资源环境科技发展态势分析平台

The impact of internal tropical Pacific variability on global mean surface temperature (GMST) is quantified using a multimodel ensemble. A tropical Pacific index (TPI) is defined to track tropical Pacific sea surface temperature (SST) variability. The simulated GMST is highly correlated with TPI on the interannual time scale but this correlation weakens on the decadal time scale. The time-scale dependency is such that the GMST regression equation derived from the observations, which are dominated by interannual variability, would underestimate the magnitude of decadal GMST response to tropical Pacific variability. The surface air temperature response to tropical Pacific variability is strong in the tropics but weakens in the extratropics. The regression coefficient of GMST against TPI shows considerable intermodel variations, primarily because of differences in high latitudes. The results have important implications for the planned intercomparison of pacemaker experiments that force Pacific variability to follow the observed evolution. The model dependency of the GMST regression suggests that in pacemaker experiments-model performance in simulating the recent "slowdown'' in global warming-will vary substantially among models. It also highlights the need to develop observational constraints and to quantify the TPI effect on the decadal variability of GMST.

Compared to GMST, the correlation between global mean tropospheric temperature and TPI is high on both interannual and decadal time scales because of a common structure in the tropical tropospheric temperature response that is upward amplified and meridionally broad.