2022 年 18 巻 p. 96-103
The technique for composing a small subset of global climate models is critical to provide climate scenarios for impact and adaptation studies of regional climate changes. A recent study developed a novel statistical method for selecting a mini-ensemble of five climate models from the Coupled Model Intercomparison Project Phase 6 for widely capturing different future projections of Japanese climate across eight atmospheric variables at the surface. However, it remains unclear which mini-ensemble model contributes to efficiently covering the full projection ranges. Here, we rank each mini-ensemble projection around Japan among a full ensemble, showing that the selected five models capture the full ranges without systematic biases, except for relative humidity. Furthermore, we find that the widespread global warming level contributes to covering well the projection uncertainties in the daily-mean, maximum and minimum air temperatures and downward longwave radiation but not in precipitation, solar radiation, relative humidity, or wind speed. As the last four variables are sensitive to various factors, such as large-scale circulation and aerosol-forcing changes, rather than global-mean temperature changes, the model selection method featured here is preferable for capturing the wide future projection ranges in Japan.