Earth Observation via remote sensing provides the main source of information about the Earth system. Satellites constantly capture observation data from the Earth. This data is processed and complemented by other auxiliary data in order to monitor natural resources, the biophysical processes, geo-information or climate models.
The complexity of many of these applications and the growing number of records, makes necessary that this data relies in a quality indicator that describes the compatibility between different sensor data and the suitability for a certain application. The quality assured data, ideally, should be SI traceable and accompanied with uncertainty estimates. The latter is the core of this research for which a software and scientifically rigorous solution is pursued.
This project seeks to estimate the radiometric uncertainties associated to each pixel of the Sentinel-2 L1C product (orthorectified Top-Of-Atmosphere (TOA) reflectance images provided by ESA). The project focuses in both the methodologies for the estimation of the radiometric uncertainty and the software implementation strategy for an optimised and distributed software. The Sentinel-2 radiometric uncertainty analysis focuses on the review of the pre- and post-launch characterisations in order to specify the uncertainty contributors at a pixel level. Auxiliary information for the uncertainty calculation is extracted from the metadata and quality masks integrated in the L1C product. The analysis also focuses on the establishment of a set of validation procedures for some of these uncertainty contributors and a validation of the radiometric uncertainty combination. The software implementation has been integrated in the Sentinels Toolbox for direct access to the EO community.