2.1.1 Surface aerosol properties

The optical properties of surface aerosols were measured by a suite of instruments making up the Aerosol Observing System (AOS), which is the primary ARM platform for in situ aerosol observations. The TSI-3010 condensation particle counter was used to obtain the total number concentration of condensation particles (N_a) with diameters larger than 10 nm and smaller than 3 µm. A TSI-3653 nephelometer and a Radiance Research particle soot absorption photometer (PSAP) measured the scattering (σ_s) coefficients at three wavelengths (450, 550, and 700 nm) and the absorption (σ_a) coefficients at three wavelengths (470, 528, and 660 nm), respectively, of total (≤10 µm) and fine-mode (≤1 µm) aerosol particles (Jefferson, 2011). Nephelometer and PSAP measurements have undergone calibration and quality control using the methods developed by Anderson and Ogren (1998) and Anderson et al. (1999), respectively. Measurements of σ_a at 470 nm were normalized to 450 nm to match σ_s measurements. The single-scattering albedo (SSA) of surface aerosol particles is then calculated as σ_s/(σ_s+σ_a) using σ_s and σ_a at 450 nm. The time resolution of the N_a, σ_s, and σ_a measurements is 1 min.

The aerosol size distribution ranging from 15 to 450 nm was measured by a scanning mobility particle sizer (SMPS) with 5 min averaging. The SMPS contains a cylindrical differential mobility analyzer (TSI Inc., model 3081) and a condensation particle counter (TSI Inc., model 3010) and is calibrated using polystyrene latex standards (Wang et al., 2003). An aerosol chemical speciation monitor (ACSM) measured the bulk chemical composition of the nonrefractory components of submicron (aerodynamic diameter ranging from ∼40 to 1000 nm) aerosol particles (organics, sulfate, nitrate, ammonium, and chloride). The ACSM is a thermal vaporization electron impact ionization mass spectrometer built upon the same technology as the widely used aerosol mass spectrometer. Under ambient conditions, the detection limit of the mass concentration of particles is less than 0.2 µg m⁻³ for 30 min signal averaging. The ACSM is calibrated with ammonium nitrate following the method of Ng et al. (2011).

2.1.2 Columnar aerosol properties

Columnar aerosol optical depths (AODs) and Ångström exponents (AEs) were obtained from the Aerosol Robotic Network (AERONET) database (Holben et al., 1998). AODs are retrieved from direct Sun measurements with an uncertainty of 0.01–0.02 (Eck et al., 1999). This study uses Level 2.0 quality-assured and cloud-screened data.

2.2.1 Cloud boundaries

Cloud-base and cloud-top heights were identified using a combination of observations from the 95 GHz W-band ARM cloud radar (WACR), the micropulse lidar (MPL), and the ceilometer (Kollias et al., 2007). The algorithm used in the cloud boundary retrieval is similar to the method developed by Clothiaux et al. (2000) based on 35 GHz millimeter cloud radar observations. Cloud and precipitation masks are determined from the WACR based on the signal-to-noise ratio thresholds determined for each time profile. An MPL cloud mask is combined with ceilometer cloud-base estimates to produce a best-estimate cloud base for each time point. The MPL and WACR cloud masks are then merged with an additional filter to remove insect returns in the lower troposphere. Insects are identified using a combination of the WACR linear depolarization ratio and reflectivity measurements. The temporal and vertical resolutions of the cloud boundary product are 5 s and 42.856 m, respectively. Cloud-base and cloud-top heights were temporally averaged to generate data at a 1 min time resolution. The cloud-base and cloud-top height uncertainties are ∼7.5 and ∼45 m, respectively (Zhao et al., 2012a; Garrett and Zhao, 2013).

