Reproducibility of seasonal ensemble integrations with ECMWF GCM and its association with ENSO

Summary This study evaluates seasonal climate potential predictability with a dataset of nine-member ensemble seasonal integrations produced by the ECMWF GCM for the ERA-15 period (1979–1993). The methodology used here is the measure of the ensemble reproducibility for a particular season defined by Yang et al (1998). High reproducibility reflects the dominant role of the lower boundary forcing in seasonal climate anomaly, indicating good potential predictability. Spatial patterns of the reproducibility for selected variables are documented, which exhibit obvious regionality and seasonality. Such variables are always highly reproducible over most of the tropical regions. Over the northern extratropics, primary reproducible information, taking the 500hPa geopotential height for example, is found over the PNA region during winter while over most of Asia during summer. Winter has the largest reproducible area; autumn has the lowest, while summer and spring are in between. Association of the reproducibility with ENSO events was examined regionally. Internal variances due to ensemble spread were broken down for individual years, and the reproducibility was computed for four categories: El Niño, La Niña, ENSO, and non-ENSO years. The reproducibility during winter especially over the PNA region is insensitive to ENSO events, with exceptions over the tropical western Pacific, central Siberia and Western Europe. Contrarily, ENSO events have significant impacts on the reproducibility over the southwest USA and most of Asian monsoon region during summer. These results suggest that ENSO events may not be helpful to the seasonal climate predictability over the PNA region during winter, but they may increase predictable information over many regions of the northern extratropical continents during summer.