Z37 m

Proposed Kw320 (fw)=0.044 Crater Lake Kd,0=0.050 Antarctic Lake (Vanda) Kw320 (fw)=0.092 S&B 1981 Sargassso Sea Mediterranean Sea Pacific, equatorial S. America lakes Red Sea

W. Greenland ocean Austrian Alps lakes Coastal Japan Southern Ocean, Antarctica Low elev. Lakes Arctic Ocean Coral reefs North Sea Gulf St. Lawrence Arctic lake Baltic Sea

Figure 3. Maximum 37% attenuation depths for 320 nm (depth where irradiance is attenuated to 37% of the value just beneath the surface, computed as 1/Kd320)- Values have been calculated for the lowest Kd's reported for each category from Table 1.

ent component scattering and is proportional to the concentration of scattering substances. These three IOPS's are expressed in units of m_1.

A property of the light field that relates AOPs with IOPs is the mean cosine (/I). It summarizes the angular distribution of photons according to equation (5). Direct measurement of (i combines downwelling irradiance (Ed, with cosine response to solar zenith angle, maximal to vertical light from above and no response to horizontal light or light from below), upwelling irradiance (£u), and scalar irradiance (E0, equal response to light from any direction):

Underwater ft varies with sky and water conditions, the angle of the sun from vertical (solar zenith angle), wavelength, and depth [17]. From equation (5) one can establish a theoretical value (assuming no scattering) of 1.0 for a collimated nz cz c c [

beam from above, —1.0 for a beam from below, and zero for a completely diffuse light field. Direct measurements of scalar irradiance have not been reported using commercial underwater UY instruments although Danish scientists have reported upwelling and downwelling scalar irradiance and corresponding values for fit and ¿uu [117]. Commercial underwater radiometers are available for determining spectral reflectance ratios for visible and UV wavelengths (Ed, Eu, and also upwelling radiance Lu). Stramska et al. [30] have proposed and tested a method to calculate p, as well as a and bh (the backscattered portion of b) from field measurements of Ed, Eu, and Lu in the wavelengths from 400-560 nm. This approach is promising for determining p. over UV wavelengths but it will require validation beyond the currently specified range of wavelengths.

The mean cosine relates "a" and Kd in natural waters when these optical properties describe a narrow waveband. An exact relationship valid for all depths in the absence of any "internal light source" [17,31] is

where R is irradiance reflectance (EJEd) and Ku is the diffuse attenuation coefficient for upwelling irradiance. Internal light sources include Raman scattering and fluorescence emitted from DOM or chlorophyll after absorbing light at shorter wavelengths. When Eu Ed, and thus when R becomes very small, equation (6) becomes

where ji is less than 1 and thus Kd is greater than "a" to account for the longer mean path traveled by either diffuse or off-vertical sunlight for each vertical metre in the water column. The exact relationship in this form represents Gershun's Law [32]:

where KE is the attenuation coefficient for the net downward irradiance, Ed — Eu. From the modeling work of Gordon [17], the mean cosine for downwelling irradiance (/¡d) also relates Kdfi (measured just below the surface) to IOPs at that depth in a useful empirical equation:

In equation (9) ¿¡d,o is the fraction of downwelling scalar irradiance contributed by downwelling cosine irradiance (EJE0^), measured just below the surface, and bh is the backscattered fraction of b. This relationship was developed for Case 1 waters (described in Section 2) and should be tested for validity in non-Case 1 waters. Other empirical relationships for predicting Kd from IOPs are described in Kirk [6,10],

Underwater spectral UVR measurements have sometimes been summarized by integration over a broad waveband (e.g., for UV-B and UV-A bands in [33] and [34]). The response of a broadband attenuation measurement, whether calculated from a detailed spectrum or measured with a broadband sensor (e.g., PAR), is subject to errors when used to characterize the optical properties of the water. These errors occur in attempting to characterize a uniform section of the water column (where IOPs are constant) because the effective Kd and effective "central wavelength"' for the waveband will shift with depth and with the magnitude of attenuation [11,35]. In the UV-B waveband, for example, a simple spectral attenuation model can demonstrate that the wideband /Quvb calculated for a specific depth deviates from the surface KdUVb (full solar spectrum) by 12-19% over a range of Kd (Kd32o = 0.1-22 m_1) and depths {Z^-Z-^yX For XdUVR the effect is even greater; using published coastal ocean spectral data Booth and Morrow [11] calculated that XduvR would change 36% with depth (from 0.32 m_1 at the surface to 0.25 m-1 at 15 m depth, with an asymptote of 0.21 m-1 at much greater depth). Instruments with sensor bandwidth of <8 nm appear to perform well throughout the UV-A and UV-B ranges [36]. Spectral modeling has confirmed this: errors caused by spectral shift for the 8-10 nm bandwidth sensors of a widely-used UV radiometer (the PUV-500 from Biospherical Instruments, Inc.) are in the range of 1 % from the surface down to Z io% and less than 5% down to the limit of detection [35]. If data reduction from full spectral data is required, a better strategy than broadband integration is to present attenuation or irradiance values for several narrow wavebands within the UV-B and UV-A wavelengths.

0 0

Post a comment