Aerosol optical and hygroscopic properties during TexAQS-GoMACCS 2006 and their impact on aerosol direct radiative forcing


  • P. Massoli,

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

    1. Pacific Marine Environment Laboratory, NOAA, Seattle, Washington, USA
    Search for more papers by this author
  • P. K. Quinn,

    1. Pacific Marine Environment Laboratory, NOAA, Seattle, Washington, USA
    Search for more papers by this author
  • D. A. Lack,

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

    1. Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado, USA
    2. Chemical Sciences Division, Earth System Research Laboratory, NOAA, Boulder, Colorado, USA
    3. Now at Lockheed Martin Coherent Technologies, Louisville, Colorado, USA.
    Search for more papers by this author
  • B. M. Lerner,

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

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

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

    1. Department of Regional and Global Pollution Issues, Norwegian Institute for Air Research, Kjeller, Norway
    Search for more papers by this author
  • E. J. Williams

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


[1] In situ measurements of aerosol optical and hygroscopic properties were made on board the National Oceanic and Atmospheric Administration R/V Ronald H. Brown during the Texas Air Quality Study–Gulf of Mexico Atmospheric Composition and Climate Study (TexAQS-GoMACCS). The aerosol light extinction coefficient (σep) was measured at 355, 532, and 1064 nm at 25%, 60%, and 85% relative humidity (RH) for both sub-1- and sub-10-μm-diameter particles with a cavity ring–down aerosol extinction spectrometer. The 532-nm σep was coupled with the 532-nm light absorption coefficient (σap) measured with a photoacoustic absorption spectrometer to calculate the aerosol single scattering albedo (ω) with absolute uncertainty <0.01. The σep dependence on RH was expressed in terms of gamma (γ). The sampled aerosols covered a broad spectrum of γ and ω values; aerosols from traffic emissions were hydrophobic and highly light-absorbing with γ ∼ 0.4 and ω ∼ 0.6, whereas the regional aerosols exhibited variable values of both γ and ω. Aerosols with the highest sulfate content also had the highest γ and ω values (>0.65 and >0.9, respectively). The optical data were used to estimate local, top of atmosphere aerosol-induced climate forcing (ΔFR). The ΔFR calculations were performed using both ω values measured at 25% RH and ω values converted to ambient RH. The calculated ambient ΔFR ranged from −7 to −40 W/m2 with absolute uncertainty between 0.7 and 2.5 W/m2. The results show that including aerosol hygroscopic properties in climate calculations is critical for improving estimates of aerosol forcing on climate.

1. Introduction

[2] The Texas Air Quality Study–Gulf of Mexico Atmospheric Composition and Climate Study (TexAQS–GoMACCS) took place in summer 2006 to identify the sources, transport processes, and radiative properties of aerosols over the Gulf of Mexico and the Houston/Galveston area. The NOAA research vessel Ronald H. Brown (RHB) spent approximately 5 weeks in this region (2 August thru 11 September 2006) with a number of gas and aerosol instruments deployed to measure air quality and climate relevant parameters. The Houston metropolitan area and surrounding region are heavily industrialized and urbanized, and they often experience high mass concentrations of particulate matter with diameters <2.5 μm, or PM2.5 [Allen and Fraser, 2006] and high ozone levels [Kleinman et al., 2002]. Typical industrial activities in the area include refineries, petrochemical facilities and power plants, with emissions rich in reactive volatile organic compounds (VOCs) and NOX [Ryerson et al., 2003, and references therein]. In this environment, secondary organic aerosol (SOA) resulting from gas-to-particle conversion of VOCs are likely to constitute a substantial fraction of the aerosol organic mass [de Gouw et al., 2005; R. Bahreini et al., Organic aerosol formation in urban and industrial plumes near Houston and Dallas, Texas, submitted to Journal of Geophysical Research, 2008]. Similarly, high concentrations of sulfate aerosols are expected from the oxidation of SO2 that is emitted by diesel-fueled mobile sources and from the numerous coal-fired industries and power generation stations located both in the Houston area and elsewhere in Texas [Brock et al., 2003; Bates et al., 2008]. On the basis of 2006 emission inventories around 0.1 Tgr a−1 of SO2 are emitted in the Houston region (G. Frost, personal communication, 2008, The heavy commercial vessel traffic in the area is also a significant source of SO2 (E. J. Williams et al., Emissions of NOX, CO, SO2, H2CO and C2H4 from commercial marine vessels during TexAQS 2006, manuscript in preparation, 2009) and particulate matter including black carbon or soot [Lack et al., 2008, 2009]. The Houston/Galveston area experiences substantial air quality degradation especially in summertime, when photochemistry is at its maximum and land/sea breezes favor air stagnation in Galveston Bay and along the Texas coastline [Banta et al., 2005]. These conditions lead to frequent episodes of elevated surface ozone concentrations, with summer daytime levels often exceeding the U.S. national ambient air quality standards [Zhang et al., 2004].

[3] Accurate measurements of aerosol optical properties (light extinction coefficient (σep), light scattering coefficient (σsp) and light absorption coefficient (σap)) are necessary to calculate particle single scattering albedo (ω) and estimate direct aerosol radiative forcing on climate. Aerosol optical properties can be strongly dependent upon relative humidity (RH). Water uptake affects aerosol atmospheric lifetime and composition, which in turn affect atmospheric visibility, direct radiative forcing of climate [Intergovernmental Panel on Climate Change (IPCC), 2007] and cloud microphysics [Lohmann and Feichter, 2005]. The dependence of σep (or σsp) on relative humidity, (RH), is determined empirically by simultaneous measurements of σep (or σsp) at two different RH values, typically at a high (RH > 75%) and a low (RH < 40%) value. Many studies have reported the humidity response of both laboratory-generated and atmospheric aerosols [Doherty et al., 2005; Baynard et al., 2006; Garland et al., 2007] and empirically derived f(RH) are now incorporated into most chemical transport and radiative transfer models; however, some difficulties still exist in modeling the changes of aerosol optical responses with water uptake for species such as dust and soot, and in generalizing the treatment of complex organic aerosols [Malm and Kreidenweis, 1997; Kanakidou et al., 2005; Mircea et al., 2005]. Furthermore, most satellite algorithms retrievals still widely omit hygroscopic properties of ambient aerosols [Bates et al., 2006; Wang and Martin, 2007]. Recent work has provided empirically based parameterizations to generalize and simplify the treatment of the RH dependence of σep (or σsp) as function of aerosol size, composition and mixing state [Malm et al., 2005; Quinn et al., 2005].