2.2.2 Cloud microphysical properties

A two-channel narrow-field-of-view (NFOV) radiometer and a microwave radiometer profiler (MWRP) generated cloud optical depth (COD) and LWP retrievals. The cloud droplet effective radius (r_e) was calculated using the following equation:

where ρ_w is the density of liquid water and τ is the COD in the visible. The NFOV radiometer with a 5.7^∘ field of view measuring downwelling zenith radiances at 673 and 870 nm at a 1 s time resolution is used to retrieve COD using the method described by Chiu et al. (2010) and Liu et al. (2013). Simultaneous highly accurate AERONET sun-photometer-measured radiances (Holben et al., 1998) quantified the biases in the NFOV radiance measurements (Fig. 1). AERONET and NFOV radiances agree well at 673 and 870 nm (coefficient of correlation, r, equal to 0.99 in both cases). However, NFOV-measured zenith radiances at 673 nm are underestimated by ∼15 %. Consequently, NFOV measurements at 673 nm were adjusted using the following formula:

where F_673,obs represents measured zenith radiances and F_673,adj represents adjusted radiances at 673 nm. The total uncertainty in COD retrievals using this method is ∼17 % (Chiu et al., 2010). The COD retrievals were averaged to generate data at a 1 min resolution for matching the time resolution of the LWP retrievals. The MWRP built by the Radiometrics Corporation measures atmospheric brightness temperatures at 12 frequencies. LWPs were retrieved using brightness temperatures measured at the five K-band channels (22.235, 23.035, 23.835, 26.235, and 30.0 GHz) with a 1 min time resolution based on the statistical retrieval algorithm developed by Liljegren and Lesht (2004). Typical uncertainties in LWP retrievals from microwave radiometers are ∼20 g m⁻² for LWP < 200 g m⁻² and ∼10 % for LWP > 200 g m⁻² (Liljegren and Lesht, 2004; Dong et al., 2008).

Figure 1CIMEL sun-photometer-measured radiance as a function of narrow-field-of-view (NFOV) radiometer-measured radiance at 673 nm (black dots) and 870 nm (gray dots). The diagonal line represents the 1:1 line. Units are watts per steradian per squared meter (W sr⁻¹ m⁻²). The legend gives the coefficient of correlation (r) and the slope of the best-fit linear regression line through each dataset.

Download

This study considers only nonprecipitating, low warm clouds with cloud-top heights less than 3 km. LWP observations less than 40 g m⁻² and greater than 300 g m⁻² were excluded to avoid very thin clouds, broken cloud cover, postprecipitation conditions (McComiskey et al., 2009), and potential precipitation contamination (Dong et al., 2008).

3.1.1 Seasonal variations in aerosol optical properties and number concentration

Figure 2 shows monthly statistics describing surface-measured σ_s for total (σ₁₀) and fine-mode (σ₁) aerosol particles and N_a. Table 1 summarizes their seasonal and annual mean values. Maxima in σ₁ and σ₁₀ are found in the summer months, and minima in σ₁ and σ₁₀ are found in the winter months. Fine particles dominate aerosol scattering in the summertime and are responsible for ∼75 % of the total particle scattering. The contribution of fine particle scattering to total particle scattering in other seasons ranges from ∼46 % to ∼54 %, indicating that particles with diameters ≤1 µm and ranging from 1 to 10 µm play a similar role in aerosol scattering extinction. Monthly and seasonal variations in N_a show that maximum and minimum seasonally mean N_a values occur in spring and autumn, respectively, inconsistent with the variations in aerosol scattering coefficient. This inconsistency is probably due to seasonal differences in aerosol particle size distribution and chemical composition since aerosol extinction properties depend strongly on particle size and chemical composition. The total particle SSA shows a slight seasonal variation, suggesting smaller changes in aerosol particle absorption properties. Figure 3 shows monthly statistics describing columnar AOD and AE. Table 1 summarizes their seasonal and annual mean values. The variations in AOD and AE are consistent with the variations in surface-measured σ_s and the ratio σ₁∕σ₁₀, indicating that surface aerosol properties can represent columnar aerosol properties very well at this site. Figure 4 shows monthly mean wind speeds and wind directions during the campaign period. Monthly mean wind speeds ranged from ∼3.8 to 6.6 m s⁻¹, and southwesterly winds dominated throughout the whole year over the area. Months in summer and winter with the strongest mean surface wind speeds (e.g., June and January/February, respectively) are generally times when the contribution of fine particles to the total scattering extinction is small.

Figure 2Monthly variations in (a) aerosol scattering coefficient at 450 nm (σ₄₅₀) for total (in blue, particle diameter, D_p, less than 10 µm) and fine-mode (in red, D_p less than 1 µm) aerosol particles and (b) aerosol particle number concentration (N_a). Box-and-whisker plots include median values (horizontal lines inside boxes), 25th and 75th percentiles (ends of boxes), 5th and 95th percentiles (ends of whiskers), and mean values (black dots). Months from left to right start at July 2012 and end at June 2013.

