The Role of Small Soluble Aerosols in the Microphysics of Deep Maritime Clouds

A. P. Khain Department of Atmospheric Sciences, The Hebrew University of Jerusalem, Jerusalem, Israel

Search for other papers by A. P. Khain in
Current site
Google Scholar
PubMed
Close
,
V. Phillips The Department of Meteorology, University of Hawaii at Manoa, Honolulu, Hawaii

Search for other papers by V. Phillips in
Current site
Google Scholar
PubMed
Close
,
N. Benmoshe Department of Atmospheric Sciences, The Hebrew University of Jerusalem, Jerusalem, Israel

Search for other papers by N. Benmoshe in
Current site
Google Scholar
PubMed
Close
, and
A. Pokrovsky Department of Atmospheric Sciences, The Hebrew University of Jerusalem, Jerusalem, Israel

Search for other papers by A. Pokrovsky in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

Some observational evidence—such as bimodal drop size distributions, comparatively high concentrations of supercooled drops at upper levels, high concentrations of small ice crystals in cloud anvils leading to high optical depth, and lightning in the eyewalls of hurricanes—indicates that the traditional view of the microphysics of deep tropical maritime clouds requires, possibly, some revisions. In the present study it is shown that the observed phenomena listed above can be attributed to the presence of small cloud condensation nuclei (CCN) with diameters less than about 0.05 μm. An increase in vertical velocity above cloud base can lead to an increase in supersaturation and to activation of the smallest CCN, resulting in production of new droplets several kilometers above the cloud base. A significant increase in supersaturation can be also caused by a decrease in droplet concentration during intense warm rain formation accompanied by an intense vertical velocity. This increase in supersaturation also can trigger in-cloud nucleation and formation of small droplets. Another reason for an increase in supersaturation and in-cloud nucleation can be riming, resulting in a decrease in droplet concentration. It has been shown that successive growth of new nucleated droplets increases supercooled water content and leads to significant ice crystal concentrations aloft. The analysis of the synergetic effect of the smallest CCN and giant CCN on production of supercooled water and ice crystals in cloud anvils allows reconsideration of the role of giant CCN. Significant effects of small aerosols on precipitation and cloud updrafts have been found. The possible role of these small aerosols as well as small aerosols with combination of giant CCN in creating conditions favorable for lightning in deep maritime clouds is discussed.

Corresponding author address: Prof. Alexander Khain, Department of Atmospheric Sciences, The Hebrew University of Jerusalem, Jerusalem, Givat Ram 91904, Israel. E-mail: khain@vms.huji.ac.il

Abstract

Some observational evidence—such as bimodal drop size distributions, comparatively high concentrations of supercooled drops at upper levels, high concentrations of small ice crystals in cloud anvils leading to high optical depth, and lightning in the eyewalls of hurricanes—indicates that the traditional view of the microphysics of deep tropical maritime clouds requires, possibly, some revisions. In the present study it is shown that the observed phenomena listed above can be attributed to the presence of small cloud condensation nuclei (CCN) with diameters less than about 0.05 μm. An increase in vertical velocity above cloud base can lead to an increase in supersaturation and to activation of the smallest CCN, resulting in production of new droplets several kilometers above the cloud base. A significant increase in supersaturation can be also caused by a decrease in droplet concentration during intense warm rain formation accompanied by an intense vertical velocity. This increase in supersaturation also can trigger in-cloud nucleation and formation of small droplets. Another reason for an increase in supersaturation and in-cloud nucleation can be riming, resulting in a decrease in droplet concentration. It has been shown that successive growth of new nucleated droplets increases supercooled water content and leads to significant ice crystal concentrations aloft. The analysis of the synergetic effect of the smallest CCN and giant CCN on production of supercooled water and ice crystals in cloud anvils allows reconsideration of the role of giant CCN. Significant effects of small aerosols on precipitation and cloud updrafts have been found. The possible role of these small aerosols as well as small aerosols with combination of giant CCN in creating conditions favorable for lightning in deep maritime clouds is discussed.

Corresponding author address: Prof. Alexander Khain, Department of Atmospheric Sciences, The Hebrew University of Jerusalem, Jerusalem, Givat Ram 91904, Israel. E-mail: khain@vms.huji.ac.il

1. Introduction

There are several thermodynamic and microphysical features that specify deep convective maritime clouds in the tropics: (a) cloud-base height between 1 and 2 km altitude; (b) high freezing level height of 4–4.5 km; (c) concentration of cloud condensation nuclei (CCN) of 50–200 cm−3 at 1% supersaturation Sw (Pruppacher and Klett 1997; Levin and Cotton 2009), which is about an order of magnitude lower than that for clouds developing over continents; and (d) significant concentrations of giant CCN (GCCN), with dry radii exceeding 1 μm, especially pronounced under strong wind conditions due to sea spray production. Low CCN concentrations and the presence of GCCN determine typical maritime properties of these clouds characterized by low droplet concentrations of about 50–150 cm−3 and rapid formation of raindrops below the 4–5-km level. The raindrops collect most of the droplets nucleated near the cloud base, leading to intense warm rain. Comparatively rare drops that do not fall out below the 4–5-km level freeze within an altitude of 5–6 km, forming frozen drops or graupel of a comparatively low concentration (e.g., Blyth et al. 1998; Yin et al. 2000; Ovtchinnikov et al. 2000; Rosenfeld et al. 2002; Khain et al. 2004; Phillips et al. 2005; Freud et al. 2008).

It has been widely accepted that at altitudes of about 8–10 km these clouds do not have any significant supercooled cloud-liquid content or any significant concentration of ice crystals (see reviews by Rosenfeld et al. 2007; Khain 2009). According to this view, drop concentrations at levels of homogeneous freezing of about 9.5–10 km should be negligible, so these clouds should contain very low concentrations of ice crystals in anvils of deep convective clouds and related cirrus clouds.

Contrary to this view of the microphysics of maritime clouds, there is both indirect and direct evidence indicating the existence of small supercooled droplets at the upper levels of deep tropical clouds, which can be the source of high ice crystal concentration in the cloud anvils. The indirect evidence includes the existence of lightning in tropical cyclone (TC) eyewalls and maritime deep convection in the intertropical convergence zone (ITCZ). It is widely accepted that the charge separation in clouds takes place at temperatures below about −13°C with maximum intensity near −20°C, where collisions between low-density and high-density ice particles occur in the presence of a significant amount of supercooled liquid water (e.g., Takahashi 1978; Black and Hallett 1999; Saunders 1993; Cecil et al. 2002a,b; Sherwood et al. 2006). According to this view, it should be expected that the lack of supercooled water at low temperatures inhibits the charge separation process. Such a decrease in the amount of supercooled water aloft should be especially strong in clouds within TCs eyewalls where a huge amount of GCCN lead to a dramatic intensification of warm rain (in this sense, clouds in TC eyewalls can be regarded as extremely maritime clouds).

Yet, contrary to these widely accepted views, lightning within rainbands and eyewalls of TCs is a regular phenomenon (e.g., Black and Hallett 1999; Orville and Coyne 1999; Cecil et al. 2002a; Molinari et al. 1999; Rodgers et al. 2000; Shao et al. 2005; Demetriades and Holle 2006; Fierro et al. 2007, 2010; Squires and Businger 2008; Leary and Ritchie 2009; Price et al. 2009). With the help of The World Wide Lightning Location Network (WWLLN), intense lightning in TCs and tropical depressions over the open ocean were registered (Lay et al. 2007). Although WWLLN enables registering only the strongest flashes, the lightning activity and its variations in TC were investigated in order to find ways for using lightning intensity as a predictor of TC intensity changes (e.g., Price et al. 2009). In most studies it was found that the closer the lightning is to the storm center, the more likely the TC is to intensify. If the general concept of lightning formation is correct, these observations imply that at upper levels, even in extremely maritime clouds in a TC eyewall, supercooled droplets exist in the presence of spray droplets and dramatically high concentrations of GCCN up to tens per cubic centimeter at the cloud-base level (e.g., Toba 1965; Clarke et al. 2006; Shpund et al. 2011).

The direct evidence indicating a significant concentration of supercooled droplets at the upper levels of deep convective maritime clouds is provided by aircraft observations of supercooled water of about 0.3 g m−3 in deep convective updrafts with peak vertical velocities up to about 15–20 m s−1 over the sea near Florida (Fridlind et al. 2004; Heymsfield et al. 2005; Phillips et al. 2005). These observations were part of The Cirrus Regional Study of Tropical Anvils and Cirrus Layers–Florida Area Cirrus Experiment (CRYSTAL-FACE). The aircraft observations in the Kwajalein Experiment (KWAJEX) showed droplet concentrations of about 50 cm−3 at altitudes up to 9 km at temperatures of about −30°C throughout most of the mixed-phase region, averaged over deep highly visible cloud decks (Phillips et al. 2007a).

Moreover, a detailed analysis of aircraft observations from six field experiments by Heymsfield et al. (2009) shows the existence of significant concentrations of small supercooled droplets above about 6 km in maritime deep convective clouds. Supercooled droplets existing at temperatures below about −20°C are inevitably small, since raindrops almost immediately freeze at such low temperatures, partly due to activation of their immersed ice nuclei or to collisions with crystals and other ice particles (Pruppacher and Klett 1997). Indeed, Heymsfield et al. (2009, p. 3551) reported that the sizes of these droplets aloft are typically below 50 μm. In agreement with the presence of a significant concentration of small supercooled droplets at the upper levels of deep convective maritime clouds, Heymsfield et al. (2009) reported the concentration of ice crystals with diameters below 50 μm in anvils of deep maritime convective clouds at temperatures of −40°C to be as high as 30 cm−3 in relatively clean air and 300 cm−3 in polluted clouds sampled off the west coast of Africa. Takahashi and Kuhara (1993) found significant concentrations of ice crystals in anvils of deep tropical convective clouds using video sondes.

Such concentrations of ice crystals exceed the maximal concentration of active ice nuclei (IN) by several orders of magnitude and can be attributed to homogeneous freezing of cloud droplets at temperatures as low as −38°C (Rosenfeld and Woodley 2000; Khain et al. 2001b; Phillips et al. 2005, 2007a; Heymsfield et al. 2009). Concentration of ice crystals is of high importance as it determines the radiative properties of cloud anvils and of the related cirrus clouds (e.g., Carrió et al. 2007).

It is interesting that according to Heymsfield et al. (2009) the mass content and concentration of supercooled droplets decrease in maritime clouds over height above the freezing level. Supercooled droplets almost disappear at temperatures of about −15° to −20°C and then arise again at colder temperatures, their concentration increasing upward.

