Regional empirical algorithm for an improved retrieval of chlorophyll a concentrations in the Black Sea using Sentinel 3 ocean color data

The Black Sea is an extraordinarily complex water body for ocean color remote sensing, as it belong to Case 2 waters, which are characterized by relatively high absorption by Colored Dissolved Organic Matter (CDOM) while the concentration of non-pigmented particulate matter does not co-vary in a predictable manner with chlorophyll a . The optical complexity of the Black Sea is the reason why the standard bio-optical algorithms developed for Case 1 waters, are the source of large uncertainties (of the order of hundreds of percent) of chlorophyll a concentration in the coastal and shelf regions. In the framework of ESA contract “BIO-OPTICS FOR OCEAN COLOR REMOTE SENSING OF THE BLACK SEA - Black Sea Color” we developed empirical ocean color algorithm for chlorophyll a retrieval from Sentinel 3A/OLCI primary ocean color products using the in situ reference bio-optical datasets collected in the Black Sea in the period 2012-2019. Results obtained from the assessment of operational S3A/OLCI chlorophyll products, highlighted and confirmed that the specific regional algorithm is essential for the Black Sea. The coefficients of the regional algorithm were derived from the regression of log-transformed pigment concentrations and remote sensing reflectance ratio at 490nm and 560 nm with determination coefficient R² =0.88 and number of samples N=186. The algorithm predicts chlorophyll a values using a cubic polynomial formulation. The result of assessment of the regional chlorophyll a product against independent in situ measurements from the data utilized for algorithm development, showed relatively high accuracy (31.7%), fewer underestimations (MPD=-9.2%) and a good agreement (R²=0.66) between datasets indicating that the regional algorithm is more effective in reproducing the  pigment concentration in the Black Sea waters in comparison to the standard Sentinel 3A/OLCI algorithms. Our analysis revealed the importance of providing regional algorithms strictly required to suit the peculiar bio-optical properties featuring this basin. However, this requires collection of accurate in situ measurements in the different parts of the Black Sea. The validity of the reported empirical algorithm obviously depends on the size of the dataset used for its development. The Black Sea waters vary at a basin level due to the sub-regional features, environmental factors and seasonal variability, consequently the presented regional algorithm might have a limited generalization capability. Clearly, more in situ data with improved spatial and temporal coverage are critically needed for further calibration and validation of the ocean color products in the Black Sea.