Download

Figure 3Monthly variations in columnar (a) aerosol optical depth at 440 nm (AOD₄₄₀) and (b) Ångström exponent (AE). Box-and-whisker plots include median values (horizontal lines inside boxes), 25th and 75th percentiles (ends of boxes), 5th and 95th percentiles (ends of whiskers), and mean values (black dots). Months from left to right start at July 2012 and end at June 2013.

Download

Table 1Seasonally averaged aerosol properties during the campaign period.

σ₁: scattering coefficient, fine-mode particles; σ₁₀: scattering coefficient, total; N_a: aerosol number concentration; SSA: single-scattering albedo; AOD₄₄₀: aerosol optical depth at 440 nm; AE: Ångström exponent.

Download Print Version | Download XLSX

Figure 4Monthly mean (a) wind speed (Wspd) and (b) wind direction (Wdir) during the campaign period. Months from left to right start at July 2012 and end at June 2013.

Download

3.1.2 Aerosol optical properties under different air-mass conditions

Table 2 gives the discrepancies in aerosol properties when different air masses are in place over the site. The mean value of σ₁ is the largest/smallest under continental/marine air-mass conditions. However, σ₁₀ is the largest under cluster III conditions and has similar values under cluster I and II conditions. The inconsistent variations in σ₁ and σ₁₀ under different air-mass conditions can be explained by dominant particle size as indicated by the ratio σ₁∕σ₁₀. When the continental air mass influenced the site, fine particles dominated aerosol scattering and were responsible for ∼65 % of the total particle scattering, indicating that more anthropogenic aerosols with small particle sizes were transported to the site from continental regions to the west. The values of σ₁∕σ₁₀ under cluster II and III air-mass conditions show that both fine-mode and coarse-mode particles played similar roles on the total particle scattering. The variation in N_a is consistent with that in σ₁ with the largest and smallest values under cluster I and II conditions, respectively. Smaller SSA values are found under continental air-mass conditions, suggesting that more absorbing aerosols were present in this air mass than in other air masses due to the anthropogenic influence. AOD values in each air mass are similar, and the variation in AE is consistent with the variation in the ratio σ₁∕σ₁₀.

Table 2Means and standard deviations of aerosol and cloud properties for each air-mass cluster.

σ₁: scattering coefficient, fine-mode particles; σ₁₀: scattering coefficient, total; N_a: aerosol number concentration; SSA: single-scattering albedo; AOD₄₄₀: aerosol optical depth at 440 nm; AE: Ångström exponent; COD: cloud optical depth; LWP: liquid water path; DER: cloud droplet effective radius.

Download Print Version | Download XLSX

3.1.3 Aerosol chemical composition and size distribution

Figure 5 shows the size distribution and the corresponding mass fraction of organics, sulfate, ammonium, and nitrate of surface aerosol particles sampled in July and August 2012. New-particle-formation and growth periods were detected and are outlined by red rectangles in the figure. During the measurement period, fine particles containing more organics were dominant with a mean particle radius of 91.4±20.6 nm and a mean organic mass fraction of 0.67±0.16. Mean mass fractions of sulfate, ammonium, and nitrate are 0.18±0.11, 0.10±0.09, and 0.04±0.02, respectively. At the beginning of new-particle-formation and growth periods, organics contributed the most to small particle sizes. Their contribution decreased as the growth period progressed to be replaced by contributions from inorganics, in particular sulfate. This is possible because sulfate ions are formed during nucleation involving neutral gaseous species like ammonia and sulfuric acid (Crilley et al., 2014). Small aerosol particles generally contribute more organics to the total aerosol mass over the study site, which can also be seen in the relation between mean aerosol particle radii and organic mass fraction (Fig. 6). The strong decrease in aerosol particle size with increasing organic mass fraction has also been reported by others (Broekhuizen et al., 2006; McFiggans et al., 2006).

