Impact of nesting methods on model performance

Abstract

Nesting means to use time dependent boundary values from a coarser grid simulation in a high-resolution model run. The aim of this work is to evaluate how often the forcing data have to be updated for representing the large-scale development in a realistic way. For our study, the mesoscale model METRAS is used in different resolutions: A coarse grid run provides the forcing fields with update intervals of different length for high-resolution simulations. Besides intervals of a constant length, we used a method to adapt the output times from the coarse-grid run to the time scales, in which the simulated fields change. As first results, we have seen that the simulations are very sensitive to the nesting and the forcing data. Additionally, shorter update intervals for the forcing fields lead to higher model performance and the interval length should be similar for consecutive intervals. © 2011 Springer Berlin Heidelberg.