On the Characterization of Sen2Like Surface Reflectance Data Harmonization and Fusion Processes.

There are growing and evolving expectations from the Earth Observation (EO) Community for merging an increasing number of input data streams recorded with the various space based Multi-Spectral (MS) High Resolution (HR) EO mission instruments. The various agencies and industries are now delivering Analysis Ready Data (ARD) aiming at stacking together multi source data. One critical objective of ARD is to enable the monitoring of rapid and subtle changes. This category of applications sets some strong requirements regarding the quality of the surface reflectance harmonization in the time series. In the context of Sentinel-2 / Landsat 8, this paper proposes a comprehensive analysis of the Sen2Like processing. It shows that it is possible to increase significantly the geometric coregistration accuracy and to ensure an appropriate cross calibration. However, some issues persist and are due to directional effects. These are partially corrected with the current approach but remain a major source of noise in time series.