Investigating the simulation of cloud microphysical processes in numerical models using a one-dimensional dynamical framework

Transcription

1 ATMOSPHERIC SCIENCE LETTERS Atmos. Sci. Let. 10: (2009) Published online in Wiley InterScience ( Investigating the simulation of cloud microphysical processes in numerical models using a one-dimensional dynamical framework Christopher Dearden* School of Earth, Atmospheric and Environmental Science, University of Manchester, Manchester, UK *Correspondence to: Christopher Dearden, School of Earth, Atmospheric and Environmental Science, University of Manchester, Room 3.16, Simon Building, Oxford Road, Manchester M13 9PL, UK. christopher.dearden@ postgrad.manchester.ac.uk Received: 25 January 2009 Revised: 7 July 2009 Accepted: 20 August 2009 Abstract This paper describes the method by which the performance of a suite of microphysics schemes of varying levels of complexity can be compared within an idealised framework. The purpose is to establish the level of microphysical sophistication required for the successful simulation of liquid clouds in operational models, paying particular attention to the required level of coupling with aerosols. Initial results from a lagrangian parcel model are able to demonstrate the importance of the treatment of droplet activation in dual moment schemes for predicting droplet number qualitatively. Subsequent testing within a one-dimensional (1D) column model using the existing factorial method (FM) will aim to quantify the importance of microphysical complexity on precipitation and cloud albedo relative to the effects of meteorology. Copyright 2009 Royal Meteorological Society Keywords: clouds; microphysics; parcel model 1. Introduction The level of microphysical complexity implemented within a numerical weather prediction (NWP) model or climate model is an important consideration because model complexity has to be balanced alongside cost implications, ultimately determined by computational limitations. It is important to capture liquid phase processes accurately not just because they determine the structure of warm clouds but also because liquid droplets can exist at temperatures below 0 Cinthe form of supercooled water, which plays a role in ice formation through homogeneous and heterogeneous freezing mechanisms (Cantrell and Heymsfield, 2005). Consequently, an accurate representation of the liquid phase should be achieved first, as a precursor to an eventual improved treatment of the ice phase. Because aerosols in the atmosphere are known to reduce the size of the energy barrier required for the phase transition of water, a study of liquid phase microphysics is incomplete without consideration of the role of cloud condensation nuclei (CCN). The effect of CCN can lead to the modification of cloud properties on the large scale: the albedo of warm clouds is enhanced in the presence of CCN by virtue of an increase in the surface area of droplets (Twomey, 1959). Such interactions, known as aerosol indirect effects, can have important consequences for the net cloud radiative forcing and hence climate. Aerosol indirect effects may also be important for weather, as they can prolong the lifetime of a cloud by increasing the liquid water content due to a suppression in the drizzle rate (Albrecht, 1989). However, observational evidence is not always consistent with this (Coakley and Walsh, 2002), suggesting that the specifics of the meteorology may be important. This makes it difficult to isolate the aerosol effects and identify those that may be worthy of parameterization. In recent years, considerable advances in understanding have been made in accounting for CCN properties on droplet activation (McFiggans et al., 2006), and detailed process models have been designed to investigate the remaining uncertainty surrounding the role of organic compounds (Topping et al., 2005a, 2005b). Such process models can be used to inform and constrain the development of microphysics schemes that either describe the bulk properties of hydrometeors ( bulk schemes ) or resolve the hydrometeor size distribution explicitly by breaking it down into discrete size bins ( bin schemes ). However, the performance of these schemes still needs to be considered under a range of thermodynamic environments to establish whether the additional microphysical complexity can consistently add any skill to the model. Bulk schemes work by assuming some form of the particle size distribution based on aircraft observations. Typically, the distribution is based on a gamma function, which is in turn a function of particle diameter, D: n x (D) = n x0 D α x exp( λ x D) (1) where the class of hydrometeor is denoted by x, n x is the number distribution, n x0 is the intercept parameter, α x is the shape parameter [which when set to zero, reduces Equation (1) to an exponential distribution] and λ x is the slope parameter. The Copyright 2009 Royal Meteorological Society

