Figure 12. Spectral diffuse attenuation coefficients (Kd J calculated at 10 nm intervals for Crater Lake Oregon, from downwelling irradiance scans at fixed depths, 12:00-13:00 local time, 20 August 2001 (SZA = 31°, clear sky) using a LI-COR LI-1800UW spectral radiometer (8 nm bandwidth single monochromator). Kw X for freshwater estimated by Smith and Baker  is also compared to two new estimates computed by subtracting particulate absorption (measured similarly to spectra in Figure 10) from Kd>A averaged over 0-20 m and 30-40 m, adjusted for sky and sun angle (Figure 4B).
also been demonstrated in a field comparison . Details of Crater Lake UV attenuation data will be published separately (B. Hargreaves, G. Larsen, J. Morrow, and others).
In Crater Lake during August, phytoplankton are extremely scarce in the upper 15 m (chlorophyll concentration ca. 0.1-0.2 mg m~3 [94 and Emmanuel Boss, personal communication]), a depth that also represents Z37o/o for 320 nm. Typically the Crater Lake chlorophyll concentration in August increases with depth to a maximum concentration near 130 m (ca. 0.6 mg m-3 [94 and Emmanuel Boss, personal communication]). As the phytoplankton pigment increases with depth, the phytoplankton optical effects (direct and via CDOM) on UY and PAR diffuse attenuation increase as well (Figure 12). By comparing the curves for Kd32o and Xw32o in Figure 12 it is apparent that phytoplankton and associated CDOM absorption are contributing about 50% of the diffuse 320 nm attenuation at 50 m (Kd320 = 0.092 m-1, Xw320 = 0.045 m"1). Ayoub et al.  used the QFT to determine relative importance and seasonal variations for UV absorption by phytoplankton in the two lakes represented in Figure 10. The contribution of particulates was also high for oligotrophic L. Giles (particles contributed 20-55% of total absorption at 320 nm) but was low for the humic L. Lacawac (8-18% of total absorption, with particles dominated by CDOM-like detritus rather than phytoplankton). Other cases of within-lake variation in UV attenuation attributed to phytoplankton include high mountain lakes [62,63] and Lake Biwa, Japan . More work will be required to determine if depth-variation in phytoplankton biomass and in chlorophyll-specific absorption represent the same biotic mechanisms and range of values reported for horizontal trophic gradients in ocean waters  and among lakes.
Absorption in the UV-A wavelengths includes accessory pigments associated with photosynthesis while UV-B absorption by living cells is caused in part by proteins and nucleic acids. Compounds that absorb with various peaks in the UV range accumulate in some organisms subject to UVR exposure and may serve as UV-B screening compounds: myco-sporine-like amino acids (MAA's) in algae and invertebrates and scytonemin and its derivatives in cyanobacteria [45,50, Chapter 10]. Chlorophyll-specific absorption is known to change with depth in the water column, but while the specific absorption of photosynthetic pigments tends to increase in dim light as depth increases, the response of UV-screening pigments may be opposite. Helbling et al.  adapted the QFT to measure UV and visible chlorophyll-specific absorption by marine phytoplankton (a*chi) in marine waters of Antarctica and observed a UV-protective pigment whenever algae were exposed to at least 1% of incident UV-B (320 nm) irradiance. They observed a UV absorption peak between 310 and 330 nm for algae within the upper mixed layer as deep as 20 m (roughly the Zv/o attenuation depth for 320 nm) and down to 90 m when thermal stratification was weak enough to allow deep mixing. The value for a*Chi at 327 nm varied inversely with the depth of the mixed layer for a large number of samples, especially when diatoms were dominant, suggesting that some algae produce a UV screening pigment in proportion to their UV-B exposure. A similar UV-B particulate absorption peak was observed in Crater Lake, Oregon above the Zio/O)320 depth
(Hargreaves, unpublished). In this case the peak was evident at 50 m (Figure 10), but not at 100 m, when the mixing depth was approximately 10 m and Z\%t 320 was 60 m (Figure 11).
3.3.3 Ultra-low attenuation (Kirk type W, WA, or WG natural waters)
Early studies of UV attenuation identified several ocean regions as unusually transparent. Table 1 shows three regions (Sargasso Sea, East Mediterranean, and central equatorial Pacific) with Kd<0.15 in the range of 310-320 nm, including the earliest radiometer measurement of underwater UV-B attenuation . Crater Lake Oregon is the only freshwater site to receive early attention for its "near distilled water" transparency near the surface. Crater Lake spectra for visible (Tyler ) and UV-A (Smith and Tyler ) wavelengths were recorded by the first underwater scanning spectroradiometer (Tyler and Smith ). An improved instrument recorded UV-A and visible spectral irradiance underwater in Crater Lake during July 1969 . The authors commented on the similarities of Crater Lake water optical properties to those of pure water. There are no published data for Crater Lake using later versions of the spectroradiometer [8,100] that included the capacity to record UV-B wavelengths.
In the 1990's new commercial UV instruments made possible a large number of lake measurements, with the result that high mountain lakes in Austria, North America, and South America have been identified with Xd32o<0.17 (Table 1). Vincent et al.  measured UV attenuation in several Antarctic lakes and reported record low values for the depth range 10-20 m below the ice-covered surface of Lake Vanda (Xd320 = 0.055), values smaller than the attenuation estimated for pure water reported by Smith and Baker . Recent UVR measurements at Crater Lake OR revealed similar low values in surface waters (Xd320 from 0.050 to 0.071, Table 1). As indicated by equation (10), Kd values include contributions from water, phytoplankton, and DOC. These recent low values for two lakes suggest that CDOM and phytoplankton concentrations are very low. The implication that the Smith and Baker's  values for Kw are too high in the UV wavelengths will be addressed in the next section. In both Crater Lake and Lake Vanda, UV attenuation changes with depth and is minimal near the surface. From the limited information available it appears that both phytoplankton and CDOM contribute to this change in UV attenuation with depth, as discussed in the previous section.
