Info

where esw is the saturation vapor pressure and D'v is the diffusion coefficient Dv corrected with the gas kinetic effect:

in which a is the mass accommodation coefficient, and v and A are the mean thermal velocity and the mean free path of water vapor, respectively. The gas kinetic effect (also called the "vapor jump effect") arises because within one mean free path close to the droplet surface, the

Rv T a r transport of water vapor (as well as heat conduction) is considered to proceed not by diffusion but by the gas kinetic process. FFCs retard droplet growth through the influence on a, which represents the probability of a water vapor molecule staying when it hits the droplet surface. From Eq. (8) one can see that the effect of a on diffusion is most significant when either the droplet size or a is small, and thus the FFC is able to retard condensation/evaporation by drastically reducing the value of a.

Past studies (e.g. Derjaguin et al., 1985; Feingold and Chuang, 2002) often assumed that the effect of FFCs on a occurs only when the surface coverage by the FFCs exceeds a threshold value. So the value of a is either 0.035 for pure water (aw) or for full FFC

coverage (aH). However, Chen and Hsieh (2004) re-examined the experimental data in relevant articles (e.g., Archer and La Mer, 1955; Rubel and Gentry, 1984; Seaver et al., 1992) and suggested that the effect should still exist under partial coverage, as shown in Fig. 14. Based on molecule flux theory, they developed a new method to describe the relationship between a and FFC coverage, then fitted the experimental data of a with extrapolation to that of pure water at zero coverage.

The above parametrization was incorporated into a detailed microphysical parcel model of Chen and Lamb (1994) to simulate the FFC effects. For a mass fraction more than just 1%, FFCs may drastically retard the condensation growth and elevate Sa. The more FFCs there are, the higher Sa may reach. However, CDNC does not respond so linearly. Intuitively, a higher value of the peak Sa should allow more smaller aerosols to be activated. Yet, the retardation of condensation alters (lengthens) the characteristic time of growth such that some of the aerosols whose critical Sd is below the peak Sa (and may be activated) do not have time to grow to their critical size before Sa starts to taper off. An analogy is that someone buys an airplane ticket but does not arrive before the gate closes.

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Surface coverage, 0

Figure 14. Parametrization of the overall mass accommodation coefficient as a function of FFC coverage. Filled triangles are the experimental data from Rubel and Gentry (1984), which show a sharp transition of FFCs from a liquid film to a solid film under surface coverage exceeding about 97.5%; whereas circles are data from La Mer et al. (1964), which show the state of the solid film extending to lower coverage. The dotted curve is a fitting of the overall a for FFC coverage up to 97.5%, above which we assume that a is at a solid state and aH = 2 X 10-5.

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Surface coverage, 0

Figure 14. Parametrization of the overall mass accommodation coefficient as a function of FFC coverage. Filled triangles are the experimental data from Rubel and Gentry (1984), which show a sharp transition of FFCs from a liquid film to a solid film under surface coverage exceeding about 97.5%; whereas circles are data from La Mer et al. (1964), which show the state of the solid film extending to lower coverage. The dotted curve is a fitting of the overall a for FFC coverage up to 97.5%, above which we assume that a is at a solid state and aH = 2 X 10-5.

The above effect can be seen from the evolution of the droplet size spectrum shown in Fig. 15. For normal conditions (i.e. without the FFC effect), the initially continuous size spectrum breaks up into two segments with distinct gaps during and after the activation stage [Fig. 15(a)]. On the left are the unacti-vated haze particles, which swell only a little bit initially; on the right are those activated into cloud drops, which continue to grow with time. The cutoff size of activation here is around 0.1 ¡m. For the case with 1% FFCs [Fig. 15(b)], the cutoff size is reduced to 0.07¡m because of the elevated Sa, and so more smaller droplets are activated. However, taking the spectrum at 600 s for example, the largest droplets do not seem to grow at all, let alone become activated. But why can the smaller droplets grow under the influence of FFCs? The reason is that the dilution rate (mass acquired compared to existing mass) is faster for smaller droplets. Also, the amount of the FFC in each droplet is

Figure 15. The evolution of droplet size distribution (DSD) as a function of the corresponding droplet radius for (a) no FFC effect and (b) an FFC mass fraction of 1%. Each dot represents the number concentration of droplets in each size bin. The black circle, light-gray square and dark triangle represent the spectra at the initial time, 600 s and 1200 s of simulation.

