KH2 O [H2 0CH 2O

where ki and kH2 o are the precipitation frequency (or rate) of a chemical species i and water, respectively; variables in square brackets denote the species concentration in the precipitation; and C represents the concentrations in the air. Figures 20(b)-20(c) show the ER of various trace chemicals due to scavenging by liquid phase and ice phase precipitation, respectively, from the orographic clouds shown in Fig. 19. Note that since essentially all sulfate exists in aerosol or cloud droplets (see Fig. 18), and sulfate in cloud drops will return to aerosols when clouds dissipate, the curves labeled S(VI) may represent the ER of aerosols. One can see that the aerosol ER is the highest value among all chemicals. Furthermore, the aerosol Er in rainfall [Fig. 20(b)] is significantly higher than unity, which means that they can be removed more efficiently than water from the air. Between 25 and 60 km, the aerosol ER of the aqueous phase decreases gradually, due to two factors: (1) the rimed crystals accrete smaller and smaller cloud drops containing progressively less solute, and (2) the less-rimed ice actually relies more on depositional growth and obtains more clear water. At 60-65 km, the main form of aqueous phase precipitation switches from melted ice to intercepted cloud water, so the aerosol ER is elevated somewhat. The only place in Fig. 20(b) where the aerosol ER is lower than unity is at 90-100 km; at these locations the "rain" is actually ground interception of cloud drops, and these cloud drops were activated from smaller CCNs which contain less sulfate. Cloud drops activated from larger CCNs tend to grow larger and get accreted more easily by ice crystals. Ammonia is highly, but not completely, condensable in aerosol and cloud drops, so it also has a rather high Er. S(IV) has a moderate Er, and the value is higher further up the first hill because the intercepted cloud drops are more diluted and have higher pH values to allow high solubility of SO2.

The atmospheric chemistry model and air pollution models often neglect the scavenging of chemicals by ice phase precipitation, which is commonly regarded as a "clean" form of falling water. Yet, from Fig. 20(c) one can see that ice phase precipitation does contain chemicals. In fact, the removal efficiency of aerosols by ice phase precipitation can be several times higher than that of water, shown for example by the high Er at distances before 80 km. This, of course, is due to the riming process through which ice crystals obtain solutes in cloud drops. Note that the ER of S(IV) may approach that of aerosols (between 105 and 135km), as S(IV) can be incorporated into ice by entrapment during riming and by sorption during vapor deposition (see Chen and Lamb, 1990).

From the above discussions we may summarize that ER is determined by solubility and microphysical mechanisms by which the chemicals are incorporated into the precipitation. Furthermore, precipitation is very effective in removing aerosols from the atmosphere even in a weak-lifting orographic cloud. In the previous section we saw that clouds can also be strong sources of aerosols. The tug of war between in-cloud production and removal by precipitation causes strong variability of atmospheric CCNs.

3.3. Cloud venting

Cloud venting is the process of exchanging mixed-layer air masses with the free atmosphere through updrafts and downdrafts associated with cloud activities. It plays a major role in the vertical redistribution of atmospheric trace chemicals, including water vapor and aerosols. Trace chemicals produced near the Earth's surface, if not already scavenged, can reach the upper troposphere rather quickly with the assistance of strong updrafts. Strong convection may even significantly contribute to cross-tropopause transport into the stratosphere (e.g. Wang, 2003; Sherwood and Dessler, 2003). Stratospheric chemicals can also be transported downward by cloud turbulence and precipitation-induced downdrafts.

Cloud venting may proceed not only through cloud scale convections but also through synoptic cloud systems. For instance, the warm conveyor belts (WCBs), which are ascending air-streams ahead of cold fronts, are the primary mechanism for rapidly transporting air pollution from the continental planetary boundary layer to the upper troposphere and from one continent to another. Cooper et al. (2004) estimated in a case study that 8% of the WCB mass originated in the stratosphere and 44% passed through the lower troposphere. Cotton et al. (1995) performed a global analysis and found that the extratropical cyclone (e.g. WCBs) has the highest boundary layer mass flux of all cloud venting systems, followed by the general class of mesoscale convective systems, ordinary thunderstorms, tropical cyclones and mesoscale convective complexes. According to their estimation, these cloud ventings may pump the entire atmospheric boundary layer into the free troposphere 90 times a year. Note that the proportion of chemicals being transported by clouds varies from species to species because of chemical fractionation in scavenging processes. So, cloud venting and precipitation scavenging are actually two sides of the same coin.

