurement period, SO2 plumes were only successfully quantified in the HSC industrial area. Figure 2c shows the spatial distribution of differential SO2 columns (color coded) quantified during a measurement conducted at the HSC on 31 August 2006 starting at 1037 LT. The known SO2 emission sources from the EI inside the HSC are shown as gray dots. Figure 2d shows the differential SO2 columns quantified during the measurement. The traverse started at the northeast part of the HSC (black square in Figure 2c) and continued toward the west. Two main SO2 peaks were found during the traverse, coinciding with the previously discussed NO2 peaks. As for the case of NO2, the highest SO2 peak found at the southwest is thought to include most of the accumulated emissions of the HSC, while the second, smaller peak most likely represents emissions from the eastern part of the HSC.

[32] Table 2 shows a comparison between measured SO2 fluxes and reported values in the inventory. The reported daily emissions from the EI are also presented for comparison. In addition, a full summary of daily SO2 measurements carried out during the field campaign is presented in Table 3. Table 2 indicates that the quantified SO2 fluxes at the HSC industrial area were 34% higher than the emissions reported in the inventory, with a range of uncertainty, however, that encompasses the inventory.

3.3. HCHO Event

[33] On only 1 day during the field campaign was it possible to clearly identify the presence of HCHO in the plumes from the HGA. On 31 August 2006, HCHO was quantified during three consecutive measurements performed at the HSC industrial site between 0946 and 1245 LT (1446-1745 UTC). Figure 2e shows the spatial distribution of differential HCHO columns (color coded) quantified during a measurement conducted at the HSC on 31 August 2006 starting at 1037 LT. The known HCHO emission sources from the EI inside the HSC are shown as gray dots. Figure 2f shows the differential HCHO columns quantified during the measurement. The traverse started in the northeastern part of the HSC (black square in Figure 2e) and continued toward the west. As for the case of the NO2 and SO2 emissions measured during the same traverse, two main HCHO peaks were found during this traverse; the smaller peak found in the southeastern corner is thought to include the accumulated flux of the entire HSC, while the second, larger peak most likely represents the flux from the eastern part of the HSC where most of the HCHO sources are located.

[34] The three consecutive transects yielded a median flux of 481 kg h−1 with a MAD of 21 kg h−1. From HCHO fluxes data reported in Table 3, it can be seen that the quantified HCHO fluxes during the three consecutive traverses give very similar values. The high HCHO quantified flux at the HSC is unexpected, especially when compared to the emissions inventory of 14 kg h−1 of HCHO for the HSC area (Table 2), representing primary emissions from the industries. It is important to note that only a small fraction of industrial facilities specify, in the data reported in the EI, the type of VOC emitted. In this case we have added all sources located inside the HSC that specifically report HCHO emissions; however, it is possible that some facilities have been missed because of the way of reporting HCHO as VOCs in general.

[35] During the morning flight of 31 August 2006, the Piper Aztec aircraft measured elevated mixing ratios of HCHO over the HSC. Figure 2g shows the spatial distribution of HCHO along the HSC measured by the Piper Aztec aircraft. The peak on the HCHO mixing ratio measured by the aircraft at the southern part of the HSC (Figure 2g) coincides with measurements reported in this study conducted by the mobile miniDOAS instrument.

[36] It is highly likely that the quantified HCHO in the HSC during this field observation was not directly emitted but rather was formed by oxidation of emitted VOCs. During this flight, the Piper Aztec aircraft also measured elevated mixing ratios of reactive alkenes at the southwestern part of the HSC [Alvarez et al., 2007], coinciding with HCHO measurements reported in this study conducted by the mobile miniDOAS instrument. In addition, Mellqvist et al. [2007, 2010] measured large and highly variable emissions of propene, 200–2000 kg h−1, on the same day and upwind of the location where the high formaldehyde differential columns were measured with the mobile miniDOAS instrument [Mellqvist et al., 2007]. In addition, 31 August 2006 was characterized by low wind speeds (about 2 m s−1), suggesting both more concentrated VOC levels and a longer photochemical processing time between the VOC emission and sampling time.

[37] We consider that the presence of HCHO in the HSC on 31 August 2006 is an isolated event rather than a permanent situation since it was only observed for a short period of time (3 h) and only once during 5 weeks of measurements conducted in the HGA. The occurrence of industrial emission events in the HGA has been previously identified and studied. They typically last hours to days and are largely made up of ethene and propene. Some events are reported to release a few hundred kilograms [Murphy and Allen, 2005], while others have been reported to lead to emission rates of up to several thousand kilograms per hour [Vizuete et al., 2008]. More than half of the mass of highly reactive VOC emission events has been attributed to ethene and almost one third to propene; the remainder consists of isomers of butene and 1,3-butadiene [Murphy and Allen, 2005]. In this case, the HCHO quantified in the HSC cannot be regarded as an emission but rather as the photochemical production of HCHO by the oxidation of emitted alkenes, which tend to efficiently produce HCHO [Dodge, 1990; Goldan et al., 2000; Lee et al., 1998; Wert et al., 2003].

[38] In addition, as part of the TexAQS 2006 field experiment, Brooks et al. [2009] report extreme pollution events in the early morning hours of 31 August 2006 during measurements conducted atop the Moody Tower TexAQS II Radical Measurement Project site. Their findings include the highest fine particulate mercury value observed on 31 August 2006, along with extremely high values of VOCs, SO2, aerosol mass, and most other primary pollutants. At 1430 UTC spikes in HCHO (23 ppb), PAN (3.6 ppb), and O3 (∼100 ppb) were reported as well. Back trajectories and wind direction showed air parcels coming from the Houston Ship Channel [Brooks et al., 2009].

4. Conclusions

[39] In summary, measurements conducted during the TexAQS 2006 field experiment show that measured SO2 emissions at HSC were 34% higher than the emissions reported in the inventory, with a range of uncertainty, however, that encompasses the inventory. The derived NOx emissions at the HSC and TC industrial areas were 70% and 43%, respectively, above the reported inventory values, with 33% uncertainty. This should be compared to mobile solar infrared measurements and airborne measurements showing an order of magnitude difference in measured and reported ethene emissions from industrial sources in the HGA [De Gouw et al., 2009]. Short-term variability of NOx and SO2 emissions were also found. On 31 August 2006, the measured emissions of NO2 and SO2 at the HSC dropped by half between a traverse conducted at 1153 LT when compared to traverses conducted previously during the same day.

[40] During the TexAQS 2006 field experiment, it was possible to identify a HCHO event occurring on 31 August 2006 yielding HCHO fluxes an order of magnitude greater than the reported primary emissions in the inventory. For this reason and due to the fact that during daytime, generally the amount of primary emitted HCHO is considerably smaller than the amount that is secondarily photochemically produced [Wert et al., 2003], it is highly likely that the quantified HCHO in the HSC was formed by oxidation of emitted alkenes since these tend to efficiently produce HCHO, implying a very rapid photochemical conversion.

Acknowledgments

[41] Funding for this work was provided by Texas Environmental Research Consortium (TERC) under project H53. The authors would like to thank Craig Clements for conducting some of the GPS soundings and John Johansson for estimation of NO2/NOx ratios. We thank TCEQ for supplying wind profiler data. The Baylor Piper Aztec data were funded by TERC under project H63. We thank Levi Kauffman, Tim Compton, Grazia Zanin, Maxwell Shauck, and Martin Buhr for conducting part of the work with the Piper Aztec and Noor Gillani for flight planning.

Ancillary