Two mechanisms can be assumed to account for the formation of small supercooled droplets at temperatures below −15° to −20°C: (a) in-cloud nucleation of CCN laterally entrained from the free troposphere into deep convective updrafts and (b) the presence of unactivated CCN coming through cloud base [following Pruppacher and Klett (1997), we use here the terms “activated” and “nonactivated” CCN, whose sum represents the concentration of condensation nuclei (CN)]. The first mechanism was discussed by Phillips et al. (2005), Fridlind et al. (2004), and Yin et al. (2005). It is possible because of the entrainment even when vertical velocities are relatively weak in clouds. In the case of a vigorous Florida thunderstorm simulated by Phillips et al. (2005) and Fridlind et al. (2004), the entrainment of aerosol particles (APs) aloft was found to be highly efficient in production of small droplets.

In many cases, concentration of CCN exponentially decreases with height (Levin and Cotton 2009). In these cases the mechanism of the CCN entrainment through lateral cloud boundaries is less efficient at high levels. It is reasonable to assume that the role of lateral entrainment of background aerosols in clouds within hurricane eyewalls is negligible. Hence, the formation of small droplets can be attributed to in-cloud activation of CCN ascending from cloud base together with growing droplets. In-cloud nucleation in stratocumulus and cumulus clouds was observed and discussed in a number of studies (e.g., Ochs 1978; Ludlam 1980; Korolev 1994; Pinsky and Khain 2002; Segal et al. 2003; Phillips et al. 2005; Prabha et al. 2011). Using a spectral (bin) microphysics model, Phillips et al. (2005) simulated a vigorous deep convective cell observed over the sea near Florida and found that the vast majority (99%) of supercooled cloud droplets upwelled to the top of the mixed-phase region were activated in cloud (“secondary droplet nucleation”), far above the cloud base. The droplets froze homogeneously near −37°C, accounting for almost all cell ice crystals at anvil levels.

The conceptual scheme of this process is illustrated in Fig. 1 (after Pinsky and Khain 2002, with some changes). Let us assume that near cloud base the size distribution of aerosols (nonactivated CCN) is wide and includes particles with dry radii ranging from 0.001 to about 2 μm (Hobbs 1993; Pruppacher and Klett 1997). The first CCN activation takes place at the supersaturation maximum located within a few tens of meters above the cloud base. The supersaturation maximum near the cloud base depends on the vertical velocity at this level, which does not exceed a few meters per second in maritime clouds (e.g., Heymsfield et al. 2009, their Fig. 1), and typically does not exceed 0.5%. Given these low values, only CCN with radii larger than 0.02 μm are activated at the cloud base. The droplets formed near the cloud base trigger the formation of the first mode in droplet size distribution (DSD) and play a dominant role in warm rain formation. Above the local supersaturation maximum near the cloud base Sw decreases with height but then begins growing again during in-cloud ascent because of an increase of the vertical velocity with increasing height.

Fig. 1.
Fig. 1.

The conceptual scheme of droplet spectra formation in deep convective clouds [after Pinsky and Khain (2002), with changes explained in the text].

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

In case the value of Sw in the updraft exceeds the local maximum reached below, a new portion of CCN of smaller size is activated, giving rise to formation of the second mode of DSD. According to observations (Warner 1969a,b; Prabha et al. 2011) and simulations (Pinsky and Khain 2002; Segal et al. 2003), the second mode (i.e., bimodal DSD) forms at a few hundred meters to a few kilometers above cloud base. In zones of efficient interdrop collisions and rapid rain fallout, the droplet concentration decreases and the vertical velocity increases. Therefore, Sw can dramatically increase, triggering a new portion of the smallest CCN to be activated, producing small droplets at heights of about 6 km above the cloud base. Since the process of in-cloud nucleation of droplets often takes place at high supersaturation, nucleation of the smallest droplets is accompanied by their fast growth. As a result, the gaps between the modes can disappear, leading to a unimodal DSD with a dispersion of 0.2–0.3, containing the smallest droplets at high levels aloft. Indeed, in-cloud activation of droplets can occur continuously over a deep layer of the cloud updraft in which the vertical velocity increases with height, causing a monotonic increase in supersaturation with height. This process was investigated in detail by Prabha et al. (2011), who analyzed in situ measurements in premonsoon and monsoon clouds over central India.

In the present study, we argue that all the phenomena mentioned above—formation of small supercooled droplets at temperatures below −20°C, formation of high concentration of ice crystals in anvils of deep convective clouds, and lightning in extremely maritime clouds in the hurricane eyewall—can be caused by the same mechanism of in-cloud droplet nucleation. The rise in Sw can be caused by an increase in the vertical velocity above cloud base and the depletion of cloud droplets caused by the coalescence during deep maritime ascent (Ochs 1978; Pinsky and Khain 2002; Phillips et al. 2005). Pinsky and Khain (2002) showed that the combined effects of vertical velocity increase and the depletion of cloud droplets by coalescence can lead to formation of three-mode DSDs.

The hypothesis of in-cloud nucleation in air ascending from the cloud base implies the existence of supersaturations exceeding the local maximum at cloud base. The values of supersaturation in clouds are impossible to measure. It is known, however (Squires 1952; Korolev and Mazin 2003), that in warm and mixed-phase clouds supersaturation rapidly tends to a quasi-steady value of , where A is the coefficient slightly depending on temperature, νm is the coefficient of molecular diffusion of water vapor, Nd is droplet concentration, is the mean droplet radius, and W is the vertical velocity. For instance, at within the zone of efficient collisions and just above it, at and W ~ 12 m s−1, supersaturation may even exceed 15%–20% (Korolev and Mazin 2003). The values of Sqs reaching 8% were calculated using in situ measurements in polluted premonsoon and monsoon clouds over central India during the Cloud Aerosol Interaction and Precipitation Enhancement Experiment (CAIPEEX; Prabha et al. 2011). Fukuta (1993) found that theoretically supersaturation may exceed 10% in cumulus clouds.

The hypothesis of in-cloud nucleation in air ascending from the cloud base also implies the existence of very small CCN with radii below about 0.02 μm, which cannot be activated at the cloud base of maritime clouds and ascend within updrafts above cloud base. According to Pruppacher and Klett (1997) and Levin and Cotton (2009), the concentration of activated CCN in the maritime atmosphere increases monotonically with Sw increasing up to 8%. The radius of soluble CCN that can be activated at Sw of 8% is about 0.003 μm. The dependence of CCN concentration on supersaturation is traditionally described by the semiempirical formula
e1
where Nccn is the concentration of activated AP at supersaturation Sw (%) with respect to water, and No and k are measured parameters. The slope parameter k varies from 0.3 to 1.3 in different zones over the ocean (see Pruppacher and Klett 1997). According to Pruppacher and Klett (1997), the averaged value of k for all maritime clouds is close to 0.9, which indicates the existence of a significant amount of small CCN in the maritime atmosphere. According to Hudson and Yum (2002) and Levin and Cotton (2009), the typical value of k is close to 0.3, which also indicates the existence of small CCN, but with concentration lower than the data reviewed by Pruppacher and Klett (1997). The data obtained by Levin and Cotton (2009) correspond to remote maritime conditions.

Small aerosols less than 15 nm were directly observed in the free troposphere over the remote Pacific (Clarke and Kapustin 2002). The small CCN belonging to the Aitken mode can be of continental nature (e.g., originating from fossil fuel combustion or Saharan dust, which is typically slightly hygroscopic) (e.g., Twohy et al. 2009). Such CCN are often found in convective storms near the eastern African coast as well and in storms and hurricanes reaching the American coast, or they can form via various chemical reactions over the sea (Covert et al. 1992; Hobbs 1993; Pruppacher and Klett 1997; Clarke and Kapustin 2002). Significant concentrations (of a few hundred cubic centimeters) of CCN with radii ranging from 0.002 to 0.03 μm in the marine atmosphere were reported by Jaenicke (1993).

Many numerical simulations of aerosol effects on cloud microphysics, dynamics, and precipitation [see overviews by Levin and Cotton (2009) and Khain (2009)] were carried out under different No values varying between a few tens and several thousand per cubic centimeter. The role of the slope parameter is typically not discussed, being implicitly assumed not to be decisive. However, in several studies the slope parameter was assumed to decrease rapidly with increasing Sw, which substantially decreases the concentration of the smallest CCN and, correspondingly, the efficiency of in-cloud nucleation. In many mesoscale models and in global circulation models the nucleation of cloud droplets is assumed to be restricted to the cloud-base level.

The main goal of the present study is to show that small CCN can dramatically modify the microphysics and dynamics of maritime deep convection, where supersaturation can be very high in convective updrafts. It is also shown that in maritime clouds the combined effect of the smallest CCN and GCCN increases the concentrations of supercooled droplets aloft and of ice crystals in cloud anvils. In addition, the development of deep clouds within polluted air was simulated to compare the ice concentration simulated by the model with that observed by Heymsfield et al. (2009) near the western coast of Africa. Numerical simulations are performed using an updated version of the Hebrew University Cloud Model (HUCM).

2. Model description

The HUCM is a 2D mixed-phase model with spectral bin microphysics based on solving a system of kinetic equations for size distribution functions for water drops, ice crystals (plate, columnar, and branch types), aggregates (snow), graupel, and hail/frozen drops, as well as APs (Khain et al. 2004, 2008a). Each size distribution is described using 43 doubling mass bins allowing the simulation of hailstones with diameters up to 6.8 cm. The minimum particle mass is equal to that of a droplet with a radius of 2 μm. The radii of dry APs serving as CCN range from 0.003 to 2 μm. Ice particles are characterized by their mass and shape. To improve the representation of melting as well as riming by snow, several auxiliary size distributions have been recently introduced: (a) distributions of liquid water mass within snow, graupel, and hail and (b) distributions of rimed masses in snowflakes. The model is specially designed to take into account the AP effects on the cloud microphysics, dynamics, and precipitation. The initial (at t = 0) CCN size distribution is prescribed. At t > 0, the prognostic equation for the size distribution of nonactivated CCN is solved. Using the Sw values, the critical CCN radius is calculated according to Kohler theory. CCN with radii exceeding the critical value are activated and new droplets are nucleated. The corresponding bins of the CCN size distribution then become empty.