Figure 15. The evolution of droplet size distribution (DSD) as a function of the corresponding droplet radius for (a) no FFC effect and (b) an FFC mass fraction of 1%. Each dot represents the number concentration of droplets in each size bin. The black circle, light-gray square and dark triangle represent the spectra at the initial time, 600 s and 1200 s of simulation.

fixed, and initially is enough to form more than one layer of surface film for this case. At some point the slow swelling will eventually break off the FFC coverage and induce stronger condensation and thus accelerated dilution. As the dilution process is faster for smaller droplets, only the smaller aerosols are able to grow. Given a longer time, as shown in the spectrum at 1200s simulation time, the smaller of the previously un-activated larger droplets were able to break through the FFC shielding and try to catch up in growth. Some of the remaining ones will never have the chance of activation, because the Sa falls off rather fast to a value that is below their critical Sd. Note that FFCs may enter hygroscopic aerosols by deposition from the gas phase or by coagulation. So, theoretically, their mass fraction may be proportional to the surface area or mass of the aerosol, respectively, and the calculations for Fig. 15 assume the former.

The FFC effect on the CDNC can be summarized as: (1) the retarded condensation growth elevates the ambient supersaturation, thus reducing the activation cutoff size and increasing the CDNC; (2) large aerosols cannot break off the FFC shielding to grow into cloud drops, causing a decrease in the CDNC. Which effect is stronger depends on the FFC content and aerosol size distribution, as well as various environment parameters. Chen and Hsieh (2004) performed a comprehensive calculation of the change of CDNC as a function of FFC content and updraft speed for different aerosol size distributions. In general, for maritime aerosols containing FFCs, the influence on the CDNC is negative for high FFC contents, and positive for low FFC contents. The maximum change is a-few-ten percent. For urban aerosol distributions, the changes in the CDNC are all positive, reaching over 200% in some situations.

Besides modifying the CDNC by directly influencing the activation process, there are other indirect effects that FFCs may have on cloud microphysics and chemistry. First of all, they change the size distribution of cloud drops. From Fig. 15 one may notice the significant reduction of large cloud drops when the FFC is in operation. Those large cloud drops are important to the initiation of rain. They also have the function of decreasing the CDNC by accretion. In this regard, the FFC yields consequences equivalent to Twomey's first and second indirect effects. From the figure one may also realize that, with the FFC effect, smaller CCNs will replace large CCNs to activate into cloud drops. This means that there are less hygroscopic materials in cloud water, which may have a substantial effect on the cloud chemistry. Furthermore, the FFC retards not only the condensation and evaporation of cloud drops but also the mass exchange (absorption/desorption) of other gases. This again has potential influences on cloud chemistry and acid rain formation.

Surfactants can also affect cloud drop activation through the influence on water activity (Raymond and Pandis, 2002) or surface tension (Facchini et al., 1999a; Anttila and Kerminen, 2002). Charlson et al. (2001) reported measurements on samples of cloud water, which showed a large decrease in the surface tension due to surface-active organics. They suggested that if a large surface tension depression occurs in cloud droplets near the critical size for activation, it will lead to an increase in droplet concentration and hence in cloud albedo. They estimated that if there is such an effect on all stratus clouds, a global mean forcing of almost — 1Wm-2 will arise. The change of surface tension may also affect other microphysical processes. Garrett (1978) suggested that partially covered surface film tends to concentrate at the downwind portion of the free-falling droplet surface and thus gives rise to a surface tension gradient, which then influences internal circulation. This is another way to affect mass transfer across the phase boundary and the chemical reactions inside the droplets. The influence of surfactants on surface tension may also affect the dynamics of droplet coalescence and breakup during collision (Ryan, 1976), processes that are very important to rain formation.

