Satellite-based time series calibration of GF-1 WFV sensors for large view zenith angle observations

Abstract

ABSTRACTA novel satellite cross-calibration method is proposed using image pairs with large view zenith angle observations. Previous studies on cross calibration have focused primarily on image pairs with nadir or small view zenith angles, while this study introduces a technique for large view zenith angle observations using bidirectional reflectance distribution function (BRDF) correction. This technique was applied to images from the wide field of view (WFV) onboard the GAO FEN-1 (GF-1) satellite and Terra Moderate Resolution Imaging Spectroradiometer (MODIS) sensors as it transited over the Dunhuang site in China. A total of 62 image pairs, collected from 2013 to 2016, were used in the cross-calibration process. However, only 12 of 62 image pairs included a small view zenith angle difference (less than 10°), while the other image pairs featured a large view zenith angle difference. An experimental time series calibration was conducted using the proposed technique as well as conventional cross-calibration methods for comparison purposes. Time series calibration results also showed an attenuation trend for GF-1 WFV sensor and produced a calibration equation for day since launch (DSL). Then, the calibration equation by DSL was compared with four vicarious calibration results carried out over the Dunhuang test site from 2013 to 2016 to demonstrate its consistency. The calibration coefficients using the proposed method were also compared with the coefficients from other cross-calibration methods, taking the vicarious calibration coefficient as the reference. The comparison results showed that the calibration coefficients using the proposed method were with higher calibration accuracy and higher calibration frequency than other cross-calibration methods. The cross-calibration uncertainties caused by difference influence factors were analysed quantitatively. The results showed that the calibration uncertainties were decreased from 35.03–44.60% to 5.95–8.26% after BRDF correction, which demonstrated the proposed method to be superior for processing large view zenith angle images. These results suggest the proposed technique could be an effective new tool for future time series satellite calibration.