[4] This paper reports the in situ optical and hygroscopic aerosol properties measured onboard RHB using a three-wavelength cavity ring-down aerosol extinction spectrometer, CRD [Baynard et al., 2007] and a photoacoustic aerosol absorption spectrometer, PAS [Lack et al., 2006]. The CRD was configured to provide 532 nm σep measurements at 25%, 65% and 85% RH, and 355 and 1064 nm σep measurements at 25% and 85% RH. The PAS measured the σap coefficient at 532 nm and 25% RH, which was used with measurements from the CRD to obtain ω. A detailed description of the optical measurements is given in section 2. The measured and derived aerosol optical properties are evaluated in the context of the aerosol chemical composition and meteorology to better understand the factors affecting the optical properties (section 3). We find that both the specific meteorological conditions of the area and episodes of long-range transport can play a decisive role in impacting the region air quality. Finally, in section 4 we use the measured optical parameters in simple radiative forcing calculations to estimate the aerosol-induced forcing (ΔFR) at the top of atmosphere (TOA). We quantify the uncertainty in forcing on the basis of current measurement uncertainties of the experimental ω obtained from the CRD and PAS, and demonstrate the sensitivity of radiative forcing to the uncertainty in the ω measurements. By performing these calculations using both the measured ω values (25% RH) and ω values adjusted to ambient RH, we show the sensitivity of forcing to changes in relative humidity.

2. Instruments and Methods

2.1. Aerosol Sampling

[5] Air was sampled 18 m above sea level through a 6-m-long mast drawing 1 m3/min of air into the laboratory container. The mast was positioned forward of the ship's stack and automatically pointed into the relative wind to maximize sampling efficiency and avoid RHB's exhaust. The mast was heated in its lowest 1.5 m to reduce the RH to ∼60%. Further details about the sampling mast are given by Bates et al. [2002]. At the bottom of the mast, several 1.6 cm inner diameter conductive tubes were used to supply 30 L/min flow to the instruments used for aerosol optical, chemical and physical measurements. Upstream of the CRD, an automated, computer controlled valve switched the sample flow between two Berner-type impactors [Berner et al., 1979] to select either sub-10- or sub-1-μm-diameter particles (aerodynamic 50% cutoff diameter). Flexible conductive silicone tubing (TSI Inc.) was used between the impactor and CRD to avoid sharp bends. On the basis of previous laboratory studies with the same setup, losses through the inlet feeding the CRD were insignificant for sub-1-μm-diameter particles, and <2% for 1- to 5-μm-diameter particles. In this paper, we report only the measurements of sub-1-μm aerosols.

2.2. Aerosol Optical Instrumentation

[6] The CRD was equipped with six channels for measurements of σep at different RH values. The σep at 532 nm was measured at 25%, 60% and 85% RH (three separated CRD cavities with independent flow and RH control). The 355 and 1064 nm CRD cavities were connected in series, with the upstream aerosol flow alternating between 25% and 85% RH. The 85% RH was reached by natural cooling of the sample stream when moving from the heated mast (∼32°C, 60% RH) to the CRD cavities insulated at the laboratory temperature (∼25°C). The midpoint RH (60%) was obtained by heating a second 532-nm CRD cell to ∼32°C. For the measurements at 25% RH, the flow was dried by passing the sample air through Nafion diffusion driers (Permapure Inc.). These RH values were stable through the experiment (standard deviation of ±3% RH). Temperature was monitored in several critical points of the CRD, and the RH values were measured either immediately downstream or inside each CRD cavity. Humidity in the 85% and 25% RH cavities was measured using high-precision Rotronics probes (Rotronics Inc.) having an accuracy of ±0.5°C and ±1% RH. RH in the other cavities was measured by Vaisala probes (Vaisala Inc.) with accuracy ±2% RH. The probes were checked after the field mission, and minimal drift was found from the previous calibration. RH was also calculated in each CRD cavity using temperature measured with thermistors (Omega Inc.).

[7] A fourth CRD cavity measured particle free air to determine the interference from gas phase species at 532 nm, mainly NO2 [Baynard et al., 2007]. Gas phase interference contributed, on average, 4% to σep and 15% to σap, and was subtracted from total aerosol extinction and total absorption. For 1-min time resolution, the uncertainty in the submicrometer σep at 532 nm and 25% RH was δσep = 1%. For measurements at elevated RH levels, the δσep was generally higher mostly owing to the uncertainty in the measured RH values. With an average accuracy of ±1% RH, δσep was approximately 5% and 2% for measurements at 85% and 60% RH, respectively.

[8] We express the RH dependence of σep at 532 nm,

equation image

where in our study RHref is 25% RH. The γ parameter in our case was obtained by combining the three σep measurements at 85%, 65% and 25% RH. The use of γ has the advantage of describing the hygroscopic behavior of aerosols in a linear manner over a broad range of RH values; it also implies that particles are deliquesced [Quinn et al., 2005], a reasonable assumption for this data set due to the high ambient relative humidity during the field study. The γ parameter is dimensionless, and it increases with increasing particle water uptake. From previous studies, typical values of γ for ambient aerosol ranged between 0.1 and 1.5 [Gassó et al., 2000; Quinn et al., 2005; Clarke et al., 2007]. The absolute uncertainty in the γ values reported in this work, δγ, is ∼0.03 for submicron sizes and 1 min or longer averaging times.

[9] The values of 355- and 1064-nm submicrometer σep at 25% RH were used to calculate the Ångström exponent of light extinction (Åσep) according to [Ångström, 1929]

equation image

where λ1 = 355 nm and λ2 = 1064 nm. The absolute uncertainty in Åσep is 0.02 based on δσep values of 1.5% for the 355 and 1064 nm σep measured at 25% RH and for sub-1-μm sizes, signal levels >5 Mm−1, and time resolution of 1 min or higher. We also report Åσep for ambient RH conditions (Åσep_ambient), obtained using the 355- and 1064-nm σep data converted to ambient RH by using the ep,355(RH) and the ep,1064(RH) as scaling factors, respectively. The absolute uncertainty in Åσep_ambient is 0.08 based on a 5% δσep value for the 355- and 1064-nm σep coefficients converted to ambient RH.

[10] The σap was inferred by PAS on the basis of aerosol light absorption technique (PAS), which provides a measure of the actual heat released by particles from the absorption of power modulated light. The heat release creates a pressure wave that resonates in a frequency modulated cavity and it can be detected as a sound wave; the acoustic response is in turn converted into particle light absorption response. Further details of the PAS instrument and principles are described by Lack et al. [2006]. The PAS was connected downstream of the CRD and measured the sub-1-μm σap at 532 nm and 25% RH with an estimated uncertainty of 5%. Combination of the 532-nm σep and σap at 25% RH provided a direct measure of ω as

equation image

The absolute uncertainty for the ω values measured at 25% RH (δω) is estimated by propagating the uncertainty of the independent values δσep = 1% and δσap = 5%. The δω values varied with ω from 0.01 (for ω = 0.8) to 0.03 (for ω = 0.4). Values of ω are also reported for ambient RH conditions (ωambient), using σep converted to ambient RH and assuming σap is not a function of RH [Nessler et al., 2005]. The uncertainty in the ambient ω values (δωambient) is based on δσep = 5% and δσap = 5%, and is δωambient = 0.015 (for ω = 0.8), and 0.042 (for ω = 0.4). All parameters are reported at 298 K and 1013.25 hPa.