2.2.4. Bioaerosols

Primary biogenic organic aerosols (hereafter called "bioaerosols") include whole organisms (e.g. bacteria, fungi and phytoplankton), reproductive materials (e.g. pollen) and fragments (e.g. plant waxes). Air masses that are influenced in rural areas generally have large amounts of giant biological particles such as pollen spores, whereas in urban and industrial air there tend to be higher concentrations of smaller biological particles such as bacteria (Matthias-Maser and Jaenicke, 1995). Over the ocean, primary production results from the ejection and dispersion of saltwater droplets from bursting bubbles at the sea surface (Woodcock, 1953). Matthias-Maser et al. (2000) suggested that the proportion by volume of atmospheric particles made up by biological material in remote continental, populated continental and remote maritime environments is respectively 28%, 22% and 10%.

Bioaerosols may lead to the formation of ice crystals by serving as INs (Schnell and Vali, 1973; Hazra et al., 2006), among which bacteria are possibly the most abundant and effective in ice nucleation. Decaying plant leaves can also act as INs, but the ice nucleation has been identified to be of bacterial origin. In fact, some plant frost injury has been shown to involve an interaction of certain leaf-surface bacteria that cause the frost-sensitive plants on which they reside to become more susceptible to freezing damage (Lindow, 1983). The worldwide availability of such nuclei was established by finding ice-forming nuclei in plant litters collected in different climatic zones. The studies by Vali et al. (1976), Levin et al. (1987) and Hazra et al. (2004) showed that biogenic INs might be released from the Earth's surface to the atmosphere and be active in initiating ice formation at temperatures as high as —2°C. Interestingly, some bacteria also serve as CCNs (Bauer et al., 2003), and have indeed been observed in the active form in clouds (Sattler et al., 2001).

Here, we demonstrate the effect of bacteria on precipitation formation using the model of Cheng et al. (2007), but add in the mechanism of ice nucleation from bacteria. The rate equation of bacteria ice nucleation is derived from the experimental data of Hazra et al. (2004) for a common bacterium — Pseudomonas aeruginosa. A typical bacteria concentration of

20 per liter is applied to the whole model domain. For comparison, we also apply the heterogeneous (deposition-condensation) nucleation rate from Hoffman (1973) to represent natural ice nucleation. A springtime cold-front system passing through northern Taiwan is selected for simulation. Since bacteria INs are effective at high subzero temperatures, one would expect more ice particles to be formed in the cloud. But there seems to be less snow and graupel in the case of bacteria INs (not shown). In fact, the formation of ice from bacteria is so effective that ice particles formed from them largely fall and melt to form raindrops. Interestingly, although bacteria cause more rainwater in the air, the precipitation intensity on the ground is actually weaker before the middle of the second day, as shown in Fig. 16. This is because there are so many raindrops formed in the bacteria IN case that each raindrop becomes much smaller. Smaller raindrops not only fall slower (hence lower precipitation intensity) but also evaporate faster. Nevertheless, bacteria INs seem to be able to produce precipitation with a similar order of magnitude to that by natural INs. Furthermore, toward the end of the second day, rain formation in the bacteria INs case seems to pick up, and produces more rain on the ground. The 48-hour cumulative rainfall of 67.0 mm is, in fact, slightly

Time, hour

Figure 16. Time series of surface precipitation intensity simulated with natural INs (solid) and bacteria INs (dashed).

Time, hour

Figure 16. Time series of surface precipitation intensity simulated with natural INs (solid) and bacteria INs (dashed).

higher than the 65.5 mm in the natural INs case. Note that we also run the simulation with all ice processes turned off, and find that warm rain processes are also effective but less prominent (49.2 mm rainfall in 48 hours) than ice processes.

