Intrinsic error growth in a large-domain Eta regional model

[...]as far as the instability and error growth are concerned, boundary conditions (which are necessary to define the planetary-scale solution inside the domain) are felt by the system as perturbations that do not modify substantially the internal dynamics. [...]a domain size could probably be considered as optimal with respect to the competing requirements of having a realistic solution of the model at hand and a low computer time demand. [...]an important check of the quality of this model size should be provided by introducing the natural variability at the boundaries, using, for instance, the reanalyses that are provided at NCEP and at the European Centre for Medium-Range Weather Forecasts.