Performance of a three-dimensional semi-Lagrangian advection scheme for idealized steep topography

A three-dimensional, semi-Lagrangian (3DSL) advection scheme is introduced into the Advanced Research Weather Research and Forecasting (WRF-ARW) model. Current operational numerical weather prediction models are running at increasingly high model resolution, but those using an Eulerian advection scheme are required to satisfy a numerical stability criterion and require a smaller time step than is needed to accurately resolve phenomena at the scale of interest. With increasing resolution, steep terrain is also better resolved. Steep topography enhances vertical velocities, and requires either the model's vertical resolution to decrease (which is undesirable) or the time step to be reduced in order to attain numerical stability. The performance, including the computational time and the representation of several idealized atmospheric processes over steep topography, of the 3DSL scheme will be compared with the existing WRF-ARW Eulerian scheme. The focus will be on the performance of the vertical trajectory computational algorithm in the 3DSL scheme.