Light scattering enhancement factors in the marine boundary layer (Mace Head, Ireland)

Authors


Abstract

[1] Direct climate aerosol radiative forcing is influenced by the light scattering of atmospheric aerosols. The chemical composition, the size distribution, and the ambient relative humidity (RH) determine the amount of visible light scattered by aerosols. We measured the aerosol light scattering coefficients at RH varying from 30% to 90% of the marine atmosphere at the Mace Head Atmospheric Research Station on the west coast of Ireland. At this site, two major air mass types can be distinguished: clean marine and polluted air. In this paper, we present measurements of light scattering enhancement factors f(RH) = σsp(RH)/σsp(dry) from a 1 month field campaign (January–February 2009). At this site in winter, the mean f(RH = 85%) (standard deviation) for marine air masses at the wavelength of 550 nm was 2.22 (±0.17) and 1.77 (±0.31) for polluted air. Measured σsp(RH) and f(RH) agreed well with calculations from Mie theory using measurements of the size distribution and hygroscopic diameter growth factors as input. In addition, we investigated the RH influence on additional intensive optical properties: the backscatter fraction and the single scattering albedo. The backscatter fraction decreased by about 20%, and the single scattering albedo increased on average by 1%–5% at 85% RH compared to dry conditions.

1. Introduction

[2] Atmospheric aerosols influence the Earth's radiation budget by two effects: the direct and indirect effect. Aerosols indirectly influence the Earth's radiation budget by modifying the microphysical and thus the aerosol radiative properties, and the water content and lifetime of clouds. The direct effect is the mechanism by which aerosols scatter and absorb sunlight, thereby altering the radiative balance of the Earth-atmosphere system [Intergovernmental Panel on Climate Change, 2007]. Three aerosol optical properties are important to determine the aerosol direct radiative forcing: the aerosol extinction coefficient which is the sum of aerosol scattering coefficient (σsp) and aerosol absorption coefficient (σap) and which specifies how much aerosol particles attenuate electromagnetic radiation; the single scattering albedo (ω0), which is defined as the ratio of light scattering to total light extinction; and the angular distribution of light scattering which can be parameterized by the asymmetry parameter, the upscatter fraction or backscatter fraction [Kiehl and Briegleb, 1993].

[3] These three aerosol optical properties can be measured, at least approximately, with a nephelometer, which measures the light scattering and backscattering coefficient (σbsp), and combined with an aethalometer, which measures the light absorption coefficient, like it is done at various World Meteorological Organisation (WMO) Global Atmosphere Watch (GAW) stations. To better compare between different measurements at these stations, the WMO recommends measurements to be below 40% relative humidity (RH), which is considered a dry measurement [WMO/GAW, 2003]. However, many global circulation model studies found that the negative radiative forcing of the aerosols is significantly higher when the ambient RH was accounted for [Haywood and Ramaswamy, 1998; Penner et al., 1998; Grant et al., 1999; Kiehl et al., 2000].

[4] To measure σsp and σbsp at different RH, we used a humidification system upstream of a commercial nephelometer (TSI Inc., model 3563) which allows for the measurement of σsp and σbsp at a defined humidity below 90% RH [Fierz-Schmidhauser et al., 2010a]. The system is able to measure how hygroscopic properties and hysteresis effects of the atmospheric aerosol influence the scattering properties. The light scattering enhancement factor f(RH) = σsp(RH)/σsp(dry) was used to quantify the dependence of σsp on RH.

[5] Several measurements of f(RH) of various aerosol types and at different locations in the world have been made. Kotchenruther and Hobbs [1998], Gras et al. [1999], and Kim et al. [2006] measured f(RH) of biomass burning aerosol. Dust and pollution aerosol was measured by Carrico et al. [2003] and Kim et al. [2006]. Fitzgerald et al. [1982] and Yan et al. [2009] studied urban aerosol. Fierz-Schmidhauser et al. [2010b] and Zieger et al. [2010] measured f(RH) in the free troposphere and the Arctic, respectively, with the same instrument as we used in this study.

