Many coastal-spawning fish utilize multiple estuaries as nursery grounds, which guards against the population as a whole being adversely affected by recruitment failure in any single estuary. Some such species also undertake coastal migrations that extend further afield each year as the adults age. In these species, the location and timing of spawning associated with age-dependent behavior, coupled with natural variability in ocean currents that transport eggs and young larvae, can affect the supply of larvae to nursery areas and ultimately structure the pattern of overall population recruitment. Atlantic menhaden (Brevoortia tyrannus), sometimes called "bunker" in Mid-Atlantic States, is one such species. In this collaborative study, physical oceanographers at Rutgers University developed a hydrodynamic computer simulation model of ocean circulation in the Mid-Atlantic Bight that was utlized by fisheries science colleagues at the University of Maryland to underpin a coupled physical-biological model of menhaden larval dispersal and survival post-spawning. The hydrodynamic model ROMS (Regional Ocean Modeling System) is used by an international user community of some 4000 ocean scientists, and has been widely applied to studies of coastal ocean biogeochemistry, geomorphology, and ecosystems to deduce transport pathways for nutrients, sediments and pollutants, in addition to larvae. The ROMS computer software is open source, and distributed and maintained through a registered user portal www.myroms.org hosted at Rutgers University. Simulation models for limited geographic sub-regions of the ocean – here, the Mid-Atlantic Bight continental shelf – must be constrained at their perimeter by open boundary conditions. When this information (ocean temperature, salinity, velocity and sea level) is derived from larger domain basin or global model, as we do here using NOAAÆs Integrated Ocean Observing System (IOOS) HYCOM model, this procedure is typically referred to as "downscaling". Whilst downscaled hydrodynamic models are generally skillful at reproducing the statistics of ocean current variability, much greater skill at reproducing individual circulation events is achieved if direct ocean observations are formally assimilated into the model in the same fashion that atmospheric data are incorporated into Numerical Weather Prediction (NWP) models. This project achieved significant improvements in the methodology of 4-Dimensional Variational data assimilation (4DVAR) in ROMS to incorporate observations of sea level anomaly (SLA) from radar altimeter satellites, sea surface temperature (SST) from infrared and passive microwave radiometer satellites, and surface currents from land-based high-frequency radar (CODAR) in the model-based analysis. Furthermore, a procedure was developed by which 4DVAR could be applied to estimate the mean climatological state of the ocean constrained by data from an historical archive of moored current-meters, more than 10 years of shipboard acoustic current meter data acquired by a commercial vessel that regularly traverses the route from New York to Bermuda, and more that 600,000 vertical profiles of temperature and salinity held in archives at the National Ocean Data Center and National Marine Fisheries Service. This estimate of the true ocean mean state was used to remove regional biases in the global HYCOM model that would otherwise have introduced errors into the downscaling process, and to define an accurate mean sea surface height map to sum with the anomaly sea level data from satellites. With these aspects of the 4DVAR assimilation improved, a retrospective reanalysis of the period 2006-2012 was conducted assimilating all available satellite SLA and SST data, and CODAR currents. The outcome is the most accurate reanalysis of coastal ocean circulation thus far achieved for this region, as assessed by a comparison of the model results to those from 6 other models. We therefore have...
©2020 Biological and Chemical Oceanography Data Management Office.