Improved protocol for determining the spectral absorption coefficient of aquatic particles using the filter-pad technique

Dariusz Stramski, University of California San Diego, La Jolla, CA, United States, Rick A Reynolds, Scripps Institution of Oceanography, La Jolla, CA, United States, Julia Uitz, Sorbonne Universités, UPMC Univ Paris 06, INSU-CNRS, Laboratoire d’Océanographie de Villefranche-sur-Mer (LOV), Villefranche-sur-mer, France and Guangming Zheng, NOAA, NESDIS/STAR, College Park, MD, United States
Spectrophotometric measurement of light absorption by aquatic particles collected on filters has long been the most common method for routine determinations of the spectral absorption coefficient of particles, ap(λ), with high spectral resolution (~1 nm) over a broad spectral range from the ultraviolet through visible to near-infrared (λ is light wavelength in vacuo). The pathlength amplification resulting from light scattering within the sample/filter matrix is one of most important sources of uncertainty in this filter-pad technique. The use of differing geometrical measurement configurations and large variations in the reported correction for pathlength amplification have hindered adoption of an established measurement protocol. We present results of laboratory experiments with various types of particulate samples to examine the pathlength amplification factor for three filter measurement geometries; the filter in the transmittance configuration (T), the filter in the transmittance-reflectance configuration (T-R), and the filter placed inside an integrating sphere (IS). Relationships between optical density measured on suspensions (ODs) and filters (ODf) were evaluated for the formulation of pathlength amplification correction within the visible portion of the spectrum. Power functions provided the best functional representation of the relationship for all three geometries. The IS method exhibited the least sample-to-sample variability and the smallest uncertainties in the relationship between ODs and ODf. For six different samples, a median error of 7.1% is observed for predicted values of ODs using the IS method. The largest uncertainties occur in the T method. The relationships established for the three filter-pad methods are applicable to historical and ongoing measurements. For future work, the use of the IS method with the established correction for pathlength amplification is recommended whenever feasible.