R-CCS Cafe 第206回 第3部

Deep convective clouds often cause severe weather disasters and, at the same time, can organize into large convective systems which impact the Earth’s climate significantly. However, their representation in numerical atmospheric models is not perfect, and thus, accurate prediction of weather and climate is still challenging. As a team member of Computational Climate Science Research team, I work on studies on numerical simulations of deep convective clouds. In this presentation, I talk about two topics related to this issue. The first topic is the numerical convergence of simulations of organized convective systems. To investigate the resolution necessary for reasonable simulations of convective systems, we conducted a grid-refinement experiment using the K-computer. We found that the structure of an updraft ensemble in an organized convective system converges at progressively smaller scales as the grid spacing is reduced. The numerical convergence is achieved when the grid spacing becomes as small as 1/20–1/40 of the updraft radius. The second topic is parameter estimation for a cloud microphysics scheme which plays a major role in generating convective clouds in atmospheric simulations. To establish a procedure for estimating optimal parameter values for the cloud microphysics scheme, we performed an idealized experiment for parameter estimation by using an ensemble Kalman filter-based method. In the experiment, the true parameter value was successfully estimated by assimilating radar reflectivity data calculated from a convective system simulation. We show that the precision of parameter estimation can be maximized by giving an optimal ensemble spread for the parameter to be estimated.