Abstract: The heterogeneity of urban surface leads to high intra-city spatial variability of air temperature (Ta), impacting the accurate estimation of canopy urban heat island (CUHI) intensity and heat mitigation strategies. Using the hourly air temperature observation data from 153 automatic weather stations in Nanjing during the summer of 2019, the spatial variability of Ta and its influencing factors are analyzed, and the impact of Ta spatial variability on the calculation of the CUHI intensity is quantified. The results show significant summertime intra-city spatial variability of Ta in Nanjing, especially at nighttime, with the coefficient of spatial variation being 2.2 times that of daytime. Area fraction of buildings (R＝ 0.44, p < 0.001) and vegetation coverage (R＝－ 0.74, p < 0.001) are the most critical factors influencing the spatial patterns of daytime and nighttime Ta, respectively. In terms of building form factors, daytime Ta correlates significantly with the fraction of building roof projection area (R＝ 0.35, p < 0.001), while nighttime Ta is more correlated with building height (R＝ 0.69, p < 0.001). Population density effects on the spatial distribution of Ta are nonlinear. Ta is more sensitive to anthropogenic activities when population density is less than 150 people/km². Uncertainty of CUHI intensity caused by the selection of urban and suburban stations is 0.15 ℃ during the daytime and 0.28 ℃ at nighttime, accounting for 44% and 37% of CUHI intensity (daytime: 0.34 ℃, nighttime: 0.76 ℃), respectively. When using the local climate zone (LCZ) method, the strongest CUHI intensity occurs in LCZ 10 (industry, 1.85 ℃) during the daytime and LCZ 1 (dense high-rise building, 0.61 ℃) at nighttime. When the building density is the same, the CUHI intensity in the high-rise building area is 0.55－ 1.15 ℃ stronger than that in low-rise building area. For high-rise building, the CUHI intensity in the dense building area is 0.43 ℃ stronger than that in the open areas.