1. Introduction
Cloud microphysical processes are at the heart of aerosol, cloud, and climate interaction. Through microphysical processes, aerosols, serving as cloud condensation nuclei (CCN), can affect the precipitation efficiency, cloud lifetime, cloud radiative properties, and thus climate (Twomey 1977; Albrecht 1989; Rosenfeld and Lensky 1998; Lohmann and Feichter 2005; Solomon et al. 2007). Significant progress has been made to improve the treatment of stratiform (large scale) cloud microphysics in global climate models (GCMs) (Liu et al. 2007; Morrison and Gettelman 2008, hereafter MG08; Gettelman et al. 2010; Salzmann et al. 2010). For example, the new four-class (liquid, ice, rain, and snow), two-moment large-scale cloud microphysics scheme of MG08 has been implemented in the latest version of the National Center for Atmospheric Research (NCAR) Community Atmosphere Model version 5 (CAM5) and the Geophysical Fluid Dynamics Laboratory (GFDL) Atmospheric Model version 3 (AM3) GCM (Salzmann et al. 2010). This improvement makes it possible for the models to simulate the full aerosol–large-scale cloud interaction, including cloud droplet activation by aerosols, precipitation processes due to particle-size-dependent behavior, and explicit radiative interaction of cloud particles. As such, the models are able to simulate the indirect radiative effects of aerosols on large-scale clouds.
In convective clouds, the microphysical processes play a particularly important role because the resulting latent heating and condensate loading directly influence the cumulus dynamics. The interactions between microphysical processes and cumulus dynamics further complicate the aerosol impacts on cloud, precipitation, and climate. In warm cumuli, an increase of aerosol loading increases the number concentration and decreases the size of cloud droplets. As a result, the coalescence of cloud droplets into raindrops is greatly reduced and thus the warm rain production is suppressed (Rosenfeld 1999). However, in deep convective clouds, the suppression of warm rain formation in the lower part of the cloud can cause greater amounts of cloud water to ascend above the freezing level. The subsequent enhanced ice formation increases the latent heat release and invigorates the convection (Khain et al. 2005; Rosenfeld et al. 2008). Cloud-resolving model simulations (Tao et al. 2007) show that in a tropical cloud more aerosols may cause more and smaller raindrops, which evaporate faster and cause stronger downdrafts. In addition, more and smaller cloud droplets may also enhance evaporation in downdrafts near cloud edges and strengthen the downdrafts (Lee et al. 2008). The stronger low-level convergence induced by the enhanced downdrafts then may intensify subsequent convection. Yet, there are other modeling studies showing that convection may be weakened in polluted deep convective clouds due to greater radiative heating in the upper troposphere associated with more and smaller ice crystals (Morrison and Grabowski 2011) or weaker immersion freezing near the cloud top caused by enhanced evaporation of cloud droplets (Cui et al. 2006). Seifert and Beheng (2006), Lee et al. (2008), and Khain (2009) further show that aerosol effects on convective clouds depend on the environmental conditions (e.g., relative humidity, vertical wind shear) and cloud type. These studies clearly demonstrate that the impact of aerosols on convective clouds is a complex scientific issue relying on convective microphysics.
Because of the extreme complexity and incomplete knowledge of microphysics in convective updrafts and computational limitations, the microphysical processes in convective clouds describing the formation of cloud and precipitating particles are ignored or parameterized crudely in many convective parameterization schemes in GCMs. A number of mass-flux convection schemes currently used in GCMs (e.g., Arakawa and Schubert 1974; Tiedtke 1989; Zhang and McFarlane 1995, hereafter ZM) omit the details of the formation of convective hydrometeors and assume that the conversion of cloud liquid water to precipitation is proportional to the amount of cloud liquid water in the updrafts. Without detailed microphysics parameterization in convective clouds, GCMs are unable to assess the impacts of aerosol–convection interactions, such as those described above. In addition, since the proportionality constant for cloud water to rainwater conversion is a tunable parameter, it can lead to large uncertainty in the partitioning of convective condensate between precipitation particles and detrained condensate in convection schemes. An evaluation of simulations of CAM3 (Boville et al. 2006), a predecessor of CAM5, showed a much lower stratiform precipitation ratio (≤10%) in tropical precipitation compared to the Tropical Rainfall Measuring Mission (TRMM) satellite observations (~40%; Schumacher and Houze 2003), which can be partly attributed to the significantly underestimated detrainment of condensate in the ZM convection scheme into the stratiform clouds in CAM3.
Convective detrainment is an important water source for tropical cirrus and upper tropospheric moisture. It has great implications for global climate change (Betts 1990; Ramanathan and Collins 1991; Lindzen et al. 2001). Because the amount of hydrometeors detrained from convection highly depends on the convective microphysical properties (e.g., mixing ratio and number concentration of cloud ice), proper treatment of convective microphysical processes in GCMs is crucial to reliable climate simulation and projection. Rennó et al. (1994) showed that the climate is very sensitive to cloud microphysics and associated precipitation efficiency in a convective–radiative equilibrium model. They suggest that convective parameterization schemes without microphysics may not be adequate for studying future climate change.
Lohmann (2008) first extended a double-moment stratiform cloud microphysics scheme to convective clouds in the ECHAM5 GCM to investigate the global aerosol effects on convective clouds. In her approach, a constant updraft velocity (2 m s−1) for convective cores was used in the microphysics scheme to obtain cloud fraction. She found that convective precipitation increases with aerosol loading. To improve the representation of convective clouds and its interactions with large-scale clouds and aerosols in GCMs, Song and Zhang (2011, hereafter SZ11) developed an efficient two-moment diagnostic convective microphysics parameterization scheme, in which the convective updraft vertical velocity is estimated from the equation for updraft kinetic energy. The scheme explicitly treats mass mixing ratio and number concentration of four hydrometeor species (cloud water, cloud ice, rain, and snow) and describes several microphysical processes, including autoconversion, self-collection, collection between hydrometeor species, freezing, sedimentation, cloud ice nucleation, and droplet activation by aerosols. An evaluation of the scheme in the single-column version of the NCAR Community Atmosphere Model version 3.5 (SCAM3.5) shows that the simulation of cloud microphysical properties in convective updrafts is significantly improved. With more realistic convective cloud microphysical properties and their detrainment, the simulations of surface large-scale precipitation, shortwave and longwave radiation fluxes, specific humidity, and temperature were all improved. Motivated by this improvement in the SCAM3.5, the present study implements the SZ11 microphysics scheme in the CAM5 and evaluates its performance and impact on the global climate simulation.
The paper is organized as follows: Section 2 briefly describes the model, microphysics parameterization scheme, and experiment design. Section 3 validates the scheme by comparing with the observations. Section 4 presents the impact of microphysics scheme on climate simulations. A summary and conclusions are given in section 5.
2. Model and microphysics parameterization
a. Model
The Community Atmosphere Model version 5.0.1 (CAM5) is the atmospheric component of the Community Earth System Model, version 1 (CESM1). In the CAM5, the radiation scheme has been updated to the Rapid Radiative Transfer Method for GCMs (RRTMG) described by Iacono et al. (2008). The moist boundary layer scheme is based on Bretherton and Park (2009), and the large-scale cloud and precipitation processes are parameterized with a prognostic two-moment bulk cloud microphysics scheme (MG08). The Park and Bretherton (2009) scheme is employed for shallow convection, and the ZM convection scheme with a dilution approximation for the calculation of convective available potential energy (Neale et al. 2008) is used for deep convection. The ZM scheme bypasses all the details of the formation mechanisms of cloud water and precipitation inside the convective updrafts, and the conversion of cloud water to rainwater is determined through a tunable parameter. The rainwater is removed immediately from the updrafts either as surface precipitation or through evaporation in the atmosphere. Since ice phase is not taken into account, all detrained hydrometeor is in liquid phase. Because the MG08 two-moment bulk cloud microphysics scheme in CAM5 takes convective detrainment of cloud liquid and ice water contents and their number concentrations as input, the detrained liquid condensate is first linearly partitioned into liquid and ice over the temperature range of −5°C < T < −35°C, and then the detrained number is estimated from the detrained mass by assuming a mean volume radius of 8 and 25 μm for droplet and cloud ice, respectively.
b. Microphysics parameterization for convective clouds
The microphysics parameterization scheme described in SZ11 is a four-class (liquid, ice, rain, and snow), two-moment (mass and number) scheme for convective clouds. It is designed for use in bulk mass-flux convection parameterization schemes in GCMs. Figure 1 shows the schematic of the processes considered in the microphysics scheme. It includes droplet activation and ice nucleation by aerosols; autoconversion of cloud water/ice to rain/snow; accretion of cloud water by rain; accretion of cloud water, cloud ice, and rain by snow; homogeneous and heterogeneous freezing of rain to form snow; Bergeron–Findeisen process; fallout of rain and snow; condensation/deposition; self-collection of rain drops; and self-aggregation of snow. Here we will highlight the droplet activation and ice nucleation. For a detailed description of other processes, the reader is referred to SZ11.
