AUGUST 2010 HOOSE ET AL. 2483 A Classical-Theory-Based Parameterization of Heterogeneous Ice Nucleation by Mineral Dust, Soot, and Biological Particles in a Global Climate Model CORINNA HOOSE* AND JÓN EGILL KRISTJÁNSSON Department of Geosciences, University of Oslo, Oslo, Norway JEN-PING CHEN AND ANUPAM HAZRA1 Department of Atmospheric Sciences, National Taiwan University, Taipei, Taiwan (Manuscript received 24 December 2009, in final form 8 March 2010) ABSTRACT An ice nucleation parameterization based on classical nucleation theory, with aerosol-specific parameters derived from experiments, has been implemented into a global climate model—the Community Atmosphere Model (CAM)-Oslo. The parameterization treats immersion, contact, and deposition nucleation by mineral dust, soot, bacteria, fungal spores, and pollen in mixed-phase clouds at temperatures between 08 and 2388C. Immersion freezing is considered for insoluble particles that are activated to cloud droplets, and deposition and contact nucleation are only allowed for uncoated, unactivated aerosols. Immersion freezing by mineral dust is found to be the dominant ice formation process, followed by immersion and contact freezing by soot. The simulated biological aerosol contribution to global atmospheric ice formation is marginal, even with high estimates of their ice nucleation activity, because the number concentration of ice nucleation active biological particles in the atmosphere is low compared to other ice nucleating aerosols. Because of the dominance of mineral dust, the simulated ice nuclei concentrations at temperatures below 2208C are found to correlate with coarse-mode aerosol particle concentrations. The ice nuclei (IN) concentrations in the model agree well overall with in situ continuous flow diffusion chamber measurements. At individual locations, the model exhibits a stronger temperature dependence on IN concentrations than what is observed. The simulated IN composition (77% mineral dust, 23% soot, and 1025% biological particles) lies in the range of observed ice nuclei and ice crystal residue compositions. 1. Introduction Ice in tropospheric clouds is important for cloud radiative properties and precipitation formation, but its formation is neither theoretically fully understood nor empirically well constrained (Cantrell and Heymsfield 2005). At temperatures between 08 and 2388C, aerosol particles are required as ice nuclei (IN) to initiate either freezing of supercooled cloud droplets or ice nucleation * Current affiliation: Institute for Meteorology and Climate Research (IMK-AAF), Karlsruhe Institute of Technology, Karlsruhe, Germany. 1 Current affiliation: Indian Institute of Tropical Meteorology, Pune, India. Corresponding author address: Corinna Hoose, Department of Geosciences, University of Oslo, P.O. Box 1022, Blindern, 0315 Oslo, Norway. E-mail: corinna.hoose@kit.edu DOI: 10.1175/2010JAS3425.1 Ó 2010 American Meteorological Society from the vapor phase. Various insoluble particles such as mineral dust, soot, metallic particles, volcanic ash, or primary biological particles can act as IN (Pruppacher and Klett 1997; Szyrmer and Zawadzki 1997). IN concentrations are usually low (0.01–100 L21) compared to total aerosol concentrations. The dependence of heterogeneous ice nucleation on temperature, particle composition, size, coating, and various other parameters has been the subject of numerous laboratory experiments (e.g., Schaller and Fukuta 1979; Levin and Yankofsky 1983; Knopf and Koop 2006; Bundke et al. 2008; Durant et al. 2008; Welti et al. 2009). In general, it is found that some bacteria and the artificial IN silver iodide nucleate ice at the warmest temperatures, followed by other biological particles and mineral dust; combustion particles are relatively inefficient IN. Atmospheric in situ observation of ice nucleation and the involved particles is very difficult. One possibility is to examine the ice nucleation properties of particles 2484 JOURNAL OF THE ATMOSPHERIC SCIENCES at cloud altitude under controlled conditions in an aircraft-borne continuous flow diffusion chamber and to relate the IN counts to the ambient particle properties (DeMott et al. 2003b). Alternatively, sampling of cloud ice crystals and investigation of the residual aerosol particles after evaporation can give information on the IN composition (Cziczo et al. 2004; Targino et al. 2006; Cozic et al. 2008; Cziczo et al. 2009b; Pratt et al. 2009). As a combination of both methods, the composition of the subset of ambient aerosol particles that formed ice in a continuous flow diffusion chamber has been characterized in a number of studies (DeMott et al. 2003a; Richardson et al. 2007; Prenni et al. 2009a,b). The compilation of a large number of such data by Phillips et al. (2008) suggests that mineral dust is the dominant atmospheric IN. Additionally, Phillips et al. (2008) report a large portion of carbonaceous IN, but their exact composition (elemental or organic carbon) was not determined. Recently, Prenni et al. (2009b) and Pratt et al. (2009) observed high percentages of biological IN in the Amazon basin and in a wave cloud over North America, respectively. The variety of different IN types, and their scarcity, complicates the measurement and simulation of heterogeneous ice nucleation. In addition, heterogeneous ice nucleation can occur via several different mechanisms, called nucleation modes (Vali 1985). For ‘‘immersion freezing’’ an ice nucleus within a supercooled cloud droplet initiates the freezing process. The term ‘‘contact freezing’’ commonly refers to a supercooled droplet colliding with a dry ice nucleus, such that the freezing process is initiated from the outside. In addition, ‘‘inside out’’ contact freezing has been observed when the immersed ice nucleus contacts the droplet surface from the inside (Durant and Shaw 2005; Fornea et al. 2009). Contact freezing is often observed at higher temperatures than immersion freezing (e.g., Pitter and Pruppacher 1973). ‘‘Deposition nucleation’’ refers to the direct growth of ice from the vapor phase on a dry ice nucleus. The term ‘‘condensation freezing’’ is used for the process when a (at least partially insoluble) cloud condensation nucleus subsequently initiates the freezing. From a mechanistic standpoint, the differentiation between condensation and immersion freezing is vague, and in the following only the term ‘‘immersion freezing’’ is used. Contact and immersion freezing involve liquid droplets and are therefore the most commonly accepted ice nucleation mechanisms in supercooled liquid clouds. Meanwhile, the atmospheric relevance of deposition nucleation at temperatures above 2388C is uncertain. Favorable for the occurrence of deposition nucleation is the availability of uncoated IN (e.g., dust particles) in VOLUME 67 regions with low temperatures and supersaturation over ice. Wiacek and Peter (2009) performed trajectory calculations originating near the surface of the Chinese Taklimakan desert and found that most trajectories pass through ice-saturated (but water-subsaturated) regions (where deposition nucleation is the only possible ice formation mechanism) before reaching water saturation. At this stage, the dust particles have not undergone cloud processing and are possibly uncoated, such that deposition nucleation would be relatively efficient. However, observations of whether deposition nucleation occurs in mixed-phase conditions are ambiguous. On the one hand, Ansmann et al. (2009) observed that tropical altocumulus clouds over Cape Verde, investigated with ground-based lidar, almost always had a liquid cloud top and concluded that deposition nucleation was unimportant during the initial phase of altocumulus glaciation. Also in the presence of high dust concentrations in Morocco, cloud temperatures needed to be lower than approximately 2208C and liquid clouds were required before ice formed (Ansmann et al. 2008). On the other hand, lidar observations of a cloud influenced by boreal forest fire smoke (Sassen and Khvorostyanov 2008) showed ice nucleation prior to liquid cloud formation (i.e., below water saturation) at approximately 2158C. Theoretical formulations of heterogeneous ice nucleation include the so-called classical nucleation theory (CNT; e.g., Fletcher 1962), which treats nucleation as a stochastic process, and the semiempirical singular hypothesis (e.g., Levine 1950), which assigns a defined spontaneous freezing temperature to every aerosol particle. Parameterizations that are used in large-scale models are mostly empirical (Lohmann 2002; Lohmann and Diehl 2006; Hoose et al. 2008; Morrison and Gettelman 2008; Phillips et al. 2008; Storelvmo et al. 2008a). Biological particles have so far not been considered as IN in global models. In this article, an ice nucleation parameterization based on classical nucleation theory is formulated for use in a global model. Immersion and contact freezing as well as deposition nucleation are included. The necessary aerosol-related parameters are derived from laboratory experiments. Mineral dust and soot are considered as possible ice nuclei, as well as several primary biological particles: bacteria, fungal spores, and pollen. Section 2 describes the model and the new ice nucleation parameterization. In section 3, the relative importance of the different freezing processes is presented. Ice nuclei concentrations and composition are compared to observations. Finally, implications and uncertainties are discussed in section 4. AUGUST 2010 HOOSE ET AL. 2485 TABLE 1. Simulation descriptions. Simulation CTL CNT CNT-highbact CNT-lowdust CNT-nosootct CNT-nobio Description CAM-Oslo with warm cloud microphysics as in Hoose et al. (2009) and cold cloud microphysics as in simulation WBF in Storelvmo et al. (2008b). Freezing parameterizations after Lohmann and Diehl (2006). As in CTL, but with the new freezing parameterizations from classical nucleation theory. As in CNT, but with a higher fraction of bacteria assumed to be ice-nucleation active. As in CNT, but with higher wet scavenging of dust. As in CNT, but without contact freezing by soot. As in CNT, but without freezing by biological aerosol particles. 