[6] In this paper we describe field measurements of aerosol extensive (concentration dependent) and intensive (independent of the amount of aerosol present) properties. The measured properties are σsp, σbsp, σap, the single scattering albedo, the backscatter fraction, the wavelength dependence of the scattering coefficients (scattering Ångström exponent), the hygroscopic growth factor, the light scattering enhancement factor and the number size distribution. Measurements were conducted at the GAW Atmospheric Research Station at Mace Head, Ireland. This site is considered representative of the midlatitude marine troposphere [O'Dowd et al., 2001]. We discuss the f(RH) and compare it to other measurements of f(RH) of marine aerosol from Tasmania [Carrico et al., 1998], from the southwest coast of Portugal [Carrico et al., 2000], from the Canary Islands [Gasso et al., 2000], from the Pacific between Hawaii and Japan [Carrico et al., 2003] and the northeast coast of North America [McInnes et al., 1998; Wang et al., 2007]. Using measured size distributions and hygroscopic growth factors, we perform a closure/comparative study using a model based on Mie theory, which calculates the f(RH) at different RH.

2. Experimental Procedure

2.1. Measurement Site

[7] The GAW Atmospheric Research Station at Mace Head is an excellent site for studying marine aerosols. It is located at the west coast of Ireland (53°19′N, 9°54′W) on a peninsula, which is surrounded by coastline and tidal areas except for a small sector (where there is landmass) ranging from 20° to 40°. The research station is about 70 to 120 m from the shoreline (tide dependent) at 5 m above sea level. Air that arrives at the Mace Head site from the direction between 180° and 300° is classified as marine or clean air [Jennings et al., 1997].

[8] Most of the instruments are connected via the community air-sampling system of a 10 cm diameter stainless steel pipe, reaching 10 m above ground level, with a flow rate of 150 liter per minute (lpm) to ensure laminar flow. While size fractionation of the aerosol at the sample inlet was not conducted [Kleefeld et al., 2002; Yoon et al., 2007], the whole particle size range was not measured in this work. We estimated the 50% cutoff diameter of the inlet system, using the calculations of the sampling system efficiency of Kleefeld et al. [2002], to be on average about 2 μm at a mean wind speed of 13.8 m/s (standard deviation: 6.6 m/s) in January/February 2009.

2.2. Instrumentation

2.2.1. Light Scattering Coefficient Under Dry Conditions and at High RH

[9] Since July 2001 a three-wavelength (λ = 450, 550 and 700 nm) integrating nephelometer (TSI Inc., model 3563) has measured the dry scattering coefficients σsp and dry backscattering coefficients σbsp at Mace Head [O'Connor et al., 2008]. The RH in the dry nephelometer was on average 26.7 ± 4.5% during our measurement campaign. No drying of the air was needed to achieve this RH, due to the internal heating of the nephelometer to 24.2°C (average) and mainly due to the temperature difference between inside and outside the laboratory (campaign average ambient T = 5.95°C).

[10] We used a humidification system for a second integrating nephelometer to measure the RH dependence of σsp and σbsp at a defined RH in the range of 30%–90% RH (described in the work of Fierz-Schmidhauser et al. [2010a]). It consists of a humidifier to raise the RH of the aerosol up to 90% RH, followed by a dryer, which dries the aerosol to the desired RH. This system enables us to measure the hysteresis behavior of deliquescent aerosol particles. The light scattering enhancement factor f(RH) is defined as the ratio of σsp at high and low RH:

equation image

There are two different operating conditions for the humidified nephelometer: hydration and dehydration. Hydration is defined when the dryer is turned off and the RH within the instrument is monotonically rising from the humidifier to the entrance of the nephelometer. We define dehydration in a similar manner: at the exit of the humidifier the aerosol particles are exposed to RH > 80%. The dryer is then turned on, which results in a monotonic decrease of RH in the dryer line. With these two operating conditions it is possible to measure a deliquescent aerosol, which can exist in the liquid and solid phases at the same RH, known as hysteresis [Orr et al., 1958].

[11] The humidified nephelometer measured in parallel with the dry nephelometer of the regular GAW program from 19 January to 15 February 2009. The dry nephelometer measured at a frequency of 5 min, whereas the humidified nephelometer detected scattering coefficients every minute. These values were then averaged to 5 min values. In the beginning of the measurement campaign (19 January 15:00 to 21 January 13:00) both nephelometers measured at dry conditions and agreed well with each other, with a slope of 1.054, an intercept of 7.4·10−7 m−1 and a squared correlation coefficient of R2 = 0.97 (at λ = 550 nm). We corrected σsp of the humidified nephelometer with this linear relationship. During almost 12 days (285 h) (21 to 26 January and 29 January to 5 February) the humidified nephelometer determined humidity cycles of the scattering enhancement, commonly referred to as humidograms. Measuring one humidogram took 2 h. On 26/27 January and from 5 to 15 February the humidified nephelometer measured at 85% RH (±5% RH). All values between 80% and 90% RH were fitted to 85% with one free parameter γ, already used in other work on f(RH) [Kotchenruther and Hobbs, 1998; Gasso et al., 2000; Zieger et al., 2010]:

equation image

We corrected σsp and σbsp for nephelometer nonidealities (angular truncation and non lambertian light source) according to Anderson and Ogren [1998].

