[1] Irradiance spectra were measured at vertical increments within the bottommost layers of landfast sea ice with the aid of divers in Franklin Bay, Canada, in an effort to obtain input parameters for bio-optical modeling of sea ice. The study took place between 22 April and 9 May 2004 during the overwintering stage of CASES ( Canadian Arctic Shelf Exchange Study). The ice was about 1.8 m thick with a snow cover of variable thickness (similar to 0.04 to 0.4 m). Ice surface temperatures increased from about - 12 degrees to -6.4 degrees C during the sampling period, while ice temperatures within the bottommost portion under study ranged from -3.0 degrees to -1.2 degrees C. Ice algae were visible within the bottommost centimeters of the sea ice. This algae layer had a marked effect on the spectral distribution of transmitted irradiance beneath the ice. Particulate absorption spectra, a(p)(lambda), measured from melted ice samples showed evidence of chloroplastic pigment degradation and could not fully explain the shape of the in situ diffuse attenuation coefficient, K-d(lambda), for the algal layer. Interior ice layers, however, did show absorption curves similar to a(p)(lambda) from samples, suggesting the presence of degraded algal pigments within these layers. The discrete ordinates radiative transfer ( DISORT) code was iterated in an inverse approach to estimate a(p)(lambda) and the scattering coefficient, b(tot), from the irradiance profiles. For the bottom 0.1 m of the sea ice, b(tot) was around 400 m(-1), while at the 0.1- to 0.2-m layer from the ice bottom it decreased to 165 m(-1). Using a(p)(lambda) combined with wavelength independent btot as inputs to DISORT seem to adequately explain the radiative transfer near the bottom of first-year sea ice provided that adjustments were made to the brine volume fraction.