AbstractA Cimel Sun‐sky‐lunar photometer (CE318‐T) is designed to perform daytime and nighttime photometric measurements and calculate diurnal cycle of aerosol optical depth (AOD). Nevertheless, the determination of nocturnal AOD from CE318‐T requires a precise knowledge of extraterrestrial lunar irradiance, which significantly changes with moon's phase angle (MPA) and lunar libration in a single night. This study evaluated the 1‐year nocturnal AODs at Lanzhou by using three different methods, which were validated by collocated measurements of DIAL Lidar (as a reference) and Cimel software (as a proxy). The results indicated that three independent approaches could implement a similar performance to compute nocturnal AOD near full moon phase (i.e., MPA = 0°) under moderate aerosol loading. The spectral AOD values at nighttime calculated by ROLO Lunar Langley (Robotic Lunar Observatory model) and Sun‐moon Gain Factor (SMGF) methods are significantly underestimated under low moon's illumination or high MPA (MPA < −47° or MPA > 47°) and distinctly dependent on MPA. The RIMO correction factor (RCF) (ROLO Implementation for Moon photometry Observation correction factor) method could compensate the underestimated extraterrestrial lunar irradiances of ROLO model for about 6.76%–9.78%, and thus greatly improve the calculation accuracy of nighttime AOD. The day/night coherence transition test has demonstrated that we would obtain a good diurnal variation of AODs at Lanzhou after RCF correction. The overall averages of nocturnal AOD440nm differences between Cimel software and ROLO model and SMGF method are 0.064 ± 0.024 and 0.052 ± 0.022, respectively, while the corresponding difference of RCF is less than 0.021 ± 0.014 for all wavelengths, falling within uncertainty range of AERONET AOD products (∼±0.02). The diurnal variations of AODs determined from RCF method agree well with synchronous results of DIAL Lidar, with total mean AOD532nm differences of 0.038 ± 0.024 and 0.023 ± 0.017 in daytime and nighttime, respectively. The spectral AODs computed from RCF method are well consistent with Cimel software, although there are some discrepancies under low AOD cases (AOD440nm < 0.30), and the overall average of AOD440nm differences are less than −0.0053 ± 0.002 and −0.0185 ± 0.013 in daytime and nighttime, respectively. Our results confirmed that the CE318‐T photometer can be reasonably calculated nighttime AOD and Ångström exponent (AE440–870nm) at urban Lanzhou by using three independent methods, although the former two need to be greatly improved under low moon's illumination. The RCF method was proved to reliably calculate nocturnal AOD from moonlight irradiance, which didn't rely on MPA. A more accurate lunar irradiance model needs to be developed to improve the underestimation of current ROLO model. Long‐term climatological information of nocturnal AOD is crucial for comprehensively characterizing the diurnal variations of aerosol optical properties and atmospheric boundary layer structure during the winter at typical northern city of China, which deserves to be further investigated in the future.