2.3. Aerosol and Gas-Phase Chemistry Measurements

[11] Nonrefractory (NR) aerosol chemical composition was measured onboard RHB with a quadrupole aerosol mass spectrometer (Q-AMS, Aerodyne Research Inc., Billerica, Massachusetts) [Jayne et al., 2000]. The Q-AMS provides concentrations in μg m−3 of NR NH4+, SO42−, NO3 and POM (particulate organic matter). Details of the Q-AMS performance and operation during TeXAQS-GoMACCS are reported by Bates et al. [2008]. The SO42− and POM concentrations were used to calculate the parameter FPOM, as

equation image

where the sum of POM and SO42− is assumed to represent the majority of the submicrometer aerosol mass [Quinn et al., 2005]. The hydrocarbon-like (HOA) and the oxygenated (OOA) organic aerosol fractions based on the m/z of 57 and 44, respectively [Zhang et al., 2005] were used to calculate FOOA,

equation image

where the reconstructed sum of OOA and HOA concentrations is approximately equal to NR POM [Bates et al., 2008].

[12] Tropospheric ozone mixing ratios (O3, ppb) were measured with two commercial TECO 49 instruments, calibrated with a NIST traceable standard prior to the field experiment. The O3 detection limit was 2 ppbv with a 10-s time resolution. Carbon monoxide (CO, ppbv) was measured via vacuum UV fluorescence with a commercial CO analyzer (AL 5002, AeroLaser GmbH, Germany). The CO detection limit was 1.5 ppbv for 1-Hz data averaged to 1-min time resolution. Sulfur dioxide (SO2, ppbv) was measured with a pulsed UV fluorescence analyzer (Thermo Environmental Instruments1, TEI, Model 43C) with a detection limit of 100 pptv with a 1-min time resolution. Measured concentrations of SO2 and NR SO42− were used to calculate the fraction of sulfur as sulfate, FSO4,

equation image

FSO4 is an indicator of the level of aerosol “relative age” or oxidation in absence of significant fresher emissions [Quinn et al., 2005], with higher FSO4 indicating older aerosols due to oxidation of SO2 into SO42−. The use of FSO4 as an indicator of aerosol relative age in the TeXAQS-GoMACCS study region was limited by the existence of both onshore and offshore sources of SO2 and SO42−. However, FSO4 remains useful as a qualitative indicator of the aerosol oxidation level. High FSO4 values unambiguously indicate that the aerosol is oxidized and there were no recent inputs of SO2. A low FSO4 value is more ambiguous, because it might indicate either recent addition of SO2 to an aged aerosol or recent emission of SO2 that has not yet oxidized to SO42−. We also use measurements of NR NH4+, NO3 and SO42− to estimate the degree of aerosol neutralization (or acidification) defined as the equivalence ratio (ER),

equation image

Aerosol optical depth (AOD) was measured using three five-channel handheld Microtops Sun photometers (Solar Light Co., units SN 4080, 3803 and 5355) at wavelengths of 340, 380, 440, 500, 675, and 870 nm [Quinn et al., 2002]. The Sun photometers were calibrated prior to the field deployment. Data reduction followed the protocol of Knobelspiesse et al. [2003]. Ozone column amounts used to calculate the ozone optical depth were obtained from daily ozone sondes launched from the ship. AOD was measured on RHB twice a day between 1000 and 1400 local standard time during periods of clear sky. The absolute uncertainty for AOD values is 0.015.

2.4. FLEXPART and Mixing Height Estimates

[13] The Lagrangian dispersion model for particle transport and diffusion FLEXPART version 6.2 [Stohl et al., 1998, 2005] provides air mass transport pathways and information regarding how anthropogenic emissions of CO, NOX and SO2 are likely to impact such air masses. FLEXPART backward transport simulations were calculated for each hour along the RHB cruise track or whenever the ship changed position by more than 0.1 degrees in either latitude or longitude. Each simulation results from 40,000 particles being released in a certain volume of sampled air and transported back in time. The spatial distribution of the air mass in the day(s) prior to sampling, expressed as its sensitivity to emission input, is given with 1-day resolution [Seibert and Frank, 2004]. FLEXPART uses meteorological input data from the European Centre for Medium Range Weather Forecast (ECMWF) provided with standard resolution of 0.36° × 0.36°. Current emission inventories are used to estimate the contribution of anthropogenic sources emitted into the modeled air parcels. The FLEXPART product used here is the footprint emission sensitivity (ns kg−1) which is a measure of the residence time of the particles in the lowest 100 m of the vertical column mapped at a resolution of 0.1° × 0.1°. Additional information on the FLEXPART model and other output products are described at

[14] Higher-resolution FLEXPART backward simulations were also made available by using the meteorological field outputs from the Weather Research and Forecasting (WRF) model [Doran et al., 2008] with a horizontal resolution of 5 km and 60 vertical levels. In this case, the simulations are based on 20,000 particles released every 15 min from boxes of 20 by 20 km in horizontal size and 100 m in vertical size, and centered along the RHB pathway. The FLEXPART output grid had a resolution of 5 km, and included a turbulent parameterization based on the turbulent kinetic energy from WRF.

[15] NOAA's high-resolution Doppler lidar (HDRL) provided detailed information about three-dimensional wind speed and direction as well as turbulence profiles from which mixing heights could be estimated following the procedure described by Tucker et al. [2009].

2.5. Meteorological Conditions During TexAQS–GoMACCS

[16] The RHB cruise tracks in the Gulf of Mexico and the various locations sampled during the study are shown in Figure 1a. Figure 1b depicts the ship track in Galveston Bay and along the industrial ship channel, and the location of Barbours Cut, a container port at the south entrance of the ship channel where the RHB spent ∼200 h during the study.

Figure 1.

(a) Ronald H. Brown cruise track in the Gulf of Mexico during the period 2 August to 11 September 2006. (b) Enlarged map of Galveston Bay.

[17] Southerly winds with either westerly or easterly components (ESE-WSW) were the prevailing synoptic meteorological condition until late August. Under these conditions, the submicrometer aerosol was largely from local sources with variable aerosol amounts and composition depending on wind direction and sampling location. Occasionally, the Galveston Bay meteorology was dominated by the local land/sea breeze circulation typical of this area in summer. From late August into mid September, air masses came predominantly from the continental United States owing to stronger northerly winds often with an easterly component (N-NE). The aerosols in these air masses were mixtures from local urban/industrial sources; rural sources in Texas, Arkansas and Louisiana; and distant sources in the northeast United States.

[18] The marine boundary layer height was highly dependent on location and time; it extended to 500–600 m without a diurnal cycle offshore in the Gulf of Mexico, whereas inland it ranged from 200 m during the night to 800 to 1200 m during the day. Bates et al. [2008] demonstrated that the daily variation in mixing height had a primary role in shaping the composition of the surface aerosol in the Galveston Bay area, to the point that daily trends observed in the aerosol composition could be largely explained by vertical mixing of aerosol precursors and particles emitted and mixed at different times of the day.

