Assessment of the dependence of the skill and predictability of seasonal forecasts on boundary conditions of the model

A comparison of estimates of the root-mean-square error (RMSE) and potential predictability index (PPI) is carried out between experiments with observed and "persistent" anomalies of sea surface temperature (SST). The results obtained point to a possible significant bias of seasonal forecasting results in some regions when boundary conditions are introduced by a "persistence" procedure, particularly for summer T (850). Indirect evidence of the influence of extratropical SST anomalies points to their possible role in seasonal forecasts, which is more substantial in the summer season. Although the conclusions should rather be regarded as preliminary ones because of a limited size of the sample, it is nonetheless certain that the influence of boundary conditions governing the signal becomes more significant in summer because of a decrease in the instability of the internal atmospheric dynamics.