2. Model description and treatment of ice nucleation a. CAM-Oslo The aerosol–climate model CAM-Oslo is based on version 3 of the Community Atmosphere Model (CAM3; Collins et al. 2006). It has been extended to include a detailed aerosol module (Seland et al. 2008) and a prognostic double-moment cloud microphysics scheme (Storelvmo et al. 2006; Hoose et al. 2009). The microphysical scheme for mixed-phase clouds (Storelvmo et al. 2008a,b) has now been modified by a new treatment of ice nucleation in mixed-phase clouds (see below). The aerosol concentrations, mixing states, and the fractions activated to cloud droplets are simulated online in CAM-Oslo and provide the input parameters for the ice nucleation parameterization. The CAM-Oslo aerosol scheme treats sea salt, mineral dust, sulfate, black carbon, and organic aerosols in 16 modes and 44 size bins with process-determined mixing states. Aerosol and precursor gas emissions are taken from the AeroCom inventory (Dentener et al. 2006). Compared to the original scheme by Seland et al. (2008), Hoose et al. (2009) reduced the in-cloud scavenging ratio for mineral dust from 1 to 0.1 for a better agreement of background dust concentrations and cloud droplet numbers over land with observations. The primary biological particles are treated as described in Hoose et al. (2010), with emissions based on Burrows et al. (2009) for bacteria, Heald and Spracklen (2009) for fungal spores, and Jacobson and Streets (2009) for pollen. These particles are assumed to be spherical and monodisperse, with diameters of 1 mm for bacteria (Burrows et al. 2009), 5 mm for fungal spores (Elbert et al. 2007), and 30 mm for pollen (Jacobson and Streets 2009). b. Simulation setup A control experiment (CTL) with the previously used freezing parameterization (Lohmann and Diehl 2006), a simulation (CNT) with the new freezing parameterization as described below, and several sensitivity experiments have been conducted (Table 1). The sensitivity experiments explore different assumptions about the ice nucleation active fraction of bacteria and fungal spores, suppression of contact freezing by soot, and mineral dust scavenging. In the simulation CNThighbact, the ice nucleation active fraction (discussed below) of bacteria and fungal spores is increased from 1% to 10%, and the parameter fi,max is increased from 0.1% to 1%. This results in a maximum increase of bacteria and fungal spore ice nuclei by up to a factor of 100. In CNT-lowdust, the scavenging ratio for mineral dust is raised from 0.1 to 0.5, resulting in lower background dust concentrations. Simulation CNT-nosootct excludes contact freezing by soot particles, which is considered the most uncertain freezing process. Finally, in CNT-nobio all biological freezing processes are set to 0. All simulations are run in T42 resolution (2.81258 3 2.81258) with 26 vertical levels. The simulations are integrated for 5 yr after 4 months of spinup, for both presentday and preindustrial aerosol emissions (Dentener et al. 2006). c. Ice nucleation active aerosol particles This study considers ice nucleation for mineral dust, soot, and primary biological particles (bacteria, fungal spores, and pollen). Because particles of these categories are in reality of varying chemical composition and morphology, representative ice nucleation properties have to be assigned. Mineral dust is assumed to have the ice nucleation properties of montmorillonite/illite (i.e., rather efficient ice nuclei). This has been shown to give similar results to simulations with mixed-mineralogy particles depending on the source region (Hoose et al. 2008) because the most efficient ice nucleating dust component determines the average freezing rate. For soot, not many suitable experimental data are available. The data selected here are for laboratory-generated soot from an acetylene burner and a graphite spark generator, respectively. 2486 JOURNAL OF THE ATMOSPHERIC SCIENCES For biological particles, the variability in ice nucleation properties is largest. Only a small fraction of the atmospheric primary biological particles belong to ice nucleation active species. Lindemann et al. (1982) found that the ratio of bacteria that produce colonies active as IN to the total number of colony-forming units ranged between 0.04% and 4%, measured over bare soil and different crops. Maki and Willoughby (1978) identified 2 out of 13 (15%) bacteria strains in snow samples as ice nucleation active, but 0 out of 5 strains isolated from rain. Constantinidou et al. (1990) measured a fraction of 5.5% ice nucleation active bacteria strains during a rain event over a soybean field. A relative abundance on the order of ,5% for the Pseudomonadaceae family, to which the Pseudomonas genus with several ice nucleation active species belongs, was found in several air and snow samples from a high elevation site in Colorado (Bowers et al. 2009). Based on these observations, we assume that on global average 1% of all bacteria belong to ice nucleation active species (see also Phillips et al. 2009), represented by Pseudomonas syringae. These are called ‘‘Pseudomonas syringae–like’’ in the following. Note that also for Pseudomonas syringae–like bacteria species, only a small fraction of all cells of this species—not all cells—can nucleate ice (Hirano and Upper 1995). Concentration measurements of atmospheric concentrations of ice nucleation active fungi are rarer. Two (out of 14 investigated) species of the Fusarium genus have been found to nucleate ice with characteristics similar to Pseudomonas bacteria (Pouleur et al. 1992). The whole Fusarium genus again contributed to less than 3% of the total airborne fungal flora measured on Finnish farms (Lappalainen et al. 1996). In addition, some lichen fungi have been identified as ice nucleators (Kieft 1988; Henderson-Begg et al. 2009). We therefore assume that, as with the bacteria, 1% of all fungal spores belong to ice nucleation active (Pseudomonas syringae– like) species, which probably gives an upper estimate of the contribution of fungal spores to ice nucleation in the atmosphere. Ideally, the fraction of ice nucleation active bacteria and fungi species would be simulated as a function of climatic zones (Schnell and Vali 1973), but at present observations are too scarce to take this variation into account. A wide variety of pollen species have been found to nucleate ice (Diehl et al. 2002; von Blohn et al. 2005; Chen et al. 2008). Von Blohn et al. (2005) concluded that the ice nucleating ability seems to be a general pollen property. Therefore we assume that 100% of all pollen have ice nucleation properties similar to birch pollen (Diehl et al. 2002), which gives a high estimate of the pollen ice nucleation. VOLUME 67 d. Ice nucleation parameterizations The ice nucleation parameterization used in this study is based on CNT. Similar parameterizations based on CNT have been applied successfully in models on different scales (Khvorostyanov and Curry 2005; Morrison et al. 2005; Liu and Penner 2005), but so far the determination of the required aerosol-specific parameters has been uncertain. Chen et al. (2008) have presented a method for derivation of these parameters from laboratory experiments, and this method is applied here. Because of missing information about some experimental parameters (in particular, the number of particles per droplet for immersion freezing experiments, and the observation time), the conversion of the observed onset or median freezing temperatures into freezing rates is associated with considerable uncertainty, which is translated into the derived parameters. Similar derivations (e.g., Marcolli et al. 2007; Eastwood et al. 2008; Fornea et al. 2009; Welti et al. 2009; Kanji and Abbatt 2010; Kulkarni and Dobbie 2010; Luond et al. 2010), some with simplified formulations of classical nucleation theory, have demonstrated a large spread associated with the derived parameters, in particular for contact angles. The parameterization presented here has the advantage that other experimental results can be easily incorporated. In classical theory, the ice nucleation is seen as a stochastic process (Pruppacher and Klett 1997). An energy barrier has to be passed to add more molecules to small agglomerates of ice (subcritical germs) on the ice nucleus surface, until a critical germ size is reached. Following the notation in Chen et al. (2008), both immersion and deposition nucleation can be expressed in the same general form. The rate of heterogeneous nucleation per aerosol particle and time J is given by J 5 A9r2N ! pffiffiffi Dg# f Dg8g f exp , kT (1) where A9 is a prefactor depending only on ambient parameters (specified below for immersion and deposition nucleation), rN is the aerosol particle (nucleus) radius, f is a form factor containing information about the aerosol’s ice nucleation ability, Dg# is the activation energy (aerosol dependent and with different values for immersion and deposition nucleation), Dg8g is the homogeneous energy of germ formation (specified below for immersion and deposition nucleation), k is the Boltzmann constant, and T is the temperature in K. Taking into account the effect of curved surfaces, f has the general form AUGUST 2010 2487 HOOSE ET AL. 8 33 2 3 ! ! > > r r > 6 7 6 7 > !3> 1 cosu N cosu N > 6 7 6 7 rg rg 6 7 6 7 rN < 1 71 6 vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi7 v ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi f5 1 16 2 3 6 7 6 7 ! ! ! ! u u 2 2 > 2 rg > 6u 6u > rN rN 7 rN rN 7 > 4t 5 4t 5 > > 1 1 1 2 cosu 1 2 cosu : rg rg rg rg 2 33 9 2 3 ! ! > > r r > 6 7> 7 !26 cosu N cosu N > 6 7> 6 7 rg rg 6 7= 7 rN 6 6 7 6 7 v ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi v ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi 16 ! !2 7 > 1 3 cosu r 6 u ! !2 17 , u > 6u 7 6 7 g u > r r r r 4t 5> 4t 5 > > 1 2 cosu N 1 N 1 2 cosu N 1 N ; rg rg rg rg h 2 i with the critical germ size rg 5 rg,imm for immersion freezing or rg 5 rg,dep for deposition nucleation (parameterized below). The ice nucleus surface properties are contained in the contact angle u. Small contact angles facilitate the formation of ice germs on the particle surface. Highly efficient ice nuclei have the lowest values of u. In general, the contact angle for a specific aerosol has different values for immersion uimm and deposition nucleation udep. All parameterizations described below are applied in the temperature range of 08 to 2388C. Note that in this temperature range, heterogeneous freezing is the trigger for cloud glaciation via the Wegener–Bergeron–Findeisen process (e.g., Storelvmo et al. 2008b). (2) where n1,w is the number of molecules in contact with a unit area of particle surface and h is the Planck constant. Having calculated the nucleation rate per particle for all considered ice nuclei, the total change in ice crystal concentration Ni through immersion freezing can be obtained by summing up the contributions from the different aerosol species x, multiplied by the aerosol number concentration Naer,x and the fraction of these particles that is activated to liquid droplets fl, x: dN i 5 dt imm åx d( f l,x N aer,x ) dt 5 åx Jimm,x f l,x N aer,x , (6) 1) IMMERSION FREEZING In the liquid phase, the critical germ size is given by 2y w si/w . (3) rg,imm 5 kT ln(aw esw /esi ) The parameters contained here (see also Tables 2 and 3) are the volume of a water molecule y w, the surface tension between ice and liquid water si/w, the water activity aw, and the saturation vapor pressures over liquid water esw and ice esi. The freezing point depression through the solute effect is included by taking into account the water activity (,1) of the cloud droplet. Next, the homogeneous energy of germ formation is calculated from rg,imm: 4p . (4) s r2 Dgg8,imm 5 3 i/w g,imm In Eq. (1), which includes parts of the Zeldovich factor and the molecule flux toward the ice germ, A9 can be parameterized from ambient parameters: rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi yw n1,w 3 kTDgg8,imm A9imm 5 3 hrg,imm p rffiffiffiffiffiffiffiffiffiffiffi n1,w k3 T 3 ln2(aw esw /esi ) 4si/w 5 , (5) kT 4hy w s2i/w where x stands for three different modes of soot (two process-tagged Aitken