Hermann, A.J., G.A. Gibson, N.A. Bond, E.N. Curchitser, K. Hedstrom, W. Cheng, M. Wang, P.J. Stabeno, L. Eisner, and K.D. Cieciel (2013): A multivariate analysis of observed and modeled biophysical variability on the Bering Sea shelf: Multidecadal hindcasts (1970–2009) and forecasts (2010–2040).
It is a safe bet that the future will include a warmer Bering Sea. But it is uncertain exactly how climate change will be manifested, and in particular, how fast it will warm in summer versus winter, and in the north versus the south. Nevertheless, these details in the climate forcing are key in terms of their impacts on plankton distributions and types, and ultimately the entire marine ecosystem. The formidable problem of how climate change is liable to impact lower-trophic levels, i.e., the base of the food web, was tackled under the auspices of the Bering Sea Project using novel methods and massive computing resources. These results were reported in a special issue of Deep-Sea Research. This was the third special issue of DSR devoted to the Bering Sea Project; a fourth and final special issue is currently in development.
Our approach featured high-resolution ocean model simulations using the Regional Ocean Modeling System (ROMS). The regional biophysical model includes interactions among physical water properties, nutrient concentrations, and the growth and consumption of groups of plankton crucial to fish, sea birds and marine mammals. The regional simulations were embedded in large-scale atmosphere and ocean conditions from global climate model predictions carried out on behalf of the Intergovernmental Panel of Climate Change (IPCC). ROMS is much more realistic than the global models in representing smaller-scale effects of bottom topography on the currents and temperature. Such regional, "downscaling" model predictions have an inherent uncertainty, partly inherited from the global simulations. Nonetheless, they are showing some important changes may be in store for the Bering Sea, such as a northward shift in both pelagic and benthic biomass.
As with nature itself, such complex biophysical models include multiple interacting factors, so the interpretation of results is a significant challenge. For our analysis, we adapted a multivariate statistical technique to look simultaneously at the covariance among different factors (e.g., temperature and nutrients), and the covariance among regions (e.g., the southern shelf vs. the northern shelf). This procedure helped us to identify which regions could experience a particular mode of coupled biophysical change - for example, a rise in temperature accompanied by a decline in large crustacean zooplankton on the southern Bering Sea shelf.