Some of the cloud systems are too small to be represented in global scale models. So, cloud venting of trace chemicals is parametrized in essentially all current global chemical transport models (e.g. Jacob et al., 1997), in which the vertical transport of soluble gases is parametrized by normalizing against the transport of tracers such as water. However, as pointed out by Yin et al. (2001), there is no straightforward scaling factor, particularly if small concentrations of highly soluble gases in the upper troposphere need to be defined. The transport of aerosols is no exception. In fact, the amount of chemicals being transported vertically depends not only on the strength of convection but also on the scavenging processes involving many cloud-microphysical mechanisms, as mentioned in the previous section. So the parametrization of aerosol scavenging and venting in large-scale models are particularly difficult, because these models do not truly resolve the cloud microphy-sical mechanisms.

3.4. Influences on aerosol production

In Subsec. 2.1.1 we mentioned the DMS-cloud-climate cycle, in which clouds may indirectly influence the formation of aerosols by impacting marine biological activities. Similarly, clouds may influence the productivity of land plants and the generation of ice nucleation bioaerosols (see Subsec. 2.2.4). The generation of atmospheric sea salt particles and of mineral dust are also related to clouds, as they favor the conditions of strong gust winds associated with cloud activities. All these indirect influences have strong feedback cycles, and the cycles may link to each other, forming a convoluted process network. In this Subsection we will introduce only a few additional mechanisms, with focus on aerosol nucleation, to exemplify the complexity of aerosol-cloud interactions.

3.4.1. Actinic flux effect

Although clouds may increase the total aerosol mass by aqueous phase chemical production, they normally reduce the number of aerosols either by the coalescence and accretion processes or by precipitation scavenging. Yet, some observational studies have found evidence of strong new aerosol production near clouds (e.g. Dinger et al., 1970; Hegg et al., 1990; Frick and Hoppel,

1993; Saxena and Gravenstein, 1994; Wiedensohler et al., 1997). Earlier postulations of the causes of such particle production included the shattering of salt crystals formed by rapid evaporation of cloud drops (e.g. Dessens, 1949; Twomey and McMaster, 1955; Radke and Hegg, 1972), but Mitra et al. (1992) disproved this mechanism with wind tunnel experiments. More advanced aerosol measurements later revealed that the new particles are actually quite small in the nuclei mode size range. A more plausible cause of the generation of such nuclei mode aerosols is the homogeneous nucleation from the gas phase, which involves photochemical reactions. The remaining question is: How do clouds get involved in the process?

Homogeneous nucleation of atmospheric particles usually proceeds via the H20-H2S04 binary interaction. Ternary nucleation with the participation of an additional gas such as ammonia or nitric acid may also enhance the process. The main conditions favorable for binary or ternary homogeneous nucleation are a high saturation ratio of water vapor Sw (relative humidity) and a high saturation ratio of sulfuric acid vapor Sa (relative acidity). High Sw and Sa may be caused by either increasing the vapor concentration or lowering the temperature. The increase of sulfuric acid concentration is particularly complex, because the amount of sul-furic acid in the gas phase is normally very low (sec Fig. 18). It requires either a strong production (such as via photochemical reactions) or a low concentration of existing aerosols which may otherwise consume the vapor quickly. Shaw (1989) postulated that only in very clean air could binary nucleation of H2 0-H2S04 particles occur, because the existing aerosols may consume sulfuric acid vapor and inhibit high Sa. Hegg et al. (1990) then pointed out that high water vapor concentrations detrained from the cloud coupled with high actinic flux due to backscatter from the cloud droplets are additional controlling factors. A high actinic flux may boost photochemical reactions that lead to sulfuric acid production and high relative acidity. Crawford et al. (2003) measured enhancements in UV actinic flux of up to 40% over clear-sky values, whereas Kylling et al. (2005) found a 60-100% increase. Los et al. (1997) even estimated that photolysis rate coefficients can be 300% higher at cloud tops than in clear-sky conditions. Perry and Hobbs (1994) further drew the picture that convective clouds bring up the precursor of sulfuric acid, SO2, by cloud venting and remove a large portion of the existing aerosols by cloud and precipitation scavenging. Then, high relative humidity and reflected sunlight near the clouds together form a favorable nurturing environment for new particles.

