Abstract
A review is presented of spatial aspects of the numerical methods used in weather prediction models. Considerations that have and are being made in distinguishing among various finite-difference techniques are summarized. Properties that have been examined in order to arrive at an estimate and/or at an understanding of the skill numerical schemes have in simulating atmospheric dynamical processes are listed and discussed. A number of points are illustrated by brief presentations of some as yet unpublished results of the authors. One of these concerns the difference in spectral properties of the fields simulated by two horizontal advection schemes. Another is addressed at error calculations done for a recent “θ-conserving” sigma system pressure gradient force scheme. Finally, issues presently attracting interest in horizontal as well as in vertical differencing are reviewed, and some recent results are summarized.
