Radiometric Cross-Calibration of GF-6 PMS/WFV Using a TOA Reflectance Model Over the Dunhuang Site

Abstract

Radiometric calibration is essential for quantitative remote sensing, and pseudo-invariant calibration sites (PICSs) provide an effective means for vicarious calibration without in situ measurements. This letter presents a comprehensive top-of-atmosphere (TOA) reflectance model for the Dunhuang PICS using five years (2020–2025) of Landsat-8 and Landsat-9 observations over a carefully selected stable target (spatial CV < 0.05 and temporal CV < 0.1). The model explicitly accounts for solar zenith angle (SZA), scattering angle, and seasonal atmospheric variations through a coupled function. Validation with independent Landsat data demonstrates high accuracy, with relative errors ranging from −3.07% to 1.35% and root mean square error (RMSE) below 0.021 across the blue, green, red, and near-infrared (NIR) bands. Based on this model, we perform a cross-calibration of the GF-6/wide-field view (WFV) sensor against Landsat-8/OLI, incorporating the spectral band adjustment factor (SBAF) to correct for spectral response differences. The derived calibration gains show relative errors of 2.75%–6.87% for WFV and 4.99%–6.31% for panchromatic and multispectral (PMS) compared to official values, with the blue band performing best and the NIR band showing larger uncertainties due to atmospheric effects. The GF-6 calibration results serve as a preliminary case study demonstrating the feasibility and transferability of the proposed TOA reflectance model to Chinese optical sensors, paving the way for future multitemporal operational calibration and multisensor consistency.