MERIS Case II water processor comparison on coastal sites of the northern Baltic Sea

Three MERIS Case II water processors included in the BEAM software package were studied for estimating the water quality in the coastal waters of the northern Baltic Sea. The processors, named Case II Regional (C2R), boreal (BOR), and eutrophic (EUT), for the associated lake types, have been developed for different types of coastal or inland (Case II) waters. Chlorophyll-a (chl-a), total suspended matter (TSM), absorption of colored dissolved organic matter (aCDOM(443)), and signal depth (Z90) products of the BEAM processors were compared with in situ data. In addition, total absorption (aTOT) and scattering of TSM (bTSM) from different BEAM processors were compared against the results of coastal field campaign measurements. The in situ water quality data consisted of monitoring station data gathered by the Finnish environmental administration during 2006–2009 and data from coastal field campaigns with a flow-through system. AERONET-OC (SeaPRISM) data from the Helsinki Lighthouse station were used to validate the BEAM reflectance products. The comparison with the BEAM processor results and in situ data showed that the bias of the original BEAM algorithms can be decreased through adjustment of the coefficients that relate IOPs such as the absorption of pigments and the scattering of TSM to water quality constituents such as chl-a and turbidity. The TSM products of the BEAM processors can be used to estimate the turbidity measured at monitoring stations with an r² of 0.76–0.84 and an RMSE of 0.7–0.85 FNUs (Formazin Nephelometric Units) on the coast of Finland. The best functionality for turbidity estimation was observed with the EUT processor, but the C2R processor also gave a sound performance. The BOR and C2R processors proved to be the best for deriving chl-a concentration. However, the accuracy of chl-a estimations was low with both processors (r² ranged from 0.45 to 0.47 and RMSE was between 44 and 45%). Chl-a products, particularly during the phytoplankton bloom seasons of spring and summer, require further development. The Z90 product from the BOR processor was used to derive an algorithm for Secchi disk depth estimation with r2 0.48 and RMSE 0.97 m. The BOR processor was the most successful at CDOM estimation (r² 0.6 and RMSE 0.49 1/m), but a simple reflectance ratio was actually able to perform better (r² 0.75 and RMSE 0.39 1/m). In many cases, the differences between the outcomes of processors were small and related only to a part of the in situ dataset.