3. Results: Aerosol Optical Properties and Source Apportionment

[19] The aerosols sampled on RHB during TexAQS were extremely diverse owing to the large number of nearby urban and industrial sources of particles and particle precursors, many different regional sources, long-range transport, and the complex meteorological patterns associated with the coastal geography of the Houston area. Bates et al. [2008] and D. Covert et al. (Aerosol optical properties in the Texas Gulf coast region during TexAQS-GoMACCS 2006, manuscript in preparation, 2008) describe the chemical and optical properties of the TexAQS-GoMACCS aerosols sampled within three broad air mass categories identified as southerly flow (ESE to WSW) Gulf, southerly flow inland, and northerly flow (WNW to ENE). In the analysis presented here, we focus on portions of the data that are associated with identifiable sources and with specific aerosol characteristics (Table 1). For each of these case studies we report the optical and chemical properties of the aerosol (γ, ω, the organic fraction FPOM and the oxygenated organic fraction FOOA), and gas phase parameters used for source apportionment (Table 2).

Table 1. Meteorological Conditions, Ship Position, and Upwind Source Region for the Case Studies Described in Section 3
 Time Period (Day, UTC)Wind DirectionAmbient RH, %Mixing HeightRHB LocationAir Mass Source
Traffic emissions3,14,15 August 1100–1400W (250°)80200 mBarbours CutLocal Highway
Galveston Bay, 114 August 1600–2400W-E-SW (250°–100°–200°)70250 m (1500 UTC) 1300 m (1800 UTC)Barbours CutUrban/Industrial
Galveston Bay, 215 August 1600–2400W-E-SW (250°–100°–200°)70250 m (1500 UTC) 1300 m (1800 UTC)Galveston BayUrban/Industrial
Galveston Bay, 316–17 August 1500–0130W-E-SW (300°–100°–200°)70400 m (1500 UTC) 800 m (1800 UTC)Galveston BayUrban/Industrial
Galveston Bay, 4a17 August 1500–2400N-SE (0°–150°)65400 m (1500 UTC) 800 m (1800 UTC)Galveston BayUrban/Industrial
Galveston Bay, 4b18 August 0000–1300N-E (0°–100°)75500–700 from 0000 until 1300 UTCGulf, Galveston IslandLocal SO2 Source
Long-range transport2 September, 2100 5 September, 0830N-NE (20°–100°)70∼600 mGulfOhio River Valley
Industrial air massesseveral days (200 h total)NE (40°–100°)70VariableBarbours CutIndustrial
Table 2. Average Values and 1σ Standard Deviations of γ, ω, FPOM, FOOA, FSO4, ER, Åσep, and Åσep_ambient During the Case Studiesa
 Traffic EmissionsGalveston Bay, 1Galveston Bay, 2Galveston Bay, 3Galveston Bay, 4aGalveston Bay, 4bORVIndustrial
  • a

    The values of CO and O3 (ppb) are the peak values reached during those events.

CO (ppbv)300120120140200150200150–200
O3 (ppbv)<10406080115959050–100
γ0.39 ± 0.080.46 ± 0.140.57 ± 0.100.60 ± 0.050.56 ± 0.040.73 ± 0.060.66 ± 0.070.55 ± 0.08
ω0.60 ± 0.120.56 ± 0.170.80 ± 0.110.81 ± 0.070.89 ± 0.040.94 ± 0.020.93 ± 0.180.87 ± 0.10
FPOM0.59 ± 0.090.40 ± 0.060.42 ± 0.160.44 ± 0.070.54 ± 0.040.34 ± 0.050.37 ± 0.060.57 ± 0.14
FOOA0.28 ± 0.010.52 ± 0.120.65 ± 0.200.76 ± 0.130.84 ± 0.070.92 ± 0.050.90 ± 0.040.70 ± 0.18
Åσep1.6 ± 0.21.65 ± 0.21.82 ± 0.22.05 ± 0.132.22 ± 0.122.40 ± 0.182.0 ± 0.162.05 ± 0.25
Åσep_ambientn/a1.76 ± 0.21.87 ± 0.141.98 ± 0.261.90 ± 0.152.0 ± 0.171.80 ± 0.151.99 ± 0.17
FSO40.23 ± 0.300.36 ± 0.240.34 ± 0.250.32 ± 0.140.36 ± 0.150.53 ± 0.180.64 ± 0.180.30 ± 0.20
ER0.95 ± 0.190.68 ± 0.200.71 ± 0.180.69 ± 0.170.75 ± 0.110.41 ± 0.100.53 ± 0.080.90 ± 0.24

3.1. Urban Traffic Emissions

[20] RHB was often stationed overnight in Barbours Cut, south of the industrial ship channel (Figure 1b). A highway that crosses the ship channel approximately 1.6 km west of Barbours Cut was directly upwind of RHB on 13–15 August during the early morning hours, when the boundary layer was shallow and fresh emissions were confined to near the surface. Figure 2a shows time series of wind direction (degrees), 25% RH submicron σep coefficient (Mm−1), CO (ppbv) and NO2 (ppbv) for this period. On 14 and 15 August, the CO, NO2 and σep levels began to increase around 1100 UTC (0600 local time) under westerly winds (250°), peaked between ∼1200 and 1300 UTC (∼0700–0800 local time) and decreased to previous levels by 1400 UTC (0900 local time). Such evolution is consistent with traffic emissions from the upwind highway. The lack of similar CO and NOx plumes on a Sunday (13 August) supports this interpretation. The optical (γ, ω), physical (Åσep) and chemical (FPOM, FOOA) properties of the aerosols associated with the traffic emissions averaged over three episodes (14, 15 August and 3 August, not shown in the time series) are depicted in Figure 2b. The aerosol sampled during morning rush hour shows strong hydrophobic and absorbing properties, with γ < 0.6 and ω < 0.8. The lowest γ and ω values (0.3 and 0.5, respectively) occur at the peak of the rush-hour plume between ∼1200 and ∼1300 UTC. The Åσep values decrease from 1.8 at the beginning of rush hour to 1.3–1.4 at the peak, and then rise again to 1.7 by 1400 UTC. Such a trend in the Åσep values suggests a shift in the particle size toward larger diameters (not due to humidity effect) during the period of maximum traffic impact. Simultaneously measured number size distribution data, not shown here, indicate a significant increase in the number of particle in the 150- to 300-nm-diameter size range during the traffic peak, while sub-100-nm particles were continuously elevated during this period. These results are consistent with previous studies that report accumulation mode soot components (i.e., absorbing and hydrophobic) associated with vehicle traffic [Imhof et al., 2006, and references therein]. Particle composition was dominated by POM, with FPOM increasing from ∼0.45 (1100 UTC) to 0.7 (1200–1300 UTC) and decreasing back to 0.5 by 1400 UTC. The FOOA values were ∼0.3 and did not show significant variations, suggesting that the POM during the rush hours was mostly in the form of hydrocarbon-like organics (HOA) as expected for fresh and relatively unprocessed vehicular sources.

Figure 2.

