A regional climate model, HIRHAM4, has been used with perfect boundary conditions from the ECMWF reanalysis project with an integration area covering the Mediterranean area. The model has been applied iteratively to one particular year, 1982, in the sense that boundary conditions corresponding to that year were applied for several "years" (iterations); soil temperature and moisture have been allowed to evolve through the simulation, whereas all atmospheric variables have been reinitialized for each iteration. This way the soil variables, in particular temperature, approach a stationary slate with a vanishing net energy and water exchange with the atmosphere. This approach over most of the integration area follows an exponential function after a transient period of not more than 1 year. This has the consequence that the asymptotic soil temperature can be calculated after just 4 iterations of the model, resulting in an estimate much closer to equilibrium than the value after 4 years. Soil water is much more noisy and reaches approximate equilibrium after a few iterations; the extrapolation procedure for this quantity is only recommended for arid regions. In general, the present experiment constitutes a method to distinguish between the transient behavior of slowly varying fields and the interannual variability. The asymptotic stale is independent of the initial soil state before the first iteration and is therefore a property only of the model and of the forcing through the integration period. This way of obtaining a "canonical" soil state may, for instance, have uses for model inter-comparisons, as the arbitrary influence from initial transients has been removed.