2 208 C. Dearden size mass relationship is specified as: M x (D) = c x D d x (2) where M x is the particle mass and c x and d x are constants. The moments M of the particle size distribution are given by: M k = D k n x (D)dD (3) where k can take any real value and denotes the k th moment of the size distribution. The number concentration corresponds to k = 0, and mass mixing ratio to k = d x, where d x for a spherical hydrometeor is equal to three. Thus, the moments return useful information regarding bulk cloud properties. Single moment schemes solve for the mixing ratio, related to the third moment, whereas dual moment schemes predict number concentration as well and so are doubly expensive because they must hold twice the number of prognostic variables. By predicting number as well as mass, dual moment treatments of cloud water must include a rate equation describing the activation of droplets and thus have the ability to parameterize the effects of aerosol. This study will attempt to quantify this advantage and decide whether the application of a dual moment bulk scheme can be warranted for the simulation of the liquid phase. 2. Methodology The methodology adopted for this work is based on testing the performance of a suite of microphysics schemes of increasing levels of sophistication within a common 1D framework. Such a framework enables control over atmospheric thermodynamic variables, while allowing the performance of the schemes to be analysed in isolation from systematic errors that can arise from elsewhere. Because CCN lead to the formation of liquid droplets in the atmosphere, specific consideration will be given to establishing the sensitivity of cloud properties to changes in aerosol properties such as concentration and composition. This will be achieved through the factorial method (FM), which involves systematically exploring the phase space of thermodynamic and microphysical variables to assess the impact on precipitation reaching the ground. This will enable an assessment to be made concerning the importance of aerosol effects in the context of changing meteorological conditions, and therefore whether any robust conclusions concerning cloud aerosol interactions can be reached. Equally as importantly, the minimum requirements needed to simulate these aerosol effects in numerical models will also be ascertained. The FM is described in Teller and Levin (2008), where its application is demonstrated using the Tel Aviv University 2D (TAU 2D) cloud resolving model to investigate the relative contributions of a number of factors to changes in surface precipitation from mixed phase convective cloud. For brevity, the key aspects of the method are now described. Factors are chosen to reflect those variables whose effects require evaluation, for example, CCN concentration, or initial temperature profile. If k factors are considered at two levels (corresponding to a high value and a low value), this would give a 2 k factorial design. In the case of three factors, labelled A, B and C respectively, eight simulations would be needed. Each simulation is labelled according to the level of the factors used, such that a high value of the factors A, B and C is denoted by a lowercase letter a, b and c respectively, and the low value is denoted by the absence of the corresponding letter. The case when all three factors are considered at their low levels is denoted as (1). Thus, the eight treatment combinations in standard order can be written as (1), a, b, ab, c, ac, bc and abc. From this it can be seen that three degrees of freedom are associated with the main effects of A, B and C, and four degrees of freedom are associated with the interactions between AB, AC, BC and ABC. The main effect of A can be obtained from the average of the four treatment combinations where A is at the high level, minus the average of the four treatment combinations where A is at the low level. In standard notation, this can be written as: A = (a + ab + ac + abc) [(1) + b + c + bc] = 1 [a + ab + ac + abc (1) b c bc] () The effects of B and C are obtained in a similar manner, yielding: B = 1 [b + ab + bc + abc (1) a c ac] (5) C = 1 [c + ac + bc + abc (1) a b ab] (6) The effects of the two-factor interactions (namely AB, AC and BC ) are computed thus. The AB interaction can be thought of as one-half of the difference between the average A effects at the two levels of B: AB = 1 [ ] (average A effect at high B value) 2 (average A effect at low B value) AB = 1 [abc bc + ab b ac + c a + (1)] (7) Following similar logic, the AC and BC interactions are given by: AC = 1 [(1) a + b ab c + ac bc + abc] (8) BC = 1 [(1) + a b ab c ac + bc + abc] (9) Copyright 2009 Royal Meteorological Society Atmos. Sci. Let. 10: (2009)