modes and an internally mixed accumulation mode), two modes of dust (accumulation and coarse mode), and bacteria, fungal spores, and pollen. The modal size of the particles is used for rN in Eqs. (1) and (2). Here, fl,x is calculated for the soot and dust modes in the cloud droplet activation parameterization (Abdul-Razzak and Ghan 2000). In general, particles that are coated with soluble material are more easily activated to cloud droplets (higher fl,x) than uncoated particles. The biological particles are assumed to be 100% activated to cloud droplets (fl,bacteria 5 fl,fungi 5 fl,pollen 5 1) because of their large sizes and high ‘‘wettability’’ (Ariya et al. 2009). Possible immersion nuclei that have entered droplets via collision scavenging are not considered because this would require a separate tracking of in-droplet particles. Integrating Eq. (6) over one model time step Dt, we obtain DN i,imm 5 åx Minff l,x N aer,x f i,max,x , f l,x N aer,x [1 exp(J imm,x Dt)]g. (7) 2488 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 67 TABLE 2. List of symbols. Symbol (units) 22 21 A9 (m s ) A9dep (m22 s21) A9imm (m22 s21) aw e (Pa) esi (Pa) esw (Pa) f fi,max,x fl,x fx,coated Dg# (J) Dg# d,dep (J) Dgg8 (J) Dgg8,dep (J) Dgg8,imm (J) h (J s) J (s21) Jcontact (s21) Jdep (s21) Jimm (s21) Kcoll (m23 s21) k (J K21) mw (kg) Naer,x (m23) Ng,contact Ni (m23) DNi,contact (m23) DNi,dep (m23) DNi,imm (m23) Nl (m23) n1,w (m22) rg (m) rg,dep (m) rg,imm (m) rN (m) rN,x (m) Si T (K) Tc (8C) t (s) Dt (s) y w (m3) x u (8) udep (8) uimm (8) ns (s21) ri (kg m23) si/y (J m22) si/w (J m22) Description Prefactor in the nucleation rate calculation [Eq. (1)] Prefactor for deposition nucleation Prefactor for immersion nucleation Water activity [parameterized following Chen (1994)] Water vapor pressure Saturation vapor pressure over ice (Murphy and Koop 2005) Saturation vapor pressure over water (Murphy and Koop 2005) Form factor Maximum ice nucleating fraction for particles of species x Fraction of particles of species x activated to cloud droplets Coated fraction for particles of species x Activation energy (Table 3) Activation energy for deposition nucleation (Table 3) Homogeneous energy for germ formation Homogeneous energy for germ formation in the vapor phase Homogeneous energy for germ formation in the liquid phase Planck constant (56.63 3 10234) Ice nucleation rate per particle and time Contact freezing rate per particle and time Deposition nucleation rate per particle and time Immersion freezing rate per particle and time Collision kernel Boltzmann constant (51.38 3 10223) Mass of a water molecule (52.99 3 10226) Aerosol number concentration for species x Number of contact ice germs per aerosol particle Ice crystal concentration Change in ice crystal concentration due to contact nucleation Change in ice crystal concentration due to deposition nucleation Change in ice crystal concentration due to immersion nucleation Cloud droplet concentration Number of single molecules in contact with unit area of the substrate (in liquid water; 5 1019; Chen et al. 2008) Critical germ radius Critical germ radius for deposition nucleation Critical germ radius for immersion freezing Nucleus (aerosol particle) radius Nucleus (aerosol particle) radius for species x Supersaturation over ice Temperature Temperature Time Model time step (52400) Volume of a water molecule in ice (5mw/ri) Aerosol species index Contact angle (Table 3) Contact angle for deposition nucleation (Table 3) Contact angle for immersion freezing (Table 3) Frequency of vibration of water vapor molecule adsorbed on solid substrate (51013; Pruppacher and Klett 1997, p. 299) Density of ice [5916.7 2 0.175Tc 2 5 3 1024T c2; Pruppacher and Klett 1997, Eq. (3–2)] Surface tension between ice and vapor f5 si/w 1 sw/y 5 [(76.1 2 0.155Tc) 1 (28.5 1 0.25Tc)] 3 1023; Pruppacher and Klett 1997, Eqs. (5–46), (5–47a), and (5–12)g Surface tension between ice and water [5(28 1 0.25Tc) 3 1023; Pruppacher and Klett 1997, Eq. (5–47a)] The fraction of particles acting as immersion nuclei per model time step of 40 min fi,max,x is limited to 1% for soot and 0.1% for Pseudomonas syringae–like bacteria and fungal spores, based on typically observed maximum values (DeMott 1990; Möhler et al. 2008; Yankofsky et al. 1981; Phillips et al. 2009). These limits are reached at T & 250 K for soot and T & 268 K for Pseudomonas syringae–like bacteria and fungal spores. For mineral AUGUST 2010 2489 HOOSE ET AL. TABLE 3. Parameters for the ice nucleation parameterization, derived from laboratory data. Aerosol Reference Nucleation mode u (8) Dg# (10220 J) fi,max,x Soot Dust 1% of all bacteria 1% of all fungal spores Pollen Soot Dust DeMott (1990) Montmorillonite (Pitter and Pruppacher 1973) Pseudomonas syringae (Yankofsky et al. 1981) Pseudomonas syringae (Yankofsky et al. 1981) Birch (Diehl et al. 2002) Möhler et al. (2005) Illite (Zimmermann et al. 2008) Immersion Immersion Immersion Immersion Immersion Deposition Deposition 40.17 30.98 14.82 14.82 25.16 28.00 12.70 14.4 15.7 17.6 17.6 17.3 220.0 20.621 0.01 1.0 0.001 0.001 1.0 0.01 1.0 dust, no limit is imposed (fi,max,dust 5 1), but the simulated IN fractions in CAM-Oslo never exceed 25% before the Wegener–Bergeron–Findeisen process sets in and further nucleation is suppressed. Pollen (fi,max,pollen 5 1) reach IN fractions of 100% at T & 258 K. By imposing upper limits for the IN fractions, we account for the probably limited validity of the stochastic assumption of classical nucleation theory over the global model time step length (e.g., Vali 1994). The aerosol-specific immersion nucleation parameters are based on measurements by DeMott (1990), Pitter and Pruppacher (1973), Yankofsky et al. (1981), and Diehl et al. (2002) and are listed in Table 3. The derivation follows the fitting method by Chen et al. (2008). Figure 1 shows the parameterized nucleation rate Jimm as a function of temperature. Note that Jimm is calculated from Eq. (1), with rg,imm, Dgg8, imm , and A9imm as specified in Eqs. (3)–(5). Also included in Fig. 1 are the measurements used to derive u and Dg#. The freezing onset, defined as Jimm . 1025 s21, is approximately 288C for birch pollen, 2138C for montmorillonite, and 2248C for soot. The maximum nucleation rate is highest (meaning that freezing is fastest) for dust and birch pollen. For Pseudomonas syringae, the freezing rate does not exceed 1025 s21. At 258C, the freezing rate exceeds 1027 s21, which corresponds to a typical freezing onset temperature in experiments with relatively large liquid samples (e.g., Vali et al. 1976). e2 y w A9dep 5 mw kTns rffiffiffiffiffiffiffi s i/y . kT (10) Here si/y is the surface tension between ice and water vapor, mw is the mass of a water molecule, and ns is the vibration frequency of a water molecule attached to a surface. All constants and temperature-dependent parameters are listed in Tables 2 and 3. The isolines of constant Jdep in the T 2 Si space (Fig. 2) can be compared to the alignment of nucleation onset points (e.g., Zimmermann et al. 2008) and to the threshold (T, Si) values required for a certain activated fraction (e.g., Schaller and Fukuta 1979; Welti et al. 2009) from 2) DEPOSITION NUCLEATION The variables entering Eq. (1) for deposition nucleation are given below by analogy to immersion freezing. For a detailed derivation, see Chen et al. (2008). The critical germ size rg,dep, Dgg8, dep , and A9dep are functions of the temperature and of the water vapor pressure e (equivalent: the supersaturation over ice Si 5 e/esi): 2yw si/y rg,dep 5 , kT ln(e/esi ) Dgg8,dep 5 4p s r2 , 3 i/y g,dep (8) and (9) FIG. 1. Parameterized immersion freezing rates for soot (for rN 5 40 nm, as in the experiment), montmorillonite dust (750 nm), birch pollen (12.5 mm), and Pseudomonas syringae bacteria (500 nm). The crosses indicate freezing rates derived from measurements (see Table 3). The sizes of the CAM-Oslo aerosols are variable, depending on the mode and aging processes, and deviate from the radii shown here. The immersion freezing rates change accordingly. 2490 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 67 FIG. 2. Isolines of Jdep for illite dust (rN 5 2.5 mm) and soot (90 nm). The red crosses indicate the onset of nucleation in measurements (see Table 3). The black lines are parameterized deposition nucleation. The red line is simulated isoline of Jdep, which corresponds to the observed data. The blue lines are isolines of constant Jcontact, assuming a collision rate of 1023 s21. The blue lines correspond to nucleation rates of 1026, 1023, and 1022 s21 and the black lines to nucleation rates of 1026, 1023, 1022, 1021, and 1 s21 (from bottom to top). laboratory studies. The isolines calculated from the above formulas are either parallel to lines of constant Si (i.e., Jdep is independent of T), or are bent to higher Si at lower T (i.e., at constant Si, Jdep decreases with decreasing temperature). The latter behavior seems unexpected and contradicts most observations but can be physically explained with the lower absolute value of e at lower temperatures and the slowing down of the deposition process. At lower temperatures, some observations (Möhler et al. 2005; Shilling et al. 2006; Stetzer et al. 2008; Welti et al. 2009) reflect a slight decrease of Jdep with decreasing T, but closer to water saturation most data show the opposite behavior (see, e.g., Schaller and Fukuta 1979; Möhler et al. 2006; Bundke et al. 2008; Welti et al. 2009). This feature cannot be explained by the classical description for deposition nucleation on a dry substrate. A possible explanation is hygroscopic growth or surface wetting of the particles close to water saturation, such that the formation of ice germs from the vapor phase is inhibited. Some empirical formulations have been developed to cover this regime (Fukuta and Schaller 1982; DeMott 1995), but no general theoretical description is available. For derivation of the parameters used in this study, data for illite (Zimmermann et al. 2008) and for soot (Möhler et al. 2005) have been used. The illite data show a constant onset Si, independent of temperature, and can therefore be matched well by the theoretical description (Fig. 2). This is not true for the soot data. We have selected parameters that match the observations close to water saturation, but at lower temperatures the parameterization severely underestimates the deposition ice nucleation on soot. Because deposition nucleation before the formation of a liquid cloud is questionable, we consider here only in-cloud deposition nucleation. Based on observations by Korolev and Isaac (2006), a relative humidity of 98% (over water) is assumed inside mixed-phase clouds. The particles available for this process are uncoated dust and soot particles, which are not activated to liquid droplets. We assume here that coated particles are completely deactivated, which is a simplification of recent experimental results (Eastwood et al. 2009; Cziczo et al. 2009a). These studies demonstrated that coated dust particles require higher supersaturation (or lower temperatures) than uncoated particles to be activated. The change of ice crystal number with time by deposition nucleation