A Bayesian hierarchical model for estimating the phytoplankton physiology parameter PBopt in the equatorial Pacific

The Vertically Generalized Production Model (VGPM) is one of the most widely used algorithms to derive primary production rates from satellite ocean color data. Its output is highly dependent on the parameter PBopt, the maximum biomass-specific daily photosynthetic rate in a water column. When the VGPM uses measured field values of PBopt it predicts 80% of primary production variability, while when it uses estimated values it accounts only for 58%. Currently, there is no method available to measure PBopt from space and empirical relationships are required to relate it to other remotely-sensed environmental parameters. The standard relationship used to estimate PBopt is a high-order polynomial defined with remotely-sensed sea surface temperature alone. In this analysis we used Bayesian hierarchical modeling and the Climatology of Primary Productivity (ClimPP) dataset to provide an improved regional estimate of PBopt for the equatorial Pacific. ClimPP was collected in the years 1983 to 1995 in 31 research cruises that covered the area, and includes profiles of primary production and chlorophyll, and measurements of surface photosynthetically active radiation (PAR), sea surface temperature (SST), and mixed layer depth (MLD). Our Bayesian hierarchical approach captured the interactive nature of the region’s marine systems at different spatial scales. MLD had a dominant effect at large spatial scales, while SST and PAR were important at the sampling station level. These relationships had different significance in the Western Pacific Warm Pool and the Pacific Equatorial Divergence. This improved PBopt model combined field observations, prior knowledge of phytoplankton physiology, and ecological geography, and resulted in a regional characterization of PBopt that improved VGPM-derived primary production rates in the equatorial Pacific.