Abstract
A case study of long-range forecasting with the use of the Meteorological Research Institute general circulation model (MRI·GCM-I) is described. In this study, not only the predictability of anomalies on the monthly mean basis but also the time evolution of the initial condition (IC) impact and of the sea-surface temperature anomaly (SSTA) impact are discussed up to two months. May 1 of 1983 is chosen as the initial time, when El Niño was still active over the equatorial eastern Pacific Ocean.
In the tropics, the SSTA impact dominates over the IC impact already in the first month (May). Large-scale monthly mean anomalies are predicted accurately. In the second month (June), good anomaly correlation is confined in the area from the equatorial eastern Pacific Ocean to the equatorial Atlantic Ocean. The IC impact is almost negligible at this stage.
In the extratropics of the Northern Hemisphere (NH), the IC impact is larger than the SSTA impact in the first month, although the SSTA impact is already comparable with the IC impact in the Southern Hemisphere. Many monthly mean anomalies are judged significant with a 95% confidence level in the vicinity of their peaks. However, the predicted anomalies in the NH do not have good correspondence with the observed ones. In the second month, the SSTA impact dominates over the IC impact locally. The anomaly correlation coefficient in such an area is better in the second month than in the first.
