MRI•GCMによる1984年夏の長期予報実験

Abstract

The impact of the sea surface temperature (SST) anomalies and cumulus parameterizations is investigated for the case of the 1984 northern summer with the use of the MRI general circulation model. Integrations are performed either with the observed SST for 1984 or with the chmatological SST, either with the original Arakawa-Schubert cumulus parameterization (the A-model) or with a modified scheme (the M-model) which imposes an additional constraint between the minimum entrainment rate and the depth of the predicted planetary boundary layer.Intraseasonal oscillations in the low latitudes are better simulated with the M than with the A. However, the forecast skill with the M depends sensitively on the initial conditions. In one case, the observed eastward moving wave in the velocity potential field at 200mb over the equator is forecasted well, both in its phase and amplitude up to 20 days. Using different initial conditions, the forecast skill for the transient fields was poor. The M-model also has many advantages over the A-model in its climate simulation. It has simulated the Pacific subtropical high in the proper position and succeeded in simulating a Baiu-like rain band. The M-model has a stronger monsoon activity and a stronger sensitivity to the boundary forcing (here the SST anomalies) than the A-model, due to more unstable stratification. The impact of the SST anomaly is better simulated with the M than with the A.Monthly mean forecast skill in the low latitudes varies from month to month, but overall useful skill is found in the three-month mean forecast with the M over the tropics except at the very beginning of the forecast. The SST anomaly impact is not clearly determined, mainly because of the limited forecast samples and partially because of the modest SST anomalies in 1984.