Site-based absolute radiometric calibration is an important guarantee for quantitatively application of aerial remote sensing data. The key to aerial camera calibration is to reduce the influence of field measurement environment and various measurement errors and improve the stability and accuracy of calibration results. In this paper, five gray scale tarps with different reflectance were set up at the Pu'er test site in Yunnan Province from 26th to 28th December 2020, and a reflectance-based method, which requires synchronous measurements including the surface reflectance, atmospheric parameters and geometry information when the aircraft overpass the calibration test site, was used for absolute radiometric calibration of Lecia DMC JR airborne multispectral camera. These parameters were input into the MODerate resolution atmospheric TRANsmission (MODTRAN) atmospheric radiative transfer model to obtain the spectral radiance at the entrance pupil at the aircraft height. Then, combined with the average DN of the selected area of the image, the different absolute radiometric calibration coefficients were derived for the DMC III camera via three consecutive calibration experiments based on single-, dual- and multi-site methods. By systematically comparing the calibration results of single-, dual- and multi-site methods and analyzing various error sources, proposing a high-precision absolute radiometric calibration method based on multiple observations with multi-scale tarps. Moreover, the calibration uncertainty of each band is 7. 24% (blue), 6. 20% (green), 5. 35% (red) and 4. 68% (near infrared), respectively. In order to verify the radiometric calibration results, the reflectance inversion validation was adopted to prove the rationality of the three different calibration coefficients obtained by single- dual- and multi-site methods. The different absolute calibration coefficients obtained by single-, dual- and multi-site methods were used to conduct atmospheric correction for various typical ground objects in the test site with Atmospheric/Topographic Correction for Airborne Imagery ( ATCOR 4) atmospheric correction software, and the surface reflectance obtained by inversion was compared with the measured surface reflectance for verification. The results show that multiple experiments based on multi-scale tarps in a time are strongly critical for improving the calibration accuracy. The single-site method with 20% and 60% tarps and single multi-site for single experiment method have relatively poor calibration accuracy. The calibration errors of the single-site method with 40% tarp and dual-site method decrease obviously, while the multi-site for multiple experiments method has relatively high calibration accuracy. The average relative errors of the three methods are 10%, 5. 43% and 3. 18%, respectively. The calibration method of aerial multispectral camera based on multiple experiments with multi-scale tarps put forward in this paper reduces the calibration uncertainty of single-site method, dual-site method and single experiment, which has the high reference for the high-accuracy site-based calibration of aerial cameras and the quantitative application of aerial data in the future.