At Crater Lake, DOC levels have been reported to be less than 16 ¿¿M (equivalent to 0.2 mg l-1) with algal biomass in surface waters <0.2 mg m-3 chlorophyll [94 and Emmanuel Boss, personal communication]. At Lake Vanda, DOC in the UV-transparent surface water is reported to be 0.3 g m-3 with algal biomass <0.1 mg m-3 chlorophyll . The role of CDOM at such low attenuation levels is difficult to measure by spectrophotometer and even the concentration of DOC is at the level of "blank" values for high temperature oxidation instruments . If one assumes that the surface value for DOC in Crater Lake is actually 0.2 g m~3 at the time of low Kd measurements, the absorption by CDOM can be estimated using the DOC specific absorption relationship published for other sites. Marine DOC concentrated from Gulf of Mexico and Mississippi River plume water by Carder et al.  would give acdorn32o = 0.02 m_1 if all the DOC consisted of marine-like fulvic acids (equivalent to acdom4i2~0.001), or acdom32o = 0.11 m-1 if it were all marine-like humic acids. The lower estimate, 0.02 m_1 at 320 nm, could be a component of the minimum value in Crater Lake of 0.05 m_1, depending on the absorption and scattering attributed to water molecules and phytoplankton. Using the reported DOC value of 0.3 g m_3 for Lake Vanda surface waters, the predicted Kd320 is 0.035 m_1 (again, assuming marine-like fulvic acids), compared to measured Xd32o = 0.055 m-1. Xd32o can also be estimated using a regression of Xd32o versus DOC from a series of high latitude lakes (Figure 8) that included Lake Vanda and other Antarctic lakes ( and Table 3). Predicted Xd32o ranges from 0.060 to 0.090 m"1 for lakes with DOC in the range 0.2-0.3 mg l"1 (after adding Kw32o = 0.04). To avoid introducing a latitudinal bias (caused by sun angle differences) future relationships of this sort should be adjusted using equation (11).
The extremely low UV attenuation in surface waters of Crater Lake provides an opportunity to improve the Smith and Baker  "upper bounds estimate" of UV attenuation in pure water, Kw. The recent near-surface measurements of spectral Kd (Figure 12) and spectral absorption of particles (Figure 10) can be combined to make a new "upper bounds estimate" of Kw. Spectral Kw; values averaged over depths of 0-20 m and 30-40 m are shown in Figure 12, along with the freshwater Kw spectrum from Smith and Baker . The new "Kw" estimates have been computed by first adjusting Kd's with equation (11) for diffuse and direct sunlight (see Figure 4B) and then subtracting particle absorption appropriate to the depth. Because of uncertainties with the QFT calibration for this lake, the particulate absorption values were adjusted downward from the values calculated by the Roesler method  until the two estimates of spectral (0-20 m and 30-40 m) converged. The final adjustment was to 37% of the original particulate absorption coefficient values. A justification for this approach is that unpublished measurements of acdom made several weeks later by Emmanuel Boss (personal communication) using a Wetlabs AC-9 in situ absorption meter showed acdom440 essentially uniform in the top 40 m and just above the limit of detection for the instrument. Another supporting argument is that the same adjustment improves an estimate for chlorophyll a concentration at 25 m depth calculated from the particulate absorption peak at 675 nm (Figure 10). By using a*ch])675 = 0.012 derived from September data (2001 AC-9 particle absorption and extracted chlorophyll a concentration; Emmanuel Boss, personal communication), the uncorrected estimate for chlorophyll a concentration (0.5 mg m-3) was reduced to a reasonable value within the normal range for that depth in Crater Lake (0.18 mg m-3). Tassan et al.  have suggested that small phytoplankton (such as those abundant in the surface of Crater Lake) cause the Roesler method  to overestimate particulate absorption. The resulting "Kw" estimates may still include scattering effects of particles present plus absorption by any cdom present that was not correlated with particle absorption. For UV wavelengths the Crater Lake "high" estimates for Kw32q from 0-20 m and 30-40 m range from 0.043 to 0.047 m_1. These are substantially lower than the Smith and Baker  value of Kw32o = 0.09 m-1. Note that these freshwater values of x from Smith and Baker  include slightly smaller backscatter terms than the seawater Kw<x values used in Figure 2.
A laboratory study of spectral absorption by pure water , which extended only into the UV-A wavelengths, reported pure water absorption as aw380 = 0.011 m-1. This value can be converted into ^w380 using equation (9) (with &bw380 = 0.007 from Mobley ) to yield Xw380 = 0.017 m_1, compared to the Crater Lake estimate, Kw38o = 0.018 m_1 (using near surface Kd380 averaged from several instruments, August 2001). Both values are lower than the Smith and Baker  estimate, Kw380 = 0.026. Further refinements in estimates for spectral Kw and aw for UV wavelengths could result from improved measurements of spectra for Kd, particle absorption, CDOM absorption, and particle scattering in the remarkably clear waters of Crater Lake.
Was this article helpful?