(a) Time series of CO, NO2, 25% RH σep and wind direction from 13–15 August when RHB was in Barbours Cut. Boxed regions indicate traffic emissions sampled during morning rush hours; 13 August was a Sunday with much lighter traffic and it is not included in the analysis. (b) Time series of the average ω, γ, Åσep, FPOM, and FOOA values during 3, 14, and 15 August.

3.2. Air Masses in Galveston Bay

[21] RHB spent a significant amount of time transiting Galveston Bay between Barbours Cut and the Gulf of Mexico to sample air masses transported downwind of Houston and the surrounding industrial centers. On five consecutive days spent in this area (14–18 August), the local meteorology was mostly dominated by the typical offshore land breeze in the morning (northerly to easterly winds) and onshore sea breeze in the afternoon (southeasterly). Previous studies have demonstrated that this recirculation pattern can significantly increase pollutant concentrations in the Houston/Galveston Bay area [Banta et al., 2005]. Figure 3a shows the time series of wind direction, 25% RH submicron σep coefficient, CO and O3 between 14 August 0600 UTC and 18 August 1400 UTC. The wind direction varied daily from westerly (300°) at 1500 UTC, to easterly (100°) at 1800 UTC and back to southwesterly (200°) at approximately 2100 UTC. The Doppler lidar data (not shown) indicate that the offshore winds were light on 14 and 15 August and stronger on 16 and 17 August. Afternoon return flow was light on all four days. On the night of 17 August, winds were particularly light (<5 m/s), contributing to stagnation at the surface. Dates 15, 16 and 17 August were also characterized by very weak nighttime winds above the shallow nocturnal boundary layer, a condition that allows pollutants and precursors to persist aloft until the next day when they can be mixed down with fresh nighttime−early morning emissions.

Figure 3.

(a) Time series of CO, O3, 25% RH σep, and wind direction from 14–18 August when air masses were sampled in Galveston Bay. Boxed regions are discussed in the text. (b) Time series of γ, ω, Åσep, FPOM, and FOOA for these events.

[22] Ozone levels increased each day between ∼1500 and 2400 UTC (1000–1900 local time) as a result of daily photochemistry [e.g., Trainer et al., 2000]. Afternoon O3 mixing ratios increased gradually over the five day period from 40 ppbv on 14 August to 115 ppbv on 17 August. After 1600 UTC on 17 August ozone levels were sustained above 75 ppbv for ∼11 h. Carbon monoxide levels remained between 100 and 150 ppbv for the first three days, and were higher (200 ppbv) and more variable on 18 August. The submicron σep levels associated with the periods of elevated ozone (hereafter periods 1, 2, 3 and 4 as highlighted in Figure 3a) were between 25 and 40 Mm−1 in the first three periods, but were substantially higher (45 to 65 Mm−1) during period 4. The properties of the aerosols during these events are shown in Figure 3b. On days with a weak land breeze and relatively low ozone (periods 1 and 2), the optical properties of the aerosols were variable but typical of a relatively unoxidized air mass; aerosols were not hygroscopic (γ = 0.46 and 0.57, respectively) and were absorbing (ω = 0.56 and 0.8, respectively). POM and sulfate contributed roughly equally to the submicron particle mass (FPOM ∼ 0.4). During period 1 the fraction of organic mass that was oxygenated (FOOA) was ∼0.5, while during period 2 FOOA was ∼0.65. During periods 3 and 4 (afternoons on 16 and 17 August, respectively), the aerosol was more hygroscopic and the organics were more oxygenated, with γ > 0.55, ω > 0.80, and FOOA > 0.8.

[23] Around 0000 UTC on 18 August, while the ship transited the southern end of Galveston Bay toward the Gulf, a sudden and substantial change occurred in aerosol properties: FPOM suddenly decreased from 0.54 to 0.34 and γ increased from 0.56 to 0.73. FOOA and ω gradually increased toward higher values as well. These changes were likely due to a shift in the sampled air mass. Around 2345 UTC on 17 August, preceding the substantial changes observed in the aerosol properties, the wind direction shifted by 360° counterclockwise; Doppler lidar observations show a rapid increase in the mixing height (from ∼500 m to ∼1500 m), and deep turbulent mixing around the same time. The high-resolution WRF-FLEXPART footprint emission sensitivity shows air mass transport from the west edge of Galveston Bay coming to RHB around 0000 UTC on 18 August (Figures 4a and 4b), and then wrapping around Galveston Bay afterward (Figure 4c). After this time, both HRDL and FLEXPART show greater transport from the east side of Galveston bay rather than the west edge.

Figure 4.

(a, b, c) Footprint emission sensitivities from the WRF-FLEXPART transport model initialized between 2330 and 0015 UTC on 18 August. RHB position is indicated by the black circles. (d) Time series of SO2, CO, CO2, and NR sulfate for the time period immediately following the wind shift on 18 August at 0000 UTC.

[24] A broad increase in SO2 (up to 10 ppbv) was observed after 0000 UTC with two more concentrated plumes reaching mixing ratios of 25 ppbv at 0130 and 0300 UTC (Figure 4d). During this same period, sulfate aerosol concentrations increased to ∼10 μg m−3, with a peak at 0300 UTC of ∼17 μg m−3. The CO and CO2 traces correlate well with sulfate, but are less structured than the SO2 trace. These data indicate a shift from the oxidized urban/industrial plume of Houston to a sulfate-rich plume associated with a significant SO2 source. Several sources of SO2 in the area surrounding Galveston Bay and Houston, and mixtures of these sources, might account for the observed sulfate and SO2 concentrations. Among these, the W. A. Parish power plant on the southwest edge of the Houston area emits substantial amounts of SO2 [Brock et al., 2003]. Marine vessels have also been shown to be a significant source of SO2 and sulfate aerosols in the Gulf of Mexico under southeasterly flow [Bates et al., 2008], and might have contributed to the high load of sulfate observed on 18 August later in the day under southeasterly winds.

3.3. Long-Range Transport of Aerosols From the Ohio River Valley

[25] On 2 September at approximately 2100 UTC the submicron extinction coefficient increased abruptly to 90 Mm−1 and the concentration of SO42− doubled from 5 to 10 μg m−3 while POM remained near 5 μg m−3 (Figure 5a). These conditions of high aerosol loads persisted for 59 h until 0800 UTC on 5 September over which period RHB sampled in an area extending 50 miles along the coast and up to 10 miles out in the Gulf. Aerosol optical depth exceeded 0.3 during this period, and subjectively, visibility was diminished. The observed aerosol properties along with the duration and extent of the event suggest that it was probably associated with long-range transport from a regional-scale source [Quinn and Bates, 2003]. Nearly invariant mixing ratios of both CO (160 ppb) and O3 (65–70 ppb) measured on RHB suggest that contributions from local emissions were minor, except for a 3-h period on 3 September around 1200 UTC (not included in the analysis) when local sources were sampled.

Figure 5.