Because of their strong ice-nucleating capability and large abundance, bioaerosols may play an important role in exerting control over cloud development and precipitation. In turn, the quantity and the type of bioaerosols are strongly influenced by clouds, which alter the availability of solar radiation for photosynthesis and surface temperature, as well as precipitation that provides the main water supply. Such a feedback process is very complicated and highly uncertain, but the potential importance cannot be ignored. As summarized by Barth et al. (2005), an investigation into this coupled cycle is necessary for a better understanding of the Earth system, including weather and climate change, regional and global atmospheric chemistry, as well as changes in land cover ecology.

The concentration of bacteria in the atmosphere depends on transportation from the surface boundary layer. The bacteria flux is closely connected with the reproduction rate of bacteria. Living and dead bacteria, including ice-nucleating species, have been found in clouds and fogs (Fuzzi et al., 1997), raindrops (Maki and Willoughby, 1978) and hailstones (Man-drioli et al., 1973). Bacteria have been observed in the boundary layer, in the upper troposphere (Lindemann et al., 1982), and even in the stratosphere at altitudes of up to 41 km above the sea surface (Wainwright et al., 2003). The flux of bacteria entering the atmosphere is recognized as originating from two types of temperate vegetation zones: (a) high-primary-production row crop areas and (b) relatively-low-production desert areas. The highest flux was 1.95 x 106 colony-forming units (CFU) m~2 h_1, measured in the crop area (Lindemann et al., 1982) and 1.7 x 103 CFU m~2 h^1 over the desert area (Lighthart and Sharffer, 1994). Interestingly, bacteria not only have capabilities as CCNs and INs, but may also play a role in the modification of other atmospheric OCs that act as CCNs. Herlihy et al. (1987) showed that the bacterial utilization of formic and acetic acid is in rainwater. Bacteria and fungi transformed dicarboxilic acid (DCA) efficiently in the boundary layer, with estimated degradation lifetimes comparable to those of major atmospheric oxidants (Ariya and Amyot, 2004). The authors also showed that different fungus species drive microbiological degradation of several atmospherically active OCs at different rates. This degradation is also a function of several environmental factors, namely pH, temperature and nutrient levels.

2.3. Mineral dust

Atmospheric mineral dusts originate mainly from deserts, semiarid areas and, to a lesser extent, cultivated lands, sandy seashores, river-banks and grasslands. These airborne dust particles (which are often termed "mineral aerosols," as well as "yellow dust" or "kosa" in some regions) have potential effects on cloud and precipitation formation, atmospheric radiative transfer, air quality and atmospheric chemistry, as well as ecology over land and in the ocean (Eppley, 1980; Duce, 1986; Guieu et al., 2002). Besides the air quality and ecological issues, a lot of dust studies emphasize the climate forcing aspect of mineral aerosols due to their ability to scatter and absorb solar radiation. The report of the IPCC (2001) specifically points out that the influence of mineral dust on radiative forcing can be both positive and negative, and ranges from — 0.6 to +0.4Wm~2. However, these are just the direct effects. Evidence is mounting that mineral dust may influence the climate indirectly by influencing cloud processes. There are also influences on the hydrological cycle through similar processes mentioned in previous subsections.

2.3.1. Roles as CCNs and INs

Mineral dust particles from the arid regions of the Asiatic continent have been implicated as possible heterogeneous INs, more than 40 years ago (Isono et al., 1959). More recently, emerging evidence has indicated that mineral particles may reach the upper troposphere, where they may serve as INs for cirrus cloud formation. In an aircraft campaign conducted over the Alps, Heintzenberg et al. (1996) found that minerals were common constituents in cirrus crystal residues. Upper-tropospheric INs activated in a diffusion chamber and subsequently collected by impaction (Chen et al., 1998) also had increased number fractions of crustal particles when compared with the ambient aerosol population. A lidar study by Sassen (2002) suggested that cirrus-like ice clouds form at the top of layers of transported Asian aerosols, at temperatures considerably warmer than the cli-matological mean temperature for midlatitude cirrus formation. Sassen et al. (2003) documented glaciations of altocumuli forming at the top of the dust layer at — 8°C during one of the same episodes. DeMott et al. (2003) detected concentrations of heterogeneous INs exceeding 1 — 3 cm-3, up to 100 times higher than typical background values in and above the marine boundary layer in Florida during Saharan dust episodes. These INs were within the air layer that feeds thunderstorm development and subsequent cirrus anvil formation. Measurements of high mineral dust fractions in residual particles from anvil cirrus during that study (Cziczo et al., 2004) and cloud model simulations of these cases (Van den Heever et al., 2006) all supported the hypothesis that the dust modified the cloud microphysics and dynamics.

