Search

Sun-induced fluorescence (SIF) retrieved from satellite measurements has been widely used as proxy for chlorophyll-a concentration and as indicator of phytoplankton physiological status in oceans. The practical use of this naturally occurring light signal in environmental research is, however, under-exploited, particularly in research focusing on optically complex waters such as inland and coastal waters. In this study, we investigated methodological and knowledge gaps in remote sensing of chlorophyll-a SIF in optically complex waters by reviewing the theory behind SIF occurrence, the availability of existing and upcoming instrumentation, the availability of SIF retrieval schemes, and the applications for aquatic research. Starting with an overview of factors that influence SIF leaving the water body, we further investigated available and upcoming observational capacity by in situ, airborne and satellite sensors. We discuss requirements for spatial, spectral, temporal, and radiometric resolution of observing systems in the context of SIF dynamics. We assessed viable retrieval techniques able to disentangle SIF from non-SIF contribution to the upwelling radiance, ranging from the established multispectral Fluorescence Line Height algorithm (FLH) approach to hyperspectral approaches including model inversion, spectral fitting methods and machine learning regression procedures. Finally, we provide an overview of applications, which could potentially benefit from improved SIF emission estimates such as biomass estimation, algal bloom investigation and primary productivity modelling.

article i nfo We provide a comprehensive overview of water constituent retrieval algorithms and underlying definitions and models for optically deep and complex (i.e. case 2) waters using earth observation data. The performance of constituent retrieval algorithms is assessed based on matchup validation experiments published between January 2006 and May 2011. Validation practices range from singular vicarious calibration experiments to comparisons using extensive in situ time series. Band arithmetic and spectral inversion algorithms for all water types are classified using a method based scheme that supports the interpretation of algorithm validity ranges. Based on these ranges we discuss groups of similar algorithms in view of their strengths and weak- nesses. Such quantitative literature analysis reveals clear application boundaries. With regard to chlorophyll retrieval, validation of blue-green band ratios in coastal waters is limited to oligotrophic, predominantly ocean waters, while red-NIR ratios apply only at more than 10 mg/m 3 . Spectral inversion techniques — although not validated to the same extent — are necessary to cover all other conditions. Suspended matter retrieval is the least critical, as long as the wavelengths used in empirical models are increased with concen- trations. The retrieval of dissolved organic matter however remains relatively inaccurate and inconsistent, with large differences in the accuracy of comparable methods in similar validation experiments. We conclude that substantial progress has been made in understanding and improving retrieval of constituents in optically deep and complex waters, enabling specific solutions to almost any type of optically complex water. Further validation and intercomparison of spectral inversion procedures are however needed to learn if solutions with a larger validity range are feasible.

Semi-analytical remote sensing applications for eutrophic waters are not applicable to oligo- and mesotrophic lakes in the perialpine area, since they are insensitive to chlorophyll concentration variations between 1-10 mg/m3. The neural network based Case-2-Regional algorithm for MERIS was developed to fill this gap, and a dedicated validation campaign on Lake Constance revealed promising results. For a more extensive validation, the algorithm is applied to a collection of 239 satellite images from 2003-2008, and the results are compared to experimental and official water quality data collected on seven perialpine lakes in the same period. It is shown that remote sensing estimates can provide an adequate supplementary data source to fluorescence probe monitoring data series of the top 5-10 m water layer, provided that a sufficient number of matchups for a site specific maximum temporal offset is available. Comparison with 20 m mixed profiles is less adequate. The supplemental correction of adjacency effects by the Improved Contrast between Ocean and Land algorithm performs well regarding the retrieval of accurate remote sensing reflectances, while its improvement to the accuracy of estimated chlorophyll concentrations is less significant.