[12] The backscatter fraction b is the ratio of σbsp to σsp:

equation image

and is the percentage of radiation that is scattered back at angles between 90° and 180°. b decreases with increasing particle size. Another intensive property which can be derived from the scattering coefficient is the Ångström exponent ås of the scattering coefficient:

equation image

The Ångström exponent we used is derived from the scattering coefficients at λ1 = 450 and λ2 = 700 nm wavelength.

2.2.2. Light Absorption Coefficient

[13] An AE-16 aethalometer (Magee Scientific, Berkeley, USA) was also operated during this campaign. The instrument features an automatic filter change, and sampling time was kept at 5 min. The aethalometer operates on the principle of light attenuation due to absorption by aerosol particles deposited on a prefired quartz fiber filter. This instrument is operated with a polychromatic light source (white light). The determination of the absorption coefficient (σap) is a difficult task because of the ill defined spectral sensitivity of the employed instrument. Here σap was calculated using the method presented by Weingartner et al. [2003] with the C value (accounting for multiple scattering in the filter) of 3.05 for Mace Head [Collaud Coen et al., 2010]. A filter loading dependent correction was not performed.

[14] Absorption coefficients were first calculated for λ = 855 nm which is the center of the broadband spectral instrumental response. For the calculation of the single scattering albedo, defined as

equation image

The measured absorption coefficients were extrapolated to λ = 550 nm assuming a λ−1 dependence of the absorption coefficient.

2.2.3. Particle Number Size Distribution

[15] The particle number size distribution (size range 20 nm < Dp < 500 nm) was determined with a scanning mobility particle sizer (SMPS) which consists of a differential mobility analyzer (DMA) and a condensation particle counter (CPC). The SMPS system at Mace Head uses a Krypton-85 bipolar charger for the neutralization of aerosols, along with a TSI-type long DMA in conjunction with a TSI CPC 3010. All aerosol sample flows are dried to 40% RH or lower. Three sensors monitored the flow in the sample flow, the sheath flow and excess flow. The RH of the sheath flow was also monitored, along with the pressure in the DMA.

[16] In addition, an optical particle counter (OPC, Grimm Dustmonitor 1.108) measured the dry size distribution of the larger particles in the optical diameter range 0.3 μm < Dp < 20 μm. In the OPC, the individual particles are classified according to their light scattering intensity, which depends on the particle size, morphology and refractive index. We used the factory calibration of the OPC size bins (polystyrene latex spheres with a refractive index of 1.59 at λ = 683 nm). With the OPC output a coarse mode fraction (cmf) is defined as

equation image

We only used data from the OPC to calculate the cmf, since in this way discrepancies between instruments (SMPS and OPC) did not influence this factor.

[17] The combined SMPS and OPC data were used as input for the Mie calculation, even if they did not exactly match each other (see Figure 2). All data from the OPC were taken, whereas the SMPS data were just used up to Dp = 340 nm, to avoid the influence of doubly and triply charged particles for larger diameters.

2.2.4. Hygroscopic Growth Factor

[18] The ratio of the wet to the dry diameters is referred to as the particle's hygroscopic growth factor g(RH). The growth factor g(RH) was measured with a hygroscopicity tandem differential mobility analyzer (H-TDMA), which is described in a recent review by Swietlicki et al. [2008]. The Mace Head H-TDMA follows the recommendations for the design of an H-TDMA by Duplissy et al. [2009]. It is composed of two DMAs and a humidifier. The aerosol sample enters the system through a nafion dryer where the RH is lowered to ∼5%. Next it passes through a radioactive charger where it obtains a known charge distribution. In the first DMA a narrow distribution of particles with known mobility is selected at a fixed applied voltage. This quasi-monodisperse aerosol sample is then exposed to 90% RH in the humidifier. The humidifier consists of a heated Gore-Tex tube and humidifies the sample and the sheath air for the second DMA. After that, the sample enters the second DMA which together with a CPC acts as a SMPS. Unlike the first DMA, the voltage applied to the second DMA is changing allowing different sizes of particles go through to the CPC where they are counted. From the change in the size the growth factor is calculated using the inversion algorithm by Gysel et al. [2009].

