Uncertainty measurements and validation of ocean colour data in optically complex waters

Sammanfattning: The thesis addresses validation activities associated with the use of in situ and satellite-based radiometers to assess water quality parameters, such as chlorophyll-a (Chl-a), total suspended matter (TSM) and coloured dissolved organic matter (CDOM- also known as yellow substances, YEL) in CDOM-dominated waters. In paper I, an inter-comparison of in situ radiometers is presented. Here, a new hand-held radiometer, the Water Insight Spectrometer (WISP-3) was tested and evaluated for routine water monitoring against other common radiometers used for validation. The WISP-3 measures the reflectance at the surface, and thus also works in shallow depths. The WISP-3 is designed for validation in places where other radiometers are difficult to deploy. As it is hyperspectral it can be used to develop in-water algorithms for the retrieval of water-quality information. In paper II, satellite data from the 3rd reprocessing archives of the MEdium Resolution Imaging Spectrometer (MERIS) sensor on board of ENVISAT is evaluated. MERIS level 2 reflectance and water products are assessed against in situ data. This assessment is required in order to ensure the reliability of  monitoring systems based on MERIS data such as the Swedish coastal and lake monitoring system www.vattenkvalitet.se. The evaluation of reflectance data is based on a pair-wise comparison of the standard MEGS processor and three coastal processors that are provided as source free plug-ins for the VISAT BEAM software; all pairs are compared to data measured in situ. The derived water products are evaluated both on a pair-wise comparison of processors as well as on an individual comparison of some processors to sea-truthing data. The studied processors improved the retrievals of MERIS reflectance when used the latest MERIS FR 3rd reprocessing, equalized and “smile” corrected and a land adjacency effects were corrected using the improved contrast between ocean and land (ICOL). The blue spectral bands remain problematic for all processors. Chlorophyll was retrieved best using FUB with an overestimation between 18% - 26.5% (MNB) dependent on the compared pairs. At low chlorophyll < 2.5 mg m-3, random errors dominates the retrievals of MEGS. MEGS showed lower bias and random errors when deriving suspended particulate matter (SPM) with an overestimation in the range 8-16% (MNB). All processors failed to retrieve CDOM correctly, but FUB could at least resolve variations in CDOM, however with a systematic underestimation that may be corrected for by using a local correction factor . MEGS has shown already potential to be used as operational processor in the Himmerfjärden bay and adjacent areas, but it requires further improvement of the atmospheric correction for the blue bands and better definition at relatively low chlorophyll concentrations in presence of CDOM.