Implications of scatter corrections for absorption measurements on optical closure of Amazon floodplain lakes using the Spectral Absorption and Attenuation Meter (AC-S-WETLabs)

Abstract

Amazon floodplain lakes range widely in concentrations of optically active constituents (OAC) driven by seasonality in hydrological and biogeochemical processes, but in general they are characterized by high turbidity (NTU from 90 to 1645) compared to coastal waters. In this work, instruments for measuring inherent optical properties (IOPs) of water bodies were evaluated for the first time in floodplain lakes in the lower Amazon River during the falling limb of the hydrograph. Water column profiles of total attenuation and absorption were measured using the Spectral Absorption and Attenuation Meter (AC-S-WETLabs), and of backscattering using Hydroscat. These measurements, however, are subject to uncertainties and require corrections for turbid waters. In this paper, we assessed the implications of scattering correction methods for the absorption tube, proposed by the AC-S manufacturer, in the simulation of the Remote Sensing Reflectance (Rrs). The closure experiment comparing Hydrolight (Mobley & Sundman, 2001) simulated Rrs and in situ Rrs demonstrated that neither of the corrections was able to thoroughly account for the scattering errors which were propagated to the absorption measurements with AC-S and backscattering with Hydroscat. The three scattering correction methods (Flat, Proportional and “Kirk”) either under or overestimated the absorption coefficient that resulted in either under or overestimation of the simulated Rrs. Flat and Proportional Methods resulted in an underestimation of Rrs from 400 to 550nm and overestimation from 600 to 700nm, indicating that the assumption of zero (0) absorption in the near infrared does not apply to inland turbid water. The Rrs errors varied also according to water OAC composition. Overall, “Kirk” correction method provided the best results regarding the spectral shape of the Rrs, however, failed to account for magnitude. Based on the tuning tests, the errors in spectra magnitude seem to be sensitive to the constant fraction of scattering (CFS) used in the Kirk method. Tests carried out with CFS values varying from 0.18 to 0.38 indicated that magnitude error can be partially overcomed by tuning CFS according to water composition. Improvements in the scattering correction methods are required in order to obtain reliable IOPs in turbid inland Amazon lakes.