Schematic diagram of the microphysics scheme.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
The cloud droplet activation for multiple aerosol types with different size distribution is parameterized following Abdul-Razzak and Ghan (2000). It is well known that the primary activation of aerosols occurs at the cloud base because of high supersaturation and relatively low condensation rates there. Above the cloud base, the depletion of excess water vapor by condensation on previously activated particles will reduce the supersaturation in a constant updraft. However, in real-world deep convective clouds, the increasing updraft strength (e.g., Warner 1969; Pinsky and Khain 2002), the inevitable depletion of droplets formed at the cloud base by accretion (Lamb and Verlinde 2011; Phillips et al. 2005), and entrainment (Brenguier and Grabowski 1993; Su et al. 1998; Lasher-Trapp et al. 2005) may also produce supersaturation conditions and thus droplet activation in cloud updrafts. Previous model simulations (Slawinska et al. 2012; Phillips et al. 2005; Morrison and Grabowski 2008) and observations (Prabha et al. 2011) have confirmed the importance of droplet activation in cumulus updrafts. Thus, the aerosol activation parameterization is implemented both at and above the cloud base in this study. The amount of activated aerosols from the layers below is used as a proxy for the number of aerosols previously activated, and an activation time scale of 15 min is assumed in this study. The activation time scale represents the time scale for recirculation of air parcels to regions of droplet activation. For example, the time scale at the cloud base represents the time it takes to supply the air mass from below. Since the subcloud layer is about 1000 m or less and the vertical velocity is 1 m s−1, an activation time scale of 15 min appears reasonable for convective clouds. As will be shown in the next section, most of the aerosol activation occurs near the cloud base.
The ice nucleation parameterization includes both homogeneous freezing and heterogeneous freezing. Homogeneous droplet freezing is performed by instantaneous conversion of the supercooled cloud liquid water to cloud ice at temperatures below −40°C. Between −5° and −35°C, the immersion freezing of black carbon and dust is parameterized after Diehl and Wurzler (2004) and contact freezing of dust follows Liu et al. (2007). Below −35°C, ice nucleation is based on Liu et al. (2007), which includes heterogeneous immersion freezing of dust competing with homogeneous freezing of sulfate and depends on updraft velocity, air temperature, and aerosol properties. Recently, Phillips et al. (2007) assessed the relative roles of nucleation processes for two cases of deep convection in the tropical western Pacific region using a cloud-system-resolving model with two-moment bulk microphysics and found that homogeneous aerosol freezing occurs only in regions of weak ascent while homogeneous droplet freezing is dominant in stronger updrafts. To account for this competition between the two processes, the homogeneous freezing of sulfate is suppressed in this study when updraft vertical velocity is greater than 4 m s−1. In addition, the deposition/condensation nucleation on mineral dust between −37° and 0°C is represented by Meyers et al. (1992) and secondary ice production between −3° and −8°C (Hallett–Mossop process) is also included based on Cotton et al. (1986).
In SZ11, the accretion of cloud liquid/ice by rain/snow falling from above into the layer was not considered because the diagnostic equations for rain and snow were integrated from bottom up in the rising updraft. This may result in an underestimation of rain/snow in the layer and thus a lower efficiency of the accretion process. In this study, we take the effect of accretion by falling precipitation into account by integrating the hydrometeor equations twice. The first integration is the same as in SZ11 and it provides the provisional values of rain/snow. The iteration takes into account the accretion effect of precipitation falling from above using the provisional values. This improvement results in a reduced cloud liquid/ice water content due to enhanced accretion of cloud liquid/ice water by rain/snow. The maximum reduction of cloud liquid/ice water content can be as much as 50%/40% at some points over tropical ocean, while the effect is relatively small in land convection. To prevent any possible development of negative values of mixing ratio and number concentration of water species from a simple forward stepping, a conservation check is performed at each level for each water species before the upward integration. If the provisional values of mixing ratio or number concentration of water species are negative, the values of the microphysical sinks are adjusted to ensure water and energy conservation. Note that the separate treatments for mixing ratio and number concentration may occasionally result in inconsistency between them, giving unrealistic particle sizes. To avoid this pathological situation, the number concentration of each water species is adjusted if necessary at each level after upward integration to ensure that the mean particle size remains within a specified range.
To evaluate the performance of the microphysics scheme and its impact on climate simulation, the SZ11 microphysics scheme is implemented in the ZM convection scheme of CAM5 as follows: The ZM convection parameterization first estimates the net condensation and deposition rate by assuming that the cloud air in convective updrafts is saturated with respect to water or ice and feeds it to the microphysics scheme. Note that in the original ZM convection scheme, the equation of Bolton (1980) is used for calculating saturation vapor pressure (SVP) over liquid water, and the SVP over ice is not considered. In this study, the SVPs over liquid water and ice are calculated by using the formulations from WMO (2008) at temperatures above 0°C and below −35°C, respectively. In the temperature range 0° to −35°C, the SVP is assumed to be a linear combination of the two according to temperature (all water at 0°C and all ice at −35°C). The condensation and deposition rates are determined correspondingly based on the same temperature specification. The microphysics scheme calculates the microphysical properties of cloud particles and precipitation rate in saturated updrafts and feeds them back to the ZM convection scheme. With these cloud microphysical properties, the detrained cloud liquid/ice water content and their number concentrations are calculated in the ZM scheme and are provided to the two-moment large-scale cloud microphysics scheme. Note that the microphysics scheme is only implemented in saturated updrafts in the ZM scheme and not coupled to convective dynamics or radiation yet. Thus, while the impacts of aerosols on the suppression of warm rain production and the subsequent enhancement of freezing processes are included in the model, possible aerosol effects on evaporation in downdrafts or weakening of convection due to enhanced upper troposphere radiative heating are not included in this study.
Two 6-yr simulations with the CAM5 are conducted and the results from the last 5 yr are presented in this paper. The simulations start from 1 January and run for 6 yr, forced with prescribed climatological sea surface temperatures and aerosol distribution obtained from the NCAR repository (http://svn-ccsm-inputdata.cgd.ucar.edu/trunk/inputdata/) (after registration, the data can be accessed). In the control simulation (referred to as the CTL run) the standard CAM5 configuration is used, whereas in the experimental simulation (referred to as the MPHY run) the convective microphysics scheme is implemented in the ZM convection scheme. Both simulations use the finite volume dynamical core at 1.9° latitude by 2.5° longitude horizontal resolution with 30 vertical levels up to 3 hPa. The monthly climatology (2000–09) of sulfate, sea salt, dust, and carbonaceous species mass concentrations was generated by a CAM4 simulation with a chemistry model, which included prognostic bulk aerosols. The aerosol numbers were diagnosed from mass concentrations by assuming a lognormal distribution for each aerosol species except for sulfate. Following Lohmann et al. (2000), the number concentration of sulfate was estimated through an empirical relationship Nso4 = 340(Mso4)0.58, where Mso4 is the mass of sulfate. The radius and its standard deviation, density, hygroscopicity, and scaling factors for mass to number for each aerosol type are shown in Table 1.
Aerosol properties and parameters.
To validate the new microphysics scheme for convective clouds, the cloud ice water content in convective clouds from CloudSat (Stephens et al. 2002), release 4 (R04) dataset is used in this study. In addition, the following observational datasets are also used for simulation evaluation: the Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP; Xie and Arkin 1997) from 1979 to 1998, Global Precipitation Climatology Project (GPCP) version 2 combined precipitation dataset from 1979 to 2003 (Adler et al. 2003), the cloud fraction from combined measurement of CloudSat and Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO; Stephens et al. 2002; Winker et al. 2003) from 2006 to 2008, the cloud liquid water path of Moderate Resolution Imaging Spectroradiometer (MODIS; Platnick et al. 2003) from 2000 to 2004, and the outgoing longwave radiation (OLR) flux and net shortwave flux at the top of the atmosphere of Clouds and the Earth’s Radiant Energy System edition 2 (CERES2; Wielicki et al. 1996) from 2000 to 2005.
3. Evaluation of the parameterization with observations
a. Microphysical properties
Launched in April 2006, CloudSat provides the first global measurements of cloud profiles and cloud physical properties with seasonal and geographical variations. It is equipped with a 94-GHz nadir-looking cloud-profiling radar (CPR), which measures backscattered power profiles from hydrometeors along its path with a minimum detectable threshold of about −30 dBZ. CloudSat has a resolution of about 1.4 km cross track and 1.7 km along track. Its vertical resolution is 480 m but with 240-m vertical sampling intervals. In this study, a subset of convective pixel-scale cloud ice water content from CloudSat R04 dataset for 2 months (January and July 2007) is used. The CloudSat retrieval assumes a lognormal size distribution of cloud particles. It also makes assumptions about ice crystal number concentration, cloud effective radius, and cloud phase as a function of temperature, which is provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) reanalyses. The convective pixels are identified by a cloud classification based on cloud cluster, cloud height, temperature, maximum CloudSat radar signal and its height, cloud vertical and horizontal extent, and the occurrence of precipitation reaching the surface (Wang and Sassen 2001; Waliser et al. 2009; Chen et al. 2011).
