Relationships between the surface concentration of particulate organic carbon and optical properties in the eastern South Pacific and eastern Atlantic Oceans

摘要

We have examined several approaches for estimating the surface concentrationof particulate organic carbon, POC, from optical measurements of spectralremote-sensing reflectance, R_rs(λ), using field datacollected in tropical and subtropical waters of the eastern South Pacific andeastern Atlantic Oceans. These approaches include a direct empiricalrelationship between POC and the blue-to-green band ratio of reflectance,R_rs(λ_B)/R_rs(555), and two-step algorithmsthat consist of relationships linking reflectance to an inherent opticalproperty IOP (beam attenuation or backscattering coefficient) and POC to theIOP. We considered two-step empirical algorithms that exclusively includepairs of empirical relationships and two-step hybrid algorithms that consistof semianalytical models and empirical relationships. The surface POC in ourdata set ranges from about 10 mg m−3 within the South PacificSubtropical Gyre to 270 mg m−3 in the Chilean upwelling area, andancillary data suggest a considerable variation in the characteristics ofparticulate assemblages in the investigated waters. The POC algorithm basedon the direct relationship between POC andR_rs(λ_B)/R_rs(555) promises reasonably goodperformance in the vast areas of the open ocean covering different provincesfrom hyperoligotrophic and oligotrophic waters within subtropical gyres toeutrophic coastal upwelling regimes characteristic of eastern oceanboundaries. The best error statistics were found for power function fits tothe data of POC vs. R_rs(443)/R_rs(555) and POCvs. R_rs(490)/R_rs(555). For our data set thatincludes over 50 data pairs, these relationships are characterized by themean normalized bias of about 2% and the normalized root mean square errorof about 20%. We recommend that these algorithms be implemented for routineprocessing of ocean color satellite data to produce maps of surface POC withthe status of an evaluation data product for continued work on algorithmdevelopment and refinements. The two-step algorithms also deserve furtherattention because they can utilize various models for estimating IOPs fromreflectance, offer advantages for developing an understanding of bio-opticalvariability underlying the algorithms, and provide flexibility for regionalor seasonal parameterizations of the algorithms.