Mineral dust particles may trigger ice nucleation via either the deposition mode or the freezing mode. The former mode forms ice by direct deposition of water vapor, while the latter mode usually occurs on dust particles coated with solutes, which absorb water and form a solution coating. Quantitative evidence of such ice nucleation comes best from laboratory efforts. Hung et al. (2003) studied ice freezing of aqueous ammonium sulfate particles containing mineral dust cores, and found that the heterogeneous nucleation rates vary from 102 to 106 cm-2 s-1, depending on the diameter of the dust core, the temperature and the solute concentration. They also rationalized the nucleation rates by the equations of classical heterogeneous nucleation theory. Archuleta et al. (2005) measured the ice nucleation behavior of metal oxides coated with sulfuric acid and found that heterogeneous freezing occurs at lower relative humidities than those calculated for homogeneous freezing of the diluted particle coatings. In addition, for all dust types the heterogeneous freezing rates increased with particle size for the same thermodynamic conditions, whereas for the same particle size, natural mineral dust particles were the most effective INs. Salam et al. (2006) reported the ice nucleation results of two dust types that are of potential atmospheric relevance. They showed that both kaolinite and montmorillonite act as efficient INs in the deposition/condensation nucleation mode, but the latter is somewhat more efficient. Also, the fraction of active ice nuclei increases with decreasing temperature and with increasing relative humidity. Mohler et al. (2006) analyzed mineral dust particles from three deserts — Arizona, Taklamakan and Sahara — and found that deposition ice nucleation was most efficient on Arizona dust particles. For all samples the ice-activated aerosol fraction could be approximated by an exponential equation as a function of Si, but more experimental work is needed to quantify the variability of the ice activation spectra as functions of the temperature and dust particle properties. Field et al. (2006) collected and analyzed desert dust samples from the Sahara and Asia, and found two nucleation events. The primary nucleation event occurred at ice saturation ratios of 1.1-1.3, and is likely to have been in the deposition nucleation mode. The secondary nucleation event occurred at ice saturation ratios between 1.35 and 1.5, and is likely to have occurred via condensation mode nucleation. The activated fractions of desert dust ranged from 5-10% at — 20° C to 2040% at temperatures colder than —40°C. But the authors did not find an obvious difference between the nucleation behaviors of the two dust samples.

The knowledge of heterogeneous ice nuclea-tion in precipitation formation has been utilized to artificially enhance precipitation. As early as World War II, Findeisen attempted a cloud-seeding flight using sand as an ice-nucleating agent, but apparently without success (Schaefer, 1951), which indicates the limited ice-nucleating ability of natural mineral dust. More effective agents were later developed and applied in weather modification; they include the commonly used silver iodine (Schaefer, 1946) and rarely applied but perhaps more effective organic agents such as metaldehyde (Fukuta, 1963) and phloroglucinol (benzene-1,3,5-triol) (Langer et al., 1963). Even bacteria have been suggested as artificial seeding agents because of their superb nucleation efficiency (see Subsec. 2.2.4).

