Several unresolved issues in numerical modelling of geophysical flows

This paper discusses several issues in the numerical modelling of atmospheric flow which have been left unresolved. These include the influence of partial model resolution of features that are represented in a model by only a few grid points and the implication of this inadequacy in terms of the parametrization of subgrid-scale processes. As a result of nonlinear interactions, this lack of resolution requires the introduction of new parametrization techniques such as “backscatter”, as well as an assessment of the predictability and parametrizability of atmospheric flows as a result of the limited spatial resolution. Examples of these issues are presented in the paper. Among the conclusions is that for large eddy simulations a stochastic subgrid-scale parametrization is needed, while for well-defined surface forced mesoscale systems such as sea breezes, an ensemble-based subgrid-scale representation is adequate. Also introduced in this paper is an additional physical realism to atmospheric models (i.e., the compressibility effect) which provides an additional modelling degree of freedom to internal and externally imposed perturbations.