Pixel-based Uncertainty Propagation in MERIS and OLCI Atmospheric Correction and Bio-optical Algorithms
Mazeran, Constant; Lamquin, Nicolas
Pixel-by-pixel uncertainty estimate is explicitly required by ESA for its Ocean Colour and Land Imager (OLCI) onboard Sentinel-3. It should be more generally a requirement for any mission, so that remote-sensing data can be used in a relevant manner for scientific studies and downstream applications. We present here a mathematical framework for the propagation of radiometric noise in marine reflectance and three bio-optical algorithms: chlorophyll-a, diffuse attenuation coefficient and Secchi depth transparency. The method follows general guidance of BIPM and GEO (QA4EO), taking into account spectral variance-covariance matrix of top-of-atmosphere input noise. This study demonstrates it is possible to analytically propagate uncertainties in an historical atmospheric correction scheme over clear water and quantifies the impact of spectrally correlated noises. Uncertainty maps on deep waters are computed for MERIS. Because the maps are tributary of the specified sensor noise, here simplified, we emphasise the need to perfectly characterise the spectral and spatial structure of instrumental noise shortly after Sentinel-3 launch.