Primary productivity in the North Atlantic (59¿29'N, 20¿50'W) was estimated by applying a ''light-pigment'' productivity model (Kiefer and Mitchell, 1993) to mooring data collected during the spring of 1989. We show that the choice of the parametrization of the light captured by phytoplankton cells in a turbulent mixed layer has a significant effect on the calculated productivity estimates. It appears that the quality of such estimations benefits largely from using high-resolution time series data (minutes). We also examined phytoplankton dynamics by incorporating the Kiefer-Mitchell model into a one-dimensional model of the turbulent mixed layer (Mellor and Yamada, 1982).

The calculated time-depth distribution of phytoplankton biomass compares relatively well with that measured in situ. The model results indicate that small changes in the water column stability can be sufficient to initiate phytoplankton bloom even before the apparent formation of the seasonal thermocline. The model also describes the diel cycle of biomass concentration, suggesting that near the sea surface the daytime losses of biomass by vertical diffusion can be much larger than nighttime losses. Thus, if not accounted for, such losses may bias estimates of primary production from diel variations in biomass concentration, for example, when using a method based on beam attenuation measurements. These losses should also be considered for the proper interpretation of in situ primary production measurements by incubation methods. ¿ American Geophysical Union 1994