Morel and Prieur (1977) have used Crater Lake as an example of Case 1 natural water in their ocean classification scheme on the basis of autochthonous control of transparency. One indication of the oceanic nature of Crater Lake is the convergence of Kd spectra in Figure 3 and model Kd (not shown) calculated from the bio-optical model of Morel and Maritorena (2001). After increasing their values for Kwater slightly (4.5%) to match the measured Kd at longer wavelengths not affected by phytoplankton, it was possible to fit each curve in Figure 3 with a calculated Kd spectrum from 350–550 nm by adjusting a hypothetical chlorophyll-a value from which the model calculates Kd. On 20 August 2001 these fitted values for different depths ranged from 0.03–0.33 mg m-3 chlorophyll-a and compared well with chlorophyll-a concentrations estimated from particulate absorption at 675 nm (Figure 12A), which ranged from 0.04–0.38 mg m-3 using a specific absorption coefficient at 675 nm of 0.04 m-1 per mg m-3. For chlorophyll-a measurements from samples collected on 14 August 2001 and 11 September 2001 (chlorophyll-a extracted from 0.45 micron filters) the concentration of chlorophyll-a in the 0–30 m range of depths was 0.014 mg m-3 and the value near the DCM was 0.12 mg m-3.
Seasonal and interannual variations in radiometer-derived Kd,320 and Kd,380 (Figures 7A and 7B) were small but significant during the period 1996–2002 but few UV measurements are available from earlier years. The first Crater Lake UV measurement (August 1964) used an experimental instrument to characterize spectral horizontal radiance reflectance, 400–700 nm (Tyler 1965). Although the spectrum was clearly related (inversely) to absorption in the upper meter of the water column, it is difficult to compare because of the unconventional approach. Measurements in August 1966 of upwelling and downwelling spectral irradiance to 25 m from 360–700 nm were more useful (Smith & Tyler, 1967; Tyler & Smith, 1970). When we recomputed Kd from reported Ed values at 0, 5, 15, and 25 m we discovered substantial noise in the blue and UV wavelengths but used an exponential model (Bricaud et al., 1981) to estimate Kd at 380 nm from wavelengths 360–390 nm (depth-averaged Kd using 5–15 and 15–25 m Ed data pairs) and then adjusted for shallow depths using the relationship in Figure 5 to compare with Kd averaged over 0–40 m (Figure 12). After these adjustments the values were still historically at the low end of observed Kd suggesting that Crater Lake was extremely clear during the 1960s. The data collected in August 1969 included upwelling and downwelling spectral irradiance to 99 m over 360–700 nm (Smith et al., 1973) but no tables were provided; values were estimated from published graphs and showed that Crater Lake was similar in transparency to the average of the recent period (1996–2002). For both 1966 and 1969 the values for Kd,320 were not measured but were estimated from measured Kd,380 using the average relationship in Figure 6.
Larson et al. (1996a) discussed determinants of visual clarity in Crater Lake, primarily phytoplankton and abiotic particles (storm-related suspended mineral sediments). Only a few other cases of phytoplankton influencing UV attenuation in clear lakes have been reported (e.g. Sommaruga 2001). While we have argued above that phytoplankton control attenuation during dry periods (see Figures 10B, 11A, 12A), there probably exists a contribution to Kd from suspended allochthonous particles (mineral particles and pollen) at all times (Morel & Prieur, 1977). This contribution may vary in response to wind, turbulence, and proximity to shore. Even without the uncertain contribution of suspended minerals, chlorophyll-a is not a perfect predictor of UV attenuation if phytoplankton vary in their composition of accessory pigments and MAAs. The additional contribution of CDOM produced by phytoplankton requires more investigation given the observation of Boss et al. (this issue) that CDOM increases with depth and during the summer. The possible impact of fires on UV transparency has not been addressed but fires could influence Kd,UV directly (smoke particles entering the water) and indirectly (e.g. influencing turbidity and nutrients in runoff).