2.3. Mie Model to Calculate f(RH)

[19] We calculated σsp at dry and humid conditions and f(RH) with a model based on Mie theory [Mie, 1908] where the core Mie routine is based on the code of Bohren and Huffmann [2004]. The particles are assumed to be spherical and homogenously internally mixed. Assuming spherical particles leads to an error of σsp of less than ±5% for particles smaller than 1 μm and an underestimation of σsp of up to 30% for cubic salt particles larger than 1 μm [Chamaillard et al., 2006]. The number size distribution and the complex refractive index m of the measured aerosol is needed as input. We calculated the complex refractive index using the average chemical composition measurements of other measurement campaigns: For clean marine air we used the mass concentrations of low biological activity [O'Dowd et al., 2004] and January/February values from Yoon et al. [2007]. Polluted air in winter consists of about 15% of organics [O'Dowd et al., 2004], 5% black carbon (BC) [Jennings et al., 1997] and 80% of inorganic species. The inorganic chemical composition of nonclean data over a 4 year sampling period (2003–2006) shows 71% of sodium plus chloride ions, 14% sulfate ions, 10% nitrate ions and the remaining 5% ammonium ions [Ceburnis et al., 2010]. The refractive index for the two air mass types was then determined by a volume fraction averaging:

equation image

where mfi is the mass fraction, ρi is the density and mi(λ) is the wavelength-dependent complex refractive index of the compound i. We took the values for mi and ρi as listed in Table 1. This resulted in a refractive index for clean air of mclean(λ = 550 nm) = 1.539 + 0i and mpolluted(λ = 550 nm) = 1.529 + 0.024i for polluted air.

Table 1. Microphysical Properties of Selected Aerosol Compounds Used for the Model Predictionsa
λH2OOrganicsNH4NO3(NH4)2SO4NH4HSO4H2SO4NaClBC
mi 450 nm1.337b 1.559c1.536c 1.438d1.560c1.75+0.46ie
mi 550 nm1.333b1.48f,g1.556c1.530c1.473g,h1.434d1.548c1.75+0.44ie
mi 700 nm1.331b 1.553c1.524c 1.432d1.542c1.75+0.43ie
ρi(g/cm3)1.0i1.4j1.72i1.77i1.78i1.83i2.17i1.7f

[20] Hygroscopic growth was accounted for by using the size resolved H-TDMA measurements of diameter growth factors. The H-TDMA growth factor g(RH = 90%) at a dry diameter Dp of 165 nm was extrapolated to different RH using equation (3) from Gysel et al. [2009], which uses the κ model introduced by Petters and Kreidenweis [2007]. For the wet refractive index a volume weighting between the refractive indices of water and the dry aerosol was chosen [Hale and Querry, 1973].

3. Results

3.1. Overview

[21] Jennings et al. [2003] classified the air masses arriving at the Mace Head Atmospheric Research Station into two sectors: clean marine when the wind comes from 180° to 300° and polluted when the wind comes from 45° to 135°. We classified the air masses in this study accordingly; the two air mass types are indicated in Figure 1 with different underlying colors. Light blue is for clean marine and light yellow for polluted air masses.

Figure 1.

Temporal evolution of (a) the dry scattering and absorption coefficients, (b) the dry single scattering albedo, (c) the hygroscopic growth factor g(RH) at 90% RH with a dry diameter of 165 nm and the light scattering enhancement factor at 85% RH, (d) the dry backscatter fraction, (e) the dry Ångström exponent and the OPC coarse mode fraction, and (f) the clean marine or polluted wind sector according to Jennings et al. [2003]. Underlying colors represent clean marine (light blue) and polluted (light yellow) air masses. σsp, σap, ω0, f(RH), and b are shown at 550 nm wavelength.

