Title: Toward a better representation of clouds and precipitation: size-resolved microphysics model and cloud Doppler radar Presenter: Hyunho Lee Abstract: Drizzle, which is a consequence of active collision-coalescence of droplets, plays principal roles in climate and aerosol-cloud interactions. However, despite of remarkable advances in modeling of clouds and precipitation over the last decades, the leading mechanisms for drizzle formation in stratocumulus are still debated. Bulk microphysics schemes parameterize drop size distributions using a few (typically no more than three) quantities and usually adopt somewhat rough approximations to evaluate collisional growth of small droplets. Size-resolved (bin) microphysics schemes have an advantage over bulk schemes on representing drop size distributions by dividing drops into a few tens to hundreds of bins and evaluating microphysical processes associated with the drops in each size bin. Despite such an advantage, bin microphysics schemes usually face so-called Ònumerical diffusionÓ, which refers to an artificial broadening of drop size distributions, analogous to that typically occurs when solving an advection equation in an Eulerian space. Because drizzle formation is highly sensitive to the shapes of drop size distributions, it is essential to suppress the numerical diffusion in a bin microphysics scheme. As part of minimizing the numerical diffusion and obtaining a robust solution from a bin microphysics scheme, we choose three collision schemes and assess the impacts of collision scheme and bin width on numerical diffusion and characteristics of drop size distributions. Using a simple box model, we find that although all the schemes yield an identical converged solution at a very high resolution, rates of convergence are significantly different to each other. One of them converges so slowly that it shows substantial numerical diffusion, and another one violates mass conservation and is only conditionally stable. Based on the evaluation, we recommend one of the three schemes as superior over the others. We further evaluate two of the three schemes in simulating drizzling stratocumulus using a large eddy simulation model. A forward simulator is used to apples-to-apples compare model results to W- band cloud radar observations. Numerical simulations show that the diffusive scheme yields too large Doppler velocity and too dispersed and skewed Doppler spectra compared to observations. The better collision scheme substantially reduces, but not perfectly removed, the biases in Doppler spectra. By continuing assessment of numerics of other processes (e.g., condensation, evaporation, and sedimentation), we expect that we can obtain a better understanding of drizzle formation and accurately evaluate aerosol- cloud interactions.