The dissolved yellow humic substances in surface waters have (Chapter 3) an absorption spectrum rising exponentially into the blue. Where, as in many lakes, rivers and estuaries, CDOM is the dominant contributor to light absorption, there are generally found to be simple relationships between CDOM concentration and the ratio of reflectance in the blue to that at some longer wavelength. Bowers et al. (2000) studied the relationship between subsurface radiance reflectance (rrs) ratios and CDOM in shallow inshore waters of the Clyde Sea (Scotland), a region receiving a considerable input of fresh water from land run-off. The ratio of reflectance in the red (670 nm) to that in the blue-green/blue at 490,443 or 412 nm increased linearly with concentration of yellow substance, as expressed through measurement of 440 nm absorption (g440) on a filtrate. g440 could thus in principle be obtained from remotely sensed reflectance ratios. In the more turbid waters of the Conwy estuary (North Wales) Bowers et al. (2004) again found a linear relationship between rrs(670)/ rrs(490) and g440 but the slope and intercept were different from those previously found for the Clyde Sea. The authors attributed the difference to the fact that in these waters, suspended sediment particles contributed significantly to total light absorption. In Galway Bay (Ireland), Goddijn and White (2006) found a linear relationship between g440 and the red/ blue output ratio of a digital camera.
Among the world's major marine ecosystems, the Baltic Sea is unique in the extent to which optical properties are dominated by yellow substances, a consequence of its very high inflow of fresh water from northern European rivers combined with restricted interchange with the North Sea through the Straits of Denmark. Kowalczuk et al. (2005) studied the relationship between remote sensing reflectance (Rrs), measured at sea level, and CDOM absorption, at a large number of stations in the southern Baltic Sea. Using SeaWiFS wavebands they found the most satisfactory retrieval variable to be the ratio of reflectance at 490 nm to that at 590 nm. For CDOM absorption at 400 nm the data could be represented by the algorithm aCDOM(400 nm) = 10 f (X)
where f (X) = -0.20 - 0.50 X + 0.65 X2 and X = log10 (Rrs /Rra)
Schwarz (2005) using a simpler algorithm lng440 = -0.1123 - 0.8725 ln[Rrs(443)/Rrs(510)]
based on a limited number of in situ measurements, used SeaWiFS data to map dissolved organic carbon concentration in the Baltic Sea.
Kahru and Mitchell (2001) compared in situ measurements of CDOM concentration at stations in the California Current (at 300 nm because of low CDOM in these waters) with concurrently collected SeaWiFS radiances. The algorithm for CDOM concentration derived from the data was aCDOM(300) = 10(-0393-0 872 R)
where R is the ratio of normalized water-leaving radiance at 443 nm to that at 510 nm. D'Sa and Miller (2003) examined the relationships between (sea-level) Rrs(1) values in SeaWiFS wavebands, and CDOM absorption in the Mississippi outflow region of the northern Gulf of Mexico. The reflectance ratios for 412 and 510 nm, 443 and 510 nm, 510 and 555 nm, were all highly correlated with CDOM absorption, the logarithm of the reflectance ratio decreasing linearly with log aCDOM (412). For the 443/510 pair the relationship was log10 aCDOM(412) = -0.874 - 2.025log10[Rra(443)/Rra(510)]
Menon et al. (2006) used in situ measurements of water-leaving radiance (Lw) and CDOM absorption in estuarine waters in Goa, on the west coast of India to develop an algorithm for use with the Ocean Colour Monitor on board the Indian Remote Sensing (IRS) satellite. They found that the diminution of the ratio of radiance at 412 nm to that at 670 nm with increasing CDOM concentration could be expressed as acDOM (440) = 2.9393 R~2'2486
where R = Lw(412)/Lw(670), and they used this algorithm to map the distribution of CDOM in the estuaries and the nearby inshore waters. Kutser et al. (2009) made use of the high radiometric (16 bit), and spatial (30 m), resolution of the Advanced Land Imager on the EO-1 satellite, to map the distribution of CDOM in Estonian coastal waters (Baltic Sea), using the algorithm
where R4 and R5 are the reflectances (atmospherically corrected by the dark pixel method) in ALI Bands 4 (525-605 nm) and 5 (630-690 nm), respectively. Their data revealed the high local variability of CDOM concentration, and the consequent need for high spatial resolution for mapping it, in coastal waters.
When the optical properties of a surface water are dominated by a single component, such as phytoplankton or CDOM, simple algorithms such as those described above might reasonably be expected to provide a realistic picture of the distribution of the component in question. The greater part of the ocean consists of Case 1 waters in which phytoplank-ton is the major optically significant component. In humic lakes and in coastal regions receiving substantial outflows of gilvin-rich river water, CDOM is often the major contributor. In most Case 2 waters, however -coastal, typically shallow, often with turbid river inflow - phytoplankton, CDOM and suspended sediments all contribute significantly to the optical character of the water and application of the simple algorithms is likely to give highly misleading results.
A straightforward, and sometimes feasible, way of addressing this problem is to develop simple empirical algorithms that work in a given locality, but have no general application. A more challenging approach is to try to develop algorithms that can, from remotely sensed radiance values, extract the actual inherent optical properties themselves - a(l) and bb(1) - and then apportion these appropriately to the components of the medium. There have been numerous attempts to achieve this: some using purely empirical relationships, and others - referred to as semiana-lytical or quasi-analytical - which combine empirical with algebraic relationships arising out of radiative transfer theory. Many such algorithms are now available which give promising results, and they are comprehensively described in IOCCG Report No. 5 - Remote Sensing of Inherent
Optical Properties: Fundamentals, Tests of Algorithms and Applications.787 There is, however, a fundamental problem underlying this approach, namely that of ambiguity. Defoin-Platel and Chami (2007) point out that solutions arrived at may be non-unique since several combinations of IOP can lead to the same reflectance spectrum.
A special case of retrieval of an inherent optical property, which seems to work in both Case 1 and Case 2 waters, is the measurement of yellow substance absorption using airborne lidar. Gilvin, the dissolved yellow humic constituent of natural waters, like chlorophyll, emits fluorescence when excited by the laser beam: its emission spectrum extends through the visible region with a broad peak at ^530 nm when excited at 470 nm (Fig. 7.12). It has been monitored in river water in terms of the combined emission at 531 and 603 nm,165 and in coastal water in terms of the emission at 500 nm.1187 Hoge et al. (1995) used a frequency-tripled Nd: YAG laser at 355 nm flown on a P-3B aircraft to excite CDOM fluorescence over the Middle Atlantic Bight from Delaware Bay to the Sargasso Sea, in Monterey Bay and in the Gulf of Mexico. Using the CDOM fluorescence signal at 450 nm, normalized by the 404 nm water Raman spectral line height to allow for variation in attenuation by the water (see above), and expressed relative to the fluorescence of a quinine sulfate standard, they developed an algorithm for aCDOM(355), the absorption coefficient of CDOM at the exciting wavelength. Making a reasonable assumption about the value of the spectral slope parameter, S (eqn 3.5), the absorption coefficients of CDOM at other wavelengths can then be calculated.
It is worth noting that the Raman signal itself can be used for mapping the attenuation properties of the water: the reciprocal of the signal is proportional to an attenuation coefficient intermediate in value between Kd and c.
Was this article helpful?