3 Simulation of cloud microphysical processes 209 The remaining effect, the interaction of all three factors (ABC ), is defined as the average difference between the AB interaction for the two levels of C : ABC = 1 [abc bc ac + c ab + b + a (1)] (10) The terms in square brackets in the above equations are known as the contrasts in the treatment combinations. The contrasts are useful to evaluate as they are required to calculate the sums of squares for each of the effects. In general, the sum of squares for any effect is calculated by: SS = 1 2 k (Contrast)2 (11) The relative contribution of each effect to the total variance can be quantified in terms of a percentage of the total sum of squares, where the total is given by: SS T = SS A + SS B + SS C + SS AB + SS AC + SS BC + SS ABC (12) The reader is referred to Montgomery (2005) for the general case where more than three factors need to be evaluated. The factors chosen for this particular study are: (1) Updraught speed (to explore the effects on the number of droplets activated at cloud base) (2) Initial temperature profile (to explore the effect of cloud base temperature on droplet activation) (3) Aerosol composition (to explore the impact of increases in organic material in both internal and external mixing states) () Aerosol size (to establish the impact of increasingly larger CCN) (5) Aerosol concentration (to investigate the relationship between increases in aerosol number and the rate of autoconversion) A 1D column model (described in Section 3.3) with prescribed forcing and a fixed vertical resolution will be used to drive a suite of microphysics schemes to investigate the impact of each of the above factors on surface precipitation. The impact on the cloud albedo will also be considered in terms of the effective radius of droplets, an important quantity for climate models to predict as it determines the cloud radiative forcing. The importance of the microphysical factors (aerosol composition, size and concentration) relative to the effects of meteorology (updraught speed and temperature profile) will be judged in terms of their role in explaining the total variance. The FM will be applied to both bin and bulk microphysics schemes in the same 1D framework, including a dual moment liquid scheme with the option of prognostic aerosol, and a less computationally demanding single moment treatment of liquid. Results from the bulk schemes will be compared in turn with those from the bin scheme, to establish whether they are capable of producing similar sensitivities. As the complexity of the bulk schemes reduces, so does the potential to represent aerosol cloud interactions. However, this may not be a limiting factor if the effects of aerosol are shown from the bin scheme to be insignificant compared with the effects of meteorology. Ultimately, the tests will identify the level at which aerosols need to be represented within a bulk microphysics scheme for the purpose of simulating liquid phase processes. Once the comparison of the microphysics schemes has been conducted in the 1D framework, the bulk schemes will be engineered into a 3D cloud resolving model and a series of warm and mixed phase case studies will be performed in a mesoscale framework. This will qualitatively demonstrate whether any improvements in model skill can be had by increasing the complexity of the microphysics. Figure 1 shows a flowchart that summarises the methodology through from start to finish. 3. Model description 3.1. The bulk schemes The bulk microphysics will be based on the existing scheme of Morrison et al. (2005). It includes a dual moment representation of cloud liquid droplets, rain, cloud ice, snow and graupel, with assumed size distributions of the form given in Equation (1). Within the Morrison scheme, various options are available to specify the treatment of droplet activation and the representation of aerosol. This built-in flexibility will be exploited to develop a hierarchy of bulk schemes, starting from relatively simple microphysics, with each subsequent scheme benefiting from an incremental increase in complexity. This will allow any improvements in performance between schemes to be easily traceable and attributable. Details of the proposed hierarchy are now presented, along with a discussion of the added functionality each scheme provides relative to its predecessor Single moment liquid water and rain As the Morrison scheme uses a dual moment approach for all hydrometeors, a version of the scheme will be developed that simplifies the treatment of liquid water and rain to single moment; consequently droplet number must be prescribed. Typically in single moment schemes, it is specified as a constant, for example, Thompson et al. (200). Although a single moment scheme cannot account for the effects of aerosol composition or size, the effects of aerosol concentration can be explored by proxy through changes in the prescribed droplet number. Copyright 2009 Royal Meteorological Society Atmos. Sci. Let. 10: (2009)

