A Framework for the Estimation of Uncertainties and Spectral Error Correlation in Sentinel-2 Level-2A Data Products

The Copernicus Sentinel-2 (S2) satellite mission acquires high spatial resolution optical imagery over land and coastal areas. Delivering uncertainty estimates and spectral error correlation alongside S2 data products facilitates the constrain of retrieval algorithms, propagates further downstream the retrieval uncertainty, and, finally, makes informed decisions to end-users. This study presents a framework to produce uncertainty estimates and spectral error correlation associated with the S2 L2A data products (i.e., surface reflectance). This framework has been implemented in a prototype code available at https://doi.org/10.5281/zenodo.11971517. The uncertainty considers both the Level-1 (L1) uncertainty estimates for the top-of-atmosphere (TOA) reflectance factor and the atmospheric correction. The L2A error distribution cannot be systematically described as a normal distribution; the transformation can be nonlinear and without an explicit mathematical model. Thus, a multivariate Monte Carlo model (MCM) rather than the law of propagation of uncertainty (LPU) is selected for uncertainty propagation. We show results for surface reflectance uncertainty over the Amazon forest and Libya4 desert site. It illustrates the large uncertainty and spectral error correlation variations depending on the scene. The comparison of a multivariate MCM against an LPU propagation methodology indicates the limitations of the latter for scenes dominated by the atmospheric path. Its implementation as an operational per-pixel processing and dissemination of both the uncertainty and spectral error correlation becomes challenging. Therefore, this methodology is not expected to run at an operational level but serves as the basis to define a strategy for an operational one.