A data assimilative, coupled physical–biological model for the Coastal Gulf of Alaska

Abstract

A data assimilative, coupled physical–biological model for the Coastal Gulf of Alaska (CGOA) is used to investigate the extent to which improvements to oceanic circulation yield improvements to lower trophic level ecosystem predictions, especially in relation to mesoscale variability at the shelfbreak. The ocean circulation component is an implementation of the Regional Ocean Modeling System (ROMS), the lower trophic level ecosystem component is a six-compartment Nutrient-Phytoplankton-Zooplankton-Detritus (NPZD) model with iron limitation, and the data assimilation component is the adjoint-based, four-dimensional variational (4D-Var) system available in ROMS. Assimilated observations consist of weekly satellite sea surface height and temperature, as well as bimonthly in situ temperature and salinity measurements. Simulation results for 1998–2002 indicate that assimilation of physical observations significantly improves the accuracy with which the model reproduces the frequency, duration, and intensity of eddy events along the CGOA shelfbreak. Improvements to oceanic mesoscale processes lead to substantial improvements to the biological response predicted by the NPZD model. Observed and simulated correlations between eddy kinetic energy and surface chlorophyll concentrations suggest that ecosystem dynamics at the shelfbreak is tied to eddy activity in the northern CGOA (i.e., off the Kenai Peninsula and Kodiak Island). In the southern CGOA (i.e., off the Shumagin Islands), mesoscale processes and ecosystem response at the shelfbreak are uncorrelated, as eddies tend to occur during winter when phytoplankton growth is severely light-limited. Based on observation and control vector impact calculations for physical (eddy kinetic energy) and biological (surface chlorophyll concentrations) processes, improvements to oceanic circulation and ecosystem dynamics are primarily associated with the assimilation of satellite sea surface height observations, and occur mainly through adjustments of the model initial conditions. These similarities in the observation and control vector impacts lend further evidence to the linkages between mesoscale activity and primary production along the CGOA shelfbreak.