Reflectance of actinic flux by clouds may enhance particle nucleation, but only on the sunward side of the cloud. Behind the cloud, an opposite effect would occur. Figure 21 demonstrates such influences by cloud shadows. The results are from a parcel simulation using a detailed (binned) aerosol microphysical model with a multicomponent particle framework similar to that of Chen and Lamb (1994, 1999) for cloud microphysics (see Subsec. 3.2). It includes detailed treatments on binary nucleation and particle aggregation. For the production of sulfuric acid through photochemical reactions, a simplified production rate is used by applying the typical SO2 concentration and

-Q=1.0 Q=0.8 Q=0.6 - Q=0.4 -Q=0.2 -Q=1.0 Q=0.8 Q=0.6 --Q=0.4 -Q=0.2

-Q=1.0 Q=0.8 Q=0.6 - Q=0.4 -Q=0.2 -Q=1.0 Q=0.8 Q=0.6 --Q=0.4 -Q=0.2

Local Time, hour Local Time, hour

Figure 21. Simulated evolutions of aerosol number concentration under the influence of cloud cover. The top and the bottom panels are with stratus (St) and stratocumulus (Sc) cloud cover, respectively. The left panels are initialized with the Wanli aerosol type, and the right panels apply the NTU aerosols. Five values of fixed transmissivity are applied for the St clouds, and five values of minimum transmissivity for the Sc clouds. Note that the smooth curve in (c) is the same as the curve with Q = 1 in (a), and likewise for (b) and (d).

Local Time, hour Local Time, hour

Figure 21. Simulated evolutions of aerosol number concentration under the influence of cloud cover. The top and the bottom panels are with stratus (St) and stratocumulus (Sc) cloud cover, respectively. The left panels are initialized with the Wanli aerosol type, and the right panels apply the NTU aerosols. Five values of fixed transmissivity are applied for the St clouds, and five values of minimum transmissivity for the Sc clouds. Note that the smooth curve in (c) is the same as the curve with Q = 1 in (a), and likewise for (b) and (d).

diurnally varying reaction rate. Two types of initial aerosols are considered to demonstrate the effect of existing aerosols: (1) the aerosol size distribution measured at the Wanli station located at the northern tip of Taiwan, and its total number concentration of 2200 cm-3 representing a relatively clean condition; (2) the aerosol size distribution measured at the National Taiwan University (NTU), and the 12500 cm-3 total number concentration representing an average urban condition in Taipei.

We consider two types of cloud influences: stratus clouds (St), which have uniform influence on the actinic flux; and stratocumulus clouds (Sc), which appear and attenuate sunlight periodically. Their influence on the actinic flux is represented by a factor of atmospheric transmissivity, Q. For St clouds, fixed values of Q are imposed to modify the diurnally varying clear sky actinic flux. Similar treatment is applied to the Sc clouds, except that the transmissivity varies periodically between unity and a minimum value. Figure 21(a) shows a drastic increase in aerosol concentration after sunrise due to photochemical production of sulfate for the rural (Wanli) condition. The number of aerosols may increase by two orders of magnitude under a clear sky condition. Lower Q values result in an obvious reduction of aerosol production, and the time of the particle burst is somewhat delayed. For the urban (NTU) aerosol condition, however, particle production is much weaker. In fact, the production is discernible only for the clear-sky (Q = 1) condition. This result further confirms that particle nucleation is stronger in cleaner environments. Note that the general decreasing trend in Figs. 21(b) and (d) is due to particle aggregation.

Under the stratocumulus condition, as shown in Fig. 21(c), particles are generated during cloud breaks but quickly diminish due to Brownian coagulation under cloud cover (which indicates that most of the increase is in ultrafine particles). One may notice that the maximum particle concentration can be higher in the cloudy conditions than in the clear-sky situation. This again shows the effect of a lower existing aerosol condensation sink at the beginning of each burst. In the urban case [Fig. 21(d)], particle production is still much weaker than in the rural case. But unlike the St case [Fig. 21(b)], the fluctuation is discernible even for the lowest Q value of 0.2.

The above examples have demonstrated that aerosol nucleation is very sensitive to the changes of actinic flux and the amount of existing aerosols. As both factors are often highly variable, no wonder that aerosol particles are found to distribute rather unevenly in both space and time. But to what extent clouds contribute to such variations remains an unresolved issue.

