Sensitivity of a marine coupled physical biogeochemical model to time resolution, integration scheme and time splitting method

Abstract

Coupled marine biogeochemical models are composed of a hydrodynamic component with a transport model for the ecological state variables and a model for the biogeochemical dynamics. The combination of these components involves the implementation of a numerical coupling method, that performs the spatial–temporal integration of the combined system, introducing an additional source of error to the system (splitting error). In this article we demonstrate the sensitivity of a comparatively complex 1D hydrodynamical biogeochemical model to the coupling method, showing that for an inadequate choice of the coupling method, the splitting error may dominate the numerical error of the system. It is demonstrated that for this type of system the tracer transport time scale clearly dominates over the scale of the biogeochemical processes, that maybe computed on significantly coarser time scales. In between the implemented coupling schemes Operator Splitting and Source Splitting, the Source Splitting method inserting the biogeochemical rates into the transport tracer integration is to be preferred for these type of models.