Applicability of CMIP5 and CMIP6 Models in China: Reproducibility of Historical Simulation and Uncertainty of Future Projection

• Published in: Journal of climate Vol. 36; no. 17; pp. 5809 - 5824
• Main Authors: Jia, Qimeng, Jia, Haifeng, Li, Yijia, Yin, Dingkun
• Format: Journal Article
• Language: English
• Published: 01.09.2023
• ISSN: 0894-8755; 1520-0442
• DOI: 10.1175/JCLI-D-22-0375.1

Abstract General circulation model (GCM) outputs archived by the Coupled Model Intercomparison Project (CMIP) are widely used for climate projection. A comprehensive evaluation of the applicability of GCMs is crucial for generating reliable projections. However, previous GCM evaluation studies mainly focused on the reproducibility of the historical simulation to the observation data, ignoring the uncertainty involved in the future climate projection. In this study, an innovative and comprehensive index, named the history–future applicability score (HFAS), is proposed for evaluating the applicability of GCMs. Through incorporating the Taylor skill score ( S ) and the analysis of variance (ANOVA) within a general framework, the HFAS can reflect both the historical reproducibility and future uncertainty of GCMs. The HFAS framework is applied to quantitatively evaluating the skills of 28 GCMs (14 from CMIP5 and 14 from CMIP6) in modeling temperature and precipitation at 631 reference sites across China. Results indicate that (i) both CMIP5 and CMIP6 models underestimate the historical temperature of China and overestimate the historical precipitation; (ii) the total uncertainties in future temperature and precipitation projections increase along with time; compared to CMIP5, temperature uncertainty of CMIP6 is narrowed, while the precipitation uncertainty is increased; (iii) taking into account historical reproducibility and future uncertainty, CMIP6 models generally have increased applicability compared to CMIP5; and (iv) the best-performing GCMs in modeling temperature and precipitation of China are ACCESS-CM2 and MPI-ESM-LR, respectively, with the HFAS being 68.4 and 63.5. The findings are helpful for researchers to select locally appropriate GCMs and generate reliable regional climate projection.