Figure 5Time series of (a) particle size distribution and (b) mass fractions of organics (org, dark blue), sulfate (${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$, cyan), ammonium (${\mathrm{NH}}_{\mathrm{4}}^{+}$, yellow), and nitrate (${\mathrm{NO}}_{\mathrm{3}}^{-},\mathrm{red}$) in aerosols sampled during July and August of 2012. Dashed red rectangles outline periods of new particle formation and growth.

Download

Figure 6Mean aerosol particle radius (D_p) as a function of organic mass fraction (F_O). The black line is the linear regression line through all F_O bins. The gray line is the linear regression line for F_O bins ranging from 0.4 to 0.9, which have the most samples. Data are from July and August of 2012. The legend gives the coefficient of correlation (r) and the statistical probability (P).

Download

3.2.1 Aerosol effects on the dependence of cloud properties on meteorological conditions

Low-warm-cloud properties are sensitive to changes in thermodynamic conditions (Su et al., 2010; Medeiros and Stevens, 2011; Liu et al., 2016). Figure 7 shows cloud properties (LWP and DER) as functions of LTS under low- and high-scattering-aerosol-index (AI) conditions for continental and marine air masses. The scattering AI here is used as a proxy for CCN (Liu and Li, 2014; Sena et al., 2016) and is defined as the product of surface-measured aerosol scattering coefficients and surface-measured scattering Ångström exponents. Low and high scattering AIs are defined as the lowest and highest quarter of all scattering AI samples, respectively. The cloud properties were averaged over each 6 K LTS bin from 0 to 30 K under low- and high-scattering-AI conditions. The differences in meteorological parameters (e.g., temperature, wind speed, and relative humidity) at the surface and at 850 hPa, and in large-scale dynamic (ω) and thermodynamic parameters (LTS), are not significant under low- and high-scattering-AI conditions (figure not shown). Table 2 summarizes the means and standard deviations of cloud properties under each air-mass condition. Clouds influenced by marine air-mass conditions (cluster II) have the largest COD, LWP, and DER (33.0±18.3, 243±197 g m⁻², and 10.9±6.6 µm, respectively), and clouds associated with the air mass from continental areas (cluster I) have the smallest cloud properties (COD = 25.7±14.5, LWP = 127±99 g m⁻², and DER = 7.9±4.8 µm). The top panels of Fig. 7 show that LWP significantly increases with increasing LTS under low-aerosol conditions, consistent with results from studies using surface-based measurements (e.g., Liu et al., 2016), satellite measurements (e.g., Su et al., 2010), aircraft measurements (e.g., Cecchini et al., 2016), and model simulations (e.g., Johnson et al., 2004; West et al., 2014). Johnson et al. (2004) showed that an increase in stability induces increases in the buoyancy of free-tropospheric air above the temperature inversion capping the boundary later, inhibiting the entrainment of dry air through the cloud top and increasing LWP as a result. Under high-aerosol conditions, LWP changes little as LTS increases. A likely reason is inhibited cloud droplet sedimentation due to the reduced cloud droplet size, enhancing evaporation and entrainment at the cloud top and reducing LWP (Kaufman et al., 2005; Hill et al., 2009; Zhao and Garrett, 2015; Liu et al., 2016). There are similar variations in DER with increasing LTS under low- and high-aerosol conditions (bottom panels of Fig. 7), i.e., increasing DER as LTS increases under less polluted conditions and almost constant DER as LTS increases under more polluted conditions. The changes in DER with LTS possibly reflect the changes in LWP with LTS due to the high positive correlation between LWP and DER (Zhang et al., 2011; Sporre et al., 2014). The enhanced LWP under highly stable conditions can supply the water needed for cloud droplet growth (Su et al. 2010; Zhang et al., 2011). The increase in LWP is also commonly accompanied by an increase in droplet collision–coalescence, resulting in a decrease in cloud number concentration, thus leading to an increase in DER (Kim et al., 2008; McComiskey et al., 2009; Liu et al., 2016; Zhao et al., 2018, 2019; Yang et al., 2019). Differences in LWP and DER between low- and high-LTS conditions are larger under low-pollution conditions than under high-pollution conditions. This suggests that high-aerosol concentrations can significantly weaken the thermodynamic influence on the increase in LWP and DER due to the aerosol perturbation. These results imply that the development of clouds in a highly polluted environment is inhibited even though the thermodynamic conditions may be the same as those in a much less polluted environment. The chances of precipitation are thus reduced because the rainfall frequency of warm, low clouds and LWP are highly correlated (Chen et al., 2011; Liu et al., 2013).