[22] The temporal evolution of a selection of measured and derived aerosol variables is shown in Figure 1. The hourly dry σsp varied between 3.1·10−6 and 1.6·10−4 m−1 (uncertainties ±10% [Anderson et al., 1996]) within this measurement campaign in January/February 2009 (Figure 1a). The hourly averages of σap(λ = 550 nm) varied within this time period between 2·10−8 and 1.4·10−5 m−1 (uncertainties ±30%). This corresponds to equivalent BC mass concentrations (measured at λ = 855 nm) between 1.8 and 1400 ng m−3 (on average 134 ng m−3), which is within the range of measurements presented by Jennings et al. [1993]. Since intensive aerosol parameters are needed to understand the variability of the light scattering enhancement factor (f(RH)) and the hygroscopic growth factor (g(RH)), we investigated the dry intensive properties of the following parameters: single scattering albedo (ω0), backscatter fraction (b), Ångström exponent (ås) and OPC coarse mode fraction (cmf) for the two air mass types. The ω0 (Figure 1b) was high during the entire measurement campaign (mean: 0.95, see Table 2) with highest values during the beginning of the measured time period until 29 January. Generally the ω0 (λ = 550 nm) at Mace Head is very high, with monthly averages between 0.94 and 0.99 in the years 2000 to 2002 [Jennings et al., 2003]. During the clean marine air masses observed at the beginning of the campaign the aerosol tended to be dominated by large, primarily scattering aerosol as indicated by high values of ω0 (Figure 1b), low values of b (Figure 1d) and ås (Figure 1e). b and ås yield information on the dominant particle sizes contributing to the scattering signal. Low values of b and ås correspond to large aerosol particles. A value of ås of less than about 0.5 indicates the presence of coarse mode sea-salt aerosols [Smirnov et al., 2002], since coarse mode dust particles can be ruled out because the air masses did not originate from dust sources within 7 days before arrival at Mace Head. This contribution of coarse mode aerosol particles in the beginning of the measurement campaign is also consistent with the relatively high-coarse mode fraction from the OPC in Figure 1e. Table 2 lists averages of all intensive properties for the whole measurement campaign and for the two air mass types.

Table 2. Means and Standard Deviations of the Single Scattering Albedo, Backscatter Fraction, Ångström Exponent, and OPC Coarse Mode Fraction, for Dry Conditionsa
 Campaign MeanClean MarinePolluted
  • a

    The light scattering enhancement factor and the hygroscopic growth factor are also shown. ω0, b, f(RH = 85%) are at λ = 550 nm.

ω00.95 ± 0.050.99 ± 0.010.89 ± 0.04
b0.108 ± 0.0120.101 ± 0.0050.121 ± 0.017
ås0.413 ± 0.6480.014 ± 0.0831.431 ± 0.503
cmf0.061 ± 0.0290.078 ± 0.0120.017 ± 0.016
f(RH = 85%)2.10 ± 0.252.22 ± 0.171.77 ± 0.31
g(RH = 90%, Dp = 165 nm)1.64 ± 0.261.84 ± 0.141.36 ± 0.15

[23] All intensive properties are significantly different for the two air mass types. When the ω0 and the cmf are high and the b and ås are low, we observe a high f(RH) and g(RH) (clean marine air masses), most probably due to more hygroscopic aerosol particles (sea salt and other inorganic species). Low ω0 and cmf, f(RH) and g(RH) plus high b and ås correspond to polluted air masses, indicating a higher fraction of light absorbing aerosol with lower hygroscopicity. b, ås and cmf are all linked to the aerosol size distribution and give some indication about the mean particle size in the two air mass types. For a closer investigation we calculated the mean number size distribution and mean surface area size distribution for the two air mass types (Figure 2). The discontinuity in the combined number size distribution is likely to result from the difference of the refractive index (and complex morphology) between the ambient particles and the particles that were used to calibrate the OPC, and not due to RH differences between the SMPS and the OPC, which experienced a similar operational range of RH conditions. Similar deviations were also noted in other studies [Hand and Kreidenweis, 2002; Heim et al., 2008].

Figure 2.

(top) Mean dry number size distribution and (bottom) surface area size distribution of the clean marine and polluted air masses, originating from the measurements of the SMPS and OPC.

[24] Below Dp = 500 nm about five times more particles are present in polluted air masses compared to clean marine air masses. In contrast, for particle diameters above 500 nm the particle surface area of the clean marine air masses is more pronounced than the one of the polluted air masses. This is consistent with the results of b, cmf and ås described above.

3.2. Measurements at Varying Relative Humidity

[25] The intensive properties b, ås and ω0 change with RH. Since the atmosphere is not generally dry, these values at humid conditions are of higher climate relevance than the dry values (the average ambient RH was 85.3% during the campaign). Therefore we calculated the mean of b, åsand ω0 at 85% RH for the whole measurement campaign and for the clean marine and polluted air masses (see Table 3). The ω0 at 85% RH was determined from the dry σsp values and f(RH = 85%) as described above and from the dry measured σap values assuming that the absorption does not change with RH. This simplification is plausible since even though the humidity effect on absorption can be substantial, its maximum contribution to the humidity effect on the single scattering albedo is only 0.2% for aged aerosol within the wavelength range from 450 to 700 nm [Nessler et al., 2005b]. The ω0 increases from between 0.89 and 0.99 under dry conditions to very high values of between 0.93 and 0.99 for RH = 85%.

