Despite strong evidence that key ecosystem rates vary across regions and seasons, biogeochemical (BGC) models typically rely on fixed parameter values. This limits model forecast skill, contributes to divergence among primary production estimates, and lowers confidence in regional- and climate-scale projections. We present a new observation-constrained framework, developed within the European Space Agency Climate and Marine Production (ESA-CAMP) project, to estimate seasonally varying ecosystem model parameters using one-dimensional (1D) water-column simulations along long-term stations and BGC-Argo trajectories.
Our approach couples a 1D hydrodynamic model with biogeochemical models and, together with data assimilation, estimates model parameters from observations. In other words, we use the observations to identify the parameter values that allow the model to reproduce the observed biogeochemical behaviour more realistically. We focus on parameters controlling phytoplankton light-limited growth, photoinhibition, maximum growth, respiration, and zooplankton grazing and assimilation. Initial applications span contrasting oceanographic regimes, including the English Channel, the North-West European Shelf, the Mediterranean Sea, the North Atlantic and the North Pacific, based on selected BGC-Argo trajectories and ocean observing stations.
Preliminary results show coherent seasonal structure and site-to-site variability in both phytoplankton and zooplankton parameters, indicating that fixed parameterisations are unlikely to be adequate across diverse marine environments. These emerging parameter climatologies provide a basis for mapping parameter fields in space using machine-learning approaches, with the objective of improving primary production estimates, biogeochemical forecasts, and regional- to climate-scale ecosystem projections under changing environmental conditions.

暂无