Abstract
The potential predictability (PP) of seasonal-mean 200-hPa geopotential height (Z200) anomalies in the Pacific-North American (PNA) region is examined for El Nino and La Nina separately by using 50 ensemble members of twentieth-century AGCM simulations. Observed sea surface temperature (SST) is prescribed for the period 1870-2009, and 14 El Nino and La Nina years after 1900 are selected for the present study. The domain-averaged value of PP for Z200 in the PNA region, as measured by the signal-to-noise ratio, for El Nino is about 60% larger than that of La Nina. Such a large PP is mainly due to a larger signal and partly to less noise during El Nino compared to that during La Nina . The transient eddy feedback to the PNA circulation anomalies is stronger during El Nino events (about 50%) than that during La Nina, and this difference in the transients contributes significantly to the different Z200 signals in the PNA region. The noise variance of the transients during El Nino is about 17% smaller than during La Nina, and thus transients play an important role in the reduction of Z200 noise during El Nino. Idealized experiments with the same spatial pattern but different signs of SST anomalies confirm the results mentioned above. Moreover, these experiments with several different amplitudes of positive and negative phases of tropical Pacific SST anomalies show that signals of Z200 and transients are proportional to precipitation anomalies in the tropical Pacific, and noises of Z200 for El Nino cases are somewhat smaller than the corresponding values of La Nina.