Figure 2 shows the cloud ice water content profiles in convective clouds from CloudSat retrievals and model simulations. The profiles averaged in three major convective regimes, the tropical (20°S–20°N) ocean, tropical land, and midlatitude [20°–40°N(S)] land are shown separately. In boreal winter (January), the CloudSat retrievals (Fig. 2a) show that the convective ice water content (IWC) peaks at about 350 hPa in the tropics, with a maximum of 0.11 g m−3 in oceanic convection and 0.13 g m−3 in land convection, while it has a broader peak from 450 to 250 hPa of 0.09 g m−3 in midlatitude land convection. The midlatitude data are for summer hemisphere only to depict convective properties. The distribution of IWC produced by both model simulations is generally in good agreement with the observations in land convection, although a slightly larger maximum is observed over midlatitude land. The largest difference in IWC is between the control simulation and CloudSat over tropical ocean. The CTL run has a maximum of only about 0.02 g m−3, less than 20% of the observations. With the new microphysics parameterization, the peak of IWC produced in the MPHY run is increased to 0.11 g m−3, in much better agreement with the CloudSat observations. For boreal summer (July), the CloudSat-retrieved convective IWC distribution (Fig. 2d) is similar to that in winter month (Fig. 2a), but with a slightly smaller peak in the tropics and a larger maximum over midlatitude land. Again, the CTL is able to reproduce the IWC distribution observed by CloudSat over land but underestimates it by about 80% over tropical ocean. With the new microphysics parameterization, the MPHY reproduces the observed IWC distribution in all regions well, with a peak value of about 0.11 g m−3 at 300 hPa over tropical ocean. Moreover, the difference in IWC distribution between midlatitude land and tropical ocean (i.e., a slightly larger peak in midlatitude land) is also well simulated.
Cloud ice water content (g m−3) profiles in convective clouds averaged over the tropical (20°S–20°N) ocean (blue dotted), tropical land (black solid), and midlatitude [20°–40°N(S)] land (red dashed) for January from (a) CloudSat, (b) CTL, and (c) MPHY and for July from (d) CloudSat, (e) CTL, and (f) MPHY.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
Because of radar limitations, the cloud liquid water content (LWC) retrievals in intense convection from CloudSat are either missing or unreliable because of attenuation of radar signal by precipitation. Thus, here we compare the simulated LWC to that from aircraft observations and cloud-resolving model simulations. The statistics of aircraft observations by Borovikov et al. (1963) show that characteristic values for dense cumulus congestus and cumulonimbus are 0.5–3 g m−3. In cumulus clouds with very strong updrafts, LWC can be up to 5 g m−3 or more (Poellot and Pflaum 1989; Musil and Smith 1989). Recent aircraft observations by Prabha et al. (2011) in the Indian monsoon region show the horizontally averaged LWC in premonsoon and monsoon convection clouds in the range of 0.5–2 g m−3, with typical values between 1 and 1.5 g m−3 in midtroposphere. The simulation of a convective storm using the Weather Research and Forecasting (WRF) model with a two-moment bulk microphysical scheme produced peak values of LWC ranging from 1.5 to 3 g m−3 at 2–3 km, depending on the aerosol concentration (Li et al. 2008). The CAM5 simulations show that the LWC profiles do not change significantly from January (Figs. 3a,b) to July (Figs. 3c,d). The maximum of LWC in the CTL is only about 0.15 g m−3 over the tropical ocean, indicating a serious underestimation of LWC. With the new microphysics parameterization, the LWC produced by MPHY reaches a maximum of 1.3 g m−3 at 800 hPa, almost a factor of 8 larger than that of the CTL run. Over tropical land, the LWC produced by both model simulations peaks near 700 hPa, with a magnitude of 0.6 (0.5) g m−3 in the CTL and 1.1 (1.0) g m−3 in MPHY for January (July). The LWC distribution over midlatitude land peaks at a higher altitude near 600 hPa with a maximum of 1.2 g m−3 in the MPHY run and half as large in the CTL.
Cloud liquid water content (g m−3) profiles in convective clouds averaged over the tropical (20°S–20°N) ocean (blue dotted), tropical land (black solid), and midlatitude [20°–40°N(S)] land (red dashed) for January from (a) CTL and (b) MPHY and for July from (c) CTL and (d) MPHY.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
The explicit representation of cloud particle number concentration in the two-moment microphysics scheme enables the interaction between convection and aerosols in the scheme and allows for more comprehensive representation of the interaction between convection and large-scale clouds. In the CAM5, the two-moment large-scale cloud parameterization of MG08 takes the detrainment of cloud particle numbers from convection as a source term. However, since the number concentration of cloud particles in convection is not available in the ZM convection parameterization scheme, the model estimates the detrainment of cloud particle number concentration from the detrained cloud liquid/ice water mass by assuming the mean volume radius of 8 and 25 μm for cloud droplets and ice crystals, respectively. Apparently, this assumption of a single prescribed radius for all cloud particles at global scale is unrealistic. Observations show that the LWC of maritime cumulus does not differ significantly from that of continental cumulus (Squires 1956; Wallace and Hobbs 2006). On the other hand, the cloud droplet number concentration (CDNC) of continental cumulus is much higher than that of maritime cumulus (Squires 1956; Hudson and Yum 2001), indicating that continental cumulus has much smaller average droplet radius and narrower droplet size spectrum than those of maritime cumulus. Droplet number concentration in active maritime cumulus is typically 20–60 cm−3, while it is much higher in continental cumulus, in the range of 50–300 cm−3 or higher (Squires 1958; Pruppacher and Klett 1997; Wood et al. 2011; Wallace and Hobbs 2006).
As continental convection primarily occurs over Northern Hemisphere midlatitude land during boreal summer season besides tropical land, the profiles of nonzero CDNC in convective clouds averaged over tropical ocean and Northern Hemisphere midlatitude land for June–August (JJA) are shown in Fig. 4a for the MPHY simulation. There is a remarkable contrast in CDNC between maritime (tropical ocean) and continental (midlatitude land) convection. The maximum CDNC is about 40 cm−3 in maritime convection and 105 cm−3 in continental convection, both in good agreement with observations. Furthermore, the vertical distribution is also very different. In oceanic convection CDNC is largely concentrated below 800 hPa, while it peaks at about 600 hPa in land convection, likely due to higher cloud-base height often observed in land convection above high topography. In addition, there is a secondary peak of droplet number at about 350 hPa in maritime convection, which is not seen in the land convection. This difference can be attributed to the different population distribution of maritime and continental convection and conditional averaging. In maritime convection, the conditional average of droplet number from a few strong convective clouds and many relatively weak clouds in the middle troposphere may result in a relatively smaller CDNC. Since only strong convection can reach the upper troposphere, conditional averaging over these clouds is expected to result in greater CDNC at these levels. This may explain the secondary peak of droplet number at about 350 hPa in maritime convection, as also noted by Morrison and Grabowski (2011). In contrast, continental convection is generally dominated by deep convective clouds, thus no secondary peak of CDNC is observed.
JJA Mean cloud (a) droplet and (b) ice crystal number concentration (cm−3) profiles in convective clouds averaged over the tropical (20°S–20°N) ocean (solid) and Northern Hemisphere midlatitude (20°–40°N) land (dotted) from MPHY.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
The differences in the profiles of cloud ice crystal number concentration (ICNC) (Fig. 4b) are less dramatic between continental and oceanic convection, with both having a maximum near 150 hPa. However, the oceanic convection has a secondary peak in the middle troposphere below the homogeneous freezing level (~300 hPa). To date, the observation of ice particle number concentration in convective clouds remains a challenging problem because of complicated crystal properties (e.g., shapes, size), measurement uncertainties, and safety concerns arising from aircraft penetrating convective cores above the freezing level. Here we compare the model simulations with microphysical characteristics of convectively generated ice clouds from two field campaign observations, the Central Equatorial Pacific Experiment (CEPEX; McFarquhar and Heymsfield 1996) and the Kwajalein Experiment (KWAJEX; Heymsfield et al. 2002), and those from cloud-system-resolving model simulations (Phillips et al. 2007). The observed values for ICNC are 1 cm−3 above 9 km in CEPEX (McFarquhar and Heymsfield 1996) and 0.02–0.07 cm−3 below 11 km in KWAJEX (Heymsfield et al. 2002), while the WRF simulation of TOGA COARE convection gives ICNC in the range of 0.02–2.5 cm−3 (Phillips et al. 2007). Our CAM5 simulation shows that the ICNC ranges from 0.02 to 1.4 cm−3, in line with available observational and cloud-resolving model results. Note that these profiles of ICNC are conditionally averaged over convective clouds only when they exist. The vertical structures of the profiles are weighted by different cloud population at each level. For a given cloud type (e.g., convective clouds with tops above the 250-hPa level), sources and sinks from the microphysical processes and convective transport determine the vertical distribution of CDNC and ICNC.
