Atmos. Chem. Phys., 6, 2593-2649, 2006 www.atmos-chem-phys.net/6/2593/2006 Atmospheri C Author(s)2006. This work is licensed Chemistry under a Creative Commons License and Physics The effect of physical and chemical aerosol properties on warm cloud droplet activation G. MeFiggans', P. Artaxo, U. Baltensperger, H. Coe, M. C. Facchini, G. Feingold, S. Fuzzi, M. Gysel 3, A. Laaksonen, U. Lohmann, T. F Mentel, D. M. Murphy, CD.O'Dowd0, J.R. Snider,and E Weingartner I Atmospheric Sciences Group, SEAES, University of Manchester, P.O. Box 88, Manchester, M60 1QD, UK iNstituto de Fisica, Universidade de Sao Paulo, Rua do Matao, TravessaR, 187, CEP 05508-900 Sao Paulo, Brazil SPaul Scherrer Institut, Labor fur Atmospharenchemie, 5232 Villigen PSI,Switzerland 4Istituto di Scienze dell'Atmosfera e del Clima, CNR, 40129 Bologna, Italy NOAA Environmental Technology Laboratory, 325 Broadway, Boulder, Colorado 80305, USA b Department of Applied Physics, University of Kuopio, P.O. Box 1627, 70211 Kuopio, Finland "Institute for Atmospheric and Climate Science, Schafmattstr. 30, ETH Zurich, 8093 Zurich, Switzerland 8Forschungszentrum Julich GmbH, ICG-II: Troposphare, 52425 Julich, Germany 9NOAA Aeronomy Laboratory, 325 Broadway, Boulder, Colorado 80305, USA 10 Department of Physics, National University of Ireland, Galway, Ireland I University of Wyoming, Department of Atmospheric Science, Laramie, WY82071,USA Received: 7 June 2005- Published in Atmos. Chem. Phys. Discuss: 12 September 2005 Revised: 13 January 2006-Accepted: 29 May 2006- Published: 5 July 2006 Abstract. The effects of atmospheric aerosol on climate coming shortwave radiation and absorb outgoing longwave forcing may be very substantial but are quantified poorly at radiation(the"aerosol direct effect" Mc Cormick and lud- present; in particular, the effects of aerosols on cloud radia- wig, 1967, Charlson and Pilat, 1969; Hay wood and Boucher, tive properties, or theindirect effects"are credited with the 2000, Charlson et al., 1992). Aerosol particles that act as greatest range of uncertainty amongst the known causes of cloud condensation nuclei hanges in droplet number radiative forcing. This manuscript explores the effects that affecting the albedo and persistence of clouds; these are re- ne composition and properties of atmospheric aerosol can spectively termed theTwomey(first) and the cloud lifetime have on the activation of droplets in warm clouds, so poten- (second) aerosol indirect effects"(Warner, 1968: Twomey tially influencing the magnitude of the indirect effect. The 1974; Albrecht, 1989; Liou and Ou, 1989; Lohmann and effects of size, composition, mixing state and various derived Feichter, 2005). The Twomey effect refers to the aerosol properties are assessed and a range of these properties pro- induced increase in cloud number droplet for vided by atmospheric measurements in a variety of locations uid water content whereas the cloud lifetime effect is a result is briefly reviewed. The suitability of a range of process-level of the reduced precipitation efficiency of the more numer- descriptions to capture these aerosol effects is investigated ous smaller cloud droplets. Absorbing aerosol has also been by assessment of their sensitivities to uncertainties in aerosol shown to cause local warming of the atmosphere, which may properties and by their performance in closure studies. The result in stabilisation of the sub-cloud layer, and large-scale treatment of these effects within global models is reviewed burn-off of clouds. This has been termed the"semi-direct and suggestions for future investigations are made. effect(Fischer and Grassl, 1975; Hansen et al. 1997, Ack erman et al 2000, Johnson et al., 2004) The aerosol indirect effect is currently credited with the 1 Introduction greatest range of uncertainty amongst the known causes of radiative forcing(Ramaswamy et al., 2001); this range Aerosol particles affect the radiation balance of the atmo- stated as being around 4 times the uncertainty associated with sphere in a number of ways. They scatter and absorb in- forcing by radiatively active gases. That its absolute magni- tude may be comparable to that from radiatively active gases Correspondence to: G. Mcl necessitates greatly improved quantification of the factors af- (g. mcfiggans(@ manchester.a fecting and contributing to the aerosol indirect effect Published by Copernicus GmbH on behalf of the European Geosciences Union
Atmos. Chem. Phys., 6, 2593–2649, 2006 www.atmos-chem-phys.net/6/2593/2006/ © Author(s) 2006. This work is licensed under a Creative Commons License. Atmospheric Chemistry and Physics The effect of physical and chemical aerosol properties on warm cloud droplet activation G. McFiggans1 , P. Artaxo2 , U. Baltensperger3 , H. Coe1 , M. C. Facchini4 , G. Feingold5 , S. Fuzzi4 , M. Gysel1,3 , A. Laaksonen6 , U. Lohmann7 , T. F. Mentel8 , D. M. Murphy9 , C. D. O’Dowd10, J. R. Snider11, and E. Weingartner3 1Atmospheric Sciences Group, SEAES, University of Manchester, P.O. Box 88, Manchester, M60 1QD, UK 2 Instituto de Fisica, Universidade de Sao Paulo, Rua do Matao, Travessa R, 187, CEP 05508-900 Sao Paulo, Brazil 3Paul Scherrer Institut, Labor fur Atmosph ¨ arenchemie, 5232 Villigen PSI, Switzerland ¨ 4 Istituto di Scienze dell’Atmosfera e del Clima, CNR, 40129 Bologna, Italy 5NOAA Environmental Technology Laboratory, 325 Broadway, Boulder, Colorado 80305, USA 6Department of Applied Physics, University of Kuopio, P.O. Box 1627, 70211 Kuopio, Finland 7 Institute for Atmospheric and Climate Science, Schafmattstr. 