The primary nucleation of each type of ice crystals is performed within its own temperature range following Takahashi et al. (1991). The dependence of the ice nuclei concentration on supersaturation over ice is described using an empirical expression suggested by Meyers et al. (1992) and applied using the semi-Lagrangian approach (Khain et al. 2000), allowing us to use the diagnostic relationship within a time-dependent framework. Production of secondary ice is treated according to Hallett and Mossop (1974). The rates of inhomogeneous and homogeneous drop freezing were represented following Vali (1994) and Pruppacher (1995), respectively. The homogeneous freezing takes place at temperatures of about −38°C. Freezing of cloud droplets used in the model leads to formation of platelike crystals; freezing of raindrops leads to hail formation. The diffusional growth/evaporation of droplets and the deposition/sublimation of ice particles are calculated using the analytical solutions for supersaturation both over water and over ice. To increase the accuracy of supersaturation calculations, the equation for diffusion growth/evaporation was solved using a variable time step, which in some cases was decreased down to 0.1 s. The efficient and accurate method of solving the stochastic kinetic equation for drop collisions, suggested by Bott (1998), was extended to an equation system calculating water–ice and ice–ice collisions. The model uses height-dependent drop–drop and drop–graupel collision kernels following Khain et al. (2001a) and Pinsky et al. (2001). To calculate the turbulence-induced enhancement factors for collisions between droplets, the values of dissipation rate and Reynolds number were calculated at each grid point and at each time step. These values taken from the lookup tables presented by Pinsky et al. (2008) were used to calculate enhancement factors of interdroplet collisions. As a result, the time- and space-dependent kernels of interdroplet collisions were used for simulation of raindrop formation. Ice–ice collection rates are assumed to depend on temperature (Khain and Sednev 1996; Pruppacher and Klett 1997). A detailed melting procedure with calculation of liquid water fractions within melting aggregates, graupel, and hail is included following Phillips et al. (2007b). To control the transition of snow (aggregates) to graupel by riming snow, the bulk density was calculated for each mass bin. If the bulk density exceeded 0.2 g cm−3 (i.e., became close to that of graupel), the snow from this bin was converted into graupel with the bin of the same mass.

To determine the snow bulk density in each mass bin, the rimed fraction of snow in each bin was recalculated at each time step. Implementation of time- and space-dependent bulk density of snow required recalculation of the collision kernels due to the changes in particle size and fall velocity. The fall velocity of rimed snow was calculated by interpolation (proportionally to the bulk density) between dry snow velocity and hail velocity of the same mass. The bulk density of graupel was set equal to 0.4 g cm−3, which is the mean graupel density observed in cumulus clouds (Pruppacher and Klett 1997). As for gradual changes of bulk density of graupel during riming, these will be accounted for in the next model version. Advection of scalar values is performed with the positively defined conservative scheme proposed by Bott (1989).

Several issues regarding utilization of the 2D version of this cloud model are to be clarified. It is clear that any 2D simulations are inevitably idealized, which imposes certain limits on simulations. For instance, splitting of supercell storms can be simulated using 3D cloud models only, while simulation of clouds or squall lines can be performed with 2D models. A significant advantage of 2D models is the option of using very fine resolution (in our case, the grid spacing is 50 m in both directions). This resolution rate is necessary for an accurate description of microphysical models, for an accurate calculation of supersaturation and its variability within clouds, and so on. The 2D model used in this study has a significant number of size distributions allowing a description of microphysics based on the first principles. The model was tested against multiple observations. For instance, it reproduces the evolution of size distributions over height as measured in situ in the field experiment in the Amazon region [Large-Scale Biosphere–Atmosphere Experiment in Amazonia—Smoke, Aerosols, Clouds, Rainfall, and Climate (LBA-SMOCC)] (e.g., Khain et al. 2008a). This finding can serve as a validation of the model.

Generally, 3D cloud models have an advantage as regards model geometry, while their microphysics is inevitably simplified due to computer limitations and to grid spacing substantially larger than that used in our 2D model. To the best of our knowledge, the existing 3D models have never been validated by comparison of simulated droplet size distributions with DSD obtained in situ at several heights including the levels of first raindrop formation. The main objective of the present paper is the description of the fine processes of formation of supercooled droplets and ice crystals at the upper cloud levels. This can be achieved only with a high model resolution. We believe that the results obtained using the 2D model provide an adequate—at least qualitatively adequate—description of these physical processes.

3. The experimental design

All the runs were performed within a 2D computational domain 26.5 km × 16 km and a grid spacing of 50 m in both the horizontal and vertical directions. This resolution is close to that used in large-eddy simulation (LES) models and allows simulation of supersaturation and its variations at high accuracy.

The main goal of the simulations was to investigate the effects of small CCN that can be activated at supersaturation exceeding 0.7%–1%. Three main sets of simulations were performed. The purpose of the first set of simulations is to investigate the role of small CCN in creation of supercooled water droplets at upper cloud levels and of high concentration of ice crystals in cloud anvils of deep convective clouds. In this set of simulations, the value of No was assumed equal to 100 cm−3, which is typical of maritime convection. The value of the slope parameter was assumed equal to 0.9. To show the role of the smallest CCN, two simulations—the maritime small-CCN case, E100_S, and the maritime small CCN-free, E100_NS (S denotes the presence of small CCN, while NS denotes the absence of the smallest CCN)—are compared. In E100_S run, the minimum CCN radius was assumed equal to 0.003 μm, while in E100_NS the minimum CCN radius was assumed equal to 0.0125 μm, which corresponds to the absence of CCN at Sw ≥ 1%.

The second set of simulations aimed at the investigation of the combined effect of the smallest CCN aerosols and GCCN. This set consists of two simulations. One is E100_S_G in which the CCN size distribution was similar to that in E100_S with one exception: the concentration of GCCN with radii exceeding 1 μm was increased to be 3 times as large as that in E100_S. The other simulation, E100_ NS_G, is similar to E100_ NS with one exception: the concentration of GCCN with radii exceeding 1 μm was increased threefold.

The purpose of the third set of simulations was to investigate the effects of small CCN in clouds developing in polluted air. In these simulations, No was assumed equal to 3500 cm−3 and slope parameter k = 0.9 was used. In the first simulation of this set, referred to as E3500_S, the minimum radius of CCN was assumed equal to 0.006 μm. In the second simulation, the minimum CCN radius was assumed equal to 0.015 μm, which corresponds to the absence of CCN at Sw ≥ 0.7%. The parameters of all the simulations are presented in Table 1.

Table 1.

Parameters of the initial CCN spectra in the simulations.

Table 1.

The maximum dry CCN radius in the model is 2 μm. These largest CCN give rise to wet particles (droplets) of 10 μm in radius at 100% relative humidity. No GCCN with dry radii larger than 2 μm were allowed. The parameters of No and k, as well as the constraints mentioned above, were applied to calculate the initial size distributions of CCN using the algorithm described by Khain et al. (2000). The initial size distributions of CCN in all the simulations are shown in Fig. 2. These size distributions were assumed within the lower 2-km layer. Above this level, the CCN concentration in each mass bin was decreased exponentially over height at a characteristic spatial scale of 1.5 km.

Fig. 2.
Fig. 2.

The initial size distributions of aerosols near the surface in different simulations.

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

In some studies (e.g., Warner 1973) the formation of bimodal DSDs is attributed to mixing of clouds with their environment. Warner (1973) obtained his results using a 1D Lagrangian parcel model, where mixing immediately affects the microphysics of the entire cloud. In a multidimensional model with a realistic dynamics, the situation is more complicated. The model used in the present study allows both mixing of clouds with the environment and a lateral penetration of CCN into clouds at higher levels. Moreover, advection of CCN by the velocity field can increase the concentration of CCN (including small CCN) at upper levels. It means that the aerosol and dynamical conditions of E100_NS do not exclude the possibility of in-cloud nucleation at significant distances above cloud base. We hypothesize that small CCN ascending from the cloud base play the major role in production of supercooled droplets and ice crystals in cloud anvils. This assumption agrees with modeling results obtained by Phillips et al. (2005), as well as with the observations made by Heymsfield et al. (2009), who found the maximum concentrations of supercooled droplets and crystals formed by homogeneous freezing to be located within cloud cores and in the vicinity of cloud cores.

As the basis for the thermodynamic conditions, the sounding data observed during Day 261 of the Global Atmospheric Research Program (GARP) Atlantic Tropical Experiment (GATE) 1974, as applied in many simulations of deep tropical convection, were used (e.g., Ferrier and Houze 1989). Such soundings resemble the thermodynamics typical of tropical oceans during hurricane season (Jordan 1958). The sounding indicates 90% humidity near the surface. The freezing level is at 4.2 km. The atmosphere is relatively unstable under these conditions, so the maximum vertical velocity in the simulated clouds is about 16–18 m s−1; to simulate especially strong maritime clouds (with vertical updrafts of 25 m s−1 at about 10-km height), which produce crystals at the high concentrations reported by Heymsfield et al. (2009), the profile of the dewpoint was changed as shown in Fig. 3 (dashed line).

Fig. 3.
Fig. 3.

Vertical profiles of temperature and dewpoint used in the simulations. Dashed line shows profile of dewpoint used to increase the instability of the atmosphere to simulate deep convective clouds.

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

According to Jorgensen et al. (1985) and Jorgensen and LeMone (1989), these convective clouds fall within the top 5% of the most intense maritime clouds. The wind shear was assumed to be 2 m s−1 in the lowest 6-km layer. Above 6 km, the background wind was assumed to be 2 m s−1. This wind shear allowed simulation of clouds with quite strong vertical velocities and formation of new convective clouds near the primary cloud, which is a typical feature of maritime convection. Clouds simulated at this wind shear values did not reach lateral boundaries during their lifetime, thus allowing us to evaluate effects of small aerosols on rain accumulated at the surface.

Clouds were triggered using a temperature heating of a triangle form with a maximum of 0.005°C s−1 at the center of the model domain at the height of 100 m. The heating decreased exponentially over below height till 2-km level. Above this level the heating was set equal to zero. The horizontal extension of the heating zone was 2 km, and the duration of heating was 1200 s. This kind of convective triggering is obviously, idealized. Deep convective clouds in the tropics and in eyewalls of TC are triggered dynamically by air convergence within in the boundary layer, caused by downdrafts of neighboring clouds or by other factors. However, for the model geometry used, the utilization of the initial heating is a simpler way of simulation. The main requirement with such heating is to create updrafts at cloud base of a few meters per second in agreement with the wind updrafts in a vigorous maritime convective clouds observed during the Tropical Ocean and Global Atmosphere Coupled Ocean–Atmosphere Response Experiment (TOGA COARE) (Heymsfield et al. 2009, their Fig. 1).