Table 3. Means and Standard Deviations of the Intensive Properties at 85% RH (Single Scattering Albedo, Backscatter Fraction, Ångström Exponent) From Measured Values in January/February 2009 at Mace Heada
 Campaign MeanClean MarinePolluted
  • a

    ω0 and b are at λ = 550 nm.

ω0 (RH = 85%)0.97 ± 0.030.99 ± 0.020.93 ± 0.04
b (RH = 85%)0.086 ± 0.0100.082 ± 0.0050.098 ± 0.016
ås (RH = 85%)0.162 ± 0.465−0.037 ± 0.1180.931 ± 0.626

[26] b at RH = 85% was determined from σsp and σbsp values at RH between 80 and 90% recalculated to RH = 85% using equation (2). By taking up water, aerosol particles grow and therefore b decreases. We observed a decrease of about 20% for the campaign mean value of b from dry conditions to RH = 85%, which results in a backscatter fraction which is smaller than 10% for both clean marine and polluted air masses.

[27] For calculating the Ångström exponent of the scattering coefficient at 85% RH, we used equation (4) with σsp(RH = 85%). From all investigated parameters, the ås changed most with increasing RH, its values decreased by 35%–60% compared to those under dry conditions.

[28] We measured humidograms from 21 January to 5 February (with a break between 26 and 29 January). The total of 141 humidograms, each lasting for 2 h, were allocated to the two clean marine and polluted air masses. We linearly interpolated the dry scattering coefficients to 1 min values and calculated the f(RH), then grouped the f(RH) of each sector into 5% RH bins and calculated the mean f(RH) of each RH bin for hydration and dehydration. Since the RH at Mace Head is very high (RH > 80% for 71% of the time of the campaign) the values of the hydration measurement are much less important than the values of the dehydration curve. Nevertheless we report both curves, so that measurements from humidified nephelometers that are only capable of measuring hydration could be compared. Figure 3 presents all these values including standard deviations of f(RH) as y axis error bars. The x axis error bars represent the mean change of RH per time interval in the corresponding RH bin. These bars are much larger between 65 and 75% RH for dehydration than for all other RH bins, because of instrumental reasons: when the dryer is on, the RH in the nephelometer changes relatively fast.

Figure 3.

Averaged humidograms for (top) clean marine and (bottom) polluted air masses showing f(RH) versus RH in the nephelometer under hydrating and dehydrating conditions.

[29] In general the humidograms of polluted air show no deliquescence, with f(RH) smoothly increasing with RH to a maximum of 1.8 at 85% RH. The f(RH) already starts to be slightly enhanced at RH = 35% and monotonically rises with RH. In contrast, the aerosol particles of the clean marine air masses do not start to grow until 50% RH in the nephelometer. The humidograms for the clean marine air clearly indicate hysteresis behavior of the aerosol and for certain single humidograms also show a sharp deliquescence, which is however not visible in the mean humidogram presented in Figure 3 as deliquescence occurred at slightly varying RH. The main components of clean marine aerosol are inorganic salts (like NaCl) [O'Dowd et al., 2004; Yoon et al., 2007], which show deliquescence behavior. In the polluted air masses, components with no distinct phase transition like organics contribute more to the chemical composition [Jennings et al., 1997; O'Dowd et al., 2004; Ceburnis et al., 2010]. Note that the phase transition appears at the RH in the nephelometer, which is lower than the deliquescence relative humidity, because the highest RH in the system is encountered upstream of the nephelometer [Fierz-Schmidhauser et al., 2010a].

[30] The maximum f(RH) at 85% RH is 2.3 for marine air masses. Comparing this result to different studies of marine air masses, our f(RH) is in the range of the results of McInnes et al. [1998] (f(RH = 85%) = 2.7) and Carrico et al. [1998] (f(RH = 82%) = 1.98). McInnes et al. [1998] also measured anthropogenically influenced f(RH) at 85% RH of 1.7, which compares very well to our measurement of 1.8. The marine aerosol shows a clear hysteresis behavior with f(RH = 55%) of 1.66 for the upper branch and 1.23 for the lower branch. This hysteresis behavior is much smaller for polluted aerosol (f(RH = 55%) = 1.30 and 1.19). For conditions with a generally high ambient RH, the values of the upper branch should be used. Efflorescence could not be measured, since we were not able to dry the aerosol to below 50% RH after the humidification to 85% (see above).