(a) Time series of CO, O3, 25% RH σep, wind direction, and sulphate during long-range transport event from the Ohio River Valley. The dashed line shows the beginning of the event. (b) Time series of γ, ω, Åσep, FPOM, and FOOA for this event.

[26] The aerosol associated with this event (Figure 5b) was highly scattering (ω = 0.93) and moderately hygroscopic (γ = 0.66). Sulfate aerosols dominated the sub-1-μm mass (FPOM = 0.37) while the organic mass that was present was highly oxygenated (FOOA = 0.9). The average Åσep and Åσep_ambient values were 2.0 and 1.8, respectively.

[27] The FLEXPART transport model indicates that the air sampled by RHB throughout this period was transported from the northeast and had been over the Ohio River Valley (ORV) 3 to 4 days earlier (Figure 6). Clearly, the air might also have been exposed to further emissions during transport from the ORV region to Texas. While elevated SO2 concentrations (and sulfate levels) are typical of coastal/inland Texas where coal fired power plants are widespread [Bates et al., 2008], such local emissions would not have the time to fully oxidize given the relatively short transport time to RHB (∼1 day), and are inconsistent with the observed spatial and temporal invariance of the aerosol properties. The dominance of sulfate particles during this event is therefore consistent with transport from the extremely large sources of SO2 in the ORV region, which total ∼12 × 106 kg/d in the summer season [Brock et al., 2008].

Figure 6.

FLEXPART footprint emission sensitivities initialized at the location of the RHB at the (left) beginning and (right) end of the Ohio River Valley transport event. Numbers indicate the number of days of transport for the centroid location of the modeled particles.

[28] To summarize the results presented so far and emphasize the differences in the optical and chemical properties among the selected cases, the data are presented as histograms of the frequency of occurrence of γ, ω, FPOM and FOOA (Figure 7). As pointed out earlier, the fresh traffic emissions have the lowest values of γ, ω and FOOA. Aerosols from more aged Houston urban/industrial sources sampled in Galveston bay on 14, 15, 16 and 17 August had progressively larger values of γ and ω. Such changes were correlated with increase in FOOA (larger OOA fraction), whereas FPOM values were similar among these air masses (∼0.5). Finally, the sulfate-dominated aerosols exhibited a high γ (>0.65), highly scattering character (ω > 0.9), high sulfate fraction (FPOM < 0.4), and higher degree of oxidation of the organic components (FOOA > 0.8) compared with most of the locally influenced aerosols.

Figure 7.

Histograms of the percentage frequency of occurrence of 5-min averaged values of γ, ω, FPOM, and FOOA for the aerosol types discussed in the current study.

4. Discussion

4.1. Dependence of Aerosol Hygroscopic Behavior on Chemical Composition

[29] The relationship between fraction of organic material (FPOM) and aerosol hygroscopic behavior (γ) for the cases discussed so far is depicted in Figure 8. The TexAQS-GoMACCS data covered a broad range of both γ and FPOM values: some of the air masses covered larger ranges in the POM fraction than in the γ values (such as the aerosols sampled in Galveston Bay on 17 and 18 August). Moderately hygroscopic aerosols (γ ∼ 0.5–0.6, such as those observed in some of the Galveston Bay air masses) had a considerable amount of nonrefractory organic material, between 40 and 60%. Even for the most hygroscopic aerosols, the fraction of organics relative to organics and sulfate was at least 30%.

Figure 8.

The γ as a function of FPOM for each discussed aerosol type. The gray circles represent the entire TexAQS-GoMACCS data set. The blue line is a linear fit to the data, while the red dashed lines are the 95% confidence levels for the fit and the black dotted lines are the 95% prediction bands. The solid lines represent the theoretical boundaries for the γ-FPOM relationship predicted using Mie theory.

[30] As discussed earlier, the lowest FPOM and highest γ values were measured in Galveston Bay on 18 August, as well as during the episode of long-range transport from the ORV. However, even higher γ values may be found in the TexAQS-GoMACCS data set (gray circles). In general, the points on the top left of Figure 8, where γ > 0.8 and FPOM < 0.25, are associated with samples taken in the Gulf of Mexico in relatively clean marine air. In contrast, values of γ < 0.4 are mostly marine vessel emissions, which have been discussed elsewhere [Lack et al., 2009]. Finally, we report data from when RHB sampled on 10 different days in Barbours Cut under northeasterly flow, i.e., downwind of diverse sources in the industrial ship channel and surroundings. These industrial-dominated aerosols had the largest variability in FPOM values (between 0.4 and 0.9), and generally low γ, between 0.3 and 0.6.

[31] The linear least squares regression to the TexAQS-GoMACCS data set (coefficients ± 1σ standard deviation) yields

equation image

This result is very similar to that obtained from the ACE-ASIA and ICARTT data sets [Quinn et al., 2005] with γ calculated using light scattering coefficients measured with a TSI nephelometer

equation image

The TexAQS-GoMACCS data set is also well contained with very few outliers in the theoretical γ -FPOM parameterization boundaries (Figure 8, solid black lines) derived using both Mie theory-based sensitivity calculations [Quinn et al., 2005] and results from ep(RH) laboratory studies [Baynard et al., 2006].

[32] To further explore the relationship between the chemical composition of the aerosols and the γ - FPOM parameterization, the data are presented as a function of the degree of SO2 oxidation to form sulfate, FSO4 (Figure 9 and Table 2) and the degree of apparent acidity, or equivalence ratio ER (Figure 10 and Table 2). Relatively lower FSO4 (indicating a smaller fractional conversion of SO2 into sulfate) and higher ER (indicating more neutralization of the sulfate by ammonium) were generally associated with higher FPOM and lower γ. Most of the air masses sampled in Galveston bay had relatively low FSO4 values suggesting the effects of recent emissions of SO2. These local emissions may mask the presence of older, more oxidized aerosols from sources further upwind. The highest values of FSO4 were present during the long-range transport of ORV pollutants, and during sampling of SO2-rich but chemically processed air measured on 18 August (section 3.2). Some interesting findings emerge from the combined information of degree of relative sulfur oxidation and degree of neutralization by ammonium. For example, the sulfate aerosols sampled on 18 August after 0000 UTC were more acidic (ER = 0.4) than the ORV aerosols (ER = 0.53), the latter perhaps reflecting the influence of continental sources of NH3 during transport.

Figure 9.

As in Figure 8 but color coded by the fraction of sulfate oxidation FSO4.

Figure 10.

As in Figure 8 but color coded by degree of apparent acidity or equivalence ratio, ER.

