Ultraviolet Radiation and Bio-optics in Crater Lake, Oregon, 2005
In studies of underwater light in aquatic ecosystems it is customary to characterize the transparency or attenuation of natural waters instead of underwater irradiance because transparency and attenuation are persistent properties from which underwater UVR irradiance can be calculated for any given time and depth. A useful measure of UVR transparency in natural waters is Kd,?, the spectral diffuse attenuation coefficient for downwelling irradiance Ed,? (Baker & Smith, 1979). Kd can be used to characterize water transparency and factors controlling transparency or to reconstruct underwater solar spectra as a function of depth. Kd is calculated either from discrete measurements of Ed made at several depths or from a set of Ed values recorded over a continuous range of depths. An average Kd is often calculated for a range of depths considered to be optically mixed (e.g. the upper mixed layer or epilimnion) but depth-specific Kdvalues can also be derived from underwater measurements to reveal how attenuation varies with depth.
Valid spectral applications of Kd values are difficult to obtain unless the wavebands are narrow enough to be “spectrally-neutral” (where Kd,? varies little across the waveband). An example of a broad yet spectrally-neutral waveband in clear water is 400–500 nm. In contrast, when underwater Kd is calculated in clear water for the PAR (400–700 nm, PAR = photosynthetically active radiation) waveband, attenuation of solar radiation varies strongly with depth even though the water is uniformly mixed (Kirk, 1994b). This is because spectral variation in Kd,? leads to shifts with depth in relative proportions of different wavelengths within the waveband. Kd,PAR in a uniformly-mixed body of clear water is much greater at the surface than it is deeper because at the surface the strongly attenuated red part of the solar spectrum contributes to the average Kd; at deeper depths only the weakly attenuated blue and violet wavelengths are still present and only these contribute to average Kd. Another example of a problematic broadband application is when the entire UV-B range of wavelengths is used to calculate a single Kd,UVB, yielding values that are difficult to compare or interpret (Hargreaves, 2003). Underwater spectral radiometers useful for Kddeterminations have moderate bandwidths in the range of 10 nm or less (Kirk et al., 1994), although significant spectral shifts can occur with moderate bandwidth sensors at the shorter UV-B wavelengths (Patterson et al., 1997).
Other sensor properties can influence spectral Kd measurements. For downwelling irradiance (Ed) a sensor with an accurate cosine response to the angle of incident photons is needed. Accurate determinations of Kd,? are possible when a sensor is not accurately calibrated as long as the same sensor is used in all measurements, travels in a vertical plane during displacement over precisely determined depths, and maintains stability of its wavelength sensitivity, calibration, and cosine response to the angular distribution of light. In practice a correction for a dark offset signal and response to changing temperature may need to be incorporated into the measurement protocol (Kirk et al., 1994). Internal radiation sources (fluorescence and Raman scattering, also called “inelastic scattering”) can interfere with attempts to relate Kd,? to other optical properties when these contribute a significant fraction of the detected irradiance (Haltrin et al., 1997; Gordon, 1999). Measurement of spectral reflectance ratios (either irradiance reflectance, Eu/Ed, or radiance reflectance, Lu/Ed) can suggest the wavelengths and depths where such interference is occurring (Haltrin et al., 1997) but must account for self shading of the upwelling signal when a large instrument package is deployed (Dierssen & Smith, 2000).