As an example, Fig. 5 presents the JJA mean ICNC and CDNC budgets for clouds with tops above 250 hPa and clouds with bases below 900 hPa, respectively, for tropical oceanic convection. The droplet activation is a dominant source for CDNC (Fig. 5a). The primary droplet activation occurs near the cloud base and decreases rapidly upward, consistent with observations and detailed parcel model simulations. Below 900 hPa, the rapid change to near zero just above 1000 hPa is an averaging artifact. Since below the cloud base the activation rate is zero, averaging nonzero activation rates in convective clouds whose bases are below the averaging level and zeros for clouds whose bases are above the averaging level, results in much smaller activation rates. In the lower part of the convective clouds, the major sink term is accretion by rain, which is smaller than the source term (droplet activation). The surplus droplets are transported upward by convection, making convective transport a sink below 850 hPa. Detrainment becomes another important sink above 850 hPa, leading to the sum of sink terms exceeding the droplet activation. Accordingly, updrafts transport less out of the layer from the top than into the layer from the bottom, making convective transport a source above 850 hPa. Besides a maximum detrainment at 800 hPa corresponding to shallower convection, there is a deep layer of detrainment above 600 hPa, but with smaller magnitude, corresponding to deeper convection population. Between 400 and 250 hPa, collection by snow becomes an important sink of CDNC. For midlatitude land convection of the same type (not shown), the overall structures of each source and sink term is similar except with larger amplitudes. In addition, detrainment has a maximum in the middle and upper troposphere due to dominance by deep convection. Note that the mean droplet activation averaged over the entire population of midlatitude land convection peaks at about 600 hPa, due to weighting from clouds with bases above 900 hPa (not shown).
JJA mean budgets (g−1 m−1) of (a) CDNC in convective clouds with bases below 900 hPa and (b) ICNC in convective clouds with tops above 250 hPa over the tropical (20°S–20°N) ocean from MPHY.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
The Hallett–Mossop (HM) process (Hallett and Mossop 1974) of ice particle multiplication is a dominant source for ICNC below 400 hPa (Fig. 5b). This is consistent with the simulation of convection by a cloud-resolving model (Phillips et al. 2007), which shows that the HM process has a large impact on the mixed-phase microphysics of convection. For midlatitude continental convection (not shown), the HM process is also the dominant source in midtroposphere, but with smaller magnitude. This difference between maritime and continental convection is also observed in the cloud-resolving model simulation of Aleksic et al. (1989). Between 350 and 200 hPa, immersion freezing and homogenous freezing of droplets are important ICNC sources. Ice nucleation (deposition nucleation and homogeneous aerosol freezing) also has a nonnegligible contribution near 250 hPa. Numerous observational and modeling studies (Heymsfield et al. 2005, 2009; Phillips et al. 2005, 2007) have confirmed that homogeneous droplet freezing is a major process responsible for the formation of small ice particles in strong updrafts of deep convective clouds. Between 250 and 150 hPa, detrainment is a large sink of ICNC. Between 450 and 300 hPa, accretion by snow is also an important sink. Convective transport, which is the net flux convergence into a layer from the bottom and the top of the layer, simply reflects the imbalance between the sources and sinks in the layer. From 550 to 450 hPa, ice crystals produced by the HM process is transported upward. Immediately above it, when accretion by snow dominates, updrafts transport more ice crystals into the layer from the bottom and less out of the layer from the top. In the layer from 350 to 250 hPa, more ice crystals are produced than consumed. Thus updrafts transport them out of the layer from the top. Finally, near 200 hPa, the ice crystals transported from below must detrain. The ICNC budget for midlatitude land convection is similar, except with larger magnitude.
b. Impacts of aerosols on convection
The activation of droplets and nucleation of ice crystals are two key processes for successful simulation of aerosol impacts on deep convection. Since both of these two processes depend directly on convective vertical velocity, we examine the distribution of vertical velocity in convection updrafts. Figure 6 shows the probability distribution of vertical velocity in deep convection over midlatitude (20°–40°N) land and tropical (10°S–10°N) ocean for JJA. The convective vertical velocities over ocean range from 0.5 to 9 m s−1, peaking at about 3–4 km with maximum and median velocity of 9 and 7 m s−1, respectively. This is in good agreement with the observations for oceanic convection by Lucas et al. (1994) (their Fig. 2). In contrast, the vertical velocities over land peak at a higher altitude (400 hPa; ~8 km) with a greater maximum value (~12 m s−1). Lucas et al. (1994) also compared the observed vertical velocity of oceanic convection with that of land convection from the Thunderstorm Project (Byers and Braham 1949) and found a higher peak with a greater intensity (15 m s−1 at 8 km) in land convection. This indicates that the representation of vertical velocity in the microphysics scheme is reasonable when sampled over a large region and a long period, despite the fact that only a bulk vertical velocity instead of a spectrum is predicted at a given grid point and time step. The simulated vertical velocity in land convection is smaller than that observed from the Thunderstorm Project, especially in the lower troposphere, probably because Fig. 6 shows the averages of vertical velocity in all land convection, not just strong thunderstorms.
The maximum; minimum; and 10th, 25th, 50th, 75th, and 90th percentiles of vertical velocity distribution in convective clouds averaged over (a) the tropical (10°S–10°N) ocean and (b) the Northern Hemisphere midlatitude (20°–40°N) land for JJA from MPHY.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
To investigate whether the microphysics scheme realistically represents the possible impacts of aerosols on convective clouds, a 6-yr sensitivity simulation (referred to as to LOW_aero) is carried out. The simulation is the same as the MPHY run, except the aerosol concentration is reduced by a factor of 10. Thus, the difference in convection between the two simulations is entirely attributable to the aerosol effects. Figure 7 shows the JJA mean profiles of cloud droplet and ice crystal number concentration in convective updrafts for LOW_aero. Compared to MPHY (Fig. 4), it clearly shows that both droplet and ice crystal number concentrations are significantly reduced in the LOW_aero, indicating that droplet activation and ice nucleation are weakened because of reduced aerosol concentration. The liquid water content is only reduced slightly in the LOW_aero run (not shown). This, together with the dramatically reduced number concentrations, implies larger droplets in LOW_aero. Because the coalescence efficiency of cloud droplets into raindrops is greatly increased when the radius of droplet is large (Mason and Jonas 1974), the conversion of cloud droplets to raindrops and hence precipitation is enhanced in clouds with larger droplets. Indeed, Fig. 8a shows that autoconversion from droplets to raindrops in convection averaged over the major convective regime (20°S–40°N) is increased in the LOW_aero relative to the MPHY. Because of the reduced autoconversion efficiency in MPHY compared to LOW_aero, more cloud liquid water is transported to the upper troposphere. When these droplets freeze, they release more latent heat, which can result in more vigorous convection (Khain et al. 2005; Rosenfeld et al. 2008). Figure 8b shows that the ice production due to Bergeron process and homogeneous freezing are enhanced in MPHY relative to LOW_aero, indicating that the scheme is able to represent the enhancement of freezing in convection when aerosol loading is increased.
JJA mean cloud (a) droplet and (b) ice crystal number concentration (cm−3) profiles in convective clouds averaged over the tropical (20°S–20°N) ocean (solid) and Northern Hemisphere midlatitude (20°–40°N) land (dotted) from LOW_aero.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
(a) Autoconversion rate of cloud liquid water to rain (g kg−1 km−1) and (b) cloud ice production (g kg−1 km−1) through the Bergeron process (dotted lines) and homogeneous droplet freezing (dashed lines) in convective clouds averaged over 20°S–40°N for JJA from MPHY (thick lines) and LOW_aero (thin lines).
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
In summary, without a microphysics scheme, CAM5 in the current configuration tends to underestimate the LWC and IWC seriously, by up to 80% over tropical ocean. Considering that the ocean covers 76% of the earth’s surface in the tropics (20°S–20°N) and the most active convection resides in tropical ocean region, this underestimation could have profound impacts on the interaction between convection and clouds and thus the climate simulation. In contrast, with the new physically based microphysics parameterization scheme, the model produces more realistic cloud microphysical properties in convective clouds. Both the cloud liquid/ice water content and cloud droplet/crystal number concentrations are in good agreement with available observations. It should be noted that the simulation of convective cloud liquid/ice water content in the standard CAM5 strongly depends on a tunable parameter c0 for conversion from cloud water to rainwater in the ZM convection scheme. Therefore, the underestimation of LWC and IWC in the standard CAM5 can also be mitigated by optimal tuning of c0. The advantage of the new microphysics scheme is to improve the simulation by incorporating detailed cloud microphysical processes rather than tuning parameters. Furthermore, the new microphysics scheme can also enable the model to describe the interaction between convection and aerosols. The simulations show that the microphysics scheme is able to represent the suppression of warm rain formation and enhancement of freezing when aerosol loading is increased, reflecting its ability to represent the effect of aerosol on convective microphysics.
