Phenology Estimation From Meteosat Second Generation Data

Abstract

Many studies have focused on land surface phenology, for example as a means to characterize both water and carbon cycles for climate model inputs. However, the Spinning Enhanced Visible Infra-Red Imager (SEVIRI) sensor onboard Meteosat Second Generation (MSG) geostationary satellite has never been used for this goal. Here, five years of MSG-SEVIRI data have been processed to retrieve Normalized Difference Vegetation Index (NDVI) daily time series. Due to existing gaps as well as atmospheric and cloud contamination in the time series, an algorithm based on the iterative Interpolation for Data Reconstruction (IDR) has been developed and applied to SEVIRI NDVI time series, from which phenological parameters have been retrieved. The modified IDR (M-IDR) algorithm shows results of a similar quality to the original method, while dealing more efficiently with increased temporal resolution. The retrieved phenological phases were then analyzed and compared with an independent MODIS (Moderate resolution Imaging Spectrometer) dataset. Comparison of SEVIRI and MODIS-derived phenology with a pan-European ground phenology record shows a high accuracy of the SEVIRI-retrieved green-up and brown-down dates (within days) for most of the selected European validation sites, while differences with MODIS product are higher although this can be explained by differences in methodology. This confirms the potential of MSG data for phenological studies, with the advantage of a quicker availability of the data.