Real-time ocean observatories have been deployed in two locations on the southeastern coast of Nova Scotia, Canada. Both systems are equipped to measure hyperspectral surface reflectance and downwelling irradiance at discrete depths in four wavebands. These measurements allow estimation of chlorophyll concentration or phytoplankton absorption coefficients that can be utilised in a biophysical model of the area and in long-term monitoring and forecasting programs. Complementing these data is an extensive ground truth data set which includes profiles of irradiance, scattering, absorption and fluorescence and discrete determinations of pigments, absorption and suspended particles. In optically complex case 2 waters such as these, ocean colour band-ratio models, such as OC4, do not accurately retrieve chlorophyll concentration due to high concentrations of chromophoric dissolved organic matter (CDOM) and/or sediment. In situ fluorometry provides an alternative means to estimate chlorophyll, but the physiological effects of bright light and other biological sources of variability induce errors. The utility of an inverse model of hyperspectral ocean colour and attenuation, that includes an explicit assessment of solar-induced chlorophyll fluorescence, has shown its ability to accurately retrieve estimates of phytoplankton absorption coefficient, a[sub]phi, through a region-specific parameterisation and assumption of a constant relationship between fluorescence quantum yield and irradiance. On the basis of the success of this model within our observation area, we aim to simplify and optimise our estimates of chlorophyll concentration and a[sub]phi using our extensive ground truth data and by implementing a statistical time series model.