Figure 7Liquid water path (LWP) and cloud droplet effective radius (DER) as functions of lower tropospheric stability (LTS) at low (in blue) and high (in red) scattering aerosol index (AI) levels under cluster I air-mass (a, c) and cluster II air-mass (b, d) conditions. Low and high scattering AIs are defined as the lowest and highest quarter of all scattering AI samples, respectively.

Download

For all LTS bins, clouds under high-aerosol conditions have lower values of LWP and DER than clouds under low-aerosol conditions. The reduction in LWP and DER is greater in stable environments than in unstable environments, suggesting that clouds in stable environments are more affected by the aerosol perturbation than those in more unstable regimes. Studies on marine, warm clouds based on surface measurements have also shown this (Liu et al., 2016).

3.2.2 Aerosol effects on the relationships among cloud properties

Figure 8 shows the dependence of COD and DER on LWP under low- and high-scattering-AI conditions for air-mass clusters I and II. Under high-scattering-AI conditions, COD increases sharply as LWP increases, while, under low-scattering-AI conditions, COD changes little as LWP increases due to the decrease in DER influenced by the aerosol perturbation (Fig. 8a and b). Figure 8c and d suggest that the DER is sensitive to LWP. An increase in LWP leads to a significant increase in the size of cloud droplets (Zhang et al., 2011; Sporre et al., 2014). The increase in DER with LWP is more rapid under low-scattering-AI conditions than under high-scattering-AI conditions. This is because there is a limit to the size a cloud droplet can reach when a given amount of water is shared among a large number of particles (Zhang et al., 2011). High-aerosol-loading conditions weaken the increase in DER and strengthen the increase in COD as LWP increases, indicating that aerosols have an impact on the COD–LWP and DER–LWP relationships.

Figure 8Cloud optical depth (COD) and cloud droplet effective radius (DER) as a function of liquid water path (LWP) at low (in blue) and high (in red) scattering aerosol index (AI) levels under cluster I air-mass (a, c) and cluster II air-mass (b, d) conditions. Low and high scattering AIs are defined as the lowest and highest quarter of all scattering AI samples, respectively.

Download

Figure 8 also shows that, across all LWP bins, COD is larger and DER is smaller under high-scattering-AI conditions than under low-scattering-AI conditions, consistent with the Twomey effect. The large differences between COD under low- and high-scattering-AI conditions at high LWP values (Fig. 8a and b) and between DER under low- and high-scattering-AI conditions at high LWP values (Fig. 8c and d) suggest that when clouds have large LWPs, aerosols further inhibit the growth of cloud droplets. Under high-aerosol-loading conditions, more aerosol particles are activated into CCN, resulting in a rapid increase in cloud droplet concentration as LWP increases. However, under low-aerosol-loading conditions, cloud droplet concentrations increase slowly as LWP increases due to the lack of a CCN source, so the size of cloud droplets increases rapidly as LWP increases (Zhang et al., 2011).

