Abstract
The rise in surface air temperature is one of the most profound manifestations of global warming, and its annual cycle in mid-latitudes, has been a key focus in climate related research. This study aims to assess the performances of the CMIP6 models in terms of simulating historical and predicting future annual cycles of surface air temperature over China. The historical (1961 - 2014) and future monthly temperatures (2015 - 2100) under three shared socioeconomic pathways (SSP1-2.6, SSP2-4.5, and SSP5-8.5) from 13 models were analyzed. The position of the maximum (minimum) value of the temperature (shorted as phase-max and phase-min, respectively) and amplitude were obtained through a Fourier smoothing of the monthly temperature, and their long-term trends were calculated and compared with historical CMIP6 monthly data and gauge observations from meteorological stations. It was found that under the three future shared socioeconomic pathways, the amplitude of the temperature annual cycle decreased in the future period compared to historical stage. It shows an obvious north-south gradient in the long-term phase and amplitude trends. The mean trends were 0.11, 0.12 and 0.36 days/10 years under SSP1-2.6, SSP2-4.5, and SSP5-8.5 for phase-max of the study area, respectively, larger than those of 0.04, 0.04, and 0.06 days/10 years for phase-min. Regionally, the amplitude in the northeastern China (NE), northern China (NC), southern China (SC) and southwestern China (SW) regions decrease under all the scenarios, which was related to the asymmetric increase of temperature between the winter and summer. Besides, there were large discrepancies in the phase and amplitude trends among different CMIP6 models.