is given by the sum over two different modes of soot (one externally mixed mode and one partially coated mode) and two modes of dust (accumulation and coarse mode, both partially coated); only the uncoated fractions of these modes contribute to deposition nucleation. The index x runs over these four aerosol species: dN i 5 dt dep 5 åx d[(1 f l,x )(1 f x,coated )N aer,x ] dt åx Jdep,x, RH 50.98 (1 f l,x )(1 f x,coated )N aer,x . w (11) AUGUST 2010 The modal size of the particles is used for rN in Eqs. (1) and (2). We obtain fl,x from the cloud droplet activation parameterization (Abdul-Razzak and Ghan 2000). The coated fraction fx,coated is calculated by distributing the available soluble mass (organic and sulfate) over the dust and black carbon cores in the internally mixed modes, requiring a minimum coverage of one monolayer. Integrating Eq. (11) over one model Dt, we obtain DN i,dep 5 2491 HOOSE ET AL. contact nucleation is much higher than the onset temperature for deposition nucleation. The equilibrium number of possible contact nucleation germs per particle is given by N g,contact ’ 4pr2N ns " e pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2pmw kT 3 exp åx Minf(1 f l,x )(1 f x,coated )N aer,x f i,max,x, # Dg# dep 1 f Dg8g,dep (r g,imm ) kT . (13) (1 f l,x )(1 f x,coated )N aer, x The homogeneous nucleation energy for germ formation from the vapor phase [Eq. (9)] is evaluated at the 3 [1 exp(J dep,x,RH 50.98 Dt)]g. (12) w (smaller) critical size for immersion freezing germs. The fraction of particles acting as deposition nuclei per Because of the steep decrease of germ number for larger model time step of 40 min is limited to 1% for soot. sizes, the total number can be approximated by evaluatActivated fractions of up to 5% have been observed by ing the integral at rg,imm. We evaluate f for f(udep, rg,imm). The contact nucleation rate is given by the collision Petters et al. (2009), but most experimental data are rate between droplets and aerosols that contain at least reported for activated fractions of less than 1% (Möhler one contact nucleation germ. As in the case of deposiet al. 2005; Kanji and Abbatt 2006). No upper limit is tion nucleation, only uncoated, nonactivated particles imposed for mineral dust, and in rare cases the simulated are allowed to act as contact nuclei. The total contact dust deposition IN fraction reaches 60%. This is in nucleation rate is given (as for deposition nucleation) by agreement with activated fractions up to 69% reported the sum over two modes containing black carbon and by Field et al. (2006). Deposition nucleation on bitwo modes containing mineral dust, denoted by x. As the ological particles is not considered because they are all biological particles are assumed to be fully activated to assumed to be activated to cloud droplets (see immersion cloud droplets, they do not contribute to contact nufreezing description) and because the necessary obsercleation. vational data for the parameter derivation are missing. 3) CONTACT FREEZING As suggested by Chen et al. (2008), we calculate contact freezing following ‘‘Cooper’s hypothesis.’’ Cooper (1974) postulated that subcritical ice germs, formed through deposition from the vapor phase on a dry particle surface, can initiate immediate freezing upon collision with a liquid droplet, if their size is at or above the critical germ size for immersion nucleation. The critical germ radius for immersion nucleation [Eq. (3)] is approximately a factor of 4 smaller than the critical germ radius for deposition nucleation [Eq. (8)], evaluated at water saturation. Therefore, the onset temperature for DN i,contact 5 dN i 5 dt contact åx Kcoll (rN,x , rl )Nl (1 f l,x ) 3 (1 f x,coated )N aer, x Max(N g,contact,x , 1) . (14) Here Kcoll(rN, rl) is the collision kernel for aerosols of size rN and droplets of size rl; Kcoll includes Brownian movements, thermophoresis, and diffusiophoresis and is calculated following Young (1974) and Cotton et al. (1986) for 98% relative humidity over water and the modal aerosol size. As above, we obtain by integration over one model Dt: åx Minh(1 f l,x )(1 f x,coated )N aer,x f i,max,x , (1 f l,x )(1 f x,coated )N aer,x (15) 3 f1 exp[Kcoll (rN ,x , r1 )N l Max(N g,contact,x , 1)Dt]gi. As for deposition and immersion nucleation, a limit of 1% is applied for the ice nucleating fraction of soot. Figure 2 includes contact nucleation rates under the assumption of a typical collision rate KcollNl of 1023 aerosol–droplet collisions per aerosol particle per second (e.g., Croft et al. 2010). The contact nucleation probability is found to be a steep function of Si. The increase of contact nucleation with increasing relative humidity 2492 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 67 TABLE 4. Global annual mean cloud cover (CC, %), liquid water path (LWP, g m22), ice water path (IWP, g m22), shortwave cloud forcing (SWCF, W m22), longwave cloud forcing (LWCF, W m22) and net radiation at the top of the atmosphere Fnet (W m22) for the present-day simulations, and differences of these variables between present-day and preindustrial simulations. Standard errors are given for simulations CTL and CNT (and are similar for the other simulations). CC DCC LWP DLWP IWP DIWP SWCF DSWCF LWCF DLWCF Fnet DFnet CTL CNT CNT-highbact CNT-lowdust CNT-nosootct CNT-nobio 64.9 6 0.1 20.2 6 0.1 109.3 6 0.1 4.11 6 0.25 31.0 6 0.04 0.12 6 0.08 257.7 6 0.02 21.30 6 0.07 32.3 6 0.05 20.38 6 0.09 0.13 6 0.04 21.68 6 0.09 64.7 6 0.1 0.0 6 0.1 101.7 6 0.3 2.94 6 0.36 29.9 6 0.04 0.13 6 0.07 256.2 6 0.08 21.40 6 0.10 31.5 6 0.02 20.23 6 0.05 0.94 6 0.11 21.55 6 0.17 64.7 0.0 101.6 3.37 30.0 0.21 256.1 21.49 31.5 20.29 0.94 21.74 65.1 20.1 105.3 1.14 30.9 0.32 257.1 21.23 32.2 20.44 1.12 21.72 64.7 20.2 103.4 4.45 29.8 20.04 256.4 21.61 31.6 20.26 0.65 21.89 64.7 0.0 101.9 3.12 29.9 0.07 256.2 21.49 31.6 20.17 0.98 21.55 is consistent with the (qualitative) results by Svensson et al. (2009). For the deposition nucleation parameters for illite, this implies possible contact nucleation already around 258C at the assumed relative humidity inside mixed-phase clouds of 98%, and for soot at 298C. These values are rather high compared to experiments (e.g., Pitter and Pruppacher 1973; Diehl and Mitra 1998), but at present no other theoretically consistent parameterization is available. 3. Results a. Simulation of clouds and heterogeneous freezing In this section, results from the CAM-Oslo model with the new freezing parameterizations are presented. First, a brief overview over the simulated clouds is given. Second, the relevant aerosol concentrations are shown. Third, we discuss the contributions of the different aerosol particles to heterogeneous ice nucleation. The simulated ice nucleation rates are compared to observations in the next section. 1) CLOUDS AND RADIATIVE PROPERTIES Table 4 lists global mean values for cloud-related variables from the different experiments. In general, all simulations agree well with satellite retrievals (see e.g., Lohmann et al. 2007, their Table 2), except for an overestimation of the shortwave cloud forcing. The CNT simulation exhibits an approximately 7% lower global mean liquid water path (LWP) than the CTL simulation because of enhanced freezing. This also leads to a decrease in shortwave cloud forcing. The change in ice water path (IWP) between simulations CTL and CNT is roughly proportional to the change in LWP because more frequent cloud glaciation entails enhanced precipitation release. Most cloud properties in the different CNT sensitivity experiments are very similar, except in simulation CNT-lowdust. The global mean LWP is significantly higher in simulation CNT-lowdust than in simulation CNT because of reduced liquid-to-ice conversion. Also, the CNT-nosootct and CNT-nobio simulations exhibit less freezing and a higher LWP than simulation CNT. As far as global average cloud properties are concerned, the CNT-highbact simulation is not significantly different from the CNT simulation. 2) PARTICLE NUMBER CONCENTRATIONS The zonal average number concentrations of mineral dust, soot, and biological particles in simulation CNT are shown in Fig. 3. Soot particles, which originate from natural and anthropogenic combustion processes, are most numerous, with zonal average concentrations exceeding 1000 cm23 at the surface in the Northern Hemisphere. These particles are mainly in the Aitken mode. Mineral dust particles, which are in the accumulation and coarse mode size range, reach a maximum zonal average surface concentration of 65 cm23, and typical tropospheric concentrations are 1–10 cm23. As discussed by Seland et al. (2008), CAM-Oslo has a rather strong vertical mixing, linked to efficient deep convective vertical transport. This can also be seen for a previous version of CAM-Oslo in the AeroCom model intercomparison (Textor et al. 2006). Primary biological particles are present in significantly lower concentrations: typical annual average concentrations over continents are 1022–1021 cm23 for bacteria, 1023–1022 cm23 for fungal spores, and 1026–1025 cm23 for pollen, with a large seasonal variability. The biological particle concentrations are in fair agreement with measurements (Hoose et al. 2010). Note that the concentrations shown here are total aerosol concentrations, not IN AUGUST 2010 HOOSE ET AL. 2493 FIG. 3. Zonal annual mean particle number concentrations in simulation CNT. Note the two different color bars for the top and bottom rows. concentrations, and that only a small subset of all aerosol particles serves as ice nuclei. 