3.3.1 Variations in cloud properties with aerosol loading under different air-mass conditions

Figure 9 shows the probability distribution functions (PDFs) of COD, LWP, and DER under low- and high-scattering-AI conditions for air mass clusters I and II. Numbers written in each panel are the mean percentage differences in each cloud property defined as $(M_{\mathrm{ch}}-M_{\mathrm{cl}})/M_{\mathrm{cl}}\ast \mathrm{100}$ %, where M_c represents the mean value of a cloud property and subscripts h and l represent high and low scattering AI levels, respectively. The PDFs of COD, LWP, and COD under high- and low-scattering-AI conditions differ significantly for both air masses. Although the peak values of COD under low- and high-aerosol-loading conditions are similar, clouds under more polluted conditions have more large values of COD than those under less polluted conditions. There are 24.2 % and 21.9 % increases in COD for cluster I and II respectively. For the low-aerosol-loading case, the PDF of LWP shows a broad maximum with values from 50 to 180 g m⁻² (Fig. 9c) and 80 to 230 g m⁻² (Fig. 9d) for cluster I and II air masses, respectively. The high-aerosol-loading cases, conversely, have narrower PDFs with distinct peaks between 60 and 70 g m⁻². Under high-scattering-AI conditions, the LWP decreases on the order of 30 % and 45 % from the values under low-scattering-AI conditions for the cluster I and II air masses, respectively. Under both air-mass conditions, there is a sharp shift in DER toward smaller values when going from high-aerosol-loading conditions to low-aerosol-loading conditions. Under low-aerosol-loading conditions, the DER values show a broad range with generally higher values varying between 5 and 12 µm for the cluster I air mass and peaking around 15 µm for the cluster II air mass. Under high-aerosol-loading conditions, the PDF of DER for both air masses is significantly narrower, with most of the values less than 10 µm and with peak values around 5 µm. From low- to high-aerosol-loading conditions, DER decreases by ∼40 % (for cluster I) and ∼55 % (for cluster II). As mentioned before, whether low- or high-scattering-AI conditions are in place, meteorological parameters and large-scale dynamic and thermodynamic parameters show little difference, suggesting that aerosols are responsible for the changes in cloud properties. In general, clouds in a marine air mass have slightly larger decreases in LWP and DER from low to high aerosol loading than those in a continental air mass.

Figure 9From top to bottom, probability distribution functions (PDFs) of cloud optical depth (COD), liquid water path (LWP), and cloud droplet effective radius (DER) at low (L, in blue) and high (H, in red) scattering aerosol index (AI) levels for the cluster I air mass (a, c, e) and the cluster II air mass (b, d, f). The mean percentage difference in COD, LWP, and DER between high and low scattering AI levels (calculated as [(H–L/L)]*100 %) is given in each panel.

Download

3.3.2 Aerosol first indirect effect

The aerosol FIE is generally quantified as

where α represents CCN or CCN proxies. The FIE represents the relative change in mean cloud DER with respect to a relative change in aerosol loading for clouds having the same LWP (Feingold et al., 2003). In some studies, the scattering AI is used as the CCN proxy (Liu and Li, 2014). Cloud samples were categorized according to their LWP values. The LWP bins range from 40 to 200 g m⁻² in increments of 20 g m⁻². The choice of a small increment ensures that the LWP constraint is met in each bin. Since there were not enough samples under cluster II air-mass conditions, only FIE for clouds and aerosols under cluster I air-mass conditions are calculated. Only those cases with sample numbers greater than 50 per bin where the calculated values of FIE are statistically significant at the 95 % confidence level (P=0.05) are analyzed here. Figure 10a shows DER as function of scattering AI for clouds with LWP ranging from 120 to 140 g m⁻². It illustrates how the FIE is estimated. There is a significant decrease in DER as the scattering AI increases. For this case, the magnitude of the FIE is 0.26 with an uncertainty of 0.09. Figure 10b shows the magnitudes and uncertainties of FIE calculated in each LWP bin. Numbers above each bar are the number of samples that went into the calculation of the FIE in each LWP bin. The magnitude of the FIE changes from 0.07±0.03 to 0.26±0.09 with the smallest value found in the LWP bin of 40–60 g m⁻² and the largest value found in the LWP bin of 120–140 g m⁻². The mean value of FIE during the study period based on all LWP bins is 0.16±0.06. The values of FIE in each LWP bin increase with increasing LWP, especially for LWPs less than 140 g m⁻². This is consistent with results from previous studies (e.g., Pandithurai et al., 2009; Sporre et al., 2014; Harikishan et al., 2016). Enhanced aerosol activation due to the increase in LWP may explain this (Zhao et al., 2012b; Painemal and Zuidema, 2013). At higher LWP values, with the availability of more CCN, more droplets can activate. The droplet number increases, but their size decreases at fixed LWP levels (Harikishan et al., 2016). Estimates of the FIE reported from all available platforms range widely and are sensitive to the definition of the aerosol burden (Lihavainen et al., 2010; Zhao et al., 2012b), the methods for retrieving cloud properties (McComiskey et al., 2009), and meteorological conditions such as vertical velocity and atmospheric stability (Feingold et al., 2003; Matsui et al., 2004; McComiskey et al., 2009; Liu et al., 2016; Yang et al., 2019). Theoretical values of the FIE lie between 0 and 0.33 (McComiskey and Feingold, 2008) with most values falling between 0.05 and 0.25 (Zhao et al., 2012b). Based on surface retrievals, Feingold et al. (2003) derived FIE values of 0.02–0.16 with a mean value of 0.10±0.05 for a set of seven cases. In a study using 3 years of data from the United States Southern Great Plains, Kim et al. (2008) found that FIE values ranged from 0.04 to 0.17 over five LWP bins with a mean value of 0.09±0.05. Other estimates of FIE based on surface retrievals have been reported, e.g., 0.07±0.01 for warm, marine boundary-layer clouds over the Azores (Liu et al., 2016); 0.14±0.09 for continental clouds during the monsoon period over a rural continental site in Mahabubnagar, India (Harikishan et al., 2016); and a range of 0.05 to 0.16 over the coastal region at Pt. Reyes, California (McComiskey et al., 2009). The magnitude of the FIE in this study generally falls in this range.

