• Intermediate-complexity model;
  • data assimilation;
  • singular vectors;
  • predictability;
  • geostrophy


Atmospheric models of intermediate complexity play an important role when studying atmospheric phenomena. Their complexity is between highly truncated low-dimensional ‘toy’ models and modern general circulation or numerical weather prediction models. By design, computational cost associated with intermediate models is much reduced while at the same time some important aspects of atmospheric behaviour are still reasonably realistically described. Performing numerical experimentation with such models in the contexts of data assimilation, predictability, and atmospheric dynamics can produce informative results regarding those aspects for comparatively low cost. Nevertheless, as with any model-based study, the degree to which results so obtained may be generalized to more realistic conditions remains somewhat uncertain and dependent on the specific questions being considered. An intermediate-complexity model, named AMIC (Atmospheric Model of Intermediate Complexity) based on the nonlinear quasi-geostrophic potential vorticity equation is presented. This global model uses a spectral dynamical core, and contains ‘physical processes’, such as climatological forcing, diffusion, and damping, designed to reasonably match AMIC's behaviour with observed atmospheric properties. While AMIC has variable horizontal and vertical resolution, the properties of AMIC are studied here for two specific resolutions (T45L6 and T106L9) and these are compared against atmospheric properties in terms of energy spectra, time-mean and transient behaviour, and singular-vector perturbation growth. The model's behaviour is reasonably realistic, except for its transient activity being somewhat weak, especially in the southern (summer) hemisphere. AMIC is also suited for some data assimilation and predictability studies since it contains complete tangent-linear and adjoint models. Copyright © 2008 Royal Meteorological Society