4 210 C. Dearden Figure 1. Flowchart illustrating the application of the factorial method to assess the importance of cloud aerosol interactions in the context of a changing meteorology with reference to an explicit bin scheme Single moment liquid water, dual moment rain The use of a dual moment scheme for rain will facilitate the simulation of the thermodynamic indirect effect (Lohmann and Feichter, 2005), because the predicted number concentration for rain will have an effect on the collection rate of cloud droplets by falling rain drops. This could be significant for cases of deep convection, where the amount of supercooled water that freezes homogeneously leads to the release of latent heat. The degree of latent heating can result in a more vigorous convective updraft, allowing higher cloud top heights to be achieved. Comparison with the single moment rain scheme will help to decide whether a dual moment treatment of rain is necessary Dual moment liquid water, dual moment rain The transition from prescribed droplet number concentration to a prognostic treatment has consequences for the simulation of the mixed phase, because it provides the opportunity to investigate the possibility of a riming indirect effect, first proposed by Lohmann and Feichter (2005). An explicit treatment of droplet activation is also required. Particular attention will be given to assessing the sensitivity of total precipitation to the handling of cloud droplet activation. The influence of aerosols will initially be accounted for using the activation spectrum relation (Rogers and Yau, 1989), which relates droplet number N to supersaturation ratio, s: N = cs k (13) where c and k are constants that depend on the type of airmass and s = (q v /q sat 1) 100%, where q v is the (supersaturated) vapour mixing ratio and q sat is the saturation mixing ratio. This requires the Morrison scheme to be able to predict supersaturations explicitly. Currently a saturation adjustment method is used, whereby excess water vapour above saturation is instantaneously removed and converted to the liquid phase. Such a method is justified within a mesoscale framework, where model timesteps are typically longer than the growth rate of droplets through condensation of the available water vapour. However, this technique never permits a supersaturation to be maintained, thus rendering it incompatible with Equation (13). Consequently to study droplet activation based on Equation (13) in the 1D framework, the saturation Copyright 2009 Royal Meteorological Society Atmos. Sci. Let. 10: (2009)

5 Simulation of cloud microphysical processes 211 adjustment method must be replaced with a diffusional growth equation which is specified in Pruppacher and Klett (1997): dm dt = (q v/q sat 1) CF (1) ρab where dm/dt is the rate of change of mass due to evaporation/condensation, AB is a function of temperature, C = D/2 assuming spherical particles and F is the ventilation coefficient (in this study, ventilation effects are ignored for cloud droplets). To account for changes in aerosol composition in this scheme, N s relationships for different aerosol types can be constructed off-line using the binresolved microphysics scheme with droplet activation based on Kohler theory. From the N s relations, N w parameterizations for different aerosol types can then be developed in the manner of Twomey (1959), such that the droplet number will depend on updraft speed w rather than the resolved supersaturation Dual moment liquid water and rain, with prognostic treatment of aerosol While the previous schemes attempt to account for the effects of aerosols on some level, they do so noninteractively. The use of a prognostic dual moment scheme for aerosols addresses this limitation, such that the effect of aerosol transport on cloud properties is handled explicitly and so does not need to be parameterized. The dual moment aerosol scheme would hold mass and number concentration as prognostic variables, and provided that the spread was known, the number size distribution of the aerosols present could be represented by a log normal function (Jacobson, 2005). The effects of aerosol size can then be explored by changing the spread of the log normal function. A potential problem arises in that after activation of droplets, the aerosol size distribution will adjust back to a log normal function, which may not be realistic. The issue of whether or not it is sufficient to reduce number and mass of aerosol and recalculate the parameters based on a log-normal relationship will be investigated. As well as internal mixtures, it will be possible to account for external mixtures in this scheme, by simply adding more prognostic variables to reflect different aerosol species The bin scheme The bin scheme used to validate the performance of the bulk schemes is based on the aerosol cloud precipitation interaction model (ACPIM), which is in the final stages of development at the University of Manchester. It is a state of the art microphysical process model that accounts for multi-component aerosol thermodynamic properties in different mixing states, with droplet activation from Kohler theory. The microphysics is initialized assuming a log normal aerosol size distribution, thus requiring inputs for the geometric mean diameter, the spread and the total aerosol number. ACPIM is particularly suited to studying the effects of aerosol composition on cloud because it uses results from the ADDEM model (aerosol diameter-dependent equilibrium model), described in detail in Topping et al. (2005a, 2005b), to obtain the equilibrium behaviour of multi-component aerosols. ADDEM allows for a treatment of the effect of curvature (the Kelvin effect) in determining the hygroscopic properties of mixed inorganic/organic aerosols, and uses the general form of the Kohler equation housed within an iterative loop with a guaranteed convergence scheme to solve for the new equilibrium state. This allows accurate equilibrium vapour pressures to be computed for a wide range of internal and external mixtures, which can then be fed as inputs into ACPIM to study the effects of composition on cloud properties. Although the focus of this paper is on the effects of CCN, the ACPIM also allows the insoluble fraction of aerosol to be specified, thus making it a suitable scheme to study heterogeneous freezing processes (Connolly et al., 2009). This opens up the possibility of a future study to quantify the importance of ice nuclei (IN) effects on mixed phase cloud using the FM The 1D driver model This study employs the kinematic driver model (KiD; Shipway, 2009) to force the microphysics and initiate cloud formation. The KiD model operates in a 1D column with a fixed number of vertical levels and provides an ideal test bed for intercomparison of different microphysics schemes, based on a common advection component. The KiD model comes complete with a suite of test cases, the most simple of which consists of a single warm updraft, sinusoidal in time and constant in height, to advect vapour and hydrometeors in the vertical. Instantaneous cloud-related diagnostics, including hydrometeor mass mixing ratios, number concentration and surface precipitation, can be output at selected time intervals for subsequent analysis. The KiD model does not account for the effects of entrainment into the cloudy updraft; its main purpose is to provide a flexible framework that facilitates the testing and comparison of different microphysical treatments given a prescribed vertical velocity. While the importance of accurate atmospheric dynamics is recognized, it is first necessary to gain a fundamental understanding of the pure microphysical behaviour, and this is most easily achieved under relatively simple dynamical forcing. Subsequent work will then build on this understanding by introducing the effects of entrainment, in full 3D case studies of warm clouds with a cloud resolving model. Prior to inclusion within the KiD model, the microphysics schemes will first be developed and tested within a closed lagrangian parcel model. Such a framework does not allow for the effects of sedimentation to be studied, hence why the KiD model must be used Copyright 2009 Royal Meteorological Society Atmos. Sci. Let. 10: (2009)

6 212 C. Dearden to fulfil the needs of this study. However, a parcel model provides a suitable platform for comparing the different treatments of droplet activation, and how this determines the resulting number of droplets activated both at cloud base and within the cloud itself. The ability of the bulk schemes to capture accurate droplet concentrations will be a key factor in how they perform in a 1D column model; on this subject the parcel model can provide very useful insight. Some preliminary results are now discussed.. Initial parcel model results The parcel model has been used to compare droplet number as predicted by the bulk scheme with that from the bin scheme. In the bulk microphysics scheme, Equation (13) is used for droplet activation. However, parcel model tests show that the resolved treatment of supersaturation requires a timestep of 0.1 s or better for stability. To overcome this problem, activation of droplets can be predicted from the Twomey (1959) approximation, which is based solely on updraft speed and thus is not restricted by the choice of model timestep. Table I demonstrates the validity of the Twomey approximation for predicting cloud base droplet number concentrations. The predicted droplet number is compared under a range of updraft speeds and for activation parameters corresponding to both maritime and continental conditions. In all cases, the parcel model ascent was initialised with a temperature of 300 K, a relative humidity of 95% and a pressure of 1000 mb. The values predicted by the Twomey approximation lie within 10 20% of those predicted from the resolved supersaturation approach. The agreement improves at higher updraft speeds, and at lower values of c. However, while the Twomey approximation is sufficient to capture cloud base droplet number, once in cloud, the processes of condensation and subsequent collision/coalescence become important. This is demonstrated using ACPIM in parcel model mode, as shown in Figure 2. In this run, aerosols are assumed to Table I. Cloud base droplet number concentration (Nc), as a function of updraft speed for (top) maritime conditions (c = 120/cc and k = 0.) and (bottom) continental conditions (c = 1000/cc and k = 0.5). Cloud base Nc (Twomey