[33] Overall, what emerges from this analysis is a consistent picture of the way the optical properties vary with the chemistry of the aerosols. The variations in the relative amount of POM and sulfate mass fraction for typical ambient aerosol mixtures can explain most of the observed variability in γ. This result suggests that the γ-FPOM relationship is robust and broadly applicable in a systematic way to parameterize the bulk composition dependence of the aerosol hygroscopic behavior of both local and regional aerosols. It is also worth investigating the extent to which the chemical composition of our specific cases agrees with the chemical properties of the aerosols in the air mass categories discussed by Bates et al. [2008]. For instance, we find that the results obtained for the traffic emissions, with hydrophobic, absorbing, HOA-dominated aerosols and CO enhancement during the morning hours (Figure 2) are consistent with the morning composition as measured in inland locations under southerly flow, or “Category 2” of Bates et al. [2008]. The evolution of O3, FPOM and FOOA for the air masses sampled in Galveston Bay between 14 and 17 August (Figure 3) is also consistent with the diurnal variation of the Category 2 aerosols [Bates et al., 2008, Figure 11 (bottom)], which is partly explained by variation in mixing height during the day. Finally, the long-range transport event from sulfur-rich ORV sources occurring under northerly synoptic flow was an extraordinary case of regional aerosol fumigation. Indeed, the aerosol commonly sampled under northerly flow during TexAQS-GoMACCS, i.e., “Category 3” of Bates et al. [2008], was mainly composed of OOA-dominated organics rather than sulfate.

4.2. Estimates of Direct Radiative Forcing by Aerosols and Uncertainty Analysis

[34] Uncertainties in aerosol optical properties and the difficulties in resolving the spatial and temporal variability of such properties continue to limit precise quantification of the direct radiative forcing (DRF) by aerosols [IPCC, 2007]. Estimates of anthropogenic aerosol DRF based on experimental data are desirable to both evaluate the overall DRF uncertainty due to current aerosol property measurements and provide valuable information for radiative transfer models. The aerosol optical measurements collected during TexAQS-GoMACCS are used to calculate the DRF for the aerosol types discussed in the present study. We calculate the change in upwelling radiation at the top of atmosphere (TOA) as an estimate of the instantaneous forcing (ΔFR) locally induced by the measured aerosols. Although this approach is much simplified compared to full radiative transfer models, the parameterized method is useful to examine the sensitivity of DRF to the variability and uncertainty in the measured aerosol optical properties.

[35] ΔFR was calculated using the formula adapted from Haywood and Shine [1995] and Chylek and Wong [1995] for partially absorbing aerosols,

equation image

where the particle single scattering albedo (ω), the aerosol optical depth (AOD), and the average up-scatter fraction (β) are the measured aerosol optical parameters. The up-scatter fraction β, the light scattered back in the atmosphere, depends on the hemispheric backscatter fraction b, or the ratio of angular corrected backscatter coefficient to the total scattering coefficient. On RHB, b was measured using a TSI nephelometer at 550 nm and 60% RH [Quinn et al., 2004]. Following Anderson et al. [1999], we compute β using a simple empirical relationship based on the Henyey-Greenstein phase function [Wiscombe and Grams, 1976],

equation image

We assume a fractional day length (D) of 0.5, a solar constant (So) of 1366 W/m2 [Dewitte et al., 2004], a transmittance of the atmosphere above the aerosol layer (T) of 0.76 and an albedo of the underlying ocean surface (R) of 0.05 [Russell et al., 2002]. R values calculated by taking into account changes in solar zenith angles (SZA) and wind speed following Jin et al. [2004] showed little variation from an R of 0.05. The calculations were performed for clear-sky conditions, or Ac = 0.

[36] Given the recognized importance of accounting for aerosol water uptake in radiative forcing calculations [Bates et al., 2006], ΔFR was calculated using both ω and ωambientFR and ΔFR_ambient, respectively) to describe the sensitivity of aerosol DRF due to changes in ω values. The other optical parameters used in the forcing calculation such as β change only slightly with respect to RH [Kotchenruther et al., 1999]. Note that all the ω values were determined from measurements made at 532 nm, whereas the other optical parameters (b and AOD) are at 550 nm; this difference in λ is small and again it can be neglected.

[37] Table 3 lists the average values and absolute uncertainty of the parameters used to derive ΔFR (equation (10)). The calculated ωambient values are up to 0.1 units higher than the measured ω, with the largest changes occurring for the aerosols with ω < 0.7, consistent with Bates et al. [2006]. Such variations are significant because they are larger than the uncertainties in the ω values measured in this study, δω < 1%, for ω between 0.98 and 0.5. For the ωambient values, δωambient is between 0.1% and 7% over the same range (0.98−0.5) owing to larger uncertainties for σep at ambient RH.

Table 3. Value and Absolute Uncertainty of the TOA Forcing Calculated at 25% RH and Ambient RHa
 Traffic EmissionsGalveston Bay, 1Galveston Bay, 2Galveston Bay, 3Galveston Bay, 4aGalveston Bay, 4bORVIndustrial
  • a

    The 25% RH and Ambient RH are ΔFR and ΔFR_ambient, respectively. The average value and the absolute uncertainty of the parameters used in the calculations are also shown.

ω (δω)0.60 (0.02)0.56 (0.022)0.80 (0.010)0.81 (0.010)0.89 (0.0055)0.94 (0.0030)0.93 (0.0035)0.87 (0.0065)
ωambient (δωambient)0.72 (0.019)0.65 (0.025)0.86 (0.010)0.88 (0.0085)0.93 (0.0050)0.97 (0.0020)0.98 (0.0014)0.92 (0.0056)
AOD (δAOD)0.15 (0.015)0.20 (0.015)0.25 (0.015)0.30 (0.015)0.30 (0.015)0.30 (0.015)0.55 (0.015)0.30 (0.015)
β (δβ)0.3 (0.050)0.3 (0.050)0.3 (0.050)0.3 (0.050)0.3 (0.050)0.3 (0.050)0.24 (0.050)0.3 (0.050)
ΔFR (δΔFR)−3.8 (0.66)−8.5 (0.86)−19.4 (1.2)−23.2 (1.5)−27.2 (1.6)−29.4 (1.6)−36.2 (2.4)−23.3 (1.5)
ΔFR_ambient (δΔFR_ambient)−6.9 (0.72)−11.1 (0.95)−21.6 (1.3)−26.7 (1.5)−28.9 (1.6)−30.7 (1.6)−40.5 (2.4)−26.7 (1.5)

[38] Typical b values for TexAQS-GoMACCS were ∼0.15, which translates into β values of approximately 0.3. For the case of long-range transport of sulfate from the ORV region, the measured b was 0.1, resulting in a β of 0.24. The uncertainty in β (δβ) is estimated as 0.05 based on uncertainty values of b reported for continental aerosols [Anderson et al., 1999]. The AOD measured at the time closest to the each observed aerosol event was used for the calculation of ΔFR. The measured AOD ranged between 0.15 and 0.55, and has an estimated uncertainty, δAOD, of 0.015. AOD values above 0.3 reflect significant aerosol burden in the column and are typical of highly polluted regions [Quinn and Bates, 2003; Ramanathan et al., 2007]. In general, the AOD during TexAQS-GoMACCS was ∼0.2 during relatively cleaner southerly flow and >0.4 under northerly flow conditions. We do not have information on the vertical structure of aerosol amount and composition during high-AOD periods. However, under northerly flow conditions the planetary boundary layer was often well mixed during the day up to ∼1200 m and in situ surface measurements are likely representative of the boundary layer.