4. Results of simulations

a. Role of small CCN

In this section we will compare the results of simulations E100_S and E100_NS (in the latter run the CCN size distribution does not contain CCN with radii lower than 0.015 μm). During the first 20 min of cloud evolution, the fields of droplet concentration Nd, cloud water content (CWC; droplets with radii below 40 μm), rainwater content (RWC), and vertical velocity are quite similar in all the simulations. The cloud base is located at about 1-km altitude. The maximum Nd are about 110 cm−3, the maximum CWC reaches 2.4 g m−3, and rain forms at 2.5-km altitude (i.e., 1.5 km above the cloud base). Collection of cloud droplets by raindrops leads to a substantial decrease in Nd and in CWC substantially near the 3-km level. At higher levels, the monotonic decrease in Nd and CWC over height ceases. The mechanism of this effect is illustrated in Fig. 4 showing the fields of cloud droplet number concentration, RWC, W, and Sw in E100_S at t = 1860 s. Figure 5 (left panels) shows the CWC in E100_S. Figures 4 and 5 demonstrate that in E100_S the zone of new droplet concentration and CWC appears just above 4.5 km. The zone of new CWC is located slightly above the level of the RWC maximum and coincides with a sharp increase in vertical velocity over height up to 10 m s−1 at 5-km altitude. This increase in the vertical updrafts above the roughly 5-km level is typical of maritime clouds (e.g., Petersen et al. 1999) and can have several causes, including the instability of the atmosphere, rain unloading, and the start of ice processes accompanied by latent heat release due to freezing. In E100_S the maximum Sw in this region exceeds about 8%–10%. As discussed above, such supersaturation is not unusual, especially in maritime clouds. According to the Kohler law, at Sw of 8% the soluble CCN with dry radii as small as 0.003 μm are activated. Therefore, an intense in-cloud nucleation takes place above 4.4-km altitude.

Fig. 4.
Fig. 4.

Fields of CWC, RWC, W, and supersaturation at t = 1860 min in E100_S simulation. Horizontal dashed lines show the level of increase in supersaturation and in-cloud nucleation.

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

Fig. 5.
Fig. 5.

Fields of CWC in (a),(c) E100_S and (b),(d) E100_NS at t = (top) 1860 and (bottom) 2280 s.

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

In E100_NS, the absence of CCN with radii below 0.0125 μm does not allow nucleation of a significant amount of cloud droplets. The absence of intense in-cloud nucleation leads to an increase in supersaturation up to about 30% in this run (not shown). The difference in CWC values between these runs increases with time (Fig. 5). One can see that whereas in E100_S the regions with a significant CWC arise within the 6–8-km layer, in E100_NS the CWC above 6 km is quite low.

Figure 6a shows DSD at different points in the E100_S cloud. (These points are marked with circles in Fig. 5). One can see droplets with diameters as small as 5 μm at numerous cloud levels. New droplets with diameters of 5 μm arise at z = 5.8 km. At the height of 8.5 km the DSD again contains small droplets with diameters of 5 μm, while such droplets were absent at 6.75 km. In most cloud zones located at large distances above the cloud base, DSDs contain droplets smaller than those near cloud base (at 1.6 km). We see, therefore, that the process of in-cloud nucleation is widespread in maritime clouds, at least according to these simulations performed.

Fig. 6.
Fig. 6.

Droplet size distributions in (a) E100_S and (b) E100_NS (b) at t = 2280 s in points marked in Fig. 5 by circles.

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

It is noteworthy that supersaturation in mixed-phase clouds is determined by the liquid phase (Korolev and Mazin 2003). Since riming decreases the concentration of supercooled droplets, it should lead to an increase in supersaturation. It means that the conceptual scheme of in-cloud nucleation presented in Fig. 1 should be extended to the mixed-phase case where droplet concentration decreases and supersaturation increases due to riming during ascent. The newly nucleated droplets rapidly grow at high supersaturation values. As a result, the gaps between different modes seen in Fig. 1 disappear, and despite the presence of in-cloud droplet nucleation, DSDs become unimodal with a relative dispersion of 0.2–0.4. In E100_NS the DSDs in the lowest roughly 2 km above cloud base are quite similar to those in E100_S. Higher up, DSDs in E100_NS show much lower CWC and actually do not reveal existence of small droplets above altitude of 6 km (Fig. 6b). These DSDs show that cloud nucleation in E100_NS is much less pronounced than in E100_S cloud.

Figure 7 shows the vertical profiles of the maximum values of Nd, LWC, total ice content, and ice crystal concentration in clean air simulations with small CCN (E100_S), without small CCN (E100_NS), in the presence of high concentration of both low and GCCN (E100_SG), and with high concentration of GCCN but without small CCN (E100_NS_G).

Fig. 7.
Fig. 7.

Vertical profiles of the maximum values of (a) droplet concentration, (b) LWC, (c) total ice content, and (d) ice crystal concentration in the simulations with low CCN concentration (N0 = 100 cm−3). The profiles are obtained by averaging over the time period of 2520–2880 s.

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

In E100_S the droplet concentration is larger than in E100_NS. There are three peaks of droplet concentration in E100_S at altitudes of 4.5, 6.5, and 9 km. We attribute these peaks to in-cloud nucleation. At the same time the droplet concentration in E100_NS rapidly decreases over height above 5.5 km. Results of E100_S agree well with results of observations (Heymsfield et al. 2009), indicating the appearance of a peak of droplet concentration above 8-km level. Liquid water content is also larger in E100_S than in E100_NS. A rapid decrease in LWC above 6 km can be attributed to riming caused by collisions of water drops with ice particles of various kinds. The total mass ice content has its maximum at 8 km. The vertical profiles of ice crystal concentration reveal a dramatic difference between E100_S and E100_NS: whereas at about 10-km altitude the peak of the time-averaged concentration of ice crystals is 6.5 cm−3 in E100_S, it is about 1 cm−3 in E100_NS. This is due to fewer supercooled droplets in the upper half of the mixed-phase region where the smallest aerosols are not present and hence homogeneous freezing is drastically reduced.

Note that the conditions of E100_NS do not exclude in-cloud nucleation altogether; rather, they significantly decrease its intensity. Peaks in the droplet concentration with increasing of height above 3.5-km altitude in E100_NS (Fig. 7a) can be accounted for by three factors. First, only CCN with radii below 0.0125 μm are eliminated in E100_NS, so that smaller CCN can be activated. Second, the simulations with resolution of 50 m reproduce separate bubbles in cloud structure that ascend along different trajectories and have different histories of supersaturation. In some bubbles having low updraft velocity at cloud base, CCN larger than 0.0125 μm in radii can be activated at high levels. Third, CCN are transported by the velocity field, so small CCN can ascend from below, starting at, say, 2-km altitude (i.e., above cloud base) upward, and penetrate cloud bubbles at higher levels (or produce new cloud elements by activation). The main conclusion is that the concentration of supercooled droplets in E100_S is significantly higher than in E100_NS, which indicates the role of small CCN in production of supercooled droplets and ice crystals.

Figures 8a–c shows the fields of concentration of platelike crystals in, respectively, E100_S, E100_NS, and E100_SG at t = 2760 s, when the cloud top is above 10 km (the level of homogeneous freezing). One can see a dramatic difference in the ice crystal concentrations in cases of significant and low amounts of small CCN. The area covered by ice crystals in E100_S is also larger than that in E100_NS. Figure 8d shows the time dependence of the maximum concentration of plate crystals forming by homogeneous freezing in all clean air simulations. One can see that in E100_S the cloud reaches the level of homogeneous freezing earlier than in E100_NS, and the ice crystal concentration in E100_S is substantially larger during the entire simulation. Size distributions of platelike crystals at levels between 10 and 11 km show the maximum at diameters of about 70 μm, indicating that these crystals are comparatively small [in agreement with the observations made by Heymsfield et al. (2009)].

Fig. 8.
Fig. 8.

(a)–(c) Fields of plate crystal concentration in (a) E100_S, (b) E100_NS, and (c) E100_SG. (d) The time dependences of the maximum concentration of plate crystals in different simulations.

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

The analysis shows that graupel mass content is larger in E100_S than in E100_NS (an example of the fields of graupel mass contents in these simulations is shown in Fig. 9). We attribute this difference to the fact that in E100_S liquid water content at upper levels is higher, which leads to a larger graupel mass formed by riming.

Fig. 9.
Fig. 9.

Fields of graupel mass contents in simulations (left) E100_S and (right) E100_NS at t = 3300 s.

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

The effect of GCCN requires a separate analysis. In simulation E100_NS_G that was free of small aerosols, an increase in the GCCN concentration leads to changes in cloud microphysics that agree with the widely accepted scenario described in the introduction when warm rain rapidly forms and collects cloud droplets. As a result, above 8 km LWC becomes negligible in E100_NS_G. The maximum concentration of ice crystals at the 10-km level was about 1 cm−3 in E100_NS_G (i.e., an order of magnitude less than in E100_S). The graupel mass was also significantly lower than in E100_S. The vertical profiles in E100_NS are quite close to those in E100_NS_G.

In the presence of small CCN the effect of GCCN changes dramatically. The fields of droplet concentration, vertical velocity, RWC, and supersaturation in E100_SG resemble those plotted for E100_S (Fig. 4), with some important differences shown in Fig. 10: the increase in concentration of GCCN intensifies the formation of warm rain, accelerating the decrease in the droplet concentration with height. This decrease leads to a more intense in-cloud nucleation and successive drop growth by condensation (accompanied by latent heat release). Rapid unloading accompanied by formation of new droplets leads to an increase in vertical velocity maximum to 14 m s−1 at z = 5 km (as compared to 12 m s−1 in E100_S) and an increase in supersaturation up to 12%–15% (as compared to 8% in E100_S). In the presence of small CCN this increase in supersaturation leads to intensification of in-cloud nucleation aloft. As a result, the E100_SG cloud is the most intense one. Figures 7 and 8 show that in E100_SG the droplet concentration, LWC, total ice content, and concentration of ice crystals are higher than in other simulated clouds developing in clean air. The graupel mass content is also significantly higher in E100_SG as compared to E100_NS_G. As will be shown below, an increase in GCCN concentration in the presence of small CCN leads to a significant increase in precipitation. Thus, the results show the existence of a synergetic effect of GCCN and small aerosols that leads to cloud intensification.

Fig. 10.
Fig. 10.

Fields of droplet concentration, RWC, W, and supersaturation in E100_SG simulation at t = 1860 min.

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

b. The effects of small CCN on the lightning probability

A detailed simulation of the processes of charge separation and lightning formation was beyond the scope of this paper. To evaluate the probability of lightning, the accumulated number of collisions (without coalescence) between ice crystals and graupel per unit of volume was calculated. The graupel–ice crystal collisions at a particular time step were taken into account if the supercooled LWC at this grid point exceeded 0.1 g m−3. The accumulated number of ice–graupel collisions in the presence of LWC is referred to as the lightning potential (LP). The LP used in this study is quite similar to that successfully applied by Khain et al. (2010) to characterize the location and intensity of lightning in a landfalling hurricane. We assume that larger values of LP correspond to higher rates of charge separation and a higher probability of lightning. Figure 11 shows the LP in E100_S, E100-NS, and E100_SG. Figure 11d presents the evolution of the lightning potential in the simulations. Comparison of the LP fields shows that an increase amounts CCN leads to an increase in the lightning probability by increasing the concentration of supercooled droplets, as well as the concentration of ice crystals and graupel. Figure 11 shows also that the lightning potential in E100-SG is larger than in other simulations for the reasons discussed above. We suppose that this result can explain the intense lightning observed in eyewalls of hurricanes during their mature stage when production of GCCN is a maximum and a huge amount of GCCN enters clouds via the cloud base.

