Validation and improvement of processing algorithms for retrieval of water quality parameters in Norwegian Coastal waters.

Validation of satellite EO data products with in situ observations is a challenge and difficult task however needed in order to get confidence in and improve the algorithms used for retrieval of geo-bio-chemical information form satellite EO sensors. Specific validation protocols have been developed and dedicated validation campaigns have been executed in cooperation between the oceanographic research community and the space agencies for performing validation studies of various satellite based EO geo-bio-chemical products and sensors During 2008 the software routines for the processing and hardware storage capacities have been improved for the MarCoast WQ & HAB Service in order to make the service performance more robust and that the routine processing can be more efficient and with less human interaction in case of irregularities. A level-by-level processing solution has been developed, with a structure that makes it re-usable for new geographical areas and for introduction of new products and processing methods. Two of three service product categories have been validated with independent field data. The Ferrybox data available for this MarCoast validation have provided independent measurements of both Chlorophyll-a (using two methods – in vivo fluorescence measurements and water samples for HPLC laboratory analysis) and of measurements of the turbidity, which has been converted to total suspended matter (TSM) for the central Skagerrak waters. Unfortunately no data for validation of the yellow substance has been available. The three different Chlorophyll-a products based on the MERIS RR data have been validated against HPLC measurements of the phytoplankton content of water samples. All the satellite algal_1, algal_2 and lmchl products compares well with the HPLC measurements and the “regionally tuned” algorithms performs slightly, but significant, better than the “global” algal_1 algorithm. The comparison of total suspended material (TSM) values estimated from Ferrybox turbidity measurements and MERIS products show a less accurate agreement with only 84 and 46 % explained variance for respectively the total_susp and lmsm. Most likely this is due to the linear regression approach which is used for calculating TSM values from turbidity measurements does not take into account other factors influencing the turbidity (e.g. absorption by algae cells and suspended minerals) and may give errors of the same order of magnitude. The Ferrybox data can be used for validation of satellite TSM products. However, the conversion factors between in situ turbidity and in situ TSM should be determined and tuned for the area of interestNERS Technical Report no. 296. Funded by Norwegian Space Centre through Project NRS SatHav, and European Space Agency through Project GSE MarCoas