approximation) Cloud base Nc (resolved supersaturation) Cloud base Nc (Twomey approximation) Cloud base Nc (resolved supersaturation) 10 m/s 3 m/s 1 m/s 0.1 m/s 120/cc 120/cc 96/cc 5/cc 120/cc 115/cc 85/cc 7/cc 10 m/s 3 m/s 1 m/s 0.1 m/s 1000/cc 719/cc 517/cc 259/cc 900/cc 623/cc 29/cc 20/cc Figure 2. A parcel expansion from the ACPIM model, showing aerosol number concentration versus temperature. consist solely of ammonium sulphate, with a geometric mean diameter of 60.e 9m, a spread of 0.1 and a total aerosol number of 100.e+6. A fixed updraft speed of 1 m/s was prescribed, starting from an initial temperature of 300 K, a relative humidity of 95% and a pressure of 1000 mb. It can be seen how aerosol number reduces by an amount between 20 25% due to activation of droplets at cloud base, above which no further activation occurs because the supersaturation in the parcel reduces as the existing droplets grow by condensation. A similar parcel ascent was replicated with bulk microphysics, for the same initial thermodynamic conditions, and with the activation parameters c = 100 and k = 0.. Figure 3 shows how the Twomey approximation applied to in-cloud activation of droplets (blue line) would lead to a considerable overestimate of droplet number. When the supersaturation is resolved explicitly in the bulk scheme (Figure 3; red line), droplet number above cloud base stays roughly constant initially and then rapidly reduces as the droplets grow big enough for self-collection and accretion processes to become important. However, autoconversion of droplets to rain acts to reduce droplet mass and number, which in turn reduces the sink of water vapour (Figure a), allowing the supersaturation within the parcel to slowly increase again (Figure b). Because the activity spectrum relation (Equation 13) does not account for the depletion of CCN, anomalous new droplets are activated within the supersaturated environment. Figure 3 illustrates that the resolved supersaturation approach is clearly more desirable than the Twomey approach for droplet activation because it accounts for condensation/deposition as a sink of water vapour and thus allows for a more realistic treatment of droplet activation when existing condensate is present (i.e. within cloud). However, it requires a very short timestep for stablility, rendering it impractical for Copyright 2009 Royal Meteorological Society Atmos. Sci. Let. 10: (2009)

7 Simulation of cloud microphysical processes 213 Figure 3. Parcel expansion with bulk microphysics, showing droplet number concentration versus temperature. The red line corresponds to a resolved treatment of supersaturation for droplet activation; the blue line represents the predicted droplet number when the Twomey approximation is used (i.e. based solely on updraft speed). use in operational models. A compromise can be struck by restricting the use of the Twomey approach to cloud base activation only. When in cloud, it is possible to approximate the result achieved with the resolved supersaturation approach without the restriction of very small timesteps; such an approach involves consideration of the prognostic equation for supersaturation, δ (Morrison et al., 2005), which in the absence of the ice phase, is given by: [ dδ 1 dt = dq sat dt τ c + 1 τ r ] δ [( T t ) RAD gw ] c p (15) where δ = q v q sat, T is temperature, τ c and τ r are the phase relaxation timescales associated with cloud droplets and rain respectively, ( T / t) RAD is the radiative heating rate (which is zero in the parcel model), g is the gravitational acceleration, w is the vertical velocity and c p is the specific heat of air at constant pressure. Solving Equation (15) for dδ/dt = 0 yields a diagnostic relationship for the maximum supersaturation available for droplet activation within the current timestep, which is given by: δ = dq sat dt [( ) T t RAD gw ][ ] 1 (16) c p τ c τ r Equation (16) can be used for in-cloud activation of new droplets, by expressing δ as a percentage supersaturation ratio and inserting into Equation (13). Then if the potential number of newly activated droplets predicted by Equation (13) is less than the number that already exists, no additional droplets are activated within the timestep. To prevent unrealistically large droplet numbers when the phase relaxation timescale is long, the potential number of newly activated droplets is not allowed to exceed that obtained by the Twomey approach. The result is shown in Figure 5, which compares predicted droplet number with resolved supersaturation in red with the diagnosed equilibrium supersaturation for in-cloud activation given by Equation (16) in blue. Qualitatively the agreement is much better than in Figure 3, with the added advantage that the equilibrium supersaturation method is not limited to small timesteps. 5. Summary The FM of Teller and Levin (2008) is to be employed within the 1D kinematic driver model with state-ofthe-art bin microphysics and aerosol. The use of the 1D model is intended to aid in the identification of the key processes at work, providing isolation from non-linear interactions and systematic bias that may Figure. Plots from the bulk parcel model with resolved supersaturation. (a) (left) The first moment of the liquid droplet size distribution [= Nc(α c + 1)/λ c, where Nc is the droplet number concentration], plotted as a function of temperature of the rising parcel. The first moment is proportional to the sink of water vapour. (b) (right): Saturation ratio versus temperature. Copyright 2009 Royal Meteorological Society Atmos. Sci. Let. 10: (2009)

