The purpose of this research was to interpret the spectral shapes of backscattering as determined by a HOBI Labs Hydroscat-6 during cruises on the R/V Barnes on August 3rd, 4th, and 5th, 1998 in the West Sound, the Straits, and the East Sound of Washington, respectively. Spectra of particle backscattering (bbp) deviated from an exponential fit with respect to wavelength on all three cruises with spectral peaks observed between 488 and 532 nm. Deviations of bbp spectra from an exponential form can most likely be explained by the occurrence of relatively larger particles in the water column, as determined from spectral particulate attenuation (cp), Coulter counter particle size distributions, and particle backscattering efficiencies (bbp/bp).
Particle backscattering spectra were modelled in Hydrolight using profiles of scattering as determined by a WETLabs AC-9 on the same profiling package as the Hydroscat-6. The spectral shape of bbp in Hydrolight was determined by total scattering (b) and the Petzold phase function (b ) using two different particle backscattering efficiencies of 2% and 1.5%. As b is assumed to be wavelength independent in Hydrolight, the spectral shape of bbp is determined by the spectral shape of scattering with a peak observed in all modelled bbp spectra at 620 nm for both efficiencies. This shape distinctly deviated from those measured by the Hydroscat which peaked between 488 and 532 nm. The bbp computed with a backscattering efficiency of 1.5% was closest in magnitude to those measured by the Hydroscat, as expected from field values of backscattering efficiencies. Hydrolight computed bbp's ranged from one to three times greater than Hydroscat measured values across all wavelengths and depths.
Theoretical spectra of bbp were calculated using the anomalous diffraction approximation (ADA). Inputs to the anomalous diffraction model include wavelength, the index of refraction (n and n'), a size distribution function, and a phase function. The phase function was determined using Mie theory. Theoretical particle size distributions using gammas of -2, -3, -4.8, and a two size distribution with -4 for small particles (<2 microns) and -2 for large particles were used. An n of 1.05 and n' determined from spectrophotometer ap data was used in all cases except for the small size fraction of the two size distribution in which case an n of 1.08 and n' of approximately zero were used. The bbp corresponding to a gamma of -4.8 was best approximated by an exponential relationship, while bbp's obtained for a gamma of -2 and the two size distribution closely fit the spectral shapes of the Hydrolight output. The bbp determined using a gamma of -3 gave a spectral shape with multiple peaks. These results showed that a variety of spectral shapes can theoretically be obtained for bbp. One hypothesis is that the spectral shape of bbp will vary as determined by a balance between scattering by small particles and absorption by large particles, and may result in the spectral shapes obtained by the Hydroscat in this study.
Figure/Table Captions
Figure 1. Spectral shapes of bbp normalized to bbp442. These are bbp spectra from the Hydroscat calculated at each depth as the median of 2 meter bins from 5-point running averaged data. Particle backscattering (bbp) is determined by subtracting backscattering by water (bbw) from total backscattering (bb). Note that none of the curves have an exponential shape with respect to wavelength (l -n) as would be expected for a distribution of small particles. Peaks in all spectra occur between 488 and 532 nm.
Table 1. Various size parameters to determine the relative sizes of particles. Large variability is seen in the bbp angstrom exponents, however these values are a poor estimate of particle size as an exponential fit is an unrealistic fit to the bbp curves from the Hydroscat. Particle attenuation (cp) was determined as the total attenuation (c) from the AC-9 minus dissolved absorption (ag) from the spectrophotometer. Cp angstrom exponents are generally less than 0.1 to 0.2 implying that cp is spectrally flat and thus, that there are relatively large particles in the water column. Log-log slopes, gamma, determined from the particle size distributions of the Coulter counter were all less than 3, further evidence of large particles. Backscattering efficiencies of less than 1 to 2% also lent support to the hypothesis that relatively larger particles were present. Qualitatively, bbp magnitude increased with increasing concentrations of chlorophyll. However, values of ap(676)/chl as an estimate of particle size is probably not valid as it invokes the assumption of purely biological particles in the water column.
Figure 2. Hydrolight output for monday cast 3
(bbp/b=2%). Input to Hydrolight included a Petzold phase function
with bbp/b = 2% (turbid harbor), absorption and attenuation
from the AC-9, a 5 m/s wind speed, the semi-empirical sky model, Raman
scattering by water, and an infinitely deep water column. Hydrolight
calculates particle backscattering as:
where b is the total scattering and b is the phase function which is wavelength-independent in Hydrolight. Thus, the spectral shape of bbp is solely determined by the spectral shape of b.
Figure 3. Hydrolight output for wednesday cast 8 (bbp/b=2%). All of the input parameters were the same as in Figure 2.
Figure 4. Hydrolight output for monday cast 3 (bbp/b=1.5%). All of the input parameters were the same as in figure 2, except that a Petzold phase function of bbp/b = 1.5% (coastal waters) was used. This backscattering efficiency is closer to the efficiencies observed in the field, as seen in Table 1.
Table 2. Imaginary indices of refraction (n') as determined from the Monday Barnes spectrophotometer data at 2 meters depth. n' was determined by the equation n'=as*l /(4p nw) where nw equals 1.34, ap was measured from the spectrophotometer data, and the same scaling factor was applied at all wavelengths such that n' was equal to 0.005 at 671 nm. An n of 1.05 and n' determined from spectrophotometer ap data was used in all cases except for the small size fraction of the two size distribution in which case an n of 1.08 and n' of approximately zero were used.
Figure 5. Particle size distributions used in the ADAs and corresponding spectra of bbp. Particle size distributions with gammas of -2, -4.8, and a two size distribution using -4 for the small particles (<2 microns) and -2 for the large particles were used. The corresponding spectra of bbp from ADA are shown at the bottom of the figure. The bbp corresponding to a gamma of -4.8 was best approximated by an exponential relationship, while bbp's obtained for a gamma of -2 and the two size distribution closely fit the spectral shapes of the Hydrolight output.
Figure 6. The ADA bbp spectrum for a size distribution with gamma of -3. An n of 1.05 and n' determined from spectrophotometer data were used. This spectrum is shown separately of figure 5 to emphasize spectral differences which are small in magnitude. Note that bbp in this case has several peaks unlike the curves shown at the bottom of Figure 5.