30, ETH Zurich, 8093 Zurich, Switzerland 8Forschungszentrum Julich GmbH, ICG-II: Troposph ¨ are, 52425 J ¨ ulich, Germany ¨ 9NOAA Aeronomy Laboratory, 325 Broadway, Boulder, Colorado 80305, USA 10Department of Physics, National University of Ireland, Galway, Ireland 11University of Wyoming, Department of Atmospheric Science, Laramie, WY 82071, USA Received: 7 June 2005 – Published in Atmos. Chem. Phys. Discuss.: 12 September 2005 Revised: 13 January 2006 – Accepted: 29 May 2006 – Published: 5 July 2006 Abstract. The effects of atmospheric aerosol on climate forcing may be very substantial but are quantified poorly at present; in particular, the effects of aerosols on cloud radiative properties, or the “indirect effects” are credited with the greatest range of uncertainty amongst the known causes of radiative forcing. This manuscript explores the effects that the composition and properties of atmospheric aerosol can have on the activation of droplets in warm clouds, so potentially influencing the magnitude of the indirect effect. The effects of size, composition, mixing state and various derived properties are assessed and a range of these properties provided by atmospheric measurements in a variety of locations is briefly reviewed. The suitability of a range of process-level descriptions to capture these aerosol effects is investigated by assessment of their sensitivities to uncertainties in aerosol properties and by their performance in closure studies. The treatment of these effects within global models is reviewed and suggestions for future investigations are made. 1 Introduction Aerosol particles affect the radiation balance of the atmosphere in a number of ways. They scatter and absorb inCorrespondence to: G. McFiggans (g.mcfiggans@manchester.ac.uk) coming shortwave radiation and absorb outgoing longwave radiation (the “aerosol direct effect” McCormick and Ludwig, 1967; Charlson and Pilat, 1969; Haywood and Boucher, 2000, Charlson et al., 1992). Aerosol particles that act as cloud condensation nuclei cause changes in droplet number affecting the albedo and persistence of clouds; these are respectively termed the “Twomey (first) and the cloud lifetime (second) aerosol indirect effects” (Warner, 1968; Twomey, 1974; Albrecht, 1989; Liou and Ou, 1989; Lohmann and Feichter, 2005). The Twomey effect refers to the aerosolinduced increase in cloud number droplet for a constant liquid water content whereas the cloud lifetime effect is a result of the reduced precipitation efficiency of the more numereous smaller cloud droplets. Absorbing aerosol has also been shown to cause local warming of the atmosphere, which may result in stabilisation of the sub-cloud layer, and large-scale burn-off of clouds. This has been termed the “semi-direct effect” (Fischer and Grassl, 1975; Hansen et al., 1997; Ackerman et al., 2000; Johnson et al., 2004). The aerosol indirect effect is currently credited with the greatest range of uncertainty amongst the known causes of radiative forcing (Ramaswamy et al., 2001); this range is stated as being around 4 times the uncertainty associated with forcing by radiatively active gases. That its absolute magnitude may be comparable to that from radiatively active gases necessitates greatly improved quantification of the factors affecting and contributing to the aerosol indirect effect. Published by Copernicus GmbH on behalf of the European Geosciences Union
2594 G. McFiggans et al. aerosol effects on warm cloud activation The primary tool at our disposal for assessing aerosol ef- fluence(certainly in the Northern Hemisphere)-and cloud fects on clouds and future climate states is the general circu- formation in these regions may be particularly sensitive to lation model (GCM). These models are required to describe such input. In addition, the variety of natural primary and a large number of coupled processes which, because of the secondary sources can also lead to a complexity, not broadly enormous range of spatial and temporal scales that need to appreciated until recent years largely resulting from the nu- be addressed, and the associated computational burden, must merous organic species contributing to the aerosol burden to a large extent be simplified (e.g. Lohmann and Feichter, (Kanakidou et al., 2005). More obviously polluted air suc 2005 ). The difficulty in meaningfully incorporating a re- as that originating from densely-populated continental alistic description of aerosol-cloud interactions into GCMs gions(or regions otherwise influenced by human activity, e.g should not be underestimated. Large scale models have dif- biomass burning) are frequently heavily laden with an exter- ficulty in representing convection and clouds, which are typ- nal mixture of internally-mixed multicomponent aerosol dis- ically parameterised as subgrid processes. Therefore, their tributions(see Sect. 3.2.3) ability to represent the macroscale features of clouds(spatial It might be expected that cloud droplets forming in ris- coverage,depth, hydrometeor content )is itself a challenge ing air parcels containing such burdens of multicomponent In