4. Impacts on global climate simulation
The convective detrainment of cloud liquid/ice water is an important water source for stratiform anvil clouds. Thus, the increased convective LWC and IWC and associated detrainment from convection with the new convective microphysics parameterization is expected to have large impact on global climate simulation. Compared to the CTL run, the annual global mean large-scale precipitation is increased by about 14% in the MPHY, while the convective precipitation is decreased by about 11%. In the tropics (20°S–20°N), the large-scale precipitation in the MPHY is increased by more than 44% relative to the CTL, while the convective precipitation is decreased by about 4%. As a result, the contribution from large-scale clouds to the total precipitation is increased from 8% in the CTL to 11% in the MPHY, although it is still much less than what observations suggest (Schumacher and Houze 2003) because of overly active convection in the model.
To illustrate the regional changes resulting from the microphysics scheme, the 5-yr average of precipitation distribution for boreal summer (JJA) from the CMAP and GPCP observations, the CTL and MPHY simulations, and their differences is shown in Fig. 9. Two sets of observational estimates are presented to illustrate the uncertainty in observations. Compared to the CMAP observations, the precipitation in the western Pacific warm pool, South China Sea, Bay of Bengal, and eastern Pacific ITCZ is much weaker in the CTL run. There are also strong positive bias in the SPCZ, Maritime Continent, Arabian Peninsula, and Arabian Sea. With the new microphysics scheme, the precipitation biases in these regions are reduced remarkably (Fig. 9h). As indicated above the difference plots, the global mean precipitation biases relative to the CMAP is reduced to 0.16 mm day−1 from 0.19 mm day−1 in the CTL, with reduced root-mean-square error (RMSE) as well. Note that, although the precipitation over the western Pacific warm pool in the MPHY is weaker relative the CMAP observations, it is comparable to the GPCP observations.
JJA mean precipitation rates (mm day−1) from (a) CMAP, (b) GPCP (c) CTL, and (d) MPHY and the differences (e) between CTL and CMAP, (f) between MPHY and CMAP, (g) between CTL and GPCP, and (h) between MPHY and CTL.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
The deficit in JJA precipitation over the western Pacific is a long-standing issue in the NCAR Community Atmosphere Model. Comparison of convective and large-scale precipitation between CTL and MPHY shows that both components are increased in the western Pacific in the MPHY. It is instructive to examine the large-scale cloud liquid/ice water budget to see how the improvement in microphysical properties of convection leads to the enhancement of large-scale precipitation in the western Pacific. The JJA mean profiles of large-scale cloud liquid and ice water budget averaged between 130° and 150°E and between 10° and 20°N, where the improvement is most significant, from the CTL and MPHY are shown in Fig. 10. For cloud ice budget, the major source terms are the Bergeron process and deposition while the major sink term is autoconversion to snow, which dominates over other sink terms (sedimentation and accretion by snow). In the CTL, the contribution from convective detrainment is negligible. When the new microphysics scheme is used, the convective detrainment of cloud ice is enhanced, but the increase is smaller than those of Bergeron process and deposition. This is different from the single column model simulation for the Tropical Warm Pool-International Cloud Experiment (TWP-ICE) period, where considerable enhancement of cloud ice detrainment was found when the microphysics parameterization was included (SZ11). Further discussions on this will be given below. Compared to the CTL, the deposition term is strengthened significantly in the MPHY, indicating an enhancement of large-scale upward motion. The resulting increase of large-scale cloud ice leads to enhancement of autoconversion of ice to snow.
JJA large-scale cloud ice water budget from (a) CTL and (b) MPHY and cloud liquid water budget from (c) CTL and (d) MPHY averaged between 130° and 150°E and between 10° and 20°N.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
For cloud liquid water budget, above 600 hPa the main cloud water source is condensation, and the dominant sink is cloud water collection by rain, with some contributions from autoconversion of droplets to rain, cloud water collection by snow, and the Bergeron process. Below 600 hPa, the dominant water source is convective detrainment and the sink is evaporation. Comparing the CTL to MPHY, the major differences are at levels above 600 hPa. Both convective detrainment of cloud water and condensation are greatly enhanced in MPHY relative to CTL. Correspondingly, the sink terms resulting from autoconversion and accretion of cloud water by rain are also significantly larger in MPHY than in CTL. Thus the strengthening of the precipitation generation processes through enhancements of autoconversion of cloud ice/droplets to snow/rain and accretion of cloud droplets by rain, caused by enhanced deposition/condensation and convective detrainment of cloud liquid/ice water, results in an increase of surface large-scale precipitation.
The strengthening of large-scale condensation and deposition implies that the large-scale upward motion is increased. The JJA profiles of large-scale vertical velocity averaged over 130°–150°E and 10°–20°N (Fig. 11, dashed line) indeed show that the large-scale vertical velocity is increased significantly in the MPHY compared to the CTL, indicating that a thermodynamics–microphysics feedback is involved and is responsible for the precipitation increase in the western Pacific. Specifically, we postulate the following feedback process for this. The new microphysics scheme produces more cloud liquid water in convective updrafts (Fig. 3d), resulting in increased detrainment of cloud liquid water between 600 and 400 hPa. This leads to an increase of rainwater in stratiform clouds through enhanced autoconversion and accretion (Fig. 10d). The enhanced latent heating due to freezing of the rainwater (Fig. 11, dotted line) strengthens the upward motion (Fig. 11, dashed line) and thus condensation/deposition in the midtroposphere (Figs. 10b,d), resulting in more large-scale precipitation. The extra latent heating from the enhanced condensation/deposition further reinforces the upward motion and large-scale precipitation. In addition, the enhanced upward motion strengthens the convective activity, as indicated by convective mass flux (Fig. 11, solid line), and therefore the convective precipitation and convective cloud liquid water detrainment.
JJA vertical profiles of large-scale vertical velocity (mb h−1, dashed lines), mass flux in deep convection (mb h−1, solid lines), and latent heating due to the freezing of rainwater (J h−1 kg−1, dotted lines) averaged over 130°–150°E and 10°–20°N for CTL (thin lines) and MPHY (thick lines).
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
It is puzzling that in spite of much higher convective IWC in MPHY the enhancement of cloud ice detrainment from convection is insignificant in Fig. 10. Note that the amount of cloud ice detrained from convection depends on the product of IWC and airmass detrainment from convection. The airmass detrainment is measured by the rate of decrease of mass flux with height. It is shown in Fig. 11 (solid line) that most of the convective mass flux detrains below 400 mb for both CTL and MPHY, with little detrainment above. Clearly, weak convective detrainment of air above 400 hPa keeps the ice source term small for large-scale ice budget (Figs. 10a,b) in both simulations. On the other hand, the enhanced airmass detrainment between 500 and 400 hPa, together with the larger LWC (Figs. 3b,d), explains the larger contribution of convective detrainment to large-scale cloud water budget in MPHY.
The convective detrainment of hydrometeors affects not only large-scale precipitation but also cloud fraction, cloud water path, and therefore radiation. Figure 12 shows the annual grid-mean cloud liquid water path (LWP) from MODIS retrievals, the biases of CTL and MPHY, and the difference between the two simulations. Since the visible near-infrared retrievals of MODIS tend to overestimate the cloud water path at high latitudes (Seethala and Horváth 2010), here we only focus on the simulation at middle and low latitudes. Compared to the MODIS retrievals, the most noticeable biases of CTL are over subtropical oceans [10°–30°S(N)], where the CTL seriously underestimates the cloud LWP, by up to 120 and 80 g m−2 over the southern and northern subtropical oceans, respectively. When the new microphysics parameterization is used, the cloud LWP is increased markedly, by up to 120 g m−2 over the northern subtropical oceans and 180 g m−2 over the southern subtropical oceans. In the Northern Hemisphere midlatitude storm tracks, the CTL has significant positive biases in cloud LWP, and inclusion of convective microphysics increases these biases slightly. The mean bias averaged between 60°S and 60°N is reduced from −35 g m−2 in the CTL to −17 g m−2, and RMSE is reduced from 51 to 44 g m−2, making it in better agreement with the MODIS retrievals.
Annual grid-mean cloud liquid water path (g m−2) from (a) MODIS retrievals and differences (b) between CTL and MODIS, (c) between MPHY and MODIS, and (c) between MPHY and CTL.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
The annual mean cloud fraction for high-, mid-, and low-level clouds from CloudSat; the biases of CTL and MPHY; and the difference between MPHY and CTL are shown in Fig. 13. The CTL tends to underestimate the high-level cloud in the SPCZ and overestimate it in the Asian–Australian monsoon region. The new convective microphysics scheme reduces these biases in the CTL slightly, with the global mean bias reduced to 0.52% from 1.30% in the CTL and the RMSE reduced to 8.25% from 8.51%. In general, the impacts of microphysics scheme on high-level cloud are relatively small. This is not surprising considering that detrainment of cloud ice from convective updrafts is very small in both CTL and MPHY because of weak airmass detrainment in the upper troposphere, as shown earlier.
