The productivity of giant kelp forests is highly variable across time and space. Winter storms and summer periods of nutrient limitation act as bottom-up regulators of kelp abundance and growth in a geography-dependent manner. Our goal is to develop a predictive understanding of giant kelp forest dynamics in the nearshore waters of California using a combination of (i) bio-optical modeling of kelp productivity, (ii) high-resolution remote sensing of kelp cover, biomass and its physiological state. The consistent optical signature of the kelp canopy produces a Blade Area Index (BAI) that is linearly related to diver counts of kelp abundance. The slope (0.1) indicates that the canopy structure exerts a strong package effect on the optical efficiency of light absorption. The ability to predict BAI allows retrieval of standing biomass and productivity from measurements of below-canopy irradiance. Maps of giant kelp density and productivity were derived from 2.4 m spatial resolution hyperspectral imagery obtained over Carmel Bay (central California) and the Santa Barbara Channel (SBC) ranging from Gaviota east to Campus Point (~35 km). Spatial resolution of each scene was degraded to larger pixel sizes and reanalyzed to produce maps of giant kelp areal coverage, biomass and productivity. The relative bias at each resolution was calculated by normalizing the biomass and productivity of each scene to the values retrieved in the original (2.4 m resolution). Large errors in the retrieval of biomass and productivity estimates were evident in all scenes at pixel sizes above 50 m. We are beginning to explore the effects of patch dimensions, size distribution and proximity statistics on the retrieval bias generated by image resolution. Although fractal properties appear to scale with patch size at all sites, there are clear differences in these relationships that may affect area and biomass/productivity retrievals as spatial resolution decreases.