addition, they cannot represent the magnitude of the up- aerosol may not behave in the same way as those formed by draught velocities which may have an important bearing on the activation of simple salt particles. A range of techniques the local microphysical cloud properties(droplet number, ef- have been used to describe how aerosol particles behave as fective radius etc. ) The cloud-scale updraught velocity rele- relative humidity approaches and then exceeds saturation. A vant for cloud droplet activation is usually approximated, ei- summary of the fundamental theoretical approaches used to her by assuming a Gaussian distribution or relating subgrid- investigate droplet activation is presented, ranging from rel- scale fluctuations to the turbulent kinetic energy( Ghan et al atively well-established conventional application of Kohler 1997; Lohmann et al., 1999). It is not the intent of this arti- theory to recently developed theoretical and laboratory-based cle to investigate the broad question of dynamical cloud sys- extensions reflecting the increased aerosol complexity(Shul- tems, their effect on radiation and the incorporation of these, man et al. 1996: Kulmala et al., 1997). The effects of such and other coupled effects into large-scale models. Neglect extensions are discussed and the relative importance under a of these problems is not a reflection of their relative impor- range of conditions assessed( Sect. 4) tance in cloud radiative forcing. The main purpose of this There have been recent attempts to reconcile field obser- review is to identify one particular and important aspect of the aerosol indirect effect, namely, the properties of aerosol vations of aerosol composition with those of derived proper ties related to cloud activation. Such studies may be broadly which dominate their activation in warm clouds. This aspect categorised as"hygroscopic growth closure","CCN closure" of the indirect effect is accessible to investigation by obser and"droplet number closure"investigations. The difficulties tions alone. For example, the relationships between droplet associated with each level of closure are different and the number (Nd) and aerosol number(Na)and between effec- droplet number closure is significantly more difficult. An ap- tive droplet radius (refr)and Na may both be directly probed Should all dynamical considerations remain relatively con- praisal of these difficulties and the current status of progress stant, such observations can be used to derive relationships in such studies are reviewed in Sect. 5 between the aerosol distribution and cloud distribution prop- This paper therefore aims to examine the state-of-the- science to establish i) the dominant characteristics of the at- This article will review the currently-available observa- mospheric aerosol(in so far as they may influence cloud ac- tional evidence for the compositional complexity of atmo- tivation) n) which aerosol properties should be most accu spheric aerosol and the derived properties of their size and rately captured to investigate cloud formation in)how these composition distributions that are thought to affect their abil- properties can be represented at a detailed process level iv) ity to act as cloud condensation nuclei(CCN). An increas- whether current representations of these properties in larger ing body of evidence suggests that the complexity of atmo- scale models can adequately capture the important behaviour spheric aerosol may preclude realistic treatment of cloud for- of the aerosol and v)how improved representations may be mation using the levels of simplification incorporated into developed to investigate aerosol effects on cloud formation. all large-scale models(and even most process-level descrip- Suggestions of which properties should be captured to enable tions). The frequent assumptions of an externally mixed accurate representation of aerosol effects on cloud formation organic salt aerosol appear not to be applicable even in the follow from this examination most pristine regions. Even this description is more co A final note on the scope of this article: whilst the effects plex than that employed in many climate models( Wilson of atmospheric aerosol properties on warm cloud activation et al., 2001; Gong and Barrie, 2003). Long-range transport are subject to significant uncertainties, those surrounding the of polluted plumes lofted into the free troposphere followed roles of aerosol particles as ice nuclei (IN) are much greater by sporadic re-entrainment( Clarke et al., 1999)ensures that still. The article will therefore be limited to warm clouds and remote regions are not always free from anthropogenic in- will not attempt to consider mixed-phase or cold clouds tmos.Chem.Phys,6,2593-2649,2006 www.atmos-chem-phys.net/6/2593/2006/
