Light absorption properties of southeastern Bering Sea waters: Analysis, parameterization and implications for remote sensing

• An absorption budget for the southeastern Bering Sea waters is proposed
• Large variability observed in specific phytoplankton absorption (a*PHY(λ))
• Strong influence of CDOM absorption (aCDOM(λ)) on Rrs(λ) and Kd(λ)
• Chlorophyll-a at low concentrations overestimated due to higher aCDOM(λ) using OC4
• Chlorophyll-a at high concentrations underestimated due to lower a*PHY(λ) using OC4

The absorption coefficients of phytoplankton (aPHY(λ)), non-algal particles (NAP) (aNAP(λ)) and colored dissolved organic matter (CDOM) (aCDOM(λ)) were investigated and parameterized in the southeastern Bering Sea during July 2008. The absorption coefficients were well structured with respect to hydrographic and biogeochemical characteristics of the shelf. The highest values of aPHY(443) were observed offshore and the lowest values of aPHY(443) were found in the coastal domain, a low productivity region associated with limited macronutrients. Values of aDG(λ) (aCDOM(λ) + aNAP(λ)) revealed an east–west gradient pattern with higher values in the coastal domain, and lower values in the outer domain. Lower chlorophyll specific aPHY(λ) (a*PHY(λ)) observed relative to middle and lower latitude waters indicated a change in pigment composition and/or package effect, which was consistent with phytoplankton community structure. aCDOM(λ) was the dominant light absorbing coefficient at all wavelengths examined except at 676 nm. Modeling of remote-sensing reflectance (Rrs(λ)) and the diffuse attenuation coefficient (Kd(λ)) from inherent optical properties revealed the strong influence of aCDOM(λ) on Rrs(λ) and Kd(λ). Good optical closure was achieved between modeled and radiometer measured Rrs(λ) and Kd(λ) with average percent difference of less than 25% and 19% respectively, except at red wavelengths. The aCDOM(λ) accounted for > 50% of Kd(λ) which was vertically variable. Chlorophyll-a calculated by the NASA standard chlorophyll-a algorithm (OC4.v6) was overestimated due to higher aCDOM(λ) and underestimated due to lower a*PHY(λ) at low and high concentrations of chlorophyll-a, respectively.