Figure 1A. Crater Lake solar irradiance spectra (304–750 nm) at depths 20–160 m, 20 August 2001 (air scan 12:04 PDT). LI-COR LI-1800uw spectroradiometer (8 nm bandwidth; cosine response to downwelling irradiance). Incident Ed irradiance (thick line) plotted along with model Eo irradiance (thin line, distinct from the measurements only at the lowest and highest wavelengths) from RTBasic (Biospherical Instruments; 8-stream option, 2 nm bandwidth; model output adjusted using 8 nm running averages of model Log (Ed) to account for the 8 nm bandwidth of the LI-1800uw data; ozone=313 DU from satellite; Patm=774 torr from elevation). Ratios of measured:modeled data averaged 98% for 308–800 nm, and 96% for 308–400 nm (range 87–109%).

Figure 3. Crater Lake spectral diffuse attenuation from spectral irradiance in Figure 1, plus an average for 10–18 m depth calculated from profiles by PRR-800 and PUV-2500 instruments from Biospherical Instruments, Inc. (BSI). LI-1800uw Kd was calculated for each depth interval and then averaged for a range of depths (error bars are +/-S.E. of mean for depth ranges included in mean). BSI Kd was calculated for 10–18 m from polynomial regressions of Ed,Z vs depth to reduce noise from surface reflections. Note that the minimum Kd was at 412 nm. Values for wavelengths longer than 560 nm are excluded below 40 m because of Raman scatter artifact described in Figure 2B. Shorter UV wavelengths are excluded at depths below 55 m because of instrument detection limits. The lowest Kd,320 value was 0.057 m-1 (PUV-2500, 10–18m).

Figure 4. Kd versus depth for selected wavelengths in Crater Lake (20 August 2001) showing minimum values at surface and increasing values with depth to the deep chlorophyll maximum (DCM). Four wavebands and three instruments are compared (smoothed with 8 m running averages above 60 m). Narrow-band signals show parallel changes with depth while the PAR waveband increases near the surface (where highly-attenuated red wavelengths are detected). BSI instruments (PUV-2500, solid circles; PRR-800, solid triangles) give similar results to LICOR LI-1800uw (open circles) except for greater noise in the latter near the surface because of long scan times. The 320 nm curve for LI-1800uw data was limited to about 75 m because of instrument sensitivity but the PUV-2500 was only lowered to 63 m. The PRR-800 radiometer was not equipped with a 320 nm sensor during this comparison and the PUV-2500 PAR signal is not included for greater clarity. The optically-mixed depths correspond to the thermally mixed epilimnion (0–10 m, not shown). The lowest Kd,320 averages were 0.057 m-1 (PUV-2500, 10–18m) and 0.058 m-1 (LI-1800uw, 0–20 m).