2594 G. McFiggans et al.: Aerosol effects on warm cloud activation The primary tool at our disposal for assessing aerosol effects on clouds and future climate states is the general circulation model (GCM). These models are required to describe a large number of coupled processes which, because of the enormous range of spatial and temporal scales that need to be addressed, and the associated computational burden, must to a large extent be simplified (e.g. Lohmann and Feichter, 2005). The difficulty in meaningfully incorporating a realistic description of aerosol-cloud interactions into GCMs should not be underestimated. Large scale models have dif- ficulty in representing convection and clouds, which are typically parameterised as subgrid processes. Therefore, their ability to represent the macroscale features of clouds (spatial coverage, depth, hydrometeor content) is itself a challenge. In addition, they cannot represent the magnitude of the updraught velocities which may have an important bearing on the local microphysical cloud properties (droplet number, effective radius etc.). The cloud-scale updraught velocity relevant for cloud droplet activation is usually approximated, either by assuming a Gaussian distribution or relating subgridscale fluctuations to the turbulent kinetic energy (Ghan et al., 1997; Lohmann et al., 1999). It is not the intent of this article to investigate the broad question of dynamical cloud systems, their effect on radiation and the incorporation of these, and other coupled effects into large-scale models. Neglect of these problems is not a reflection of their relative importance in cloud radiative forcing. The main purpose of this review is to identify one particular and important aspect of the aerosol indirect effect, namely, the properties of aerosol which dominate their activation in warm clouds. This aspect of the indirect effect is accessible to investigation by observations alone. For example, the relationships between droplet number (Nd ) and aerosol number (Na) and between effective droplet radius (reff) and Na may both be directly probed. Should all dynamical considerations remain relatively constant, such observations can be used to derive relationships between the aerosol distribution and cloud distribution properties. This article will review the currently-available observational evidence for the compositional complexity of atmospheric aerosol and the derived properties of their size and composition distributions that are thought to affect their ability to act as cloud condensation nuclei (CCN). An increasing body of evidence suggests that the complexity of atmospheric aerosol may preclude realistic treatment of cloud formation using the levels of simplification incorporated into all large-scale models (and even most process-level descriptions). The frequent assumptions of an externally mixed inorganic salt aerosol appear not to be applicable even in the most pristine regions. Even this description is more complex than that employed in many climate models (Wilson et al., 2001; Gong and Barrie, 2003). Long-range transport of polluted plumes lofted into the free troposphere followed by sporadic re-entrainment (Clarke et al., 1999) ensures that remote regions are not always free from anthropogenic in- fluence (certainly in the Northern Hemisphere) – and cloud formation in these regions may be particularly sensitive to such input. In addition, the variety of natural primary and secondary sources can also lead to a complexity, not broadly appreciated until recent years largely resulting from the numerous organic species contributing to the aerosol burden (Kanakidou et al., 2005). More obviously polluted air such as that originating from densely-populated continental regions (or regions otherwise influenced by human activity, e.g. biomass burning) are frequently heavily laden with an external mixture of internally-mixed multicomponent aerosol distributions (see Sect. 3.2.3). It might be expected that cloud droplets forming in rising air parcels containing such burdens of multicomponent aerosol may not behave in the same way as those formed by the activation of simple salt particles. A range of techniques have been used to describe how aerosol particles behave as relative humidity approaches and then exceeds saturation. A summary of the fundamental theoretical approaches used to investigate droplet activation is presented, ranging from relatively well-established conventional application of Kohler ¨ theory to recently developed theoretical and laboratory-based extensions reflecting the increased aerosol complexity (Shulman et al., 1996; Kulmala et al., 1997). The effects of such extensions are discussed and the relative importance under a range of conditions assessed (Sect. 4). There have been recent attempts to reconcile field observations of aerosol composition with those of derived properties related to cloud activation. Such studies may be broadly categorised as “hygroscopic growth closure”, “CCN closure” and “droplet number closure” investigations. The difficulties associated with each level of closure are different and the droplet number closure is significantly more difficult. An appraisal of these difficulties and the current status of progress in such studies are reviewed in Sect. 5. This paper therefore aims to examine the state-of-thescience to establish i) the dominant characteristics of the atmospheric aerosol (in so far as they may influence cloud activation) ii) which aerosol properties should be most accurately captured to investigate cloud formation iii) how these properties can be represented at a detailed process level iv) whether current representations of these properties in larger scale models can adequately capture the important behaviour of the aerosol and v) how improved representations may be developed to investigate aerosol effects on cloud formation. Suggestions of which properties should be captured to enable accurate representation of aerosol effects on cloud formation follow from this examination. A final note on the scope of this article: whilst the effects of atmospheric aerosol properties on warm cloud activation are subject to significant uncertainties, those surrounding the roles of aerosol particles as ice nuclei (IN) are much greater still. The article will therefore be limited to warm clouds and will not attempt to consider mixed-phase or cold clouds. Atmos. Chem. Phys., 6, 2593–2649, 2006 www.atmos-chem-phys.net/6/2593/2006/