3) ICE NUCLEATION RATES Figures 4, 5, and 6 display the zonal annual mean freezing rates [i.e., DNi/Dt from Eqs. (7), (12), and (15), weighted with the cloud fraction and separated by aerosol component]. For Fig. 7, these rates are vertically integrated and globally averaged. Dust immersion freezing is found to be the dominant ice nucleation mechanism, followed by soot immersion and soot contact nucleation, which contribute approximately equally. Bacteria, fungal spore, and pollen immersion freezing rates are several orders of magnitude lower than the dust and soot freezing processes. Bacteria immersion freezing is highest in the lower troposphere at mid and high latitudes, while most other ice nucleation processes peak around 600–400 hPa in the midlatitudes and around 400–300 hPa in the tropics. In general, the processes that occur at lower temperatures (e.g., soot immersion freezing) peak at higher altitudes than the warm-temperature freezing mechanisms (e.g., dust contact freezing). The contact freezing rates exhibit two maxima (Fig. 6): one close to the surface sources, where the number concentrations of uncoated particles are highest, and one at upper levels, where low temperatures occur more often. Soot deposition nucleation, which is limited to temperatures close to the homogeneous freezing onset, mainly occurs in the upper tropical troposphere and in the lower troposphere over Siberia and Alaska. We note that the efficiency of soot deposition nucleation above 2388C is a matter of debate in recent literature (Gorbunov et al. 2001; Dymarska et al. 2006). If soot deposition nucleation was effective at higher temperatures than assumed here, the total soot (i.e., the anthropogenic) contribution to heterogeneous ice nucleation would increase. The main difference from the partitioning of the freezing processes as simulated in ECHAM5-Hamburg Aerosol Model (HAM; Hoose et al. 2008) is the lower soot contact freezing rate in CAM-Oslo. This is because the freezing parameterization in Hoose et al. (2008) (based on Diehl and Wurzler 2004) did not directly depend on the concentration of soot and dust particles but only on their fractional contribution to the total aerosol, which can lead to artifacts. Therefore, contact freezing by soot was omitted in the follow-up study by Lohmann and Hoose (2009). Here we explicitly calculate the collision rate between externally mixed, uncoated soot particles and droplets, which results in a lower frequency of contact freezing events. b. Comparison to observations Ice nucleation schemes in global models are difficult to evaluate. While laboratory and field measurements have been used for comparison with parameterizations in a parcel model framework (Eidhammer et al. 2009), 2494 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 67 FIG. 4. Zonal annual mean immersion freezing rates (DNi,imm/Dt) in simulation CNT. for global models so far only the ice crystal concentrations and ice crystal sizes, which are determined by both primary and secondary ice formation and sink processes, have been compared to observations (Lohmann and Diehl 2006; Storelvmo et al. 2008a). Here we show comparisons to available data from in situ IN observations. This comparison is possible only in a statistical sense because the model is not able to capture the exact conditions at the sampling points in both space and time. 1) IN CONCENTRATIONS The most common instrument for measuring ice nuclei concentrations in the atmosphere is the continuous-flow diffusion chamber (CFDC) (Rogers et al. 2001). In this instrument, aerosol particles enter through an inlet and are exposed to a chosen temperature and ice supersaturation. After a residence time of 5–20 s (depending on the instrument setup), the particles that have grown to ice crystals larger than 1 mm are optically detected. While the CFDC has the advantage of allowing real-time airborne measurements, some limitations have to be accounted for. Because of the short residence time, the dominant ice nucleation modes in the CFDC are deposition and condensation nucleation. The largest aerosol particles (.1.2–2 mm in diameter) have to be removed upstream of the chamber to avoid confusion with the nucleated ice crystals. For comparison to CFDC chamber measurements, the model ice nuclei concentration [hereafter termed ‘‘model IN(10s)’’] has been defined as a 10-s integral over the time-step mean, in-cloud freezing rates [sum over Eqs. (7), (12), and (15), multiplied by 10/Dt]. Classical nucleation theory predicts a constant freezing rate (i.e., the number of ice-nucleating particles would increase approximately proportionally to the sampling time) as long as the aerosol and droplet populations are not significantly depleted. Here, limitations on the maximum fraction of active particles per species are imposed (see Table 3). These upper bounds are accounted for in the values of the model IN(10s). Figure 8 shows the simulated model IN(10s) concentrations as a function of temperature, sampled at all global grid points at an arbitrary time step. The simulated IN(10s) concentrations attain significant values at temperatures below 2118C and increase strongly with decreasing temperature until around 2208C. In this temperature range, AUGUST 2010 HOOSE ET AL. 2495 FIG. 5. Zonal annual mean deposition nucleation rates (DNi,dep/Dt) in simulation CNT. model IN(10s) concentrations are mostly between 0.5 and 20 L21. Also shown in Fig. 8 are CFDC IN concentrations from a number of campaigns at different locations. The measured IN concentrations are of the same order of magnitude and reflect the same temperature dependence as the simulated concentrations. However, when the different studies are investigated individually, the observed temperature dependence is weaker. We have to keep in mind that the CFDC measurements report the ice nuclei concentration at a selected chamber temperature, which can be different from the environmental temperature, while the simulated ice nuclei concentrations are reported for the actual gridpoint temperature. Therefore the model ice nuclei concentrations for lower temperatures tend to be valid for higher altitudes and latitudes, where the aerosol concentration is also lower in general. For a more detailed comparison, this analysis is repeated for the grid boxes closest to the CFDC measurement locations (Fig. 9). Fort Collins and Storm Peak fall into the same global model grid box, but we have selected data from different vertical levels to account for the altitude of the Storm Peak laboratory (3200 m). The data at Barrow, Alaska, were collected by aircraft within the lowest 2000 m of the atmosphere, and the model data are sampled from the corresponding vertical levels. In this comparison, the model IN(10s) concentrations are diagnosed for 17 different temperatures at 28 intervals from 268 to 2388C by repeating the freezing rate calculations with specified temperature values. Note that the model IN(10s) concentrations calculated in this way still depend on the simulated cloud parameters (liquid activated fraction, cloud droplet sizes, etc.) and are not completely equivalent to the processes occurring in a CFDC. At all investigated locations, the mean model IN(10s) concentrations increase with decreasing temperatures from 2128 to 2248C and then flatten off or even decrease again. This behavior at T , 2248C is consistent with the observations at Storm Peak. The strong temperature dependence at T . 2208C is not confirmed by the Fort Collins and Barrow data, which are more scattered. At Storm Peak and at Barrow for T , 2158C, the observed data fall between the 25th and 75th percentiles of the simulated data. The majority of observations at Fort Collins show higher IN concentrations than simulated, and this is also true for T . 2158C at Barrow. Not much FIG. 6. Zonal annual mean contact freezing rates (DNi,contact/Dt) in simulation CNT. 2496 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 67 FIG. 8. Ice nuclei concentrations (calculated as 10-s integrals over the freezing rate), sampled at all global grid points at an arbitrary time step of simulation CNT (black dots). The colored symbols represent CFDC IN measurements at various locations. FIG. 7. Global annual mean vertically integrated nucleation rates in simulation CNT. can be said about regional variations. The vertical and temporal variability at the Fort Collins/Storm Peak gridpoint is as large as the difference between Colorado and Barrow. The CFDC IN concentrations have been found to correlate well with the concentration of coarse mode aerosol particles (DeMott et al. 2006) but not with total aerosol concentration, which is dominated by smaller particles. Similar results were obtained earlier by Georgii and Kleinjung (1967). This is in agreement with the nucleation rate increasing with the square of the particle radius [Eq. (1)]. Figure 10 displays the model IN(10s) concentration versus the concentration of aerosol particles with diameter .0.5 mm. If sampled at all temperatures (Fig. 10a), only a modest correlation is obtained because of low IN concentrations at warm subzero temperatures. But if sampled only at T , 2208C (Fig. 10b), the model IN(10s) concentration increases systematically with coarse mode aerosol particle concentration. The fit to the data from several campaigns by DeMott et al. (2006), which is included for comparison, shows a similar but steeper slope. Georgii and Kleinjung (1967) find a slope (measured for aerosol particles with a diameter .0.6 mm) that is similar to the model results. The reason for the high correlation in the model is that dust particles, which constitute the majority of the IN, are also the most abundant coarse mode aerosols in regions with low temperatures. Such temperatures are not common in the marine boundary layer, where sea salt is the dominant coarse mode aerosol and no such correlation can be expected. 2) COMPOSITION OF ICE NUCLEI AND ICE CRYSTAL RESIDUES The composition of particles obtained from evaporated ice or snow crystals (residues) can give indications about the composition of the ice nuclei that were responsible for the freezing. However, the residues also contain particles scavenged by collisions, complicating the interpretation of the data. Alternatively, the composition of particles that have nucleated ice in a CFDC can be analyzed, under the limitations of the CFDC measurements as discussed above. Table 5 lists a number of ice crystal residue (Targino et al. 2006; Pratt et al. 2009), snow crystal residue (Kumai 1961; Kumai and Francis 1962), and CFDC IN (Phillips et al. 2008; Prenni et al. 2009a,b) composition measurements. The dataset compiled by Phillips et al. (2008) is the most comprehensive one. Not all measurements distinguish organic and elemental carbonaceous particles. All observations agree on mineral dust as the dominant IN/ice crystal residue component (50%–88% in number), but the carbonaceous fraction is more variable (0%–47%). Biological particles were only identified in three cases, and with very different fractions: 1% (Kumai 1961), 33% (Pratt et al. 2009), and up to 47% (Prenni et al. 2009b). These numbers are compared to the global average IN composition from the model. In the CTL simulation, which includes only mineral dust and soot IN, these contribute to 84% and 16%, respectively. In the CNT AUGUST 2010 HOOSE ET AL. 2497 FIG. 9. Ice nuclei concentrations for specified temperatures (calculated as 10-s integrals over the freezing rate), sampled at the grid points closest to the measurement locations, at 10 arbitrary time steps during the specified month, in simulation CNT (black boxes and whiskers). The whiskers represent the 5th and 95th percentiles, and the boxes the 25th and 75th percentiles and the median. The asterisks mark the simulated mean concentrations. The colored symbols represent CFDC IN measurements. simulation, with the new freezing parameterization and additional contributions by biological IN, this distribution remains very similar: 77% and 23%. On global average, only 1 in 107 ice nuclei is of biological origin. For dust and soot, these values lie within the broad range of observed values. The simulated biological IN fraction is much lower than the high values reported by Pratt et al. (2009) and Prenni et al. (2009b). However, the Prenni et al. (2009b) data were sampled directly above the rain forest canopy, where temperatures are always above 08C and ice nuclei cannot be diagnosed in the model, and the Pratt et al. (2009) data stem from only one individual cloud. While these data might not be representative of the global contribution of biological particles to ice nucleation, they suggest that biological influence on clouds can be strongly enhanced on local scales. For the Arctic (north of 668N), an enhancement of the mineral dust component (to 88%) is found. This is consistent with the Arctic measurements by Prenni et al. (2009a) showing a larger mineral dust fraction (64%) than the more comprehensive dataset of Phillips et al. (2008) (52%), which is mainly based on the same instrumental method. Further regional comparisons are difficult to infer from the observations listed in Table 5 because of differences in sampling and instrumentation. The results of the sensitivity studies (also listed in Table 5) demonstrate the sensitivity of the model to assumptions entering the ice nucleation parameterization. In the simulation CNT, global ice nucleation is split among mineral dust, soot, and biological particles as 77%, 23%, and 1027%, respectively. The biological IN fraction is increased to 5 in 107 particles in the CNThighbact simulation. In the CNT-lowdust simulation, mineral dust contributes only to 39% of the ice nucleation, while the soot fraction is raised to 61%. This partitioning is in worse agreement with the observations 2498 JOURNAL OF THE ATMOSPHERIC SCIENCES VOLUME 67 FIG. 10. Ice nuclei concentrations (calculated as 10-s integrals over the freezing rate) in simulation CNT, displayed as a function of the number concentrations of aerosol particles with d . 