Annual mean (a) high-, (b) middle-, and (c) low-level cloud fraction from CloudSat; difference of (d) high-, (e) middle-, and (f) low-level cloud between CTL and CloudSat; difference of (g) high-, (h) middle-, and (i) low-level cloud between MPHY and CloudSat; and difference of (j) high-, (k) middle-, and (l) low-level cloud between MPHY and CTL.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
For midlevel clouds, the CTL tends to underestimate the cloud fraction, especially over land, the SPCZ, and the ITCZ, resulting in a bias of about −9% in global mean cloud fraction. Compared to the CTL, the MPHY produces more midlevel clouds in low latitudes. In the ITCZ and SPCZ regions, the cloud fraction is increased by up to 30% and the global mean bias is reduced to −6.7%, with RMSE reduced to 10.2% from 11.6% in the CTL.
The CTL has serious negative biases in low-level cloud fraction over the oceans between 30°S and 30°N, where the average bias is more than 20%, with a maximum up to 60% at 20°S(N). Comparing to the CloudSat observation, the global mean bias in low-level clouds is −9.9% and the RMSE is as high as 18.4%. With the new microphysics scheme, the low-level cloud fraction is increased significantly, with a maximum up to 50% (30%) at 20°S(N), resulting in a reduction of 3.8% and 3.5% in the global mean bias and RMSE, respectively.
The increase of midlevel clouds in the SPCZ and ITCZ is associated with the enhancement of convective detrainment of cloud liquid water (Fig. 10d). For low-level cloud fraction, examination of contributions from shallow and deep convection schemes found that the changes in both detrainment of cloud liquid water and precipitation from the Park and Bretherton (2009) shallow convection parameterization scheme are negligible and that the increase of low-level cloud fraction over subtropical oceans is mainly due to the deep convection parameterization scheme. Figure 14 shows the pressure–longitude cross section of the difference in detrainment of cloud liquid water from the ZM convection scheme between the MPHY and CTL runs at 20°S. It clearly shows that the detrainment of cloud liquid water is increased significantly at 700 hPa over the southern Indian Ocean, southeastern Pacific, and southern Atlantic, which coincides with the increase of low-level clouds seen in Fig. 13. The more realistic convective microphysical properties and associated detrainment lead to the improvement of low-level cloud simulation.
Pressure–longitude section of difference in annual mean detrainment of cloud liquid water (g kg−1 day−1) from the deep convection parameterization between the MPHY and CTL at 20°S.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
Figure 15 shows the annual mean net shortwave (SW) radiation flux at the top of atmosphere (TOA) from CERES2, the biases of CTL and MPHY, and the difference between MPHY and CTL. The CTL produces positive biases of up to 30 W m−2 over the subtropical southern Indian Ocean, subtropical southeastern and northeastern Pacific, and subtropical southern and northern Atlantic. These positive biases correspond to the negative biases in cloud liquid water path and low-level cloud fraction in the same areas. The negative bias in the CTL over northern Indian Ocean corresponds to the positive bias in high-level cloud fraction. The net bias in global mean SW flux at TOA is −10.5 W m−2. When the new microphysics scheme is used, the SW flux is increased by up to 20 W m−2 over the equatorial Indian Ocean, South Africa, and South America, resulting in a better simulation there. The SW flux is reduced significantly over subtropical oceans corresponding to the increase of LWP and low-level cloud fraction, which offsets the positive SW flux biases seen in the CTL. However, it seems overdone compared with CERES2, and the patterns are shifted westward. As a result, the global mean bias and RMSE are increased by 6.56 and 6.69 W m−2, respectively.
Annual mean net shortwave radiation flux at top of atmosphere (W m−2) from (a) CERES2 and differences (b) between CTL and CERES2, (c) between MPHY and CERES2, and (d) between MPHY and CTL.
Citation: Journal of Climate 25, 24; 10.1175/JCLI-D-11-00563.1
The changes in annual mean OLR flux at the top of atmosphere from the CTL to MPHY is not as significant as that of SW flux. There is an increase of OLR over equatorial Indian Ocean, eastern Asia, Africa, Australia, and South America, alleviating the negative bias in the CTL run (not shown). The global mean bias in OLR is reduced by 0.41 W m−2 from −8.29 W m−2 in the CTL, and the RMSE is reduced by 1.03 W m−2 from 13.65 W m−2 in the CTL.
5. Summary and conclusions
This study implemented an efficient physically based two-moment microphysics parameterization scheme for convective clouds (SZ11) in the NCAR CAM5 to improve the representation of convective clouds and its interactions with large-scale clouds and aerosols. In the standard CAM5, all the details of the formation mechanisms of cloud hydrometeor and precipitation inside convective updrafts are omitted. As such, the model is unable to assess the impacts of aerosol–convection interactions. The conversion of cloud water to rainwater in deep convection is represented by a tunable parameter, leading to a large uncertainty in the partitioning of convective condensate between precipitation particles and detrained condensate.
The explicit treatment of mass mixing ratio and number concentration of cloud and precipitation particles in the new microphysics scheme enables the model to account for the impacts of aerosols on convection. When implemented in the CAM5, the scheme links convection to large-scale clouds through convective detrainment of cloud liquid/ice water content and their number concentration, making the representation of convective and large-scale clouds more consistent. It makes it possible to investigate new and important scientific questions related to interactions of convection, clouds, and aerosols and cloud–climate interactions in general.
Two 5-yr simulations of CAM5 with and without the new microphysics parameterization scheme are performed and compared with observations to evaluate the performance of the scheme and its impact on climate simulation. The standard CAM5 in current configuration underestimates the LWC and IWC by up to 80% over tropical oceans compared to CloudSat retrievals, although this bias can be largely mitigated by optimally tuning the autoconversion parameter c0 in the ZM scheme. With the new microphysics parameterization scheme, both the cloud liquid/ice water content and cloud droplet and ice crystal number concentrations are generally in good agreement with available observations. In addition, the distribution of convective vertical velocity estimated in the microphysics scheme is in good agreement with observations. This gives us more confident in simulating aerosol–convection interaction. The budget analysis for cloud droplet number concentrations shows that the primary droplet activation occurs near the cloud base and is much stronger in land convection than in oceanic convection, consistent with observations and detailed parcel model simulations. For ice crystals, the Hallett–Mossop process of ice particle multiplication and homogeneous freezing are two major sources in the lower part of the mixed-phase region and upper troposphere, respectively. This is consistent with the simulation of convection by a cloud-resolving model and observations (Phillips et al. 2005, 2007; Heymsfield et al. 2009). Compared to a sensitivity simulation with reduced aerosol loading, results from the MPHY simulation shows that the microphysics scheme is able to represent the suppression of warm rain formation and enhancement of freezing when aerosol loading is increased, reflecting its ability to represent the effect of aerosols on convective microphysics.
With more realistic convective cloud microphysical properties and their detrainment, the large-scale precipitation simulated by the modified CAM5 is increased by more than 40% in the tropics and 14% in the global mean, with improved precipitation distribution as well. The JJA precipitation is enhanced significantly in the western Pacific warm pool, South China Sea, Bay of Bengal, and the eastern Pacific ITCZ and is weakened in the SPCZ, Maritime Continent, Arabian Peninsula, and Arabian Sea. All of these changes are desired for reducing precipitation biases in the simulation of standard CAM5. As a result, the global mean precipitation bias and RMS errors are reduced by 16% and 7%, respectively.
The deficit in JJA precipitation in the western Pacific, a long-standing issue in the NCAR Community Atmosphere Model, is significantly reduced with the new microphysics scheme. Examination of large-scale circulation and cloud liquid/ice water budget suggests that a thermodynamics–microphysics feedback is responsible for the increase of precipitation. Because the new microphysics scheme produces more cloud liquid water and hence more detrainment, the rainwater in stratiform clouds is increased through enhanced autoconversion and accretion. The enhanced latent heating due to freezing of rain strengthens the upward motion and thus condensation/deposition, resulting in more large-scale precipitation. The extra latent heating from the enhanced condensation/deposition further reinforces upward motion and large-scale precipitation. Additionally, the enhanced upward motion increases the convection activity and therefore convective precipitation and convective cloud liquid water detrainment; the latter in turn reinforces this feedback loop.
CAM5 underestimates the mid- and low-level cloud fraction seriously over ITCZ/SPCZ and the subtropical ocean, by up to 30% and 60%, respectively. Because of enhanced detrainment of cloud liquid water with the new microphysics scheme, the mid- and low-level cloud fraction is increased significantly. Correspondingly, the negative bias in cloud liquid water path over the subtropical ocean in the standard CAM5 is reduced. The effects on high-level cloud fraction are relatively small, due to weak convective mass detrainment above 400 hPa. Corresponding to the improvement of cloud liquid water path and low-level cloud fraction over subtropical ocean, the shortwave radiation flux at the top of atmosphere is reduced significantly, which helps to offset the positive biases of SW flux seen in the standard CAM5. However, it seems overdone compared with CERES2 observations. Note that in this study the model simulated cloud ice water content and cloud fraction are compared to the CloudSat retrievals directly. Because of large uncertainty in satellite retrievals due to the inevitable assumptions and the limitations of satellite sensors, there may be significant ambiguities in the comparisons between model variables and satellite retrievals. Recently, the Cloud Feedbacks Model Intercomparison Project (CFMIP) has developed an integrated satellite simulator, the CFMIP Observation Simulator Package (COSP; Bodas-Salcedo et al. 2011), which simulates the observations of multiple satellite instruments from model variables and facilitates the use of satellite data to evaluate models in a consistent way. In future work, the CFMIP observation simulator will be used to validate the robustness of the impacts of the microphysics scheme.
