A spectral model of the beam attenuation coefficient in the ocean and coastal areas

A large set (- 100 data points at each wavelength) of multispectral beam attenuation, c(X), data at nine wavelengths (440, 450, 490, 520, 535, 550, 565, 630, and 670 nm) is used to develop a spectral model of the beam attenuation coefficient. The relationship c(X) - cW(X) = [c(490 nm) - cW(490 nm)](1.563 - 1.149 x 1O-3 X) describes the spectral variation of c(X) where cW(X) is the pure water beam attenuation and X is the wavelength in nm. From a subset of the data a relationship of chloropyll (Chl) to ~(490) was found to be ~(490) = 0.39 Ch1°.57; however there is significant scatter in this relationship. The spectral c model was tested with independent data sets and the average percent difference of the measured to predicted values ranged from 0.4 to 5% for the different spectral bands. The spectral beam attenuation coefficient c is an important property in optical oceanography. Because c (units given in list of notation) is an inherent optical property (Preisendorfer 1976) it is an important measurement in the optical characterization of a water sample. The inverse of the beam attenuation coefficient, the attenuation length, is important as a scaling parameter for problems in imaging and radiative transfer. When calibrated correctly, c can be used to determine the suspended particulate load in a water sample. Reported measurements of c have predominantly been in a single spectral band. Most often the photopic band has been used (with an instrument such as the Martek transmissometer) or more recently in a spectral band at 660 nm (with an instrument such as the SeaTech transmissometer). Comparisons between data sets of c in different spectral bands require knowledge of the spectral nature of c(X). Previous studies, in which measurements at multiple wave