Fig. 11.
Fig. 11.

Lightning potential in (a) E100_S, (b) E100_NS, and (c) E100_SG. (d) Time dependence of maximum values of lightning potential in different simulations. Lightning potential in E100_NS_G is close to that in E100_NS.

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

c. The effect of small AP on microphysics of polluted clouds

In this section we compare the results in two simulations E3500_S and E3500_NS in which clouds were developed in highly polluted atmosphere. The thermodynamic conditions in these simulations were the same as those in E100 simulations.

Supersaturation with respect to water in polluted clouds is lower than that in clouds developing in clean air. Because of the lower supersaturation, the first raindrops form at higher altitudes, which agrees with observations (e.g., Freud et al. 2008). Indeed, the simulations show that supersaturation with respect to water in polluted clouds does not exceed about 1% below a 5–7-km altitude. In the ascent above this level the process of raindrop formation and riming leads to a decrease in CWC and droplet concentration. At the same time, the vertical velocities increase with increasing height. As a result, Sw grows above 7-km altitude, reaching a maximum of about 3% in E3500_S and of 20% in E3500_NS. The difference in the values of supersaturation is attributed to the efficient in-cloud nucleation followed up by diffusional growth of new droplets in E3500_S at high levels. The higher maximum value of Sw in E3500_NS can be attributed to the lower concentration of cloud droplets in this simulation.

The existence of in-cloud nucleation is clearly seen in Fig. 12 showing the vertical profiles of (a) maximum droplet concentration, (b) CWC, (c) LWC, and (d) total ice content in the E3500_S and E3500_NS simulations at t = 2700–3300 s. One can see that in E3500_S (where the amount of small CCN is higher) the concentration of droplets and CWC is higher than in E3500_NS above 6 km (Fig. 12a). The CWC above 5 km level is also larger in E3500_S (Fig. 12b). Note that higher supersaturation leads to more efficient collisions and larger LWC (due to rain drops) in E3500_NS at z ~ 6 km than in E3500_S. Intense riming at this level leads also to larger graupel content in E3500_NS than in E3500_S at z ~ 6–7 km. Thus, the difference between the microphysical structures of clouds developing in polluted air in the presence and absence of small aerosols is less pronounced than the difference in the case of clouds developing in clean air. We attribute this effect to the fact that in clouds developing in polluted air droplets are small and warm rain is suppressed both in cases with and without smallest CCN. As a result, droplets ascend to a high level and participate in ice processes in both cases, leading to nearly similar vertical profiles of total ice content (Fig. 12d). The microstructure of polluted clouds in the presence and absence of small aerosols differs largely in concentration of small supercooled droplets at the upper levels (Fig. 12a), as well as in concentration of ice crystals above the level of homogeneous freezing.

Fig. 12.
Fig. 12.

Vertical profiles of (a) maximum droplet concentration, (b) CWC, (c) LWC, and (d) total ice content in simulations E3500_S and E3500_NS averaged over the time period 2700–3000 s.

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

Figure 13 shows the fields of plate crystal concentration in (left) E3500_S and (right) E3500_NS at t = 3300 s. In E3500_S the concentration of crystals is 350 cm−3 at 12 km, in agreement with the measurements performed by Heymsfield et al. (2009) in highly polluted clouds. In E3500_NS the maximum concentration of ice crystals is 40–50 cm−3.

Fig. 13.
Fig. 13.

Fields of plate crystals concentration in (left) E3500_S and (right) E3500_NS at t = 3300 s. Note that scales in the panels are different.

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

Figure 11d shows that the lightning potential in E3500_S is larger than in all the other simulations. In E3500_S the simulated concentration of small CCN is higher than in the other simulations.

d. Effect of small CCN on cloud dynamics and accumulated rain

According to the results of simulations, the differences in concentration of small CCN affect cloud dynamics, microphysics and precipitation amount.

Figures 14a and 14b show the vertical profiles of maximum vertical velocity in clouds developing in a clean atmosphere and polluted air, respectively. The profiles are averaged over the time period of several minutes. One can see that in clouds developing in a clean atmosphere small CCN significantly increase the maximum speeds of the cloudy updrafts and increase the height of the maxima by about 2 km. This can be attributed to larger latent heat release during the diffusional growth of droplets nucleated via in-cloud nucleation, as well as to higher intensity of riming accompanied by latent heat of freezing. The profiles in runs with small CCN taken into account resemble those presented by Heymsfield et al. (2009, their Fig. 1a), also indicating the maxima at the 10-km level. In agreement with the results reported by Petersen et al. (1999), vertical velocity in clouds developing in clean air increases strongly with increase of altitude above the roughly 5–6-km level. In clouds developing in polluted air effects of small aerosols on dynamics is not pronounced. Vertical velocity maximum takes place at 10 km both in E3500_S and in E3500_NS. The weak dynamical effect of small CCN in these cases agrees with the small differences in microstructure of clouds in these simulations (see Fig. 12).

Fig. 14.
Fig. 14.

Vertical profiles of maximum values of vertical velocity in clouds developing in (a) a clean atmosphere and (b) polluted air. The profiles are obtained by averaging over the time period 3000–3300 s.

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

Figure 15 shows the time evolution of accumulated rain in different experiments. One can see that, in clean air, rain at the surface begins earlier because of faster formation of raindrops by coalescence. Accumulated rain in polluted clouds was found to be larger than that from clouds developing in a clean atmosphere largely because of the more efficient ice processes (Khain 2009). Small CCN lead to an increase in surface precipitation as the accumulated rain in E100_S is larger than that in E100_NS. The synergetic effect of small CCN and GCCN is also seen in the precipitation amount, since the increase in the concentration of GCCN in the presence of small CCN in E100_SG leads to a significant increase in precipitation. At the same time, unlike in E100_NS, a similar increase in GCCN concentration in the absence of small CCN in E100_NS_G does not increase accumulated rain.

Fig. 15.
Fig. 15.

Time dependencies of space-averaged accumulated rain at the surface in different simulations.

Citation: Journal of the Atmospheric Sciences 69, 9; 10.1175/2011JAS3649.1

In case of high CCN concentrations (polluted clouds), the number of small CCN is also higher than in clean air. Precipitation in E3500_S and E3500_NS was found to be quite similar in spite of the higher droplet concentration and higher CWC at upper levels in E3500_S. This result can be attributed to the fact that in the simulations with high CCN concentrations, the concentration of drops is high both in E3500_S and E3500_NS. As a result, supersaturation below 6–7 km remains low and many droplets ascend in updrafts to higher levels. In E3500_S, the in-cloud nucleation increases droplet concentration, which decreases supersaturation, thus retarding droplet growth. In E3500_NS, the supersaturation is higher and ascending droplets grow faster, efficiently producing graupel and hail by riming. As a result, the amount of graupel in E3500_NS was found to be a little larger than in E3500_S, which leads to slightly larger accumulated rain.

5. Discussion and conclusions

This study was motivated by observations of very high ice crystal concentrations in anvils of strong maritime clouds. The phenomenon of intense lightning in eyewalls of hurricanes despite the presence of huge GCCN concentrations also required an explanation. The simulations with a mixed-phase cloud microphysics model with a resolution of 50 m and accurate calculation of cloud supersaturation, conducted in this study, show that all these phenomena can be explained by in-cloud nucleation of small droplets at temperatures below about −20°C. It is shown that in maritime clouds, the efficient production of warm rain leads to a decrease in droplet concentration aloft, which is often accompanied by an increase in vertical velocity due to atmospheric instability, unloading precipitation, and ice formation. As a result, supersaturation in maritime clouds can exceed 8%–10%. To be efficient, in-cloud nucleation requires a significant concentration of small Aitken mode CCN with diameters below about 0.03–0.05 μm. It is shown that in cases where the vertical velocity in clouds reaches about 25 m s−1 at z ~ 10 km and the slope parameter k = 0.9, in-cloud nucleation at several kilometers above cloud base can produce ice crystals of about 5–15 cm−3 concentration by homogeneous freezing of small cloud droplets. This value of the slope parameter is typical of maritime conditions, according to Hobbs (1993) and Pruppacher and Klett (1997). Ice crystals of said concentration at anvil levels were measured in maritime clouds of the corresponding intensity by Heymsfield et al. (2009).

These conclusions are broadly consistent with the conceptual scheme (Fig. 1) and with the findings of a number of studies, including Pinsky and Khain (2002) and Segal et al. (2003), whose simulations demonstrated that in-cloud nucleation far above cloud base plays a crucial role in formation of structure of mixed-phase clouds. Phillips et al. (2005) showed that in-cloud nucleation of droplets far above cloud base produced most of the supercooled droplets freezing homogeneously. It is usually assumed that the effects of small CCN and GCCN are opposite: while small CCN tend to delay the warm rain formation, GCCN accelerate it. This study shows that small CCN activated at significant distances above the cloud base and GCCN can “unite” in a synergetic effect, fostering formation of new droplets at higher levels via in-cloud nucleation. GCCN also foster formation of warm rain, which then is unloaded from the updrafts and accelerates them, leading to a higher supersaturation in clouds aloft and to a greater production of supercooled droplets at upper levels via in-cloud nucleation. As a result, both the concentration of small droplets aloft and the concentration of ice crystals in cloud anvils forming by homogeneous freezing increases if the CCN spectrum at cloud base contains both Aitken CCN and GCCN. This synergetic effect leads to convective invigoration and increases accumulated rain at the surface. An increase in GCCN concentration in simulations without small CCN (no in-cloud nucleation) leads to a decrease in supercooled water, as well as to a low concentration of ice crystals at the upper cloud levels, which agrees with the widely accepted interpretation of the role of GCCN.

The process of in-cloud nucleation allows one to explain the formation of high concentrations of ice crystals, the high optical depth of anvils in deep tropical clouds, and the accompanying cirrus clouds in the ITCZ. It also explains the lightning in very strong convective clouds in the ITCZ over the oceans. The synergetic effect of the smallest CCN and GCCN allows one to explain the formation of lighting in extremely maritime clouds in eyewalls of hurricanes where spray drops with radii up to several tens of microns reach the cloud base.

