Sensitivity of Northwest North Atlantic Shelf Circulation to Surface and Boundary Forcing: A Regional Model Assessment

Abstract

Listen

Translate article

ABSTRACTThe northwest North Atlantic shelves, influenced by both North Atlantic subpolar and subtropical gyres, are among the most hydrographically variable regions in the North Atlantic Ocean and host biologically rich and productive fishing grounds. With the goal of simulating conditions in this complex and productive region, we implemented a nested regional ocean model that includes the Gulf of Maine, the Scotian Shelf, the Gulf of St. Lawrence, the Grand Banks, and the adjacent deep ocean. Configuring such a model requires choosing external data to supply surface forcing and initial and boundary conditions, as well as the consideration of nesting options. Although these selections can greatly affect model performance and results, they are rarely systematically investigated. Here we assessed the sensitivity of our regional model to a suite of atmospheric forcing datasets, to sets of initial and boundary conditions constructed from multiple global ocean models and a larger scale regional ocean model, and to two variants of the model grid — one extending farther off-shelf and resolving Flemish Cap topography. We conducted model simulations for a 6-year period (1999–2004) and assessed model performance relative to a regional climatological dataset of temperature and salinity, to observations collected from multiple monitoring stations and cruise transect lines, to satellite sea surface temperature (SST) data, to coastal sea level estimates, and to descriptions and estimates of regional currents from literature. Based on this model assessment, we determined the model configuration that best reproduces observations. We find that although all surface forcing datasets are capable of producing model SSTs close to observed, the different datasets result in significant differences in modelled sea surface salinity (SSS), with the European Centre for Medium-range Weather Forecasts’ (ECMWF) global atmospheric reanalysis (ERA-Interim) performing best. We also find that initial and boundary conditions based on global ocean models do not necessarily produce a realistic circulation, whereas using climatological initial and boundary conditions (constructed from long-term, monthly-mean output from a larger scale regional model) improves model performance. Through this model assessment, we determine the model configuration that best reproduces observations and gain generally applicable insight into the factors that are key to accurate model performance.