Two runs of 1000-year integrations were also done under the same purely periodic annual forcing to get statistically reliable frequency distributions of the interannual variations. The distributions for the topographic amplitude of 1000 m (N.H.-like) are positively skewed in autumn and bimodal in winter. On the other hand, those for 500 m (S.H.-like) have positive skewness for a longer months from autumn to spring and extremely large skewness is found in March. The time-series data analysis shows that the occurrence of the warm winters in the skewed distributions is well described by a Poisson process. That is, the warm winters occur at random from year to year, and preferred periodicity does not exist in the present internal interannual variability.
Based on these results and some perturbation experiments, the importance of nonlinear dynamical perspective is discussed, particularly to appreciate the effects of small-amplitude external interannual forcings such as the solar cycle, cooling trend in the stratosphere and so on. The existence of bimodality(or, multiple metastable states) in our troposphere-stratosphere coupled system is the key point.