0.5 mm, for (a) all T and (b) T # 2208C. The dashed lines are power-law fit to observations (DeMott et al. 2006; Georgii and Kleinjung 1967); d . 0.6 mm, T 5 2218C. than the other simulations, as all field studies listed in Table 5 show a larger mineral dust IN fraction than carbonaceous IN fraction. When contact freezing by soot is switched off, the mineral dust IN fraction increases to 88%, which is at the high end of the observed values. The soot fraction is 12% in the CNT-nosootct simulation. Finally, without biological particles as IN (simulation CNT-nobio), the partitioning of heterogeneous freezing between mineral dust and soot is very similar to that in simulation CNT. The simulated ice nucleation is to a larger extent dominated by mineral dust than most field observations, but this result is sensitive to assumptions on dust scavenging. The observed carbonaceous IN fractions are highly variable, with the simulated percentage (in all experiments except CNT-lowdust) well in the middle of the observed range. For biological particles, globally representative data are not available yet, so no definite conclusions on the model performance for these particles can be drawn. The nature and origin of the ‘‘other’’ particles, which make up 1%–34% of the measured IN/ ice crystal residues, remain to be solved. c. Aerosol indirect effect In Table 4, the global mean differences between the present-day and preindustrial cloud and radiative properties are listed. The model includes direct, semidirect, and indirect aerosol effects in warm and mixed-phase clouds. Aerosol effects on ice clouds at temperatures below the homogeneous freezing threshold are not considered in this study. The different aerosol indirect effects (cloud albedo effect and cloud lifetime effect in warm clouds and the glaciation and deactivation indirect effect in mixed-phase clouds; Lohmann and Hoose 2009) counteract each other. This complicates the interpretation of the resulting net effect. In the CTL simulation, the change in top-of-theatmosphere net radiation Fnet is 21.68 6 0.09 W m22 (5-yr average with standard error), which is less negative than if only warm-phase indirect effects are included (Hoose et al. 2009; 22.1 W m22). The differences compared to Storelvmo et al. (2008b) stem from model updates in the warm-phase physics. In simulation CNT, DFnet is slightly less negative (21.55 6 0.17 W m22) than in CTL, probably because of a different vertical distribution of cloud liquid water and a stronger glaciation indirect effect, which tends to counteract the indirect effect of warm clouds (Lohmann 2002). This hypothesis is confirmed by the more negative indirect effect in the CNT-nosootct simulation (21.89 6 0.11 W m22), in which part of the soot glaciation capability is suppressed. The similarly high indirect effect in the simulation CNT-lowdust is presumably linked to the increased LWP and corresponding increase in warm-phase indirect effects. Variation of the assumptions on biological ice nucleation (simulations CNT-highbact and CNT-nobio) results in nonlinear changes of the indirect effect. For both enhanced and reduced biological ice nucleation, a more negative DSWCF is found. This can be explained on the one hand by a higher LWP in the CNT-nobio simulation (and thus more clouds contributing to the warm cloud indirect effect) and on the other hand by natural aerosol dominating the onset of freezing in simulation CNThighbact (and thus less potential for anthropogenic soot to trigger cloud glaciation). AUGUST 2010 2499 HOOSE ET AL. TABLE 5. Ice nuclei composition from observations and simulations, given as fractional frequency of occurrence. From observations, ice and snow crystal residue composition and CFDC IN compositions from atmospheric samples at individual locations are reported. From the model, the global or regional average contributions to the total heterogeneous freezing rates are listed. ‘‘Mineral dust’’ includes the ‘‘dust/metallic’’ category in Phillips et al. (2008), ‘‘metal oxides/dust,’’ ‘‘metal oxides/dust 1 sulfates/salts,’’ and ‘‘metal oxides/dust 1 carbonaceous’’ particles sampled by Prenni et al. (2009a), and ‘‘dust’’ and ‘‘dust 1 carbonaceous’’ particles in Prenni et al. (2009b). For ‘‘carbonaceous’’, the ‘‘soot,’’ ‘‘biological,’’ and ‘‘organic carbon-nitrate’’ particles in Pratt et al. (2009), the ‘‘combustion product’’ and ‘‘micro-organism’’ particles in Kumai (1961), and the soot and bacteria, fungal spores, and pollen from the model are summed up, and the ‘‘low-Z’’ fraction from Targino et al. (2006) is reported. Prenni et al. (2009b) inferred from indirect evidence that the carbonaceous IN fraction measured in the Amazon basin was dominated by biological particles. Phillips et al. (2008), from six campaigns Prenni et al. (2009a), Mixed-Phase Arctic Clouds Experiment (MPACE), Alaska Prenni et al. (2009a), Surface Heat Budget of the Arctic Ocean (SHEBA)–First International Satellite Cloud Climatology Project (ISCCP) Regional Experiment–Arctic Cloud Experiment (FIRE.ACE), Arctic Ocean Prenni et al. (2009b), Amazon basin Pratt et al. (2009), Wyoming Targino et al. (2006), Scandinavia Kumai (1961), Hokkaido, Japan Kumai (1961), Honshu, Japan Kumai (1961), Michigan Kumai and Francis (1962), Greenland Simulation CTL, global Simulation CNT, global Simulation CNT, Arctic Simulation CNT-highbact, global Simulation CNT-lowdust, global Simulation CNT-nosootct, global Simulation CNT-nobio, global 4. Conclusions A new ice nucleation parameterization has been introduced in the CAM-Oslo model, treating more processes and ice nuclei species than previous global model studies. Primary biological particles (bacteria, fungal spores, and pollen) are included with simple emission parameterizations recently published in the literature. These emission functions and the resulting concentrations bear considerable uncertainties (e.g., with respect to seasonal variability). Further developments are required and can help to improve our estimates in the future [see, e.g., Vogel et al. (2008) for a detailed pollen emission parameterization in a regional model]. For biological particles as well as for mineral dust and soot, the simulated concentrations in the upper troposphere are sensitive to vertical transport and to assumptions on the particle mixing state and scavenging (e.g., Koch et al. 2009; Croft et al. 2010). The ice nucleation parameterization is based on classical theory, which provides a theoretically sound and consistent framework. Nevertheless, some observations are in conflict with the assumption of a stochastic nature of ice nucleation, especially with a freezing rate that is constant in time. A distribution of contact angles and Mineral dust Carbonaceous Soot Biological Other 0.52 0.64 0.37 0.35 n/a n/a n/a n/a 0.11 0.01 0.64 0.17 n/a n/a 0.19 0.50 0.50 0.58 0.57 0.88 0.87 0.85 0.84 0.77 0.88 0.77 0.39 0.88 0.77 0.47 0.41 0.23 0.09 0.04 0.02 0.00 0.16 0.23 0.12 0.23 0.61 0.12 0.23 n/a 0.04 n/a 0.08 0.04 0.02 0.00 0.16 0.23 0.12 0.23 0.61 0.12 0.23 up to 0.47 0.33 n/a 0.01 0.00 0.00 n/a 0.00 1 3 1027 1 3 1027 5 3 1027 2 3 1027 1 3 1027 0.00 0.03 0.09 0.19 0.34 0.08 0.11 0.15 0.00 0.00 0.00 0.00 0.00 0.00 0.00 activation energies instead of one fixed parameter per aerosol species could be a way to alleviate this discrepancy (Marcolli et al. 2007; Luond et al. 2010). With a distribution of the efficiency of ice nucleation within the aerosol population, the most efficient ice nuclei would be depleted after the first initiation of freezing, and further nucleation would be delayed. However, a distribution of contact angles and activation energies is difficult to derive from the measured nucleation rates and would entail complications in the implementation. Here, these problems are circumvented in a simplified way by applying upper limits to the percentage of aerosols acting as ice nuclei, but the values of these upper limits are also arguable. In-cloud deposition nucleation is included for uncoated mineral dust and soot particles. Contact freezing is the most uncertain process in our description, and further experimental and theoretical studies are required before its parameterization can be improved. Other less well-understood freezing mechanisms (e.g., inside-out contact nucleation; Durant and Shaw 2005) are not considered here. With the new ice nucleation parameterization applied for mineral dust, soot, bacteria, fungal spores, and pollen, it is found that on global average 77% of the simulated 2500 JOURNAL OF THE ATMOSPHERIC SCIENCES heterogeneous nucleation is initiated by mineral dust particles and 23% by soot, while biological particles only contribute a fraction of 1027 of all ice nucleation events. Immersion freezing is the dominant freezing mechanism, but for soot—which is often externally mixed and not activated to cloud droplets—contact freezing is also relevant. Even with more extreme assumptions as to the probability of bacteria and fungal spores acting as ice nuclei, the biological aerosol contribution to global freezing remains marginal because of their low number concentrations. Nevertheless, the simulated concentration of bacterial IN in precipitation is of the same order of magnitude or higher than the measured concentrations of biological IN in snow samples (Hoose et al. 2010). However, we cannot rule out the local importance of biological particles nor the possibility that in some cases at warm subzero temperatures the few but very active biological IN can initiate glaciation of clouds, which would have remained liquid without this trigger. The simulated ice nuclei concentrations are compared to CFDC measurements, and in a statistical sense a good agreement is found. At temperatures below 2208C, the simulated IN concentrations correlate with the coarse mode aerosol concentration, similar to observations. The effect of the new ice nucleation parameterization on the simulated indirect effect is overall small. Although the contribution of