Assessing the reliability of dynamical AGCM models in reproducing the observed atmospheric circulation given the lower boundary conditions, and thus its ability to predict climate, has been a recurrent concern in seasonal-to-interannual climate prediction. The assessments have been carried out in several ways, including the comparison between the leading modes of variability from the observations and the models’ outputs and comparison of the AGCM’s skill with the skill of statistical models. In those studies there was frequently concern about the influence of seasonal variation on the models’ skill, but no analysis has been carried out about the possible variation of the models’ performance throughout the years. Is the long-term variability of the atmosphere and the oceans prone to influence the performance of models, as is the seasonal variability?

In the present study, the seasonal responses of two AGCMs (ECHAM3 and NCEP) to prescribed observed SST are compared to the observed seasonal anomalies (from the Reanalysis NCEP/NCAR) to verify whether the performance is affected by long-term variations. The analysis is based on the simultaneous correlation between series of three-month mean model responses and reanalysis data, averaged over 20º latitude × 40º longitude regions all over the globe, for the period 1950-1994. The possible influence of the interdecadal variability on the models’ performance is assessed through the computation of the global fields of simultaneous correlation coefficients (CC) between 11-year running series of the reanalysis data and of the models’ output. The value considered for each year is a three-month mean of the analyzed parameter. Thus, the seasonal influence on the interdecadal variations can be detected. The variation of the CC is an indication of the interdecadal variation of the models’ skill all over the globe. Also to see whether the models reproduce the interdecadal modulation of the ENSO impacts on the atmospheric circulation, global fields of simultaneous correlation are computed between 11-year running series of the SST in the Niño 3 region and streamfunction, zonal and meridional components of the wind at 200 hPa.

A possible connection between interdecadal modes of SST variability and the decadal/interdecadal variations of the models’ skill was sought. The Empirical Ortogonal Functions (EOFs) of the global field of running correlations for streamfunction at 200 hPa were computed, and the two first principal components were correlated with the 11-year running mean of SST. This gives an indication about the relationship between the interdecadal variability of the models’ skill and the interdecadal variability of SST. The statistical significance of this correlation was assessed by using a Monte Carlo procedure. The interdecadal variation of the modelsx performance is shown to be seasonally dependent. The results show a clear interdecadal modulation of the modelsx skill and its relationship with known interdecadal SST modes of variability like those with maximal realization in North and East Pacific.