Figure 10(a) Cloud droplet effective radius (DER) as a function of scattering aerosol index (AI) for a sample bin with a constant liquid water path (LWP) range of 120 to 140 g m⁻², and (b) the quantified aerosol first indirect effect (FIE) for each LWP bin. Numbers above each bar in (b) are the number of samples that went into the calculation of the FIE.

Download

Examined next is the sensitivity of cloud properties to aerosol chemical composition represented by the mass fraction of organics. The aerosol number concentration is used as the CCN proxy here (Li et al., 2011; Yan et al., 2014; Liu et al., 2016) because aerosol scattering coefficient measurements were not taken during the aerosol chemical composition observation period. Three LWP bins were defined: 40–60, 60–80, and 80–100 g m⁻². Figure 11 shows DER as a function of N_a in each LWP bin when aerosol particle mass fractions of organics are low and high. Aerosols with low and high mass fractions of organics are defined as aerosols with mass fractions of organics less than and greater than, respectively, the mean value of the mass fraction of organics of all samples in each LWP bin. Mean values of ω and LTS in each aerosol particle mass fraction of organics category are given in the figure. Differences in ω and LTS between low and high mass fractions of organics are not significant in any LWP bin. FIE estimates when aerosol samples with low mass fractions of organics dominate are 0.10±0.05, 0.15±0.06, and 0.23±0.12 (Fig. 11a–c, respectively), which are greater than the FIE estimates when aerosol samples with high mass fractions of organics dominate (0.07±0.04, 0.12±0.06, and 0.07±0.05, respectively). Clouds are more susceptible to inorganics-dominant aerosol than to organic aerosols, resulting in a greater climate forcing. The mechanism behind the aerosol indirect effect is characterized by the ability of aerosol particles to act as CCN, which is primarily governed by particle size and chemical composition (McFiggans et al., 2006). The cloud-nucleating ability of aerosol particles is significantly greater when the aerosol particles are larger and composed of more inorganic compounds than when they are small and composed of more organic compounds (Dusek et al., 2006; Liu et al., 2011). This study (see Fig. 5) and others have demonstrated that aerosols containing more organic particles are generally smaller than those with more inorganic particles (Broekhuizen et al., 2006; McFiggans et al., 2006; Zhang et al., 2011) and that organic particles are generally less CCN-active than inorganic particles (Raymond and Pandis, 2002; Zhang et al., 2011). This can partly explain the smaller FIE values induced by aerosols with large mass fractions of organics.

Figure 11Cloud droplet effective radius (DER) as a function of aerosol number concentration (N_a) at low (in blue) and high (in red) levels of mass fraction of organics in three liquid water path (LWP) bins: (a) 40–60 g m⁻², (b) 60–80 g m⁻², and (c) 80–100 g m⁻². Linear regression lines through each set of data are drawn. F_o,l and F_o,h are defined as the means of values less than and greater than, respectively, the mean value of the mass fraction of organics from all samples in each LWP bin. The legend gives the mean values of F_o,l and F_o,h with their standard deviations and the magnitudes of the FIE with their uncertainties. Mean values of vertical velocity (ω) and lower tropospheric stability (LTS) corresponding to F_o,l and F_o,h levels in each LWP bin are also given.

Download