3.4.2. Turbulence effect

A factor overlooked by earlier studies is the turbulent condition in the atmospheric boundary layer (ABL), in which the air parcels rise and sink repeatedly with the so-called large eddy circulation. Clouds often form at the top of the large eddies, as in the circumstances of stratiform clouds in the marine boundary layer. In the rising part of the large eddy, the expanding air cools down adiabatically, and the lowering temperature necessarily causes an increase in both Sw and Sa. This is another way of increasing the nucleation rate, besides photochemical production of sulfuric acid as just mentioned above. Figure 22 shows the simulation of induced particle nucleation in an air parcel that goes up and down with the large eddies. One can see a pulsating production of aerosols similar to that shown in Fig. 21(c) under Sc cloud covers, but with significantly stronger amplitudes. The particle burst is strong even for the urban aerosol scenario.

However, the cause of the pulsating particle burst is not as simple as the cooling effect. Looking closely at each pulse, one may find that the variations are not as smooth as that shown

-Q=1.0 Q=0.8 Q=0.6 ------Q=0.4 -Q=0.2 -Q=1.0 Q=0.8 Q=0.6 -- Q=0.4 -Q=0.2

-Q=1.0 Q=0.8 Q=0.6 ------Q=0.4 -Q=0.2 -Q=1.0 Q=0.8 Q=0.6 -- Q=0.4 -Q=0.2

Local Time, hour Local Time, hour

Figure 22. Simulation of new particle production in an air parcel moving in the form of a large eddy circulation within the atmospheric boundary layer. The amplitude and period of the large eddy are set as 250 m and 30 minutes, respectively. The setup of the initial aerosol type is the same as those in Fig. 21.

Local Time, hour Local Time, hour

Figure 22. Simulation of new particle production in an air parcel moving in the form of a large eddy circulation within the atmospheric boundary layer. The amplitude and period of the large eddy are set as 250 m and 30 minutes, respectively. The setup of the initial aerosol type is the same as those in Fig. 21.

in Fig. 21(c), particularly for the urban situation where two peaks are visible. So, in Fig. 23, we zoom in on a couple of the cycles to show the details. Note that the left panels show the same results as for Fig. 21, and we put an additional simulation in the right panels that represents a moister condition (5% higher in Sw) in order to inspect also the effect of Sw.

The ups and downs of Sw in the top two panels of Fig. 23 reflect the imposed periodical vertical motion (with a period of 30 minutes and an amplitude of 250m). In Fig. 23(b) one may also notice the formation of clouds as indicated by the development of supersaturation at the tops of the circulation. The expansion cooling should in principle cause an increase in Sa if the mixing ratio of sulfuric acid vapor is kept constant, but in reality the aerosol (haze) drops swell upon increasing Sw (cf. Köhler curves in Subsec. 2.1.1), and thus quickly suck up sulfuric acid vapor and reduce Sa. An even more interesting phenomenon occurs during the downward motion period for the NTU aerosol scenario, where the compression warming causes haze drops to evaporate and become more and more concentrated, such that the sulfuric acid in them is forced out into the gas phase to reach a new Henry equilibrium. The outgassing from the evaporating haze is more than enough to offset the warming effect on Sa during the fastest descending period. Later, the air gets too warm such that Sa eventually falls off. As a consequence, every cycle of circular motion (and Sw) is accompanied by two peaks of Sa evolution. Such a phenomenon is less obvious in the Wanli aerosol scenario, because less aerosol sulfate is present.

The out-of-phase variations of Sw and Sa result in two peaks in the binary nucleation rate for every large-eddy cycle. So, as shown in Figs. 23(c)-23(f), the evolution of aerosol concentrations tends to exhibit a double-peak feature. Note that the two peaks are much more pronounced in the NTU aerosol scenario. The peak of the descending (evaporating) period is less prominent than that of the ascending period, particularly for the case of Wanli aerosols with lower relative humidity, where the second peak is almost indiscernible. Also worth noting is the stronger aerosol production with the Wanli aerosol scenario, which indicates that in a cleaner (less existing aerosols) environment more of the photochemically produced sulfuric acid can be retained in the gas phase, thus resulting in higher Sa, as shown in Figs. 23(a) and 23(b). As for the humidity effect, one can find higher aerosol production in higher relative humidity for the Wanli aerosol

Was this article helpful?

0 0

Post a comment