Every model is an imperfect representation of the real world. Understanding the source of errors in models assists the model developer and model user in interpreting the results of the model output. Miguez-Macho et al. (2005) examine the source of model error for simulation over the US with the regional climate model RAMS. They find that lower boundary conditions relating to soil moisture and soil temperature and the resolution of soil properties had a large influence on model error, as did the parameterization used for simulating convection. Interaction of internally generated waves with boundary conditions and long waves entering the domain through the boundaries was as additional source of model error that could be partially corrected by "spectral nudging." Precipitation RMS error compared in the initial simulation and resulting from successively applying improved surface conditions, improved convection scheme, and spectral nudging were, successively, 81 (before improvement), 76, 66, and 55 mm for a 1-month simulation in June.
Reference
Miguez-Macho, G.L. Stenchikov and A. Robock, 2005: Regional climate simulations over North America: Interaction of local processes with improved large-scale flow. J. Climate, 18, 1227-1246.