Evaluation of Landsat 9 and Sentinel 2 Spectral Reflectance Calibration Validation over Tropical Region

Abstract: Satellite image calibration and validation over tropical areas are rarely considered in the research area. Most of the processes are in the non-tropical area. In this research, we would like to perform mixed calibration and validation methods; both cross and vicarious validation are mixed to understand the case better. The data used are Sentinel-2 and Landsat-9, categorized as middle-resolution imagery. Since the data are available to the public, both are familiar and widely used by the researcher and user. Therefore, it is required to gain the radiometric validation of those data. It is proposed that a mixed method of cross and vicarious calibration and validation of the satellite imagery be introduced. To gain experience in the process, specific objects are chosen based on several conditions that fulfill previously studied criteria. After conducting the analysis based on each spectral BOA reflectance of the data, both observed and predicted reflectance values statistically have no significant differences after the processing applied. Landsat 9 and Sentinel 2 reflectance values are relatively similar according to the process result. However, the BOA reflectance values measured showed different values in small wavelength ranges. The T-test of three pairs in the mixed method shows no statistically significant difference between predicted and observed spectral reflectance. Keywords: Spectral reflectance, tropical area, calibration, validation