Comparative analysis of CMIP3 and CMIP5 global climate models for simulating the daily mean, maximum, and minimum temperatures and daily precipitation over China

• Published in: Journal of geophysical research. Atmospheres Vol. 120; no. 10; pp. 4806 - 4824
• Main Authors: Sun, Qiaohong, Miao, Chiyuan, Duan, Qingyun
• Format: Journal Article
• Language: English
• Published: 27.05.2015
• Subjects: China; CMIP3; CMIP5; precipitation; temperature
• DOI: 10.1002/2014JD022994
• ISSN: 2169-897X

This study assesses the simulations of the daily mean, maximum, and minimum temperatures and daily precipitation over China during the period 1990–1999, based on phase 3 and phase 5 of the Coupled Model Intercomparison Project (CMIP3 and CMIP5). Fourteen CMIP3 models and 14 CMIP5 models were investigated over eight regions across China. Skill scores quantifying the match between the simulated and observed probability density functions (PDFs) were applied to evaluate the performance of the models. For daily mean, maximum, and minimum temperatures, the results revealed that CMIP3 and CMIP5 models captured the basic pattern of the observed PDFs in all regions. However, the probabilities at lower values were overestimated in most models. In all regions except the west of Northwest China (region 7), all CMIP5 models captured more than 80% of the observed PDFs. Compared with performance at the annual time scale, the models tended to perform relatively worse over the period June to August. The performances of the CMIP5 and CMIP3 models were not as good for daily precipitation as for daily temperature, and the skill scores for precipitation were generally lower than 0.7 in all regions. The amount of drizzle (daily precipitation < 5 mm) was overestimated notably in all regions. The amount of very heavy precipitation (daily precipitation ≥ 20 mm) tended to be underestimated in humid regions but overestimated in arid regions. Compared with CMIP3, CMIP5 models showed some improvements in the simulation of daily mean, maximum, and minimum temperatures, but there was a lack of apparent improvement for simulation of daily precipitation. Key Points Most of the GCMs overestimate the probabilities of the lower value of variables GCMs' performances in daily precipitation are not as good as daily temperature CMIP5 improves the simulation of daily temperature when compared with CMIP3