G. McFiggans et al. Aerosol effects on warm cloud activation 2595 2 Theory of activation of aerosol particles in warm activating particles which may be more or less accurate de pending on the supersaturation, there are numerical approxi- mations such as that associated with the Taylor series expan- The description of the equilibrium size of a droplet with wa- sion of the exponential which limit the range of applicability ter saturation ratio, founded on the early work of Kohler of Eq (2). Figure 1 shows the contribution of the Kelv (1936), is now well-established and can be readily derived and Raoult terms to the activation behaviour of a 200 nm dry from the Clausius-Clapeyron equation modified to give a diameter ammonium sulphate particle general equilibrium relation between an aqueous salt sol This form of the expression shows a single characteris- tion droplet and water vapour tic maximum in supersaturation for a given dry composition and size, known as the critical supersaturation, Sc, associated =awexpKe = awexp with a unique size, denoted the critical radius, rc or diame- RTr (1) ter, De. Using the simplified expression(2), solutions for the critical quantities are the vapour pressure of wa es is the saturation vapour pressure of water, 3B12 (e/es=S, is known as the saturation ratio), A aw is the water activity Ke is the Kelvin factor Ww is the partial molar volume of water, S 27B sol/v is the surface tension of the solution at the composition of the droplet, R is the universal gas constant, For an increasing environmental value of s below Sc there T is the droplet temperature, is a unique equilibrium droplet size. Once the droplet grows is the particle radius beyond its critical size (i.e. as the environmental S increases above Sc) the droplet will exhibit unimpeded growth unless the environmental S reduces below the equilibrium value of This form of the Kohler equation is not generally accessi- S at the instantaneous value of r. In this case, with no fur- Yau, 1989; Pruppacher and Klett, 1997; Chylek and Wong, ther change in S, the droplet will evaporate to its sub-critical 1998: Seinfeld and Pandis, 1998) provide standard deriva- equilibrium size tions to yield the simplified form of the Kohler equation The Kohler expression can be envisaged as the compet tion between the two expressions of component properties a B determining activation of particles; the curvature term and (2) the solute term. The solute terms depends first on the number of solution molecules and then on the dissociation of these where A==dule and B= where v is the num- molecules. The effect can be illustrated for two frequently as- er of dissociated ions per so cule, ms is thethe se sumed cloud condensation nuclei types: ammonium sulphate lute mass and subscripts s and w relate to solute and water and sodium chloride.(NH4)2SO4 has a molecular weight properties, respectively. The term in A is denoted the Kelvin of 132 gMol- while that of NaCl is 58.5 gMol-1. Thus, in or curvature term and that in b. the raoult or solute term the absence of dissociation, a given mass of Nacl in solu- This latter form of the equation assumes that the droplet tion would yield 2. 26 times more dissolved molecules than behaves ideally, i.e. that the practical osmotic coefficient of (NH4)2SO4. Assuming full dissociation(infinite dilution) the salt, =l, where (NH4) SO4 yields 3 ions while NaCl yields 2, so the net effect of the molecular mass and dissociation is that nacl is 2. 26/1.5=1.5 times more active than(NH4)2SO4 for the ( same dry mass of particle(the Se ratio is around 1.22 due to the square root dependence). This is illustrated in Fig. 2 and that the number of ions in solution is independent of so- where the peak supersaturation is plotted versus dry diameter lution concentration. Equation(2)further assumes that the for particles comprising each electrolyte. This figure directly solute is completely soluble and it implicitly follows that the illustrates the significant differences in the critical supersat- solution droplet is assumed homogeneous-that the compo- uration as a function of both the chemical composition and sition is independent of distance from droplet centre. It is dry size of a particle(raoult and Kelvin effect further assumed that the surface tension and density of the The atmospheric aerosol does not solely com growing droplet are equal to those of water. In addition to the pended completely soluble inorganic salt solution assumptions relating to physico-chemical properties of the A modification to the Raoult term was reported www.atmos-chem-phys.net/6/2593/2006/ Atmos. Chem. Phys., 6, 2593-2649, 2006
G. McFiggans et al.: Aerosol effects on warm cloud activation 2595 2 Theory of activation of aerosol particles in warm clouds The description of the equilibrium size of a droplet with water saturation ratio, founded on the early work of Kohler ¨ (1936), is now well-established and can be readily derived from the Clausius-Clapeyron equation modified to give a general equilibrium relation between an aqueous salt solution droplet and water vapour: e es = awexpKe = awexp 2vwσsol/v RT r (1) where e is the vapour pressure of water, es is the saturation vapour pressure of water, (e/es=S, is known as the saturation ratio), aw is the water activity, Ke is the Kelvin factor, vw is the partial molar volume of water, σsol/v is the surface tension of the solution at the composition of the droplet, R is the universal gas constant, T is the droplet temperature, r is the particle radius. This form of the Kohler equation is not generally accessi- ¨ ble to analytical solution and a number of texts (Rogers and Yau, 1989; Pruppacher and Klett, 1997; Chylek and Wong ´ , 1998; Seinfeld and Pandis, 1998) provide standard derivations to yield the simplified form of the Kohler equation: ¨ S = e es ≈ 1 + A r − B r 3 (2) where A= 2Mwσw/v RT ρw and B= νmsMw Ms(4/3πρw) , where ν is the number of dissociated ions per solute molecule, ms is the the solute mass and subscripts s and w relate to solute and water properties, respectively. The term in A is denoted the Kelvin or curvature term, and that in B, the Raoult or solute term. This latter form of the equation assumes that the droplet behaves ideally, i.e. that the practical osmotic coefficient of the salt, φ=1, where aw = exp − νns nw φs (3) and that the number of ions in solution is independent of solution concentration. Equation (2) further assumes that the solute is completely soluble and it implicitly follows that the solution droplet is assumed homogeneous – that the composition is independent of distance from droplet centre. It is further assumed that the surface tension and density of the growing droplet are equal to those of water. In addition to the assumptions relating to physico-chemical properties of the activating particles which may be more or less accurate depending on the supersaturation, there are numerical approximations such as that associated with the Taylor series expansion of the exponential which limit the range of applicability of Eq. (2). Figure 1 shows the contribution of the Kelvin and Raoult terms to the activation behaviour of a 200 nm dry diameter ammonium sulphate particle. This form of the expression shows a single characteristic maximum in supersaturation for a given dry composition and size, known as the critical supersaturation, Sc, associated with a unique size, denoted the critical radius, rc or diameter, Dc. Using the simplified expression (2), the analytical solutions for the critical quantities are: rc = 3B A 1/2 (4) Sc = 4A3 27B !1/2 (5) For an increasing environmental value of S below Sc there is a unique equilibrium droplet size. Once the droplet grows beyond its critical size (i.e. as the environmental S increases above Sc) the droplet will exhibit unimpeded growth unless the environmental S reduces below the equilibrium value of S at the instantaneous value of r. In this case, with no further change in S, the droplet will evaporate to its sub-critical equilibrium size. The Kohler expression can be envisaged as the competi- ¨ tion between the two expressions of component properties determining activation of particles; the curvature term and the solute term. The solute terms depends first on the number of solution molecules and then on the dissociation of these molecules. The effect can be illustrated for two frequently assumed cloud condensation nuclei types: ammonium sulphate and sodium chloride. (NH4)2SO4 has a molecular weight of 132 gMol−1 while that of NaCl is 58.5 gMol−1 . Thus, in the absence of dissociation, a given mass of NaCl in solution would yield 2.26 times more dissolved molecules than (NH4)2SO4. Assuming full dissociation (infinite dilution), (NH4)2SO4 yields 3 ions while NaCl yields 2, so the net effect of the molecular mass and dissociation is that NaCl is 2.26/1.5=1.5 times more active than (NH4)2SO4 for the same dry mass of particle (the Sc ratio is around 1.22 due to the square root dependence). This is illustrated in Fig. 2 where the peak supersaturation is plotted versus dry diameter for particles comprising each electrolyte. This figure directly illustrates the significant differences in the critical supersaturation as a function of both the chemical composition and dry size of a particle (Raoult and Kelvin effects). The atmospheric aerosol does not solely comprise suspended completely soluble inorganic salt solution particles. A modification to the Raoult term was reported by Hanel ¨ www.atmos-chem-phys.net/6/2593/2006/ Atmos. Chem. Phys., 6, 2593–2649, 2006
