Since the 20th century, global warming has become a major climate change problem, which significantly affects the sustainable development of the world, China holds the unenviable position of greatly contributing to global warming. Based on the data of 1728 national surface meteorological stations in China from 1959 to 2018, this paper explores the temporal and spatial evolution and non-stationary characteristics of surface temperature in 612 effective grids, and conducts an empirical analysis of Beijing-Tianjin-Hebei, Yangtze River Delta and Pearl River Delta Urban Agglomeration (BTHUA, YRDUA, PRDUA). Several key conclusions are reached: (1) The annual average surface temperature rose at wave type, with an increase of 0.995 ℃ and a warming rate of 0.166 ℃ per decade in China from 1959 to 2018-global warming has become increasingly serious in the last 30 years, the areas of highest warming are mainly concentrated in North, East, Northwest and Northeast China. (2) The surface air temperature has strong auto-correlation, the auto-correlation coefficients of 49.35%-96.08% grids have significant positive correlation at the level of 0.05 in 13 lag periods. (3) The auto-correlation coefficient of 96.08% grids annual average temperature lag 1 period exceeds the critical value of the significance level, the auto-correlation coefficient showed a gradual decreasing trend with increasing lag period, indicating the temperature data had non-stationary characteristics. (4) The surface temperature showed a wave-like upward trend in BTHUA, YRDUA and PRDUA from 1959 to 2018, with temperature increases of 1.124℃, 1.029℃and 1.048℃ respectively, the warming rates were 0.187℃, 0.171℃ and 0.175℃ per decade respectively.