In conclusion, the implementation of a new microphysics parameterization for convective clouds in CAM5 enables a more accurate description of convection and its interactions with aerosols and large-scale clouds in the model. As a result, the simulation of precipitation and clouds is significantly improved. The microphysics scheme is able to represent the aerosol impacts on convective clouds such as the suppression of warm rain formation and enhancement of freezing when aerosol loading is increased. This makes it possible to investigate convection invigoration by aerosols (Rosenfeld et al. 2008) and aerosol indirect effect on climate. We will pursue this topic further in the near future.
Acknowledgments
This research was supported by the U.S. Department of Energy Office of Science (BER) under Grant DE-SC0000805, National Science Foundation Grants EaSM-1048995 and ATM-0832915, and U.S. National Oceanic and Atmospheric Administration Grant NA08OAR4320894. The computational support for this work was provided by the NCAR Computational and Information Systems Laboratory. The authors thank Ulrike Lohmann for helpful discussion and the anonymous reviewers for their valuable comments that helped improve the presentation of the paper.
REFERENCES
Abdul-Razzak, H., and S. J. Ghan, 2000: A parameterization of aerosol activation 2. Multiple aerosol types. J. Geophys. Res., 105, 6837–6844.
Adler, R. F., and Coauthors, 2003: The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor., 4, 1147–1167.
Albrecht, B., 1989: Aerosols, cloud microphysics, and fractional cloudiness. Science, 245, 1227–1230.
Aleksic N. M., R. D. Farley, and H. D. Orville, 1989: A numerical cloud model study of the Hallett-Mossop ice multiplication process in strong convection. Atmos. Res., 23, 1–30.
Arakawa, A., and W. H. Schubert, 1974: The interactions of a cumulus cloud ensemble with the large-scale environment. Part I. J. Atmos. Sci., 31, 674–701.
Betts, A. K., 1990: Greenhouse warming and the tropical water vapor budget. Bull. Amer. Meteor. Soc., 71, 1465–1467.
Bodas-Salcedo, A., and Coauthors, 2011: COSP: Satellite simulation software for model assessment. Bull. Amer. Meteor. Soc., 92, 1023–1043.
Bolton, D., 1980: The computation of equivalent potential temperature. Mon. Wea. Rev., 108, 1046–1053.
Borovikov, A. M., I. I. Gaivoronskii, V. V. Kostarev, I. P. Mazin, V. E. Minervin, A. K. Khrgian, and S. M. Shmeter, 1963: Fizika Oblakov (Cloud Physics). Israel Program for Scientific Translations, 392 pp.
Boville, B. A., P. J. Rasch, J. J. Hack, and J. R. McCaa, 2006: Representation of clouds and precipitation processes in the Community Atmosphere Model version 3 (CAM3). J. Climate, 19, 2184–2198.
Brenguier, J. L., and W. Grabowski, 1993: Cumulus entrainment and cloud droplet spectra: A numerical model within a two-dimensional dynamical framework. J. Atmos. Sci., 50, 120–136.
Bretherton, C. S., and S. Park, 2009: A new moist turbulence parameterization in the Community Atmosphere Model. J. Climate, 22, 3422–3448.
Byers, H. R., and R. R. Braham, 1949: The Thunderstorm Project. U.S. Weather Bureau, U.S. Department of Commerce Tech. Rep. NTIS PB234515, 287 pp.
Chen, W.-T., C. P. Woods, J.-L. F. Li, D. E. Waliser, J.-D. Chern, W.-K. Tao, J. H. Jiang, and A. M. Tompkins, 2011: Partitioning CloudSat ice water content for comparison with upper-tropospheric ice in global atmospheric models. J. Geophys. Res., 116, D19206, doi:10.1029/2010JD015179.
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.
Cui, Z., K. S. Carslaw, Y. Yin, and S. Davies, 2006: A numerical study of aerosol effects on the dynamics and microphysics of a deep convective cloud in a continental environment. J. Geophys. Res., 111, D05201, doi:10.1029/2005JD005981.
Diehl, K., 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.
Gettelman, A., and Coauthors, 2010: Global simulations of ice nucleation and ice supersaturation with an improved cloud scheme in the Community Atmosphere Model. J. Geophys. Res., 115, D18216, doi:10.1029/2009JD013797.
Hallett, J., and S. C. Mossop, 1974: Production of secondary ice particles during the riming process. Nature, 249, 26–28.
Heymsfield, A. J., A. Bansemer, P. R. Field, S. L. Durden, J. L. Stith, J. E. Dye, W. Hall, and C. Grainger, 2002: Observations and parameterizations of particle size distributions in deep tropical cirrus and stratiform precipitating clouds: Results from in situ observations in TRMM field campaigns. J. Atmos. Sci., 59, 3457–3491.
Heymsfield, A. J., L. M. Miloshevich, C. Schmitt, A. Bansemer, C. Twohy, M. R. Poellot, A. Fridlind, and H. Gerber, 2005: Homogeneous ice nucleation in subtropical and tropical convection and its influence on cirrus anvil microphysics. J. Atmos. Sci., 62, 41–64.
Heymsfield, A. J., A. Bansemer, G. Heymsfield, and A. O. Fierro, 2009: Microphysics of maritime tropical convective updrafts at temperatures from −20° to −60°. J. Atmos. Sci., 66, 3530–3562.
Hudson, J. G., and S. S. Yum, 2001: Maritime–continental drizzle contrasts in small cumuli. J. Atmos. Sci., 58, 915–926.
Iacono, M. J., J. S. Delamere, E. J. Mlawer, M. W. Shephard, S. A. Clough, and W. D. Collins, 2008: Radiative forcing by long-lived greenhouse gases: Calculations with the AER radiative transfer models. J. Geophys. Res., 113, D13103, doi:10.1029/2008JD009944.
Khain, A. P., 2009: Notes on state-of-the-art investigations of aerosol effects on precipitation: A critical review. Environ. Res. Lett., 4, 015004, doi:10.1088/1748-9326/4/1/015004.
Khain, A. P., D. Rosenfeld, and A. Pokrovsky, 2005: Aerosol impact on the dynamics and microphysics of deep convective clouds. Quart. J. Roy. Meteor. Soc., 131, 2639–2663, doi:10.1256/qj.04.62.
Lamb, D., and J. Verlinde, 2011: Physics and Chemistry of Clouds. Cambridge University Press, 584 pp.
Lasher-Trapp, S. G., W. A. Cooper, and A. M. Blyth, 2005: Broadening of droplet size distributions from entrainment and mixing in a cumulus cloud. Quart. J. Roy. Meteor. Soc., 131, 195–220.
Lee, S. S., L. J. Donner, V. T. J. Phillips, and Y. Ming, 2008: The dependence of aerosol effects on clouds and precipitation on cloud-system organization, shear and stability. J. Geophys. Res., 113, D16202, doi:10.1029/2007JD009224.
Li, G., Y. Wang, and R. Zhang, 2008: Implementation of a two-moment bulk microphysics scheme to the WRF model to investigate aerosol-cloud interaction. J. Geophys. Res., 113, D15211, doi:10.1029/2007JD009361.
Lindzen, R. S., M.-D. Chou, and A. Y. Hou, 2001: Does the earth have an adaptive infrared iris? Bull. Amer. Meteor. Soc., 82, 417–432.
Liu, X., J. E. Penner, S. Ghan, and M. Wang, 2007: Inclusion of ice microphysics in the NCAR Community Atmospheric Model version 3 (CAM3). J. Climate, 20, 4526–4547.
Lohmann, U., 2008: Global anthropogenic aerosol effects on convective clouds in ECHAM5-HAM. Atmos. Chem. Phys., 8, 2115–2131.
Lohmann, U., and J. Feichter, 2005: Global indirect aerosol effects: A review. Atmos. Chem. Phys., 5, 715–737.
Lohmann, U., J. Feichter, J. Penner, and & R. Leaitch, 2005: Indirect effect of sulfate and carbonaceous aerosols: A mechanistic treatment. J. Geophys. Res., 105, 12 193–12 206.
Lucas, C., E. Zipser, and M. LeMone, 1994: Vertical velocity in oceanic convection off tropical Australia. J. Atmos. Sci., 51, 3183–3193.
Mason, B. J., and P. R. Jonas, 1974: The evolution of droplet spectra and large droplets by condensation in cumulus clouds. Quart. J. Roy. Meteor. Soc., 100, 23–38.
McFarquhar, G. M., and A. J. Heymsfield, 1996: Microphysical characteristics of three cirrus anvils sampled during the Central Equatorial Pacific Experiment. J. Atmos. Sci., 53, 2401–2423.