2596 G. McFiggans et al. aerosol effects on warm cloud activation 0.06 Raoult term 0.04 9 0.04 0° plet radius, ,, um Fig. 1. The Kohler equation can be envisaged as the competition between the curvature(Kelvin) and solute(Raoult)terms fractions. a discussion of more rigorous treatments of lim- NH42504 ited subility is presented in Sect. 4.1 3 The composition and properties of atmospheric relevant to the indirect effect Suspended particulate material in the atmosphere is highly variable. Atmospheric aerosol particles cover four or five decades in size from a few nanometers to several tens or even hundreds of microns and the loading and composition are ex- tremely source and location dependent. Studies of cloud for- mation invariably rely on a simplified model of input aerosol composition distributions. The specific description is depen dent on the particular atmospheric scenario. This section Dry Diameter (um) presents an overview of the characteristics of atmospheric aerosol as they relate to cloud droplet activation. Fig. 2. Critical supersaturation as a function of dry size for NacI d(NH4)) SO4 particles 3.1 Characteristic size and composition of atmospheric (1976)to allow explicit consideration of internally mixed 3.1.1 Size distributions of ambient aeros completely insoluble inclusions It can be seen from Figs. 2 and 3 that the activation of aerosol B=Evms M particles is strongly dependent on the dry size. For any given (6) composition and supersaturation(which, around cloud base where E is the soluble mass fraction of the dry particle ticle activates is solely dependent on its dry size. At constant Figure 3 shows the form of activation curves for a range updraught velocity, a distribution of such particles of vary of particles of varying dry diameter and initial soluble mass ing sizes will activate if their corresponding critical radius is tmos.Chem.Phys,6,2593-2649,2006 www.atmos-chem-phys.net/6/2593/2006/
2596 G. McFiggans et al.: Aerosol effects on warm cloud activation 10−1 100 101 102 −0.04 −0.02 0 0.02 0.04 0.06 wet droplet radius, r, µm Supersaturation % = 100 x (S −1) Kelvin term Raoult term Total S c r c Fig. 1. The Kohler equation can be envisaged as the competition between the curvature (Kelvin) and solute (Raoult) terms. ¨ Fig. 2. Critical supersaturation as a function of dry size for NaCl and (NH4)2SO4 particles. (1976) to allow explicit consideration of internally mixed completely insoluble inclusions: B = ενmsMw Ms( 4/3πρw) (6) where ε is the soluble mass fraction of the dry particle. Figure 3 shows the form of activation curves for a range of particles of varying dry diameter and initial soluble mass fractions. A discussion of more rigorous treatments of limited component solubility is presented in Sect. 4.1. 3 The composition and properties of atmospheric aerosol relevant to the indirect effect Suspended particulate material in the atmosphere is highly variable. Atmospheric aerosol particles cover four or five decades in size from a few nanometers to several tens or even hundreds of microns and the loading and composition are extremely source and location dependent. Studies of cloud formation invariably rely on a simplified model of input aerosol composition distributions. The specific description is dependent on the particular atmospheric scenario. This section presents an overview of the characteristics of atmospheric aerosol as they relate to cloud droplet activation. 3.1 Characteristic size and composition of atmospheric aerosol 3.1.1 Size distributions of ambient aerosol It can be seen from Figs. 2 and 3 that the activation of aerosol particles is strongly dependent on the dry size. For any given composition and supersaturation (which, around cloud base, is directly proportional to updraught velocity), whether a particle activates is solely dependent on its dry size. At constant updraught velocity, a distribution of such particles of varying sizes will activate if their corresponding critical radius is Atmos. Chem. Phys., 6, 2593–2649, 2006 www.atmos-chem-phys.net/6/2593/2006/
G. McFiggans et al. Aerosol effects on warm cloud activation 2597 1.004 1 1.002 1001 50nmNH小28O4 097 200 nm(NH ).so 20m:A H2so4 50% insol 200 nm NacL. 50% insol 0.95 0° Droplet Diameter um Fig 3. Activation curves for a range of dry diameter of salt((NH4)2SO4-solid, NaCI-dashed) particles (red, green and blue curves)and for 200 nm particles containing 50% by mass insoluble core(magenta) reached. As more particles activate and grow, they will com- altitude with the maximum just above cloud base. Numer- te for available water vapour. The water supersaturation ous implementations of simple adiabatic cloud parcel mod- will continue to rise above cloud base but will slow as grow- els exist which describe heat transfer and mass transfer of g droplets scavenge the water vapour and relatively fewer water vapour between an adiabatically cooling air parcel and additional (smaller) aerosol particles will activate. When the aerosol/droplet population based on fundamental thermo- supersaturation sources and sinks balance, the peak super- dynamic principles(see Howell, 1949, Mordy, 1959, Prup- saturation is reached - usually within a few 10s of metres pacher and Klett, 1997; Seinfeld and Pandis, 1998 above cloud base. Following this point, the growing droplet Figure 4 demonstrates the predicted behaviour of an ide- population will lead to a reduction in supersaturation. No alised lognormal(NH4) SO4 aerosol population with height new particles will activate and the most recently activated above cloud base at an updraught velocity of 0.5 ms-using droplets may evaporate. Some particles will not have suf- such a model. It clearly shows how the droplets activated ficient time to reach their critical radius even though their from the larger classes of aerosol continue to grow above