Differences Between Ground-Based, Satellite, and Model-Derived Photosynthetically Available Radiation (PAR) and Impacts on Modeled Phytoplankton Biomass and Primary Production

Photosynthetically Available Radiation (PAR) incident at the sea surface penetrates into the water column and drives oceanic primary production. Biogeochemical models to estimate phytoplankton biomass and primary production typically require an estimate of PAR, which is available from satellite ocean color imagery and/or atmospheric model predictions. Accurate estimates of PAR are required to derive accurate model biomass and production values. We perform multi-year temporal and spatial comparisons between PAR values derived from ground-based stations in the Gulf of Mexico and coincident values derived from three sources: the Moderate Resolution Imaging Spectrometer (MODIS) imagery, the Navy Operational Global Atmospheric Prediction System (NOGAPS), and the Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS), in terms of scale and temporal coherence, to validate the data sets. We observe seasonal variability in the agreement between the data sets; in winter/spring, the ground-based values typically exceed the satellite values, whereas in the summer, the satellite values are larger. For the spatial comparisons, a large area covering the north central Gulf was selected as the region-of-interest, and PAR values were extracted from the three sources of spatial data (MODIS, NOGAPS, COAMPS), mapped to the same resolution. Agreement between the data sets varied. Finally, we assess the impact of the PAR differences on modeled biomass and primary production by forcing biogeochemical models (Carbon, Silicon, Nitrogen Ecosystem, CoSINE Model; Coastal General Ecosystem Model, cGEM) with scaled PAR values based on the observed differences between the data sets, using both spatially-invariant and spatially-variable PAR values. Our goal is not to validate the biogeochemical models, but rather to demonstrate the impact of PAR variability on model results.