Propagating Sentinel-2 Top-of-Atmosphere Radiometric Uncertainty into Land Surface Phenology Metrics Using a Monte Carlo Framework

Time series of optical imagery allow to derive land surface phenology metrics. These metrics are only complete with a statement about their uncertainty. A source uncertainty is the radiometry sensor. We propagated radiometric uncertainties within Monte-Carlo framework into phenological using TIMESAT approach based on time Normalized Difference Vegetation Index (NDVI), three-band Enhanced (EVI) and Green Leaf Area (GLAI) derived from radiative transfer modelling. Additionally, we studied effect scene pre-classification. focused Sentinel-2 MSI TOA data since quantitative estimates available. Propagation was carried out for growing season over an agricultural region in Switzerland. Propagated had little impact classification except spectrally mixed pixels. Effects spectral indices GLAI were more pronounced. In detail, uncertain due ill-posedness model inversion (median relative all crop pixels scenes: 4.4%) than EVI (2.7%) NDVI (1.1%). Regarding phenology, exhibited largest case GLAI. The magnitude depends inter-scene error correlation, which assumed be either zero (uncorrelated) or one (fully correlated) actual correlation unknown. If fully correlated, small (2 3 days) but take values up greater 10 days under uncorrelated assumption. Thus, our work provides guidance interpretation