Interannual variability associated with ENSO: Seasonal dependence and interdecadal change

The seasonal dependence of atmospheric short-term climate predictability is studied. This is accomplished by analyzing the output from ensemble integration of two atmospheric general circulation models (AGCM). All the AGCM's integrations analyzed in this study use observed evolution of sea surface temperature (SST) as prescribed boundary forcing. Forced by the interannual variation of SST, the short-term climate predictability of the atmospheric circulation is geographically and seasonally dependent. The predictability is larger in tropics than the extratropics and better in the winter hemisphere than in the summer hemisphere. The predictability is better in Pacific-Atlantic Ocean section compared to the section of Indian Ocean-Asian Monsoon region. Special attention is drawn to the fact that the 1982/83 strong El Nino event produced stronger atmospheric predictability than the 1988/89 strong La Nina event during the northern spring while the opposite occurred during the northern autumn. Also, both models show weakest predictability in the Northern Hemisphere during the boreal autumn.; It is hypothesized that the seasonal and geographical dependence of the SST-forced atmospheric predictability may be explained by the following mechanisms: (1) Annual-cycle/ENSO interaction over the tropical Pacific Ocean; (2) Non-linear effects of hydrological processes associated with the annual-cycle/ENSO interaction; (3) Dynamic effects of meridional wind in planetary wave propagation.; The interdecadal variation of the strength of ENSO is also studied. An attempt is made to understand why the strength of ENSO was weak during the period of a few decades from 1920s to 1950s. It is hypothesized that the interdecadal change in the strength of ENSO is caused by the interactions among climate oscillations of different time-scales associated with interannual variability of the Monsoon and the Pacific Ocean. To verify this hypothesis, characteristic time-scales of the interdecadal variations in the global SST pattern and precipitation at land-stations are first examined by applying wavelet, empirical orthogonal function (EOF), and other statistical analyses to the hundred-year data time series. It is shown that in the tropical Pacific Ocean there exist interdecadal oscillations with periods of about 12-40 years. Maximum power exists at about 20 years in the tropical Pacific. The strongest signal in the Indian monsoon and the North Pacific Ocean region is the multidecadal oscillation with the "period" of about 80 years or longer. Since the data length is only about a hundred years, it is impossible to determine whether the 80-year period is significant and persistent. However, it is clear that some very slow multidecadal oscillation occurs with a large amplitude in these two regions. In addition to the very slow multidecadal oscillation, the Indian monsoon also has a decadal oscillation with quasi-period of about 10-15 years while the North Pacific Ocean has an interdecadal oscillation with quasi-period of about 30-40 years.; Results of cross-wavelet analysis show that the phase-relation among climate variations in the tropical Pacific, Northern Pacific, and monsoon were quite different during the period from 1920s to 1950s comparing to the decades before and after.