Abstract
This study examines the skills in simulating interannual variability of northwestern Pacific (NWP) summer climate in 12 atmospheric general circulation models (AGCMs) attending the Atmospheric Model Intercomparison Project phase 2 (AMIP II). The models show a wide range of skills, among those version 1 of the Hadley Centre Global Atmosphere Model (HadGAM1) showed the highest fidelity and thus may be a better choice for studying East Asian–NWP summer climate. To understand the possible causes for the difference among the models, five models {HadGAM1; ECHAM5; the Geophysical Fluid Dynamics Laboratory Atmosphere Model, version 2.1 (AM2.1); Model for Interdisciplinary Research on Climate 3.2, high-resolution version [MIROC3.2(hires)]; and the fourth-generation National Center for Atmospheric Research Community Atmosphere Model (CAM3)} that have various skill levels, ranging from the highest to the moderate to the minor, were selected for analyses. The simulated teleconnection of NWP summer climate with sea surface temperatures (SSTs) in the tropical Indian and Pacific Oceans was first compared. HadGAM1 reproduces suppressed (intensified) rainfall during El Niño (La Niña) events and captures well the remote connection with the tropical Indian Ocean, while the other models either underestimate [ECHAM5, AM2.1, MIROC3.2(hires)] or fail to reproduce (CAM3) these teleconnections. The Walker cell and diabatic heating were further compared to shed light on the underlying physical mechanisms for the difference. Consistent with the best performance in simulating interannual rainfall, HadGAM1 exhibits the highest-level skill in capturing the observed climatology of the Walker cell and diabatic heating. These results highlight the key roles of the model’s background climatology in the Walker cell and diabatic heating, thus providing important clues to improving the model’s ability.