[31] It is interesting to note that the f(RH) at Mace Head is much lower than the values reported for laboratory measurements of monodisperse pure sodium chloride (NaCl) particles in the size range of 100 to 300 nm, where the f(RH) at 80% ranged from 15 to 5.5 [Fierz-Schmidhauser et al., 2010a]. The reasons for this discrepancy are firstly that in this study the surface area size distribution is dominated by particles that are larger than 300 nm (see Figure 2) and f(RH) decreases with increasing particle size. A second reason for the discrepancy is that the Mace Head aerosol does not consist of pure NaCl. O'Dowd et al. [2004] found that in winter 75% of the mass fraction of the accumulation mode aerosol at Mace Head is sea salt, 10% non-sea-salt sulfate and 15% organic compounds.

3.3. Model Calculation of the Scattering Coefficients and f(RH)

[32] With our model based on Mie theory we calculated σsp(dry) and σsp(RH), which depend on the aerosol size distribution, the chemical composition of the aerosol and the RH in the nephelometer. The chemical composition determines the refractive index and the hygroscopic growth factor g(RH). We calculated a refractive index for clean marine and polluted air as described in section 2.3. Using one fixed dry refractive index per air mass type is not a critical issue, since the refractive index at high RH will be dominated by the one of water. g(RH) was measured by the H-TDMA at 90% RH and Dp = 165 nm.

[33] For Figure 4 we only used measured f(RH) data points, when at the same time size distribution and H-TDMA data was available and so f(RH) could be calculated. Since there were either missing OPC or SMPS data for appreciable time fractions, we could not calculate the scattering coefficients and f(RH) for 62% of the time when the air masses were polluted (69 h out of total 111 h), therefore the humidograms of Figures 3 and 4 do not exactly look alike. The gray and white circles in Figure 4 represent the average of these data points for hydration and dehydration, respectively. The mean calculated f(RH) versus RH are shown in blue for clean marine air masses and in orange for polluted air masses.

Figure 4.

Calculated (blue and orange) and measured (gray) humidograms for (a) clean marine and (b) polluted air masses. Measured data is only shown for periods where calculations were possible. Calculations apply to the liquid state of the aerosol and use size distributions and hygroscopic growth factors as input.

[34] The calculated f(RH) at low RH (30%–45%) is much higher than the measured (especially for clean marine air) since the model assumes that particles are hydrated over the full RH range, which is not the case for a deliquescent aerosol like pure NaCl or sea salt or the aerosol we measured in clean marine air. At RH when the measured aerosol shows hysteresis behavior (55% < RH < 75%) the calculated f(RH) lies between the values for hydration and dehydration f(RH). For both clean marine and polluted air masses at RH > 80% the model either overestimates f(RH) or the humidified nephelometer measures too low values. Below 75% RH the calculated f(RH) humidogram of polluted air masses is consistent with the measured humidogram. A hypothesis for the discrepancy between model and measurement at high RH is losses of largely growing super-μm sea salt particles in and after the humidification system. We calculated particle losses in the humidification system with the program described by von der Weiden et al. [2009] and found that we have less than 10% losses for particles smaller than 5 μm and about 33% losses for particles with a diameter of 10 μm (for a flow of 10 lpm).

[35] Figure 5a shows that the model calculated the σsp(dry) very well, with a slope of 0.99 and a squared correlation coefficient R2 = 0.9. This supports our choice of the refractive index in the model assumption of the dry aerosol. We did the same calculation for refractive indices of either pure NaCl or pure organics; the resulting slopes ranged from 0.98 to 1.06. The calculation of σsp(RH) is also quite good (Figure 5b); the difference between measured and calculated σsp(RH) is within about 10%. As a consequence the calculated ratio of σsp(RH) and σsp(dry), the f(RH), correlates quite well with the measured f(RH) (Figure 5c). Figure 5d shows the same as Figure 5c but separated into clean marine (blue) and polluted (orange) air masses. The calculated f(RH) of clean marine air masses is higher than that of polluted air. In the scatterplot of f(RH) measured versus f(RH) calculated of clean air masses (Figure 5e) it is also possible to see how the f(RH) is overestimated by the model both at low and high RH. For polluted air masses, we also observe higher values of the calculated f(RH) at low RH but not so much at high RH.

Figure 5.

Measured versus calculated (a) σsp(dry), (b) σsp(RH), (c) f(RH), (d) f(RH) split into clean marine and polluted air masses, (e) f(RH) from clean marine air masses, and (f) f(RH) from polluted air masses. The color code denotes the RH in the humidified nephelometer. Only times where measurements of SMPS, OPC, and H-TDMA were available are shown here.