anthropogenic soot to heterogeneous ice nucleation is slightly higher than in the control simulation, the glaciation indirect effect is lower than in previous studies and cannot significantly offset the indirect effects of warm clouds. Numerous uncertainties remain concerning the numerical description of ice nucleation in large-scale models, especially for biological particles: emissions, size distributions, ice nucleation active fractions, hydrophilicity, wet deposition, freezing rates, the role of preactivation, the abundance of different biological species in different climatological regions, and maximum ice nucleating fractions. Furthermore, the ice nucleation efficiency of mineral dust has also been linked to biogenic contamination, which would mean that biological ice nucleation is already implicitly contained when mineral dust ice nucleation is included. The importance of further possible ice nucleators, such as volcanic ash and anthropogenic metallic particles, cannot be assessed yet because their global sources are not well known. With the uncertain parameters selected to the best of our present knowledge, we find that mineral dust dominates cloud glaciation and that the role of biological particles for ice formation in mixed-phase clouds is small in a global average. This implies that a possible aerosol influence on precipitation formation via the Wegener– Bergeron–Findeisen process varies with past or future VOLUME 67 change in dust emissions. We suggest that further laboratory and field experiments are mandatory in order to obtain a larger database for improved modeling studies. Acknowledgments. C. H. thanks Xiaohong Liu and Stephan Weinbruch for helpful discussions, Paul DeMott for providing data, Trond Iversen, Alf Kirkevåg, and Øyvind Seland for development of the CAM-Oslo aerosol module, and Trude Storelvmo for providing the double-moment cloud microphysics scheme and valuable comments. Three anonymous reviewers are acknowledged for their constructive comments, which helped to improve this article. This research was supported by the projects NorClim (Norwegian Research Council Grant 178246), EUCAARI (European Integrated Project 036833-2), and POLARCAT (Norwegian Research Council Grant 460724), and computing time was provided through a grant from the Norwegian Research Council’s Program for Supercomputing. REFERENCES Abdul-Razzak, H., and S. J. Ghan, 2000: A parameterization of aerosol activation. 2. Multiple aerosol types. J. Geophys. Res., 105 (D5), 6837–6844. Ansmann, A., and Coauthors, 2008: Influence of Saharan dust on cloud glaciation in southern Morocco during the Saharan Mineral Dust Experiment. J. Geophys. Res., 113, D04210, doi:10.1029/2007JD008785. ——, and Coauthors, 2009: Evolution of the ice phase in tropical altocumulus: SAMUM lidar observations over Cape Verde. J. Geophys. Res., 114, D17208, doi:10.1029/2008JD011659. Ariya, P., J. Sun, N. Eltouny, E. Hudson, C. Hayes, and G. Kos, 2009: Physical and chemical characterization of bioaerosols— Implications for nucleation processes. Int. Rev. Phys. Chem., 28, 1–32. Bowers, R. M., C. L. Lauber, C. Wiedinmyer, M. Hamady, A. G. Hallar, R. Fall, R. Knight, and N. Fierer, 2009: Characterization of airborne microbial communities at a high-elevation site and their potential to act as atmospheric ice nuclei. Appl. Environ. Microbiol., 75, 5121–5130. Bundke, U., B. Nillius, R. Jaenicke, T. Wetter, H. Klein, and H. Bingemer, 2008: The fast ice nucleus chamber FINCH. Atmos. Res., 90, 180–186. Burrows, S. M., T. Butler, P. Jöckel, H. Tost, A. Kerkweg, U. Pöschl, and M. G. Lawrence, 2009: Bacteria in the global atmosphere. Part 2: Modeling of emissions and transport between different ecosystems. Atmos. Chem. Phys., 9, 9281–9297. Cantrell, W., and A. Heymsfield, 2005: Production of ice in tropospheric clouds: A review. Bull. Amer. Meteor. Soc., 86, 795–807. Chen, J.-P., 1994: Theory of deliquescence and modified Köhler curves. J. Atmos. Sci., 51, 3505–3516. ——, A. Hazra, and Z. Levin, 2008: Parameterizing ice nucleation rates using contact angle and activation energy derived from laboratory data. Atmos. Chem. Phys., 8, 7431–7449. Collins, W. D., and Coauthors, 2006: The formulation and atmospheric simulation of the Community Atmosphere Model version 3 (CAM3). J. Climate, 19, 2144–2161. AUGUST 2010 HOOSE ET AL. Constantinidou, H. A., S. S. Hirano, L. S. Baker, and C. D. Upper, 1990: Atmospheric dispersal of ice nucleation-active bacteria: The role of rain. Phytopathology, 80, 934–937. Cooper, W. A., 1974: A possible mechanism for contact nucleation. J. Atmos. Sci., 31, 1832–1837. Cotton, W. R., G. J. Tripoli, R. M. Rauber, and E. A. Mulvihill, 1986: Numerical simulation of the effects of varying ice crystal nucleation rates and aggregation processes on orographic snowfall. J. Climate Appl. Meteor., 25, 1658–1680. Cozic, J., and Coauthors, 2008: Chemical composition of free tropospheric aerosol for PM1 and coarse mode at the high alpine site Jungfraujoch. Atmos. Chem. Phys., 8, 407–423. Croft, B., and Coauthors, 2010: Influences of in-cloud aerosol scavenging parameterizations on aerosol concentrations and wet deposition in ECHAM5-HAM. Atmos. Chem. Phys., 10, 1511–1543. Cziczo, D. J., D. M. Murphy, P. K. Hudson, and D. S. Thomson, 2004: Single particle measurements of the chemical composition of cirrus ice residue during CRYSTAL-FACE. J. Geophys. Res., 109, D04201, doi:10.1029/2003JD004032. ——, K. D. Froyd, S. J. Gallavardin, O. Moehler, S. Benz, H. Saathoff, and D. M. Murphy, 2009a: Deactivation of ice nuclei due to atmospherically relevant surface coatings. Environ. Res. Lett., 4, 044013, doi:10.1088/1748-9326/4/4/044013. ——, and Coauthors, 2009b: Inadvertent climate modification due to anthropogenic lead. Nat. Geosci., 2, 333–336, doi:10.1038/ ngeo499. DeMott, P. J., 1990: An exploratory study of ice nucleation by soot aerosols. J. Appl. Meteor., 29, 1072–1079. ——, 1995: Quantitative descriptions of ice formation mechanisms of silver iodide-type aerosols. Atmos. Res., 38, 63–99. ——, D. J. Cziczo, A. J. Prenni, D. M. Murphy, S. M. Kreidenweis, D. S. Thomson, R. Borys, and D. C. Rogers, 2003a: Measurements of the concentration and composition of nuclei for cirrus formation. Proc. Natl. Acad. Sci. USA, 100, 14 655–14 660. ——, K. Sassen, M. R. Poellot, D. Baumgardner, D. C. Rogers, S. D. Brooks, A. J. Prenni, and S. M. Kreidenweis, 2003b: African dust aerosols as atmospheric ice nuclei. Geophys. Res. Lett., 30, 1732, doi:10.1029/2003GL017410. ——, A. J. Prenni, M. S. Richardson, S. M. Kreidenweis, C. H. Twohy, and D. C. Rogers, 2006: Ice nuclei variability, relation to ambient aerosol properties, and impacts on mixed-phase clouds. Preprints, 12th Conf. on Cloud Physics, Madison, WI, Amer. Meteor. Soc., 2.1 [Available online at http://ams.confex.com/ams/ Madison2006/techprogram/paper_113242.htm.] Dentener, F., and Coauthors, 2006: Emissions of primary aerosol and precursor gases in the years 2000 and 1750 prescribed data-sets for AeroCom. Atmos. Chem. Phys., 6, 4321–4344. Diehl, K., and S. K. Mitra, 1998: A laboratory study of the effects of a kerosene-burner exhaust on ice nucleation and the evaporation rate of ice crystals. Atmos. Environ., 32, 3145–3151. ——, and S. Wurzler, 2004: Heterogeneous drop freezing in the immersion mode: Model calculations considering soluble and insoluble particles in the drops. J. Atmos. Sci., 61, 2063–2072. ——, S. Matthias-Maser, S. K. Mitra, and R. Jaenicke, 2002: The ice nucleating ability of pollen. Part II: Laboratory studies in immersion and contact freezing modes. Atmos. Res., 61, 125–133. Durant, A. J., and R. A. Shaw, 2005: Evaporation freezing by contact nucleation inside-out. Geophys. Res. Lett., 32, L20814, doi:10.1029/2005GL024175. ——, ——, W. I. Rose, Y. Mi, and G. G. J. Ernst, 2008: Ice nucleation and overseeding of ice in volcanic clouds. J. Geophys. Res., 113, D09206, doi:10.1029/2007JD009064. 2501 Dymarska, M., B. J. Murray, L. Sun, M. L. Eastwood, D. A. Knopf, and A. K. Bertram, 2006: Deposition ice nucleation on soot at temperatures relevant for the lower troposphere. J. Geophys. Res., 111, D04204, doi:10.1029/2005JD006627. Eastwood, M. L., S. Cremel, C. Gehrke, E. Girard, and A. K. Bertram, 2008: Ice nucleation on mineral dust particles: Onset conditions, nucleation rates and contact angles. J. Geophys. Res., 113, D22203, doi:10.1029/2008JD010639. ——, ——, M. Wheeler, B. J. Murray, E. Girard, and A. K. Bertram, 2009: Effects of sulfuric acid and ammonium sulfate coatings on the ice nucleation properties of kaolinite particles. Geophys. Res. Lett., 36, L02811, doi:10.1029/2008GL035997. Eidhammer, T., P. J. DeMott, and S. M. Kreidenweis, 2009: A comparison of heterogeneous ice nucleation parameterizations using a parcel model framework. J. Geophys. Res., 114, D06202, doi:10.1029/2008JD011095. Elbert, W., P. E. Taylor, M. O. Andreae, and U. Pöschl, 2007: Contribution of fungi to primary biogenic aerosols in the atmosphere: Wet and dry discharged spores, carbohydrates, and inorganic ions. Atmos. Chem. Phys., 7, 4569–4588. Field, P. R., O. Möhler, P. Connolly, M. Krämer, R. Cotton, A. J. Heymsfield, H. Saathoff, and M. Schnaiter, 2006: Some ice nucleation characteristics of Asian and Saharan desert dust. Atmos. Chem. Phys., 6, 2991–3006. Fletcher, N. H., 1962: Physics of Rain Clouds. Cambridge University Press, 386 pp. Fornea, A. P., S. D. Brooks, J. B. Dooley, and A. Saha, 2009: Heterogeneous freezing of ice on atmospheric aerosols containing ash, soot, and soil. J. Geophys. Res., 114, D13201, doi:10.1029/2009JD011958. Fukuta, N., and R. C. Schaller, 1982: Ice nucleation by aerosol particles: Theory of condensation-freezing nucleation. J. Atmos. Sci., 39, 648–655. Georgii, H. W., and E. Kleinjung, 1967: Relations between the chemical composition of atmospheric aerosol particles and the concentration of natural ice nuclei. J. Rech. Atmos., 3, 145–156. Gorbunov, B., A. Baklanov, N. Kakutkina, H. L. Windsor, and R. Toumi, 2001: Ice nucleation on soot particles. J. Aerosol Sci., 32, 199–215. Heald, C. L., and D. V. Spracklen, 2009: Atmospheric budget of primary biological aerosol particles from fungal spores. Geophys. Res. Lett., 36, L09806, doi:10.1029/2009GL037493. Henderson-Begg, S. K., T. Hill, R. Thyrhaug, M. Khan, and B. F. Moffett, 2009: Terrestrial and airborne non-bacterial ice nuclei. Atmos. Sci. Lett., 10, 215–219, doi:10.1002/asl.241. Hirano, S. S., and C. D. Upper, 1995: Ecology of ice nucleationactive bacteria. Biological Ice Nucleation and Its Applications, R. E. Lee, G. J. Warren, and L. V. Gusta, Eds., The American Phytopathological Society, 41–61. Hoose, C., U. Lohmann, R. Erdin, and I. Tegen, 2008: The global influence of dust mineralogical composition on heterogeneous ice nucleation in mixed-phase clouds. Environ. Res. Lett., 3, 025003, doi:10.1088/1748-9326/3/2/025003. ——, J. E. Kristjánsson, T. Iversen, A. Kirkevåg, Ø. Seland, and A. Gettelman, 2009: Constraining cloud droplet number concentration in GCMs suppresses the aerosol indirect effect. Geophys. Res. Lett., 36, L12807, doi:10.1029/2009GL038568. ——, ——, and S. M. Burrows, 2010: How important is biological ice nucleation in clouds on a global scale? Environ. Res. Lett., 5, 024009, doi:10.1088/1748-9326/5/2/024009. Jacobson, M. Z., and D. G. Streets, 2009: Influence of future anthropogenic emissions on climate, natural emissions, and air quality. J. Geophys. Res., 114, D08118, doi:10.1029/2008JD011476. 