Abstract: Objective　Against the backdrop of the growing demand for quantitative applications of high-resolution remote sensing satellites, the radiometric performance of in-orbit sensors is prone to degradation due to factors such as component aging and space radiation, so regular radiometric calibration is essential for maintaining data accuracy. Traditional calibration methods rely on manual measurements of surface and atmospheric parameters, which are costly, labor-intensive, and limited by low frequency (typically once a year), making it difficult to timely track the changes of sensor performance. Gaofen-2 (GF-2) satellite is China's first civil land satellite with sub-meter spatial resolution. Currently, it relies solely on annual calibration at the Dunhuang test site, which cannot meet the requirements of high-frequency and wide dynamic range calibration. This study aims to address these limitations by developing an automated radiometric calibration method based on the Songshan Fixed Artificial Target Site, thereby improving calibration frequency and reducing costs. Methods　This study implemented automated radiometric calibration on the Panchromatic Multispectral Sensor 2 (PMS2) aboard the GF-2 satellite using the Songshan Fixed Artificial Target Site. The site is equipped with multi-level gray-scale targets (5%, 40%, and 60% reflectivity) and automated observation instruments. The automated observation system includes three spectral radiance meters (SRM) for real-time target reflectance spectral radiance measurement, one hyperspectral irradiance meter (HIM) for ground spectral irradiance acquisition, and one precision solar radiometer (PSR) for retrieving aerosol optical depth (AOD) at 550 nm and atmospheric water vapor content. Based on the reflectance-based method, the target reflectance was calculated using synchronous measurements from SRM and HIM. Then atmospheric parameters (AOD, water vapor, ozone content from MODIS products) and geometric parameters were input into the 6S radiative transfer model to compute the top-of-atmosphere (TOA) spectral radiance. Finally, the calibration coefficients were derived by fitting the TOA radiance with the digital number (DN) values extracted from GF-2 satellite images. Valid calibration data were obtained on three clear days from September to November 2020, and the calibration accuracy was further verified using the Dunhuang automated calibration system. Results and Discussion　The automated observation system achieved stable and unattended data collection, with SRM and HIM measuring every 6 minutes and PSR every 3 minutes. The quality control criteria (clear weather, stable atmosphere, AOD < 0.3, and cloud-free satellite observations) ensured the reliability of the data. The calibration coefficients (gain and bias) were obtained for the panchromatic band and four multispectral bands of PMS2. The verification using the Dunhuang automated calibration system showed that the relative differences in TOA spectral radiance calculated by the two automated systems were within 4.3% for all bands. This confirms that the proposed method maintains comparable calibration accuracy to traditional methods, while enabling high-frequency and wide dynamic range calibration through multi-level gray-scale targets. Conclusion　This study proposes an automated wide dynamic range radiometric calibration method based on the multi-level grayscale targets and automatic observation equipments in Songshan Fixed Artificial Target Site. Compared with traditional methods, the proposed approach not only eliminates the need for on-site manual operations, but also significantly improves calibration frequency, and reduces costs. The calibration results for the GF-2 satellite PMS2 sensor were verified by the Dunhuang system with relative differences within 4.3%, demonstrating the feasibility and effectiveness of the proposed method. This study provides a reliable alternative technical solution for high-frequency and high-precision radiometric calibration of high-resolution optical satellites, and a solid support for the quantitative application of remote sensing data. Future work will focus on optimizing data quality control, enhancing precision assessment, and extending the method to more in-orbit high-resolution satellites.

Key words: automated observation, China-Songshan Artificial Target Site, radiometric calibration, GF-2 satellite