Abstract
Optical satellite sensor generally suffers from drifts and biases relative to their pre-launch calibration, caused by launch and the space environment. Radiometric cross-calibration is valuable for calibrating sensors without coincident surface measurements through transfer radiometric values between satellite sensor via stable target. Since the data quality of GF-4/VNIR data seriously depends on vicarious calibration, which is performed once every year, cross-calibration is proposed for monitoring its radiometric performance much more frequently. In this study, the GF-4/VNIR is cross-calibrated by Landsat8/OLI, Sentinel-2/MSI, and Terra/MODIS with the aid of the tandem scenes during 2017 and 2018 after correcting the discrepancy on viewing geometries and spectrum. The results show that there is test site dependency and sensor dependency, and cross-calibration coefficients in the four bands roughly has similar degradation trend as a cosine function during 2017 and 2018, varying from 0.0002 to 0.00065. Furthermore, uncertainty analysis of the cross-calibration is carried out to analyze the uncertainty contributions affecting the cross-calibration accuracy. Note that the calibration accuracy of reference sensors, the uncertainty of MODTRAN model, BRDF uncertainty are the three main factors affecting the cross-calibration accuracy, the total uncertainty is approximately 3.9%-6.64%.
Funding Information
  • National Natural Science Foundation of China (41601398, 41671370)
  • Bureau of International Cooperation, Chinese Academy of Sciences (181811KYSB20160040)
  • National Key R&D Program of China (2018YFB0504804, 2018YFB0504800)

This publication has 23 references indexed in Scilit: