Interdecadal Variation Of Enso Predictability In Multiple Models

Youmin Tang,Ziwang Deng,Xiaobing Zhou,Yanjie Cheng,Dake Chen

DOI: https://doi.org/10.1175/2008JCLI2193.1

IF: 5.38

2008-01-01

Journal of Climate

Abstract:In this study, El Nino-Southern Oscillation (ENSO) retrospective forecasts were performed for the 120 yr from 1881 to 2000 using three realistic models that assimilate the historic dataset of sea surface temperature (SST). By examining these retrospective forecasts and corresponding observations, as well as the oceanic analyses from which forecasts were initialized, several important issues related to ENSO predictability have been explored, including its interdecadal variability and the dominant factors that control the interdecadal variability.The prediction skill of the three models showed a very consistent interdecadal variation, with high skill in the late nineteenth century and in the middle-late twentieth century, and low skill during the period from 1900 to 1960. The interdecadal variation in ENSO predictability is in good agreement with that in the signal of interannual variability and in the degree of asymmetry of ENSO system. A good relationship was also identified between the degree of asymmetry and the signal of interannual variability, and the former is highly related to the latter. Generally, the high predictability is attained when ENSO signal strength and the degree of asymmetry are enhanced, and vice versa. The atmospheric noise generally degrades overall prediction skill, especially for the skill of mean square error, but is able to favor some individual prediction cases. The possible reasons why these factors control ENSO predictability were also discussed.