Sentinel-3 Land synergy Products for SPOT VEGETATION Continuity
Kempeneers, Pieter1; Swinnen, Else1; Sara, Verbeiren1; Marc, Bouvet2; Mark, Drinkwater2
1VITO, BELGIUM; 2ESA, NETHERLANDS
The VEGETATION sensors onboard SPOT (Satellites Pour l'Observation de la Terre) 4 and 5 allow a daily monitoring of terrestrial vegetation cover through remote sensing, at regional and global levels. Products are used for research and operational services, including services developed in the context of GMES (Global Monitoring for Environment and Security). By the end of 2012, the VEGETATION instruments will have reached the end of their nominal operating lifetime. Nevertheless, the VEGETATION mission will be continued by ESA's GMES Sentinel-3. Sentinel-3 intends to provide global coverage every one to three days, possibly using a multi-satellite constellation. To serve both ocean and land services, two co-registered Visible/Infrared optical instruments will be accommodated: the Ocean and Land Colour (OLC) and the Sea and Land Surface Temperature (SLST) sensor. The idea is to use a proven technology and heritage from the original Envisat sensors MERIS and A/ATSR, together with additional channels to allow co-registration and improved atmospheric correction and cloud flagging. The proposed study demonstrates the capability to generate VEGETATION-like products and is based on the results of an ESA contract "Sentinel-3 Land Synergy Products for SPOT VGT Continuity". With respect to VEGETATION, the Sentinel-3 sensors have finer spectral and spatial resolution and a superior number of spectral bands, which could enhance the existing products. As an example, atmospheric correction and cloud flagging could be improved. However, the success of the continuity will greatly rely on their similarity to the 1 km VEGETATION baseline products. A seamless transition is needed for a consistent time series, which is the main issue for this type of data. Many derived products and services are based on differences in reflectance or NDVI (Normalized Difference Vegetation Index). The impacts of the Sentinel-3 design with respect to the VEGETATION system will be presented. The critical characteristics include geometric accuracy, acquisition geometry, frequency of acquisition and radiometry. To check the characteristics for geometric accuracy, a geometric model was developed to compute all possible configurations for acquisitions (viewing and illumination conditions). The same model was used to calculate the minimum and maximum number of acquisitions obtained for OLC and SLST at different latitudes for periods of ten days, which is the basis for computation of the S10 VEGETATION baseline product. A new method for simulating the VEGETATION spectral bands from the combination of OLC and SLST is introduced. It first corrects the measured Top Of Atmosphere (TOA) radiances for atmospheric effects using a radiative transfer model (MODTRAN4). The idea is to reduce the high frequency spectral content related to the atmospheric absorption bands, resulting in a smoother signal. This can then be interpolated to obtain a hyperspectral reflectance. The atmospheric effects are re-introduced by upscaling the interpolated signal to TOA with the same radiative transfer model. To mimic the baseline VEGETATION product, this hyperspectral TOA radiance is then filtered according to the VEGETATION spectral response functions for each respective band (B0, B2, B3 and MIR). For performance reasons, the radiative transfer model is implemented by creating a lookup table. As a reference, both simulated spectra based on a spectral library and real VEGETATION satellite data were used. In anticipation of Sentinel-3 data, VEGETATION synergy products were generated from simulated data using existing sensors (MERIS/AATSR and ASTER). A total of 17 target areas were selected for validation. The areas cover different land cover types at four periods in the year.