Besides acting as INs, mineral dust may take on the role of CCNs. It is now more clearly known that mineral dust reacts with various trace gases, and through these reactions the cycles of various chemical constituents and mineral dust become linked (Dentener et al., 1996). Such processes may also affect cloud formation. For example, dust particles provide reactive sites for heterogeneous chemistry, which enhance the uptake of nitrate onto the dust particles (Phadnis and Carmi-chael, 2000; Grassian, 2002; Bauer et al., 2004; Krueger et al., 2004). The addition of nitrate gives dust particles a hygroscopic coating, which turns them into effective CCNs and thus impacts cloud formation. Hygroscopic coating on dust may also be attained through cloud-chemical and -microphysical processes, details of which are given in Subsec. 3.1. Rosenfeld et al. (2001)

noted the effect of Saharan dust in reducing precipitation in shallow convective clouds near the source. This was hypothesized to be due to lowering of the coalescence efficiency of clouds resulting from increases in dust particles that act as cloud condensation nuclei.

2.3.2. Effects on cloud chemistry

Although mineral dust aerosols are considered insoluble, they may still dissolve to a certain degree and release trace ingredients that affect chemical processes in cloud water and aerosol water. The most interesting ingredients include calcium and transition metals such as iron and magnesium. Calcium and minerals alike are base substances that can neutralize acid rain in a manner similar to the way antacids counteract excess acid in an upset stomach. Barnard et al. (1986) proposed that dust from unpaved roads, which contribute more than 90% of all open source calcium and magnesium in most of the US states, may be of potential importance to acid rain neutralization. Inoue et al. (1991) also suspected that calcite in the eolian dust may have been neutralizing acids in rainwater and snow. More quantitative evaluations were provided by Wang et al. (2002), who showed that the yellow sand in East Asia can strongly neutralize the acidic precipitation, with the highest seasonal increase of pH values by 0.1-0.4 in Japan, 0.5-1.5 in Korea, and more than 2 in northern China. Note that the elevation of pH values in turn affects many aqueous phase chemical reactions, as mentioned in Subsec. 3.1 and discussed below.

Mineral dust particles also impact the ecology of land and sea via the release of their trace elements. They may provide iron, which is known to exert a controlling influence on biological productivity in surface waters over large areas of the ocean, and may have been an important factor in the variation of the concentration of atmospheric carbon dioxide over the glacial cycle (Martin, 1990; Boyd et al., 2000).

The enhancement of phytoplankton activity through iron fertilization may also exert an influence on the DMS-cloud-climate interactions mentioned in Subsec. 2.1.1.

Dissolved iron (Fe3+) and magnesium (Mn2+) may also affect aqueous chemistry in cloud water by acting as a catalyst in the oxidation of SO2 by O2 to produce sulfate (see Seinfeld and Pandis, 2006; p. 314). There are further interesting links between iron and sulfate. Iron in mineral dust mostly exists in the form of Fe(III), whose low solubility limited its involvement in either cloud chemistry or marine biological fertilization. There are a series of reactions cycling iron between the Fe(III) and Fe(II) states that are pH-dependent (Zhuang et al., 1992; Deutsch et al., 2001). In fact, iron can be significantly mobilized below a threshold pH of ^3.6 or above a pH value of 7.1 (Mackie et al., 2005). In summary, Fe(III) affects the oxidation of SO2 into sulfate, the production of sulfate decreases pH, and pH affects the cycling of iron as well as the oxidation of SO2. The main source of SO2 in the marine area is the production of DMS by phytoplankton, whose growth may be accelerated by the input of Fe(II) which is mobilized from Fe(III) that is contained in mineral dust. If dust is the one that provides iron, its alkaline contents (e.g. calcium and magnesium) also involve the Fe-sulfur coupling by affecting pH. Such iron-sulfur cycles are illustrated in Fig. 17.

3. Cloud Effect on Aerosols

Clouds affect aerosols in many intricate ways. They can redistribute aerosols by processes such as cloud convection and scavenging. They also provide sites for chemical reactions that would not otherwise occur in a clear sky, thus influencing the production and destruction of many trace gases and aerosol particles. Clouds can also reflect or diffuse the solar actinic flux that drives many photochemical reactions that produce precursors of aerosols. Lightning generated from

phytoplankton

0 0

Post a comment