Joint use of Sentinel 2 and OLCI-Sentinel 3 for Coastal Monitoring
Mangin, Antoine; Fanton d'Andon, Odile
The observation and monitoring of coastal waters with "traditional" Earth observation means (i.e. spectro-imagers such as MERIS) are difficult because of the small geometrical scales in presence in these areas and fine radiometric sensitivity required for waters composition and sea bottom classification. Therefore, the combination of high-spatial/low-spectral resolution with high-spectral/low-spatial resolution sensors is becoming an emerging tool for Coastal Waters observations.
Looking towards the very near future, the combination of Sentinel 2 and of OLCI (Ocean and Land Color Instrument) sensor aboard Sentinel 3 products opens the way to a very challenging optimized observation tools for these coastal waters. On the one hand, spectral capabilities of OLCI offer appropriated sampling to discriminate main components of sea waters mass but with a limited spatial resolution not fully adapted to near coastal and shallow waters. On the other hand, the imagery capabilities of Sentinel-2 offer unprecedented combined spectral/spatial tradeoff for land surface classification.
A first merging of the two (simulated) sensors has been developed and has provided encouraging results for mapping at a high spatial resolution the "traditional" sea waters components (Chlorophyll-a, CDOM and SPM) which would have been not reachable with Sentinel-2 alone. Sentinel-2 and 3 scenes have been simulated by using HICO mission and which have also been used to create "in situ - truth". The inversion method is based on Inherent Optical Properties retrieval by using Sentinel-2 visible spectrum added to a Bayesian approach to include OLCI products and access to a better spectral characterization.
This method is very innovative and promising compared to classical merging techniques used, that are mainly based on merging of radiometric classification. Preliminary results that have been presented are water properties associated with error bars at a high spatial resolution over the Venezia lagoon. Based on these encouraging first results, inversion method has been improved to account for sea bottom albedo and thus, indirectly, allows classification of benthic species at the high spatial resolution of Sentinel-2.
The talk is based on the description of the method (merging of Sentinel-2 and OLCI with "weighted" contributions) and its sensitivity/robustness to inputs data (directly linked to SNR of both instruments). The improved consideration of the sea bottom during the retrieval processing and allowing classification and better characterization of water quality will be presented. Lastly, a deeper analysis of the two missions capabilities to present twin observations (Sentinel 2 and 3) that can be merged has been analysed and will be presented.
This merging method also offers opportunity to be used on inland waters characterization for a better classification of water quality types.