In our simulations focusing on the effects of the smallest CCN penetrating through cloud base, the roles of both of vertical velocity and cloud depth were of the utmost importance. In supplementary simulations (not discussed in the paper) for which the original sounding data were used, the maximum vertical velocity of 18 m s−1 was reached near 8.5–9 km, but not at 10 km. In this case the concentration of ice crystals in cloud anvils was an order of magnitude less than in the case with the maximum vertical velocity of 25 m s−1at 10 km. The lightning potential was lower by order of magnitude accordingly. In supplementary simulations where the relative humidity was decreased to get the maximum vertical velocity of 10–13 m s−1, the concentration of ice crystals produced largely by primary ice nucleation in cloud anvils was about 100 L−1. The concentration of supercooled droplets above 8 km was negligible. The cloud top was at about 10 km altitude or lower, and the lightning potential was negligibly small. This is in agreement with findings by Black and Hallett (1999) showing that lightning in hurricane clouds is possible if the vertical velocity exceeds 13 m s−1. As regards the main factors leading to lightning, the “dynamical” and “aerosol” hypotheses are widely discussed in the literature (e.g., Williams and Stanfill 2002; Williams et al. 2004, 2005; Sherwood et al. 2006). In the first case, the vertical velocity is considered as the major factor; in the second case the aerosol effects are assumed to be dominating. The results of the present study suggest a synergetic effect of these two factors. A high vertical velocity leads to a high supersaturation and to nucleation of small aerosols aloft needed for creation of supercooled droplets. At the same time, in-cloud nucleation increases vertical velocity, supposedly due to extra latent heat released in droplet diffusion growth and riming. We would like to stress that charge separation is a result of the microphysical processes inside clouds (e.g., graupel–ice crystal collisions). Storm electrification is not directly caused by vertical velocity, wind shear, and other dynamical factors. The role of the dynamical factors is to create favorable conditions for intensification of the microphysical processes of charge separation.

The in-cloud nucleation can be responsible for the well-known experimental fact that DSDs at most vertical levels reveal the existence of very small droplets with diameters below 10 μm. Because in the deep layers the updraft speed tends to increase with increasing height, supersaturation increases continuously over height, leading to nearly permanent in-cloud droplet activation. Thus, we can expect the presence of very small droplets in the DSD over much of the cloudy updraft, even though each small droplet becomes quickly larger than 10 μm because of condensation. This statement apparently agrees with the in situ measurements of the smallest droplets obtained in the LBA-SMOCC field experiment (Andreae et al. 2004; Freud et al. 2008) performed up to heights of 4.5 km, as well as in CAIPEEX (2009) (Prabha et al. 2011) where premonsoon and monsoon clouds in central India were measured up to heights of 7 km. The existence of small droplets in the DSD within a deep layer is one of the main reasons of significant DSD dispersion (~0.3) and wide occurrence of bimodal and multimodal DSD (Prabha et al. 2011). It is also shown that small aerosols allowing nucleation of droplets a few kilometers above cloud base boost the speed of the cloudy updraft and lead to precipitation enhancement in maritime clouds. Thus, small aerosols can contribute to an increase in vertical velocity in TC eyewall clouds.

The present study shows that the smallest CCN can play an important role in microphysics of clouds developing in highly polluted atmosphere where No is high. Supersaturation in polluted clouds is much lower than in clouds developing in clean air. At the same time, at cold temperatures the concentration of droplets decreases as a result of collisions of cloud droplets with raindrops and ice particles. This can lead to an increase in supersaturation by up to several percent and to in-cloud nucleation at heights above 6–7 km. According to the results obtained in this study, at the same high value of No (if it is high), the concentration of ice crystals in cloud anvils can vary from several tens to several hundreds per cubic centimeter depending on the concentration of small CCN at cloud base. Assuming that small CCN exist within the CCN spectra, we found concentrations of ice crystals of 350 cm−3 in cloud anvils, accounted for by homogeneous freezing of the numerous supercooled small droplets at about 10-km altitude, with is in agreement with the results obtained by Heymsfield et al. (2009). An increase in No without changing the value of the slope parameter also leads to an increase in the concentration of the smallest CCN. Thus, in studies where only No was increased, an increase in the concentration of small CCN was guaranteed. For instance, substantial masses of supercooled water were measured by Rosenfeld and Woodley (2000) at the temperature −37.5°C in Texas summertime clouds. These cloud droplets gave rise to formation of about 500 cm−3 ice crystals in cloud anvils due to homogeneous freezing. Such concentrations of supercooled droplets and ice crystals were simulated by Khain et al. (2001b) using the earlier version of the HUCM. The interpretation then given for the numerical results stated that droplets in polluted continental clouds remain small and do not collide with ice particles during their ascent from cloud base up to the homogeneous freezing level. With the new finding of this study, we suppose that in-cloud nucleation could have played an important role in the case of Texas clouds.

It seems, therefore, that small aerosols causing in-cloud nucleation is the important factor of DSD formation for wide range of conditions from clean maritime (eyewalls of TC) to extremely polluted continental clouds (premonsoon clouds in India). This enhances the requirements on measurements of aerosols in atmosphere. The range of supersaturation variations in the laboratory should be increased to measure the concentration of particles that can be activated at supersaturations as high as 5%–10% and even higher. The vertical variation of the droplet concentration throughout the depth of the troposphere must also be measured.

Two main sources of small CCN over the ocean can be assumed. One source is related to the chemical reactions involving trace gases condensing to form “secondary” ultrafine particles (the Aitken mode), which may later grow due to condensation or by collisions between APs and then create the accumulation mode. Clarke and Kapustin (2002) analyzed aerosol observations from the field campaigns over the Pacific Ocean and reported the existence of ultrafine aerosols with sizes smaller than 0.01 μm in the upper troposphere, especially near the outflow from deep convection (e.g., near the ITCZ). The authors assumed that these small APs, apparently generated in the outflow aloft, should subside through the free troposphere and may be eventually entrained into the boundary layer and into deep clouds through their lateral boundaries. The important role of such entrainment was stressed by Fridlind et al. (2004) and Phillips et al. (2005). These secondary sources of ultrafine aerosols were also observed in the marine boundary layer (e.g., Covert et al. 1992). As to the second source of small aerosols, we hypothesize that they can be of continental nature and penetrate the air over the ocean with the intrusion of African dust and creating favorable conditions for lightning in ITCZ near the African coast (Chronis et al. 2007). Since CCN represents only a soluble fraction of AP, dust particles that are not small can activate just like small CCN. Observations of Saharan dust over the eastern Atlantic by Twohy et al. (2009) confirm that the slightly hygroscopic nature of the dust means that it turns into CCN, with a fraction of the dust size distribution activated at cloud base. To answer the question as regards the nature and amount of small aerosols in the maritime tropical atmosphere, more microphysical measurements of deep marine clouds and aerosol spectra in the zones of lightning over the ocean are required. The existence of the bimodal cloud droplet spectra in zones of high updrafts in cloud cores would be consistent with the role of small aerosols.

If the role of small aerosols is as important as this study suggests, many concepts concerning the role of sea spray, giant CCN, and so on can be reconsidered, at least as far as the microphysics of deep convective clouds developing over the oceans is concerned.

Note that the effects from in-cloud nucleation of droplets discussed in the present study could not be found in most numerical models with bulk microphysics, partially because they perform nucleation at cloud base only. We suppose that a description of in-cloud nucleation has to be included in meteorological models [e.g., as done in the double-moment bulk scheme by Phillips et al. (2007a, 2009)] for better understanding of aerosol effects. The conclusions reached regarding the aerosol effects on precipitation from maritime clouds should possibly be revised to take into account the role of smallest aerosols.

Acknowledgments

The authors express deep gratitude to Prof. Hudson and to Prof. Hobbs (Prof. Hobbs passed away in mid-2005) for useful discussions. The study has been conducted under the support of the Israel Academy of Science (Grant 140/07) and by the Office of Science (BER), U.S. Department of Energy (Grant DE-SC0006788). The second author was supported also by the Office of Science (BER), U.S. Department of Energy with an award (Grant DE-SC0002383) for study of mechanisms for the influence from aerosols on glaciated clouds.

REFERENCES

  • Andreae, M. O., D. Rosenfeld, P. Artaxo, A. A. Costa, G. P. Frank, K. M. Longlo, and M. A. F. Silva-Dias, 2004: Smoking rain clouds over the Amazon. Science, 303, 13371342.

    • Search Google Scholar
    • Export Citation
  • Black, R. A., and J. Hallett, 1999: Electrification of the hurricane. J. Atmos. Sci., 56, 20042028.

  • Blyth, A. M., J. Zhou, and J. Latham, 1998: Influence of ultra-giant nuclei on cumulus clouds observed during the Small Cumulus Microphysics Study in Florida. Preprints, Conf. on Cloud Physics, Everett, WA, Amer. Meteor. Soc., 492–493.

    • Search Google Scholar
    • Export Citation
  • Bott, A., 1989: A positive definite advection scheme obtained by nonlinear renormalization of the advective fluxes. Mon. Wea. Rev., 117, 10061015.

    • Search Google Scholar
    • Export Citation
  • Bott, A., 1998: A flux method for the numerical solution of the stochastic collection equation. J. Atmos. Sci., 55, 22842293.

  • Carrió, G. G., S. C. van den Heever, and W. R. Cotton, 2007: Impacts of nucleating aerosol on anvil-cirrus clouds: A modeling study. Atmos. Res., 84, 111131.

    • Search Google Scholar
    • Export Citation
  • Cecil, D. J., E. J. Zipser, and S. W. Nebitt, 2002a: Reflectivity, ice scattering, and lightning characteristics of hurricane eyewalls and rainbands. Part I: Quantitative description. Mon. Wea. Rev., 130, 769784.

    • Search Google Scholar
    • Export Citation
  • Cecil, D. J., E. J. Zipser, and S. W. Nebitt, 2002b: Reflectivity, ice scattering, and lightning characteristics of hurricane eyewalls and rainbands. Part II: Intercomparison of observations. Mon. Wea. Rev., 130, 785801.

    • Search Google Scholar
    • Export Citation
  • Chronis, T., E. Williams, E. Anagnostou, and W. Petersen, 2007: African lightning: Indicator of tropical Atlantic cyclone formation. Eos, Trans. Amer. Geophys. Union, 88, 397, doi:10.1029/2007EO400001.

    • Search Google Scholar
    • Export Citation
  • Clarke, A. D., and V. Kapustin, 2002: A Pacific aerosol survey. Part I: A decade of data on particle production, transport, evolution, and mixing in the troposphere. J. Atmos. Sci., 59, 363382.

    • Search Google Scholar
    • Export Citation
  • Clarke, A. D., S. R. Owens, and J. C. Zhou, 2006: An ultrafine sea-salt flux from breaking waves: Implications for cloud condensation nuclei in the remote marine atmosphere. J. Geophys. Res., 111, D06202, doi:10.1029/2005JD006565.

    • Search Google Scholar
    • Export Citation
  • Covert, D. S., V. N. Kapustin, P. K. Quinn, and T. S. Bates, 1992: New particle formation in the marine boundary layer. J. Geophys. Res., 97 (D18), 20 58120 589.

    • Search Google Scholar
    • Export Citation
  • Demetriades, N. W. S., and R. L. Holle, 2006: Long-range lightning nowcasting applications for tropical cyclones. Preprints, Conf. on Meteorological Application of Lightning Data, Atlanta, GA, Amer. Meteor. Soc., 9 pp.

