Predictability studies of coastal marine ecosystem behavior

Abstract

The study presented in this thesis is principally meant to analyze the genericity of a deterministic, comprehensive marine ecosystem model in combination with various refined representations of hydrodynamical processes, and to evaluate the potential predictability skills of this combined modelling system with specific applications in two rather different coastal basins. This objective has been realized by first developing a modular coupling interface between the Princeton Ocean Model (POM) and the European Regional Seas Ecosystem Model (ERSEM), called High-Resolution OpenSESAME POM ERSEM (HiROPE). Secondly, this model framework, embedding a composite of 'complex' conceptual principles of the functioning of the main biogeochemical processes, has been applied to substantially different marine systems, the Baltic proper and the northern Adriatic Sea. The generic biological first principles of the ERSEM ecosystem model have been throughly controlled for consistency, and a suitable mathematical syntax has been defined in order to accomodate the various biogeochemical cycles of the resolved elements. The model has been specifically applied in the chosen basins with different temporal and spatial resolutions: a one-dimensional (vertical, 1D-V), climatological implementation in the northern Adriatic Sea; a 1D-V implementation in the Baltic proper with realistic forcing functions in the period 1979-1991 and a fully three-dimensional, high-frequency realistic implementation in the northern Adriatic Sea (October 1995). General conclusions are that the representation of hydrodynamical variability, the definition and resolution of boundary processes, the introduction of new source terms or the implementation of new biological state variables, affect the predictability of the system behavior more than the utilization of incomplete initial conditions of biological variables in a complex comprehensive ecosystem model.