critical supersaturation is reached. This results from water the supersaturation maximum at the expense of the smaller vapour scavenging by the larger droplets reducing supersatu- classes of activated particle which evaporate to below their ration to below the critical value of the smaller particles be- critical radius. The model uses the form of the Kohler equa- fore sufficient water vapour can condense(such kinetic limi- tion shown in Eq(1) tations are discussed further in Sect. 4). Only particles reach- Given this behaviour, it can be seen that both the number ing a certain size will survive and grow. Some of the largest of particles in a given size range and the gradient of the dis- particles may not actually activate, but may be large enough tribution in certain critical size ranges will determine its acti- to be considered as droplets since even at their subcritical vation behaviour moving into supersaturation. The Twomey sizes they will often be greater than 10 or 20 microns in ra-(1959)analytical solution to this problem dius, deplete water vapour, and even act as collector drops A pseudosteady-state or quasi-equilibrium is eventually =c(100s)ands=/4(7,P)m232)l/+ reached for a constant updraught velocity where the decrease kB(3/2,k/2) in saturation ratio by condensation to the droplet population and the increase in saturation ratio owing to the updraught where c is proportional to the CCN concentration at 1%su- maintains a broadly constant supersaturation with increasing persaturation, w is the updraught velocity, k is the slope parameter of the CCN size spectrum and N is the number www.atmos-chem-phys.net/6/2593/20 Atmos. Chem. Phys., 6, 2593-2649, 2006
G. McFiggans et al.: Aerosol effects on warm cloud activation 2597 10−1 100 101 102 0.95 0.96 0.97 0.98 0.99 10−1 100 101 102 1 1.001 1.002 1.003 1.004 1.005 Droplet Diameter µm Saturation Ratio 50 nm (NH4 ) 2 SO4 50 nm NaCl 100 nm (NH4 ) 2 SO4 100 nm NaCl 200 nm (NH4 ) 2 SO4 200 nm NaCl 200 nm (NH4 ) 2 SO4 , 50% insol 200 nm NaCl, 50% insol Fig. 3. Activation curves for a range of dry diameter of salt ((NH4)2SO4 – solid, NaCl – dashed) particles (red, green and blue curves) and for 200 nm particles containing 50% by mass insoluble core (magenta). reached. As more particles activate and grow, they will compete for available water vapour. The water supersaturation will continue to rise above cloud base but will slow as growing droplets scavenge the water vapour and relatively fewer additional (smaller) aerosol particles will activate. When supersaturation sources and sinks balance, the peak supersaturation is reached - usually within a few 10’s of metres above cloud base. Following this point, the growing droplet population will lead to a reduction in supersaturation. No new particles will activate and the most recently activated droplets may evaporate. Some particles will not have suf- ficient time to reach their critical radius even though their critical supersaturation is reached. This results from water vapour scavenging by the larger droplets reducing supersaturation to below the critical value of the smaller particles before sufficient water vapour can condense (such kinetic limitations are discussed further in Sect. 4). Only particles reaching a certain size will survive and grow. Some of the largest particles may not actually activate, but may be large enough to be considered as droplets since even at their subcritical sizes they will often be greater than 10 or 20 microns in radius, deplete water vapour, and even act as collector drops. A pseudo “steady-state” or quasi-equilibrium is eventually reached for a constant updraught velocity where the decrease in saturation ratio by condensation to the droplet population and the increase in saturation ratio owing to the updraught maintains a broadly constant supersaturation with increasing altitude with the maximum just above cloud base. Numerous implementations of simple adiabatic cloud parcel models exist which describe heat transfer and mass transfer of water vapour between an adiabatically cooling air parcel and the aerosol/droplet population based on fundamental thermodynamic principles (see Howell, 1949; Mordy, 1959; Pruppacher and Klett, 1997; Seinfeld and Pandis, 1998). Figure 4 demonstrates the predicted behaviour of an idealised lognormal (NH4)2SO4 aerosol population with height above cloud base at an updraught velocity of 0.5 ms−1 using such a model. It clearly shows how the droplets activated from the larger classes of aerosol continue to grow above the supersaturation maximum at the expense of the smaller classes of activated particle which evaporate to below their critical radius. The model uses the form of the Kohler equa- ¨ tion shown in Eq. (1). Given this behaviour, it can be seen that both the number of particles in a given size range and the gradient of the distribution in certain critical size ranges will determine its activation behaviour moving into supersaturation. The Twomey (1959) analytical solution to this problem: N = c(100 + S ∗ ) k and S ∗ = A(T , P )w3/2 ckβ(3/2, k/2) !1/(k+2) (7) where c is proportional to the CCN concentration at 1% supersaturation, w is the updraught velocity, k is the slope parameter of the CCN size spectrum and N is the number www.atmos-chem-phys.net/6/2593/2006/ Atmos. Chem. Phys., 6, 2593–2649, 2006