2502 JOURNAL OF THE ATMOSPHERIC SCIENCES Kanji, Z. A., and J. P. D. Abbatt, 2006: Laboratory studies of ice formation via deposition mode nucleation onto mineral dust and n-hexane soot samples. J. Geophys. Res., 111, D16204, doi:10.1029/2005JD006766. ——, and ——, 2010: Ice nucleation onto Arizona test dust at cirrus temperatures: Effect of temperature and aerosol size on onset relative humidity. J. Phys. Chem., 114A, 935–941. Khvorostyanov, V. I., and J. A. Curry, 2005: The theory of ice nucleation by heterogeneous freezing of deliquescent mixed CCN. Part II: Parcel model simulation. J. Atmos. Sci., 62, 261–285. Kieft, T. L., 1988: Ice nucleation activity in lichens. Appl. Environ. Microbiol., 54, 1678–1681. Knopf, D. A., and T. Koop, 2006: Heterogeneous nucleation of ice on surrogates of mineral dust. J. Geophys. Res., 111, D12201, doi:10.1029/2005JD006894. Koch, D., and Coauthors, 2009: Evaluation of black carbon estimations in global aerosol models. Atmos. Chem. Phys., 9, 9001–9026. Korolev, A., and G. A. Isaac, 2006: Relative humidity in liquid, mixed-phase, and ice clouds. J. Atmos. Sci., 63, 2865–2880. Kulkarni, G., and S. Dobbie, 2010: Ice nucleation properties of mineral dust particles: Determination of onset RHi, IN active fraction, nucleation time-lag, and the effect of active sites on contact angles. Atmos. Chem. Phys., 10, 95–105. Kumai, M., 1961: Snow crystals and the identification of the nuclei in the northern United States of America. J. Atmos. Sci., 18, 139–150. ——, and K. E. Francis, 1962: Nuclei in snow and ice crystals on the Greenland ice cap under natural and artificially stimulated conditions. J. Atmos. Sci., 19, 474–481. Lappalainen, S., M. Nikulin, S. Berg, P. Parikka, E.-L. Hintikka, and A.-L. Pasanen, 1996: Fusarium toxins and fungi associated with handling of grain on eight Finnish farms. Atmos. Environ., 30, 3059–3065. Levin, Z., and S. A. Yankofsky, 1983: Contact versus immersion freezing of freely suspended droplets by bacterial ice nuclei. J. Climate Appl. Meteor., 22, 1964–1966. Levine, J., 1950: Statistical explanation of spontaneous freezing of water droplets. Tech. Rep. NACA-TN-2234, National Advisory Committee for Aeronautics, 27 pp. [Available online at http:// naca.central.cranfield.ac.uk/reports/1950/naca-tn-2234.pdf.] Lindemann, J., H. A. Constantinidou, W. R. Barchet, and C. D. Upper, 1982: Plants as sources of airborne bacteria, including ice nucleation-active bacteria. Appl. Environ. Microbiol., 44, 1059–1063. Liu, X., and J. E. Penner, 2005: Ice nucleation parameterization for global models. Meteor. Z., 14, 499–514. Lohmann, U., 2002: Possible aerosol effects on ice clouds via contact nucleation. J. Atmos. Sci., 59, 647–656. ——, and K. Diehl, 2006: Sensitivity studies of the importance of dust ice nuclei for the indirect aerosol effect on stratiform mixed-phase clouds. J. Atmos. Sci., 63, 968–982. ——, and C. Hoose, 2009: Sensitivity studies of different aerosol indirect effects in mixed-phase clouds. Atmos. Chem. Phys., 9, 8917–8934. ——, P. Stier, C. Hoose, S. Ferrachat, S. Kloster, E. Roeckner, and J. Zhang, 2007: Cloud microphysics and aerosol indirect effects in the global climate model ECHAM5-HAM. Atmos. Chem. Phys., 7, 3425–3446. Luond, F., O. Stetzer, A. Welti, and U. Lohmann, 2010: Experimental study on the ice nucleation ability of size selected kaolinite particles in the immersion mode. J. Geophys. Res., 115, D14201, doi:10.1029/2009JD012959. VOLUME 67 Maki, L. R., and K. J. Willoughby, 1978: Bacteria as biogenic sources of freezing nuclei. J. Appl. Meteor., 17, 1049–1053. Marcolli, C., S. Gedamke, T. Peter, and B. Zobrist, 2007: Efficiency of immersion mode ice nucleation on surrogates of mineral dust. Atmos. Chem. Phys., 7, 5081–5091. Möhler, O., and Coauthors, 2005: Effect of sulfuric acid coating on heterogeneous ice nucleation by soot aerosol particles. J. Geophys. Res., 110, D11210, doi:10.1029/2004JD005169. ——, and Coauthors, 2006: Efficiency of the deposition mode ice nucleation on mineral dust particles. Atmos. Chem. Phys., 6, 3007–3021. ——, and Coauthors, 2008: Heterogeneous ice nucleation activity of bacteria: New laboratory experiments at simulated cloud conditions. Biogeosciences, 5, 1425–1435. Morrison, H., and A. Gettelman, 2008: A new two-moment bulk stratiform cloud microphysics scheme in the Community Atmosphere Model, version 3 (CAM3). Part I: Description and numerical tests. J. Climate, 21, 3642–3659. ——, M. D. Shupe, J. O. Pinto, and J. A. Curry, 2005: Possible roles of ice nucleation mode and ice nuclei depletion in the extended lifetime of Arctic mixed-phase clouds. Geophys. Res. Lett., 32, L18801, doi:10.1029/2005GL023614. Murphy, D. M., and T. Koop, 2005: Review of the vapour pressures of ice and supercooled water for atmospheric applications. Quart. J. Roy. Meteor. Soc., 131, 1539–1565. Petters, M. D., and Coauthors, 2009: Ice nuclei emissions from biomass burning. J. Geophys. Res., 114, D07209, doi:10.1029/ 2008JD011532. Phillips, V. T. J., P. J. DeMott, and C. Andronache, 2008: An empirical parameterization of heterogeneous ice nucleation for multiple chemical species of aerosol. J. Atmos. Sci., 65, 2757–2783. ——, and Coauthors, 2009: Potential impacts from biological aerosols on ensembles of continental clouds simulated numerically. Biogeosciences, 6, 987–1014. Pitter, R. L., and H. R. Pruppacher, 1973: A wind tunnel investigation of freezing of small water drops falling at terminal velocity in air. Quart. J. Roy. Meteor. Soc., 99, 540–550. Pouleur, S., C. Richard, J.-G. Martin, and H. Antoun, 1992: Ice nucleation activity in Fusarium acuminatum and Fusarium avenaceum. Appl. Environ. Microbiol., 58, 2960–2964. Pratt, K. A., and Coauthors, 2009: In situ detection of biological particles in cloud ice-crystals. Nat. Geosci., 2, 398–401. Prenni, A. J., and Coauthors, 2007: Can ice-nucleating aerosols affect Arctic seasonal climate? Bull. Amer. Meteor. Soc., 88, 541–550. ——, P. J. DeMott, D. C. Rogers, S. M. Kreidenweis, G. M. McFarquhar, G. Zhang, and M. R. Poellot, 2009a: Ice nuclei characteristics from M-PACE and their relation to ice formation in clouds. Tellus, 61B, 436–448, doi:10.1111/j.16000889.2009.00415.x. ——, and Coauthors, 2009b: Relative roles of biogenic emissions and Saharan dust as ice nuclei in the Amazon basin. Nat. Geosci., 2, 402–405. Pruppacher, H. R., and J. D. Klett, 1997: Microphysics of Clouds and Precipitation. Kluwer Academic, 954 pp. Richardson, M. S., and Coauthors, 2007: Measurements of heterogeneous ice nuclei in the western United States in springtime and their relation to aerosol characteristics. J. Geophys. Res., 112, D02209, doi:10.1029/2006JD007500. Rogers, D. C., and P. J. DeMott, 1995: Measurements of natural ice nuclei, CCN, and CN in winter clouds. Preprints, Conf. on Cloud Physics, Dallas, TX, Amer. Meteor. Soc. 129–144. AUGUST 2010 HOOSE ET AL. ——, ——, S. M. Kreidenweis, and Y. Chen, 2001: A continuousflow diffusion chamber for airborne measurements of ice nuclei. J. Atmos. Oceanic Technol., 18, 725–741. Sassen, K., and V. I. Khvorostyanov, 2008: Cloud effects from boreal forest fire smoke: Evidence for ice nucleation from polarization lidar data and cloud model simulations. Environ. Res. Lett., 3, 025006, doi:10.1088/1748-9326/3/2/025006. Schaller, R., and N. Fukuta, 1979: Ice nucleation by aerosolparticles–experimental studies using a wedge-shaped ice thermal-diffusion chamber. J. Atmos. Sci., 36, 1788–1802. Schnell, R. C., and G. Vali, 1973: World-wide source of leaf-derived freezing nuclei. Nature, 246, 212–213. Seland, Ø., T. Iversen, A. Kirkevåg, and T. Storelvmo, 2008: Aerosol–climate interactions in the CAM-Oslo atmospheric GCM and investigation of associated basic shortcomings. Tellus, 60A, 459–491. Shilling, J. E., T. J. Fortin, and M. A. Tolbert, 2006: Depositional ice nucleation on crystalline organic and inorganic solids. J. Geophys. Res., 111, D12204, doi:10.1029/2005JD006664. Stetzer, O., B. Baschek, F. Lüönd, and U. Lohmann, 2008: The Zurich Ice Nucleation Chamber (ZINC)—A new instrument to investigate atmospheric ice formation. Aerosol Sci. Technol., 42, 64–74. Storelvmo, T., J. E. Kristjánsson, S. J. Ghan, A. Kirkevåg, Ø. Seland, and T. Iversen, 2006: Predicting cloud droplet number concentration in Community Atmosphere Model (CAM)-Oslo. J. Geophys. Res., 111, D24208, doi:10.1029/2005JD006300. ——, ——, and U. Lohmann, 2008a: Aerosol influence on mixedphase clouds in CAM-Oslo. J. Atmos. Sci., 65, 3214–3230. ——, ——, ——, T. Iversen, A. Kirkevåg, and Ø. Seland, 2008b: Modeling of the Wegener–Bergeron–Findeisen process— Implications for aerosol indirect effects. Environ. Res. Lett., 3, 045001, doi:10.1088/1748-9326/3/4/045001; Corrigendum, 5, 019801, doi:10.1088/1748-9326/5/1/019801. Svensson, E. A., C. Delval, P. von Hessberg, M. S. Johnson, and J. B. C. Pettersson, 2009: Freezing of water droplets colliding with kaolinite particles. Atmos. Chem. Phys., 9, 4295– 4300. 2503 Szyrmer, W., and I. Zawadzki, 1997: Biogenic and anthropogenic sources of ice-forming nuclei: A review. Bull. Amer. Meteor. Soc., 78, 209–228. Targino, A. C., R. Krejci, K. J. Noone, and P. Glantz, 2006: Single particle analysis of ice crystal residuals observed in orographic wave clouds over Scandinavia during INTACC experiment. Atmos. Chem. Phys., 6, 1977–1990. Textor, C., and Coauthors, 2006: Analysis and quantification of the diversities of aerosol life cycles within AeroCom. Atmos. Chem. Phys., 6, 1777–1813. Vali, G., 1985: Nucleation terminology. J. Aerosol Sci., 16, 575–576. ——, 1994: Freezing rate due to heterogeneous nucleation. J. Atmos. Sci., 51, 1843–1856. ——, M. Christensen, R. W. Fresh, E. L. Galyan, L. R. Maki, and R. C. Schnell, 1976: Biogenic ice nuclei. Part II: Bacterial sources. J. Atmos. Sci., 33, 1565–1570. Vogel, H., A. Pauling, and B. Vogel, 2008: Numerical simulation of birch pollen dispersion with an operational weather forecast system. Int. J. Biometeor., 52, 805–814. von Blohn, N., S. K. Mitra, K. Diehl, and S. Borrmann, 2005: The ice nucleating ability of pollen. Part III: New laboratory studies in immersion and contact freezing modes including more pollen types. Atmos. Res., 78, 182–189. Welti, A., F. Lüönd, O. Stetzer, and U. Lohmann, 2009: Influence of particle size on the ice nucleating ability of mineral dusts. Atmos. Chem. Phys., 9, 6705–6715. Wiacek, A., and T. Peter, 2009: On the availability of uncoated mineral dust ice nuclei in cold cloud regions. Geophys. Res. Lett., 36, L17801, doi:10.1029/2009GL039429. Yankofsky, S. A., Z. Levin, T. Bertold, and N. Sandlerman, 1981: Some basic characteristics of bacterial freezing nuclei. J. Appl. Meteor., 20, 1013–1019. Young, K. C., 1974: The role of contact nucleation in ice phase initiation in clouds. J. Atmos. Sci., 31, 768–776. Zimmermann, F., S. Weinbruch, L. Schütz, H. Hofmann, M. Ebert, K. Kandler, and A. Worringen, 2008: Ice nucleation properties of the most abundant mineral dust phases. J. Geophys. Res., 113, D23204, doi:10.1029/2008JD010655.