Radiation calibration is a necessary prerequisite for the quantification of remote sensing data from laboratory calibration before launch to on-orbit calibration after launch throughout the entire life cycle of remote sensors. Site calibration is a common method for on-orbit alternative calibration of remote sensors with large fields of view. When a remote sensor is operating normally, it is calibrated with a large area of uniform ground objects as the calibration source, which has the advantages of a high calibration frequency and no need for synchronous measurement. The Greenland ice sheet (-75°S, 123°E) and the Antarctic ice sheet (73.375°N, -40°W) are commonly used as targets for snow and ice scenes, whose surfaces are covered with evenly distributed snow. Due to their high altitudes (usually greater than 2 km), they are less affected by the atmosphere and help obtain calibration samples with better data quality. Furthermore, ice and snow have a relatively flat spectrum in the visible range, which makes band transfer easier with other calibration methods. Thus, using the Greenland ice sheet and the Antarctic ice sheet as the calibration source of snow and ice scenes has many advantages for the calibration and verification of remote sensors.

The research method in this paper is based on the previous study of polar scene calibration methods. First, we choose Greenland as the calibration target of snow and ice scenes and select the calibration sample in the directional polarimetric camera (DPC) level 1 data. The area interfered with by cloud pixels is then eliminated after spectral channel traversal and angle traversal through the calibration sample's rows and columns. After substituting the surface bidirectional reflectance distribution function (BRDF) and atmospheric parameters (aerosol, water vapor, ozone, and other profiles) of the snow and ice scene into the radiative transfer model to obtain the zenith reflectivity and radiance, we test the on-orbit radiation response changes of the payload of the DPC on Gaofen-5 satellite. In addition, the obtained conclusions are in good agreement with the calibration results of the desert and ocean scenes, and the dispersion of the calibration results is smaller.

We compare the measured results of the DPC with the calibration results and obtain the following findings.

1) When the view zenith angle is less than 50°, the measured reflectance values of each band of the DPC have good consistency with the calibrated reflectance values. The standard deviations of their ratios are within 3%, and the root-mean-square errors were both lower than 2% (Fig. 4 and Table 3).

2) When the view zenith angle is less than 30°, the measured top-of-atmosphere (TOA) radiance values in each band of the DPC are compared with the calibrated TOA radiance values. The results show that the DPC data in the visible light band are relatively stable and have low uncertainty (Fig. 5 and Table 4).

3) The influence of the atmosphere (aerosol, water vapor, and ozone) and the surface BRDF on TOA reflectivity is analyzed. The uncertainty of the BRDF model of the snow and ice scene is 2%. Finally, we synthesize the uncertainty of each factor, and the synthetic uncertainty of each band is 2% (Table 5).

4) The research on the size of ice and snow shows that the average relative error of fine snow in each band is within 4%, and the relative error standard deviation of bands from 443 nm to 765 nm is within 2%, so the data of fine snow in each band is relatively stable (Fig. 9 and Table 6).

5) The results obtained in this paper are compared with the calibration results of the desert scene and the ocean scene. The comparison results show that the calibration coefficient of the Greenland snow and ice scene deviates from the average value of the calibration coefficients of the ocean scene and the desert scene within 5%, and the standard deviation of the Greenland polar scene is within 2% (Table 7 and Table 8).

Inspired by the idea of on-orbit alternative calibration, we propose an on-orbit radiometric calibration method based on the glacier scenes in the North and South Poles. We choose Greenland for research and analysis and conduct radiometric calibration of the remote sensor DPC with a large field of view. The measurement results of the DPC are compared with the calibration results. The comparison results show that the measured value and the calibrated value in each band of DPC are in good agreement. The standard deviation of the ratio of the measured value to the calibrated value is within 3%, and the root-mean-square error is lower than 2%, which proves that DPC has a good performance in snow and ice scenes. Moreover, through the error analysis of the atmosphere and the surface, it is shown that the BRDF on the surface has the greatest influence, and the final synthetic uncertainty in the visible light band is 2%. Finally, the comparison of the calibration results of this paper with the calibration results of the desert scene and the ocean scene also proves the stability of the calibration results of the Greenland snow and ice scene and the validity and reliability of the calibration method. The method described in this paper can provide long-term monitoring and calibration of the detection data while the payload is on orbit and contribute to the quality improvement of products for operational application.