[39] The ΔFR and ΔFR_ambient values and their respective uncertainties (δΔFR and δΔFR_ambient) are also reported in Table 3. For all aerosol types the calculated forcing is negative indicating a cooling effect. For the majority of the air masses examined here the forcing lies between −20 and −30 W/m2. The greatest calculated cooling (−36.2 W/m2) is associated with the sulfate-rich aerosols transported from the ORV. These findings are quantitatively consistent with those reported from large-scale plumes which are advected downwind of the North American, Indian and Asian continents and which contribute to a significant reduction of radiation at the surface, or dimming effect [Ramanathan et al., 2007]. The magnitude of the forcing that we calculated depends most strongly on AOD rather than on ω, for example, the most negative ΔFR occurs for the air masses with the largest AOD and not the highest ω. This is consistent with McComiskey et al. [2008] who showed that in full radiative transfer models the forcing estimates are most sensitive to AOD rather than to ω when all else is held constant. The δΔFR values are within ±2 W/m2, or 5–7%; uncertainties become significant where ΔFR is small, for example, for the traffic emissions (−3.8 ± 0.66 W/m2, or 17%). The ΔFR_ambient are up to 5 W/m2 lower than the correspondent ΔFR values owing to generally higher ωambient, and indicate greater radiative cooling from hydrated aerosols, as expected [Garland et al., 2007]. The smallest increase in cooling due to aerosol water uptake is associated with the aerosols from traffic emissions, while the greatest increase in cooling occurs for the sulfate rich aerosols. For all the cases with ω > 0.8 (common value for atmospheric aerosols), the average percentage increase in ω to ωambient is 5%, leading to an increase in forcing of ∼10%. Hence, a relatively small percentage changes in ω (often hidden behind large uncertainties within ω estimates) can propagate to a much larger percentage change in forcing. McComiskey et al. [2008] showed that in most situations the uncertainty in DRF is due to the uncertainty in ω values, indicating that improvements in determining ω especially at ambient RH conditions will lead to the largest reductions in the overall uncertainty in DRF. Incorporating the information of the measured aerosol hygroscopic properties in the forcing expression clearly provides more realistic DRF estimates and helps quantify the sensitivity of the atmospheric radiative balance to changes in the aerosol optical properties due to water uptake.

[40] Our results based on the simplified radiative calculation show that uncertainties in ΔFR are not small, but they can be held below ∼20% even for moderately absorbing aerosols (ω ∼ 0.6) if δω is ≤5% and ΔFR values are >10 W/m2. However, relative uncertainties can become much larger for very strongly absorbing aerosols and at low AODs. Currently, optical property measurements from CRD and PAS result in lower uncertainty estimates of ω than do those obtained from combination of other instruments such as the TSI nephelometer and the particle soot absorption photometer (PSAP) for comparable time resolution and ω levels (P. Massoli et al., Uncertainty in light scattering measurements by nephelometer: results from laboratory studies and implications for ambient measurements, submitted to Aerospace Science and Technology, 2009). Continued improvements in precision and accuracy in the determination of ω would further reduce the overall uncertainty in DRF from radiative transfer models that must also accommodate uncertainties related to aerosol spatial/temporal resolution [Bates et al., 2006] and from satellite-based measurements that typically hold large uncertainties [Anderson et al., 2005; Yu et al., 2006].

5. Summary and Conclusions

[41] The hygroscopic, optical and chemical properties of aerosol particles were measured on board the NOAA RV R. H. Brown during the 2006 TexAQS-GoMACCS campaign in the Gulf of Mexico. Aerosol light extinction (σep, Mm−1) and absorption (σap, Mm−1) coefficients were measured by CRD and PAS, respectively. The particle single scattering albedo (ω) was calculated at 532 nm from the combination of σep and σap with an uncertainty of <0.01 at 25% RH for submicron aerosols. The RH dependence of extinction was parameterized using a previously developed γ formulation, i.e., correlation of γ with the fraction of submicrometer organic mass relative to organic and sulfate (FPOM). The TexAQS-GoMACCS data covered a broad spectrum of γ and FPOM values, and their correlation was consistent with results obtained from previous data sets in diverse environments [Quinn et al., 2005]. The wide range of aerosol properties encountered during the experiment serve to validate the generality of the γ parameterization for describing aerosol hygroscopic behavior in a simple manner.

[42] Portions of the TexAQS-GoMACCS data set containing unique aerosol characteristics from identifiable sources were selected for further analysis. Aerosols from traffic emissions, sampled during morning rush hours in the shallow boundary layer (0600–0900 local time), were hydrophobic and light absorbing (γ = 0.39 and ω = 0.60), and the organic fraction was mostly hydrocarbon-like (FOOA = 0.28). Aerosols properties in Galveston Bay during the daytime photochemical production of ozone (1000–1800 local time) were variable as a result of combined effect of sources, oxidation state and mixing; in general, more hygroscopic and light-scattering aerosols corresponded to higher FOOA for similar values of FPOM. The role of the region's complex transport processes in shaping the aerosol properties was evident on 18 August when a sulfate-rich, hygroscopic aerosol was brought to RHB in Galveston Bay, probably from sources in the southwest Houston area. An exceptional case of multiday, regional fumigation from sulfate aerosol sources in the Ohio River valley (ORV) showed how long-range transport can greatly affect visibility, air quality and climate over a large distant region. Aerosols sampled in Barbours Cut south of the Houston ship channel, dominated by nearby industrial and shipping sources, showed a large variability in properties; however, in general the particles were hydrophobic, absorbing, and rich in POM.

[43] The optical measurements collected during the field study (ω, AOD, b) were used to estimate the instantaneous forcing (ΔFR) locally induced at the TOA. Relatively small changes in ω led to significant changes in forcing, especially in some sensitive ω ranges or for small values of ΔFR. The ability to determine the RH dependence of optical properties is critical for realistic prediction of aerosol behavior at ambient humidity levels. Furthermore, the ability to determine changes in aerosol optical properties with small uncertainties, and their changes with changing RH, is necessary to permit accurate predictions of aerosol direct radiative forcing.


[44] The authors thank the officers and crew of the NOAA RV Ronald H. Brown, D. Hamilton, and D. Coffman for their support. We thank D. Hamilton for performing the AOD measurements during TexAQS-GoMACCS. P.M. thanks C. A. Brock for helpful comments on the manuscript and A. McComiskey, W. Angevine, and G. J. Frost for useful discussions. We thank J. Meagher, F. Fehsenfeld, and A. R. Ravishankara for programmatic support. This project was funded by the NOAA Climate and Global Change Program, the NOAA Office of Oceanic and Atmospheric Research, the NOAA Health of the Atmosphere Program, and the Texas Air Quality Study. Certain commercial equipment, instruments, or materials are identified in this article in order to adequately specify the experimental procedure. Such identification does not imply recognition or endorsement by the National Oceanic and Atmospheric Administration, nor does it imply that the material or equipment identified are necessarily the best available for the purpose.