[36] This very good agreement between measurement and calculation of σsp(dry) (Figure 5a) shows that the calculation using the refractive indices of the mean chemical composition of the two air mass types works well even if the size distributions of OPC and SMPS do not agree perfectly. From the good agreement between the measured and calculated σsp(RH) we further conclude that the g(RH) value from the H-TDMA is a good input parameter for the model even if this is a mean g(RH) value (aerosol mixing state is not taken into account) and representing particles at only one diameter (165 nm).

[37] We investigated the model sensitivity of f(RH) on the input parameters hygroscopic growth factor, refractive index, particle number concentration and particle diameter. The model predictions were repeated by keeping all but one of these parameters constant. We used the mean size distribution for both air mass types and added extra supporting points for the OPC size range by fitting a 3-log normal size distribution function to the measured distribution, in order to reduce uncertainties caused by the low resolution of the OPC. For both air mass types we varied the input parameters from −20% to +20% of their mean values and show the resulting error of f(RH) on Figure 6. The f(RH) is most sensitive to variations of the hygroscopic growth factor and the refractive index, whereas a comparable percentage of variation in the particle diameter has very little influence on the f(RH). However, variations in g(RH) and m of ±20% are much less likely than variations of ±20% in the particle diameter.

Figure 6.

Sensitivity of f(RH) to variations of the model input parameters: hygroscopic growth factor, number concentration, particle diameter, and refractive index; shown as the relative error in f(RH) versus percent variation from the mean value for (a) clean and (b) polluted air masses.

4. Summary and Conclusions

[38] With two nephelometers (one operated under dry conditions and one at changing RH) light scattering enhancement factors f(RH) and other aerosol optical properties were determined at the Mace Head Atmospheric Research Station. Additional instruments provided data on the aerosol light absorption, number size distribution (Dp = 20 nm–20 μm) and hygroscopic properties. From these aerosol measurements we derived the intensive properties b, ås, and ω0 for dry and elevated RH conditions.

[39] We analyzed the data for two distinct air mass types, i.e., polluted and clean marine. The f(RH = 85%) at λ = 550 nm was on average 1.77 for polluted air and 2.22 for clean marine air, both for winter conditions. The latter values compare well with other measurements in clean marine air [Carrico et al., 2000, 2003]. However, the f(RH = 85%) of marine aerosol is lower than for example the free tropospheric aerosol [Fierz-Schmidhauser et al., 2010b] or the Arctic aerosol [Zieger et al., 2010]. This finding can be explained by the influence of relatively large sea-salt particles with a relatively high organic mass fraction of about 15% at Mace Head in winter [O'Dowd et al., 2004].

[40] No distinct hysteresis behavior was observed for the polluted aerosol but a distinct hysteresis was encountered during clean marine conditions: Here the f(RH = 55%) values for dehydration are on average 35% higher than for hydration. This shows that the hydration state of the marine aerosol needs to be correctly measured and properly included in climate models.

[41] The dry ω0 (at λ = 550 nm) at Mace Head is high with an average value of 0.99 for clean marine and 0.89 for polluted conditions. At 85% RH the ω0 increases on average to 0.99 (clean marine) and 0.93 (polluted), as compared to dry conditions (RH < 40%). The mean b varied between 0.101 and 0.121 (clean and polluted) and decreased by about 20% at RH = 85% compared to RH < 40%.

[42] Measured scattering coefficients σsp(dry), σsp(RH) and f(RH) were compared to calculations obtained from a Mie model using measured time resolved size distributions, hygroscopicity, and a fixed aerosol refractive index for each air mass type as input. The model calculated σsp(dry) very well (slope = 0.99, R2 = 0.9) and f(RH) quite well (0.91, 0.77). For clean marine air masses, the model overestimated the f(RH) at low and moderate RH (30%–55%) since it does not account for deliquescence and hysteresis effects and assumes that the aerosol particles are only present in the liquid state. The good results of this closure study demonstrate that measurements with a humidified nephelometer can provide light scattering enhancement factors (f(RH)) similar to those retrieved from combined size distribution and g(RH) measurements. The advantage of using the humidified nephelometer in the field is that it is a relatively simple and stable system which does not need a lot of maintenance in operation.

Acknowledgments

[43] Financial support for this work was received from the EC projects EUSAAR (European Supersites for Atmospheric Aerosol Research), GEOMON (Global Earth Observation and Monitoring), EUCAARI (European Integrated project on Aerosol Cloud Climate Air Quality Interactions), and MAP (Marine Aerosol Production from Natural Sources), and Science Foundation Ireland. We thank Martin Gysel for valuable discussions.

Ancillary

Advertisement