8 21 C. Dearden Acknowledgements The author would like to thank Dr. Paul Connolly, University of Manchester, for supplying the ACPIM model, and Dr. Ben Shipway, UK Met Office, for supplying the KiD model. References Figure 5. Parcel model output comparing predicted droplet number with resolved treatment of supersaturation for droplet activation (red) with the diagnosed equilibrium supersaturation method for activation (blue). arise from other aspects of a model. The method will seek to separate the significance of aerosol effects on clouds from the effects of meteorology, and thus dictate the minimum level of complexity that needs to be incorporated into a bulk microphysics scheme to successfully simulate liquid clouds. Where possible, the potential for tuning aspects of a single moment liquid scheme to a bin scheme will be explored, as a cost-effective alternative to the more expensive dual moment options. Initial results from a parcel model provide useful insight into the effects of changing the treatment of droplet activation in the bulk scheme and show a potential weakness in schemes that do not include an explicit representation of aerosol. To investigate the impacts on surface precipitation, the microphysics schemes must be engineered into the KiD model where sedimentation between vertical levels is permitted. Thus, future work will focus on applying the FM within the KiD framework for a range of updraft speeds and temperatures, to help quantify the benefits of increasing microphysical complexity based on the metrics of surface precipitation and effective radius. Particular attention will be given to establishing the benefits of a prognostic aerosol scheme in bulk schemes. The importance of entrainment will be addressed by performing 3D case studies of warm clouds, to see if the conclusions from the FM/1D experiments are upheld given more realistic dynamical forcing. Albrecht BA Aerosols, cloud microphysics, and fractional cloudiness. Science 25(923): Cantrell W, Heymsfield A Production of ice in tropospheric clouds A review. Bulletin of the American Meteorological Society 86(6): Coakley JA, Walsh CD Limits to the aerosol indirect radiative effect derived from observations of ship tracks. Journal of the Atmospheric Sciences 59: Connolly PJ, Möhler O, Field PR, Saathoff H, Burgess R, Choularton T, Gallagher M Studies of heterogeneous freezing by three different desert dust samples. Atmospheric Chemistry and Physics 9: Jacobson MK Fundamentals of Atmospheric Modelling, 2nd edn. Cambridge University Press: Cambridge. Lohmann U, Feichter J Global indirect aerosol effects: a review. Atmospheric Chemistry and Physics 5: McFiggans G, Artaxo P, Baltensperger U, Coe H, Facchini MC, Feingold G, Fuzzi S, Gysel M, Laaksonen A, Lohmann U, Mentel TF, Murphy DM, O Dowd CD, Snider JR, Weingartner E The effect of physical and chemical aerosol properties on warm cloud droplet activation. Atmospheric Chemistry and Physics 6: Montgomery DC Design and Analysis of Experiments. John Wiley: New York. Morrison H, Curry JA, Khvorostyanov VI A new doublemoment microphysics parameterization for application in cloud and climate models. Part I: description. Journal of the Atmospheric Sciences 62(6): Pruppacher HR, Klett JD Microphysics of Clouds and Precipitation. Kluwer Academic Publishers: Norwell, USA. Rogers RR, Yau MK A Short Course in Cloud Physics, 3rd edn. Butterworth and Heinemann: Oxford. Shipway BJ Kinematic driver for microphysics intercomparison, UK Met Office. shtml. Teller A, Levin Z Factorial method as a tool for estimating the relative contribution to precipitation of cloud microphysical processes and environmental conditions: method and application. Journal of Geophysical Research-Atmospheres 113(D2) 13 pp. Thompson G, Rasmussen RM, Manning K Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part I: description and sensitivity analysis. Monthly Weather Review 132(2): Topping DO, McFiggans GB, Coe H. 2005a. A curved multicomponent aerosol hygroscopicity model framework: part 1 inorganic compounds. Atmospheric Chemistry and Physics 5: Topping DO, McFiggans GB, Coe H. 2005b. A curved multicomponent aerosol hygroscopicity model framework: part 2 including organic compounds. Atmospheric Chemistry and Physics 5: Twomey SA The nuclei of natural cloud formation. Part II: the supersaturation in natural clouds and the variation of cloud droplet concentrations. Pure and Applied Geophysics 3: Copyright 2009 Royal Meteorological Society Atmos. Sci. Let. 10: (2009)