    • Search Google Scholar
    • Export Citation
  • Ferrier, B. S., and R. A. Houze, 1989: One-dimensional time-dependent modeling of GATE cumulonimbus convection. J. Atmos. Sci., 46, 330352.

    • Search Google Scholar
    • Export Citation
  • Fierro, A. O., L. Leslie, E. Mansell, J. Straka, D. MacGorman, and C. Ziegler, 2007: A high-resolution simulation of microphysics and electrification in an idealized hurricane-like vortex. Meteor. Atmos. Phys., 98, 1333, doi:10.1007/s00703-006-0237-0.

    • Search Google Scholar
    • Export Citation
  • Fierro, A. O., X.-M. Shao, T. Hamlin, J. M. Reisner, and J. Harlin, 2010: Evolution of eyewall convective events as indicated by intra-cloud and cloud-to-ground lightning activity during the rapid intensification of Hurricanes Rita and Katrina. Mon. Wea. Rev., 139, 14921504.

    • Search Google Scholar
    • Export Citation
  • Freud, E., D. Rosenfeld, M. O. Andreae, A. A. Costa, and P. Artaxo, 2008: Robust relations between CCN and the vertical evolution of cloud drop size distribution in deep convective clouds. Atmos. Chem. Phys., 8, 16611675.

    • Search Google Scholar
    • Export Citation
  • Fridlind, A., and Coauthors, 2004: Evidence for the predominance of mid-tropospheric aerosols as subtropical anvil cloud nuclei. Science, 304, 718722.

    • Search Google Scholar
    • Export Citation
  • Fukuta, N., 1993: Water supersaturation in convective clouds. Atmos. Res., 30, 105126.

  • Hallett, J., and S. C. Mossop, 1974: Production of secondary ice particles during the riming process. Nature, 249, 2628.

  • Heymsfield, A. J., L. M. Miloshevich, C. Schmitt, A. Bansemer, C. Twohy, M. R. Poellot, A. Fridlind, and H. Gerber, 2005: Homogeneous ice nucleation in subtropical and tropical convection and its influence on cirrus anvil microphysics. J. Atmos. Sci., 62, 4164.

    • Search Google Scholar
    • Export Citation
  • Heymsfield, A. J., A. Bansemer, G. Heymsfield, and A. O. Fierro, 2009: Microphysics of maritime tropical convective updrafts at temperatures from −20°C to −60°C. J. Atmos. Sci., 66, 35303562.

    • Search Google Scholar
    • Export Citation
  • Hobbs, P. V., 1993: Aerosol–Cloud–Climate Interactions. Academic Press, 236 pp.

  • Hudson, J. G., and S. S. Yum, 2002: Cloud condensation nuclei spectra and polluted and clean clouds over the Indian Ocean. J. Geophys. Res., 107, 8022, doi:10.1029/2001JD000829.

    • Search Google Scholar
    • Export Citation
  • Jaenicke, R., 1993: Tropospheric aerosols. Aerosol–Cloud–Climate Interactions, P. Hobbs, Ed., Academic Press, 1–31.

  • Jordan, C. L., 1958: Mean soundings for the West Indies area. J. Meteor., 15, 9197.

  • Jorgensen, D. P., and M. A. LeMone, 1989: Vertical velocity characteristics of oceanic convection. J. Atmos. Sci., 46, 621640.

  • Jorgensen, D. P., E. J. Zipser, and M. A. LeMone, 1985: Vertical motions in intense hurricanes. J. Atmos. Sci., 42, 839856.

  • Khain, A. P., 2009: Notes on state-of-the-art investigations of aerosol effects on precipitation: A critical review. Environ. Res. Lett., 4, 015004, doi:10.1088/1748-9326/4/1/015004.

    • Search Google Scholar
    • Export Citation
  • Khain, A. P., and I. Sednev, 1996: Simulation of precipitation formation in the eastern Mediterranean coastal zone using a spectral microphysics cloud ensemble model. Atmos. Res., 43, 77110.

    • Search Google Scholar
    • Export Citation
  • Khain, A. P., M. Ovtchinnikov, M. Pinsky, A. Pokrovsky, and H. Krugliak, 2000: Notes on the state-of-the-art numerical modeling of cloud microphysics. Atmos. Res., 55, 159224.

    • Search Google Scholar
    • Export Citation
  • Khain, A. P., M. B. Pinsky, M. Shapiro, and A. Pokrovsky, 2001a: Graupel–drop collision efficiencies. J. Atmos. Sci., 58, 25712595.

    • Search Google Scholar
    • Export Citation
  • Khain, A. P., D. Rosenfeld, and A. Pokrovsky, 2001b: Simulating convective clouds with sustained supercooled liquid water down to −37.5°C using a spectral microphysics model. Geophys. Res. Lett., 28, 38873890.

    • Search Google Scholar
    • Export Citation
  • Khain, A. P., A. Pokrovsky, M. Pinsky, A. Seifert, and V. Phillips, 2004: Effects of atmospheric aerosols on deep convective clouds as seen from simulations using a spectral microphysics mixed-phase cumulus cloud model. Part I: Model description. J. Atmos. Sci., 61, 29632982.

    • Search Google Scholar
    • Export Citation
  • Khain, A. P., N. Benmoshe, and A. Pokrovsky, 2008a: Factors determining the impact of aerosols on surface precipitation from clouds: An attempt at classification. J. Atmos. Sci., 65, 17211748.

    • Search Google Scholar
    • Export Citation
  • Khain, A. P., N. Cohen, B. Lynn, and A. Pokrovsky, 2008b: Possible aerosol effects on lightning activity and structure of hurricanes. J. Atmos. Sci., 65, 36523667.

    • Search Google Scholar
    • Export Citation
  • Khain, A. P., B. Lynn, and J. Dudhia, 2010: Aerosol effects on intensity of landfalling hurricanes as seen from simulations with WRF model with spectral bin microphysics. J. Atmos. Sci., 67, 365384.

    • Search Google Scholar
    • Export Citation
  • Korolev, A. V., 1994: A study of bimodal droplet size distributions in stratiform clouds. Atmos. Res., 32, 143170.

  • Korolev, A. V., and I. Mazin, 2003: Supersaturation of water vapor in clouds. J. Atmos. Sci., 60, 29572974.

  • Lay, E. H., A. R. Jacobson, R. H. Holzworth, C. J. Rodger, and R. L. Dowden, 2007: Local time variation in land/ocean lightning flash density as measured by the World Wide Lightning Location Network. J. Geophys. Res., 112, D13111, doi:10.1029/2006JD007944.

    • Search Google Scholar
    • Export Citation
  • Leary, L. A., and E. A. Ritchie, 2009: Lightning flash rates as an indicator of tropical cyclone genesis in the eastern North Pacific. Mon. Wea. Rev., 137, 34563470.

    • Search Google Scholar
    • Export Citation
  • Levin, Z., and W. R. Cotton, Eds., 2009: Aerosol Pollution Impact on Precipitation: A Scientific Review. Springer, 386 pp.

  • Ludlam, F. H., 1980: Clouds and Storms. The Pennsylvania State University Press, 405 pp.

  • Meyers, M. P., P. DeMott, and W. R. Cotton, 1992: Evaluation of the potential for wintertime quantitative precipitation forecasting over mountainous terrain with an explicit cloud model. Part I: Two-dimensional sensitivity experiments. J. Appl. Meteor., 31, 2650.

    • Search Google Scholar
    • Export Citation
  • Molinari, J., P. Moore, and V. Idone, 1999: Convective structure of hurricanes as revealed by lightning locations. Mon. Wea. Rev., 127, 520534.

    • Search Google Scholar
    • Export Citation
  • Ochs, H. T., 1978: Moment-conserving techniques for warm cloud microphysical computation. Part II: Model testing and results. J. Atmos. Sci., 35, 19591973.

    • Search Google Scholar
    • Export Citation
  • Orville, R. E., and J. M. Coyne, 1999: Cloud-to-ground lightning in tropical cyclones (1986–1996). Preprints, 23rd Conf. on Hurricanes and Tropical Meteorology, Dallas, TX, Amer. Meteor. Soc., 123–127.

    • Search Google Scholar
    • Export Citation
  • Ovtchinnikov, M., Y. L. Kogan, and A. M. Blyth, 2000: An investigation of ice production mechanisms in small cumuliform clouds using a 3D cloud model with explicit microphysics. Part II: Case study of New Mexico cumulus clouds. J. Atmos. Sci., 57, 30043020.

    • Search Google Scholar
    • Export Citation
  • Petersen, W. A., R. C. Cifelli, S. A. Rutledge, B. S. Ferrier, and B. F. Smull, 1999: Shipborne dual-Doppler operations during TOGA COARE: Integrated observations of storm kinematics and electrification. Bull. Amer. Meteor. Soc., 80, 8197.

    • Search Google Scholar
    • Export Citation
  • Phillips, V. T. J., and Coauthors, 2005: Anvil glaciation in a deep cumulus updraft over Florida simulated with an explicit microphysics model. I: The impact of various nucleation processes. Quart. J. Roy. Meteor. Soc., 131, 20192046.

    • Search Google Scholar
    • Export Citation
  • Phillips, V. T. J., L. J. Donner, and S. Garner, 2007a: Nucleation processes in deep convection simulated by a cloud-system resolving model with double-moment bulk microphysics. J. Atmos. Sci., 64, 738761.

    • Search Google Scholar
    • Export Citation
  • Phillips, V. T. J., A. Pokrovsky, and A. Khain, 2007b: The influence of time-dependent melting on the dynamics and precipitation production in maritime and continental storm clouds. J. Atmos. Sci., 64, 338359.

    • Search Google Scholar
    • Export Citation
  • Phillips, V. T. J., and Coauthors, 2009: Potential impacts from biological aerosols on ensembles of continental clouds simulated numerically. Biogeosciences, 6, 9871014.

    • Search Google Scholar
    • Export Citation
  • Pinsky, M., and A. P. Khain, 2002: Effects of in-cloud nucleation and turbulence on droplet spectrum formation in cumulus clouds. Quart. J. Roy. Meteor. Soc., 128, 501533.

    • Search Google Scholar
    • Export Citation
  • Pinsky, M., A. P. Khain, and M. Shapiro, 2001: Collision efficiency of drops in a wide range of Reynolds numbers: Effects of pressure on spectrum evolution. J. Atmos. Sci., 58, 742764.

    • Search Google Scholar
    • Export Citation
  • Pinsky, M., A. Khain, and H. Krugliak, 2008: Collisions of cloud droplets in a turbulent flow. Part V: Application of detailed tables of turbulent collision rate enhancement to simulation of droplet spectra evolution. J. Atmos. Sci., 63, 357374.

