Validation of Cross-Calibration for the Time-Series of HJ-2A/B CCD Sensor Based on Baotou Sandy Site

High resolution and wide coverage are important development directions for optical sensors on satellites. Multi-camera mosaic imaging technology is often used to combine high resolution with wide coverage. When using data from these sensors for quantitative remote sensing studies, radiometric calibration is important. Based on this, this study employs a cross-calibration method incorporating the random forest algorithm and spectral interpolation to derive the time-series calibration coefficients for the charge-coupled device (CCD) sensors on the HuanJing-2A/B (HJ-2A/B) satellites, and examines the applicability of this method. The results show that, compared with the official calibration coefficients (OCCs), the relative errors of the cross-calibration results are predominantly within 10%, with the majority falling below 5%. When compared with the measured reflectance from the Baotou sandy site, the results of this study show better consistency with the 1:1 line and have average relative errors less than 8.07%, while the OCCs have average relative errors less than 34.65%. Besides, for the bands exhibiting larger individual errors relative to the OCCs, the validation results indicate that the average relative errors of these bands are smaller than those of the OCCs when referenced against the Baotou sandy site. Furthermore, the influence of different input parameters on the bidirectional reflectance distribution function (BRDF) model, not considering BRDF correction of CCD2 and CCD3 sensors on the calibration results, using the single-day coefficients, which are close to the date of the Baotou sandy site on validation results, are discussed. Finally, the study also evaluates the radiometric consistency between different cameras. The result indicates that the method proposed is applicable to other CCD sensors on HJ-2A/B, and the cross-calibration coefficients obtained can effectively eliminate radiometric inconsistencies among different cameras.