Recent reports using a one-dimensional thermodynamic parcel model have suggested that increases in aerosol number concentration may invigorate deep convection by first mitigating the autoconversion process until air parcels reach the freezing level.  This, in turn, would lead to an increase in ice water aloft and the potential for enhanced upward heat transport as a result of phase changes, hence leading to invigorated convection.  Other studies have alluded to the idea that an increase in aerosol loading may act to increase cloud top height, the direct result being an increase in the liquid water content and thus in cumulative precipitation.  Furthermore, two-dimensional cloud resolving model (CRM) studies using a highly simplified droplet activation scheme have demonstrated an increase in precipitation with an increase in aerosol concentration.  Here we study the effect of aerosol perturbations on deep convection employing the Weather Research and Forecasting (WRF) model as a three-dimensional CRM with a two-moment six-class bulk microphysics scheme, and a state-of-the-art activation scheme for liquid cloud droplets based on Köhler Theory and Population Splitting, as well as a physically-based parameterization for homogeneous and heterogeneous freezing of cloud droplets.  We perform idealized simulations of deep convection over a wide range of aerosol concentrations (i.e., 10² cm^-3 to 10⁴ cm^-3), which encompasses clean maritime conditions to polluted continental conditions, respectively.  The detailed model calculations reveal that the aerosol effect on precipitation in deep convective clouds depends strongly on the ambient water vapor mixing ratio near the surface. When it is assumed that the increase in aerosol number concentration acts solely to increase the number of cloud condensation nuclei (CCN), we find, in agreement with previous studies, that there exist two regimes:  (1) In relatively dry air an increase in aerosol concentration leads to a decrease in precipitation and (2) In relatively moist air an increase in aerosol concentration leads to an increase in precipitation.  We also show that the increase in precipitation is mostly due to an increase in cloud top height and cloud depth for increased aerosol loading. 

Furthermore, the use of a physically-based freezing parameterization allows one to analyze the validity of the assumption that all cloud droplets freeze at the -4°C isotherm as well as the effect of the ambient ice nuclei (IN) concentration on deep convection.  A large fraction (i.e., ≥10%) of condensed moisture is found to remain liquid for T < -4°C in the simulated deep convective clouds.  Moreover, our results also show that although the cloud ice mass fraction changes by ~1% for a two-order of magnitude increase in the IN concentration, the cumulative precipitation is altered significantly, potentially qualitatively altering previous conclusions mentioned above for aerosol effects on precipitation from deep convective clouds.  For example, in relatively dry air, as alluded to above, if the increase in aerosol concentration serves to increase only the CCN population, simulations suggest that the increase in aerosol concentration will act to decrease the cumulative precipitation.  However, for the same relatively dry air, if the increase in aerosol concentration serves to increase both CCN and IN populations, the cloud ice mass fraction increases slightly for -20°C ≤ T ≤ -4°C, resulting in an increase in cumulative domain-averaged precipitation.  Hence, the sign of the aerosol-induced precipitation effect in deep convective clouds is strongly dependent on the availability effective IN.