    • Search Google Scholar
    • Export Citation
  • Prabha, T., A. Khain, R. S. Maheshkumar, G. Pandithurai, J. R. Kulkarni, M. Konwar, and B. N. Goswami, 2011: Microphysics of premonsoon and monsoon clouds as seen from in situ measurements during the Cloud Aerosol Interaction and Precipitation Enhancement Experiment (CAIPEEX). J. Atmos. Sci., 68, 18821901.

    • Search Google Scholar
    • Export Citation
  • Price, C., M. Asfur, and Y. Yair, 2009: Maximum hurricane intensity preceded by increase in lightning frequency. Nature Geosci. Lett., 2, 329332, doi:10.1038/NGEO477; Corrigendum, 2, 444.

    • Search Google Scholar
    • Export Citation
  • Pruppacher, H. R., 1995: A new look at homogeneous ice nucleation in supercooled water drops. J. Atmos. Sci., 52, 19241933.

  • Pruppacher, H. R., and J. D. Klett, 1997: Microphysics of Clouds and Precipitation. 2nd ed. Oxford University Press, 963 pp.

  • Rodgers, E., J. Weinman, H. Pierce, and W. Olson, 2000: Tropical cyclone lightning distribution and its relationship to convection and intensity change. Preprints, 24th Conf. on Hurricanes and Tropical Meteorology, Ft. Lauderdale, FL, Amer. Meteor. Soc., 537–541.

    • Search Google Scholar
    • Export Citation
  • Rosenfeld, D., and W. L. Woodley, 2000: Deep convective clouds with sustained highly supercooled liquid water until −37.5°C. Nature, 405, 440442.

    • Search Google Scholar
    • Export Citation
  • Rosenfeld, D., R. Lahav, A. Khain, and M. Pinsky, 2002: The role of sea spray in cleansing air pollution over ocean via cloud processes. Science, 297, 16671670.

    • Search Google Scholar
    • Export Citation
  • Rosenfeld, D., A. Khain, B. Lynn, and W. L. Woodley, 2007: Simulation of hurricane response to suppression of warm rain by sub-micron aerosols. Atmos. Chem. Phys. Discuss., 7, 56475674.

    • Search Google Scholar
    • Export Citation
  • Saunders, C. P. R., 1993: A review of thunderstorm electrification processes. J. Appl. Meteor., 32, 642655.

  • Segal, Y., M. Pinsky, A. Khain, and C. Erlick, 2003: Thermodynamic factors influencing the bimodal spectrum formation in cumulus clouds. Atmos. Res., 66, 4364.

    • Search Google Scholar
    • Export Citation
  • Shao, X. M., and Coauthors, 2005: Katrina and Rita were lit up with lightning. Eos, Trans. Amer. Geophys. Union, 86, 398399.

  • Sherwood, S. C., V. Phillips, and J. S. Wettlaufer, 2006: Small ice crystals and the climatology of lightning. Geophys. Res. Lett., 33, L05804, doi:10.1029/2005GL025242.

    • Search Google Scholar
    • Export Citation
  • Shpund, J., M. Pinsky, and A. Khain, 2011: Microphysical structure of marine boundary layer under strong wind and spray formation as seen from simulations using a 2D explicit microphysical model. Part I: The impact of large eddies. J. Atmos. Sci., 68, 23662384.

    • Search Google Scholar
    • Export Citation
  • Squires, K., and S. Businger, 2008: The morphology of eyewall lightning outbreaks in two category 5 hurricanes. Mon. Wea. Rev., 136, 17061726.

    • Search Google Scholar
    • Export Citation
  • Squires, P., 1952: The growth of cloud drops by condensation. I. General characteristics. Aust. J. Sci. Res., 5, 5986.

  • Takahashi, T., 1978: Riming electrification as a charge generation mechanism in thunderstorms. J. Atmos. Sci., 35, 15361548.

  • Takahashi, T., and K. Kuhara, 1993: Precipitation mechanisms of cumulonimbus clouds at Pohnpei, Micronesia. J. Meteor. Soc. Japan, 71, 2131.

    • Search Google Scholar
    • Export Citation
  • Takahashi, T., T. Endoh, and G. Wakahama, 1991: Vapor diffusional growth of free-falling snow crystals between −3° and −23°C. J. Meteor. Soc. Japan, 69, 1530.

    • Search Google Scholar
    • Export Citation
  • Toba, Y., 1965: On the giant sea-salt particles in the atmosphere. II: Theory of the vertical distribution in the 10-m layer over the ocean. Tellus, 17, 365382.

    • Search Google Scholar
    • Export Citation
  • Twohy, C. H., and Coauthors, 2009: Saharan dust particles nucleate droplets in eastern Atlantic clouds. Geophys. Res. Lett., 36, L01807, doi:10.1029/2008GL035846.

    • Search Google Scholar
    • Export Citation
  • Vali, G., 1994: Freezing rate due to heterogeneous nucleation. J. Atmos. Sci., 51, 18431856.

  • Warner, J., 1969a: The microstructure of cumulus cloud. Part I: General features of the droplet spectrum. J. Atmos. Sci., 26, 10491059.

    • Search Google Scholar
    • Export Citation
  • Warner, J., 1969b: The microstructure of cumulus cloud. Part II: The effect of droplet size distribution of the cloud nucleus spectrum and updraft velocity. J. Atmos. Sci., 26, 12721282.

    • Search Google Scholar
    • Export Citation
  • Warner, J., 1973: The microstructure of cumulus cloud. Part IV: The effect on the droplet spectrum of mixing between cloud and environment. J. Atmos. Sci., 30, 256261.

    • Search Google Scholar
    • Export Citation
  • Williams, E., and S. Stanfill, 2002: The physical origin of the land–ocean contrast in lightning activity. C. R. Phys., 3, 12771292.

    • Search Google Scholar
    • Export Citation
  • Williams, E., T. Chan, and D. Boccippio, 2004: Islands as miniature continents: Another look at the land–ocean lightning contrast. J. Geophys. Res., 109, D16206, doi:10.1029/2003JD003833.

    • Search Google Scholar
    • Export Citation
  • Williams, E., V. Mushtak, D. Rosenfeld, S. Goodman, and D. Boccippio, 2005: Thermodynamic conditions favorable to superlative thunderstorm updraft, mixed phase microphysics and lightning flash rate. Atmos. Res., 76, 288306.

    • Search Google Scholar
    • Export Citation
  • Yin, Y., Z. Levin, T. Reisin, and S. Tzivion, 2000: The effects of giant cloud condensational nuclei on the development of precipitation in convective clouds: A numerical study. Atmos. Res., 53, 91116.

    • Search Google Scholar
    • Export Citation
  • Yin, Y., K. S. Carslaw, and G. Feingold, 2005: Vertical transport and processing of aerosols in a mixed-phase convective cloud and the feedback on cloud development. Quart. J. Roy. Meteor. Soc., 131, 221245.

    • Search Google Scholar
    • Export Citation
Save
  • Andreae, M. O., D. Rosenfeld, P. Artaxo, A. A. Costa, G. P. Frank, K. M. Longlo, and M. A. F. Silva-Dias, 2004: Smoking rain clouds over the Amazon. Science, 303, 13371342.

    • Search Google Scholar
    • Export Citation
  • Black, R. A., and J. Hallett, 1999: Electrification of the hurricane. J. Atmos. Sci., 56, 20042028.

  • Blyth, A. M., J. Zhou, and J. Latham, 1998: Influence of ultra-giant nuclei on cumulus clouds observed during the Small Cumulus Microphysics Study in Florida. Preprints, Conf. on Cloud Physics, Everett, WA, Amer. Meteor. Soc., 492–493.

    • Search Google Scholar
    • Export Citation
  • Bott, A., 1989: A positive definite advection scheme obtained by nonlinear renormalization of the advective fluxes. Mon. Wea. Rev., 117, 10061015.

    • Search Google Scholar
    • Export Citation
  • Bott, A., 1998: A flux method for the numerical solution of the stochastic collection equation. J. Atmos. Sci., 55, 22842293.

  • Carrió, G. G., S. C. van den Heever, and W. R. Cotton, 2007: Impacts of nucleating aerosol on anvil-cirrus clouds: A modeling study. Atmos. Res., 84, 111131.

    • Search Google Scholar
    • Export Citation
  • Cecil, D. J., E. J. Zipser, and S. W. Nebitt, 2002a: Reflectivity, ice scattering, and lightning characteristics of hurricane eyewalls and rainbands. Part I: Quantitative description. Mon. Wea. Rev., 130, 769784.

    • Search Google Scholar
    • Export Citation
  • Cecil, D. J., E. J. Zipser, and S. W. Nebitt, 2002b: Reflectivity, ice scattering, and lightning characteristics of hurricane eyewalls and rainbands. Part II: Intercomparison of observations. Mon. Wea. Rev., 130, 785801.

    • Search Google Scholar
    • Export Citation
  • Chronis, T., E. Williams, E. Anagnostou, and W. Petersen, 2007: African lightning: Indicator of tropical Atlantic cyclone formation. Eos, Trans. Amer. Geophys. Union, 88, 397, doi:10.1029/2007EO400001.

    • Search Google Scholar
    • Export Citation
  • Clarke, A. D., and V. Kapustin, 2002: A Pacific aerosol survey. Part I: A decade of data on particle production, transport, evolution, and mixing in the troposphere. J. Atmos. Sci., 59, 363382.

    • Search Google Scholar
    • Export Citation
  • Clarke, A. D., S. R. Owens, and J. C. Zhou, 2006: An ultrafine sea-salt flux from breaking waves: Implications for cloud condensation nuclei in the remote marine atmosphere. J. Geophys. Res., 111, D06202, doi:10.1029/2005JD006565.

    • Search Google Scholar
    • Export Citation
  • Covert, D. S., V. N. Kapustin, P. K. Quinn, and T. S. Bates, 1992: New particle formation in the marine boundary layer. J. Geophys. Res., 97 (D18), 20 58120 589.

    • Search Google Scholar
    • Export Citation
  • Demetriades, N. W. S., and R. L. Holle, 2006: Long-range lightning nowcasting applications for tropical cyclones. Preprints, Conf. on Meteorological Application of Lightning Data, Atlanta, GA, Amer. Meteor. Soc., 9 pp.

    • Search Google Scholar
    • Export Citation
  • Ferrier, B. S., and R. A. Houze, 1989: One-dimensional time-dependent modeling of GATE cumulonimbus convection. J. Atmos. Sci., 46, 330352.

    • Search Google Scholar
    • Export Citation
  • Fierro, A. O., L. Leslie, E. Mansell, J. Straka, D. MacGorman, and C. Ziegler, 2007: A high-resolution simulation of microphysics and electrification in an idealized hurricane-like vortex. Meteor. Atmos. Phys., 98, 1333, doi:10.1007/s00703-006-0237-0.