In addition to the presentation of IOP-based relationships for the two satellite light wavelengths of 443 and 555 nm, the statistical analyses are supplemented with examples of analogous relationships but determined at the optimal bands chosen from among those original light wavelengths for which the HydroScat-4 and AC-9 instruments performed in situ measurements. To derive statistical formulas for biogeochemical properties of suspended matter as functions of remote-sensing reflectance values, the available dataset has to be extended

with the aid of radiative transfer modelling. It has been common practice in much optical modelling work that the average values of the constituent-specific optical coefficient multiplied by the assumed selleck products concentrations of these constituents learn more give the modelled absolute values of these optical coefficients, which are then used as further inputs for radiative transfer modelling. But because the very large variability of constituent-specific optical coefficients of suspended matter in the southern Baltic Sea were documented in an earlier work by S. B. Woźniak et al. (2011), it was decided not

to use averaged values as the modelling input. Instead, a different approach to the problem is taken: in each separate modelling case the real, measured optical coefficients (i.e. the values of the coefficients an(λ), cn(λ) and bbp(λ)) are used as modelling input and the corresponding and actually measured values of biogeochemical properties are also used in the subsequent statistical analyses. From the available empirical material a subset of 83 cases was selected (see the stations denoted by grey dots in Figure 2), which

consists of only those cases for which all the biogeochemical properties of the relevant particulate matter (i.e. concentrations of SPM, POM, Cisplatin cost POC and Chl a) and all the seawater IOPs (i.e. values of an(λ), cn(λ) and bb(λ)) required for further modelling were measured at the same time. For this particular data subset, the hypothetical spectra of the remote-sensing reflectance Rrs [sr− 1] were then determined on the basis of radiative transfer numerical simulations. The Hydrolight-Ecolight 5.0 (Sequoia Scientific, Inc.) model was applied with a set of simplifying assumptions. The modelled hypothetical water bodies were chosen to be infinitely deep, and all the IOPs of the modelled waters were chosen to be constant with depth. This assumption is obviously a significant simplification, but it most likely represents quite well a common situation in the Baltic Sea, where the relatively shallow subsurface layer of water penetrated by sunlight is mixed as a result of wave action and turbulence caused by surface wind stress. Another simplification was the assumption that no inelastic scattering (no Raman scattering, or chlorophyll or CDOM fluorescence) and no internal sources (no bioluminescence) were taken into account.