Meyers, M. P., P. J. DeMott, and W. R. Cotton, 1992: New primary ice nucleation parameterizations in an explicit cloud model. J. Appl. Meteor., 31, 708–721.
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.
Morrison, H., and W. W. Grabowski, 2008: Modeling supersaturation and subgrid-scale mixing with two-moment bulk warm microphysics. J. Atmos. Sci., 65, 792–812.
Morrison, H., and W. W. Grabowski, 2011: Cloud-system resolving model simulations of aerosol indirect effects on tropical deep convection and its thermodynamic environment. Atmos. Chem. Phys., 11, 10 503–10 523, doi:10.5194/acp-11-10503-2011.
Musil, D. J., and P. L. Smith, 1989: Interior characteristics at mid-levels of thunderstorms in the southeastern United States. Atmos. Res., 24, 149–167.
Neale, R. B., J. H. Richter, and M. Jochum, 2008: The impact of convection on ENSO: From a delayed oscillator to a series of events. J. Climate, 21, 5904–5924.
Park, S., and C. S. Bretherton, 2009: The University of Washington shallow convection and moist turbulence schemes and their impact on climate simulations with the Community Atmosphere Model. J. Climate, 22, 3449–3469.
Phillips, V. T. J., and Coauthors, 2005: Anvil glaciation in a deep cumulus updraft over Florida simulated with an explicit microphysics model. I: The impact of various nucleation processes. Quart. J. Roy. Meteor. Soc., 131, 2019–2046.
Phillips, V. T. J., L. J. Donner, and S. T. Garner, 2007: Nucleation processes in deep convection simulated by a cloud-system-resolving model with double-moment microphysics. J. Atmos. Sci., 64, 738–761.
Pinsky, M., and A. P. Khain, 2002: Effects of in-cloud nucleation and turbulence on droplet spectrum formation in cumulus clouds. Quart. J. Roy. Meteor. Soc., 128, 501–533.
Platnick, S., M. D. King, S. A. Ackerman, W. P. Menzel, B. A. Baum, J. C. Riedi, and R. A. Frey, 2003: The MODIS cloud products: Algorithms and examples from Terra. IEEE Trans. Geosci. Remote Sens., 41, 459–473.
Poellot, M. R., and J. C. Pflaum, 1989: Microphysical characteristics of some convective clouds over Oklahoma. Atmos. Res., 24, 123–136.
Prabha, T. V., A. Khain, R. S. Maheshkumar, G. Pandithurai, J. R. Kulkarni, M. Konwar, and B. N. Goswami, 2011: Microphysics of premonsoon and monsoon clouds as seen from in situ measurements during the Cloud Aerosol Interaction and Precipitation Enhancement Experiment (CAIPEEX). J. Atmos. Sci., 68, 1882–1901.
Pruppacher, H. R., and J. D. Klett, 1997: Microphysics of Clouds and Precipitation. 2nd ed. Kluwer, 997 pp.
Ramanathan, V., and W. Collins, 1991: Thermodynamic regulation of ocean warming by cirrus clouds deduced from observations of the 1987 El Niño. Nature, 351, 27–32.
Rasch, P. J., and J. E. Kristjansson, 1998: A comparison of the CCM3 model climate using diagnosed and predicted condensate parameterizations. J. Climate, 11, 1587–1614.
Rennó, N. O., K. A. Emanuel, and P. H. Stone, 1994: Radiative-convective model with an explicit hydrologic cycle 1. Formulation and sensitivity to model parameters. J. Geophys. Res., 99, 14 429–14 441.
Rosenfeld, D., 1999: TRMM observed first direct evidence of smoke from forest fires inhibiting rainfall. Geophys. Res. Lett., 26, 3105–3108.
Rosenfeld, D., and I. M. Lensky, 1998: Satellite-based insights into precipitation formation processes in continental and maritime convective clouds. Bull. Amer. Meteor. Soc., 79, 2457–2476.
Rosenfeld, D., U. Lohmann, G. B. Raga, C. D. O’Dowd, M. Kulmala, S. Fuzzi, A. Reissell, and M. O. Andreae, 2008: Flood or drought: How do aerosols affect precipitation? Science, 321, 1309–1313.
Salzmann, M., Y. Ming, J.-C. Golaz, P. A. Ginoux, H. Morrison, A. Gettelman, M. Krämer, and L. J. Donner, 2010: Two-moment bulk stratiform cloud microphysics in the GFDL AM3 GCM: Description, evaluation, and sensitivity tests. Atmos. Chem. Phys., 10, 8037–8064.
Schumacher, C., and R. A. Houze Jr., 2003: Stratiform rain in the tropics as seen by the TRMM precipitation radar. J. Climate, 16, 1739–1756.
Seethala, C., and Á. Horváth, 2010: Global assessment of AMSR-E and MODIS cloud liquid water path retrievals in warm oceanic clouds. J. Geophys. Res., 115, D13202, doi:10.1029/2009JD012662.
Seifert, A., and K. D. Beheng, 2006: A two-moment cloud microphysics parameterization for mixed-phase clouds. Part 2: Maritime vs. continental deep convective storms. Meteor. Atmos. Phys., 92, 45–66.
Slawinska, J., W. W. Grabowski, H. Pawlowska, and H. Morrison, 2012: Droplet activation and mixing in large-eddy simulation of a shallow cumulus field. J. Atmos. Sci., 69, 444–462.
Solomon, S., D. Qin, M. Manning, M. Marquis, K. Averyt, M. M. B. Tignor, H. L. Miller Jr., and Z. Chen, Eds., 2007: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp.
Song, X., and G. J. Zhang, 2011: Microphysics parameterization for convective clouds in a global climate model: Description and single-column model tests. J. Geophys. Res., 116, D02201, doi:10.1029/2010JD014833.
Squires, P., 1956: The microstructure of cumuli in maritime and continental air. Tellus, 8, 443–444.
Squires, P., 1958: The microstructure and colloidal stability of warm clouds. Part I—The relation between structure and stability. Tellus, 10, 256–261.
Stephens, G. L., and Coauthors, 2002: The CloudSat mission and the A-Train. Bull. Amer. Meteor. Soc., 83, 1771–1790.
Su, C.-W., S. K. Krueger, P. A. McMurtry, and P. H. Austin, 1998: Linear eddy modeling of droplet spectral evolution during entrainment and mixing in cumulus clouds. Atmos. Res., 47–48, 41–58.
Tao, W.-K., X. Li, A. Khain, T. Matsui, S. Lang, and J. Simpson, 2007: The role of atmospheric aerosol concentration on deep convective precipitation: Cloud-resolving model simulations. J. Geophys. Res., 112, D24S18, doi:10.1029/2007JD008728.
Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon. Wea. Rev., 117, 1779–1800.
Twomey, S., 1977: The influence of pollution on the shortwave albedo of clouds. J. Atmos. Sci., 34, 1149–1152.
Waliser, D. E., and Coauthors, 2009: Cloud ice: A climate model challenge with signs and expectations of progress. J. Geophys. Res., 114, D00A21, doi:10.1029/2008JD010015.
Wallace, J. M., and P. V. Hobbs, 2006: Atmospheric Science: An Introductory Survey. 2nd ed. Elsevier, 483 pp.
Wang, Z., and K. Sassen, 2001: Cloud type and macrophysical property retrieval using multiple remote sensors. J. Appl. Meteor., 40, 1665–1682.
Warner, J., 1969: The microstructure of cumulus cloud. Part II. The effect on droplet size distribution of the cloud nucleus spectrum and updraft velocity. J. Atmos. Sci., 26, 1272–1282.
Wielicki, B. A., B. R. Barkstrom, E. F. Harrison, R. B. Lee, G. L. Smith, and J. E. Cooper, 1996: Clouds and the Earth’s Radiant Energy System (CERES): An Earth Observing System experiment. Bull. Amer. Meteor. Soc., 77, 853–868.
Winker, D. M., J. Pelon, and M. P. McCormick, 2003: The CALIPSO mission: Spaceborne lidar for observation of aerosols and clouds. Lidar Remote Sensing for Industry and Environment Monitoring III, U. N. Singh, T. Itabe, and Z. Liu, Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 4893), 1–11.
WMO, 2008: Guide to meteorological instruments and methods of observation. 7th ed. World Meteorological Organization Tech. Rep. 8, 681 pp.
Wood, R., C. S. Bretherton, D. Leon, A. D. Clarke, P. Zuidema, G. Allen, and H. Coe, 2011: An aircraft case study of the spatial transition from closed to open mesoscale cellular convection over the southeast Pacific. Atmos. Chem. Phys., 11, 2341–2370, doi:10.5194/acp-11-2341-2011.
Xie, P., and P. A. Arkin, 1997: Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs. Bull. Amer. Meteor. Soc., 78, 2539–2558.
Zhang, G. J., and N. A. McFarlane, 1995: Sensitivity of climate simulations to the parameterization of cumulus convection in the Canadian Climate Centre general circulation model. Atmos.–Ocean, 33, 407–446.