Preprints
https://doi.org/10.5194/egusphere-2024-2710
https://doi.org/10.5194/egusphere-2024-2710
11 Oct 2024
 | 11 Oct 2024

Parameterisation toolbox for physical-biogeochemical model compatible with FABM. Case study: the coupled 1D GOTM-ECOSMO E2E for the Sylt-Romo Bight, North Sea

Hoa T. T. Nguyen, Ute Daewel, Neil Banas, and Corinna Schrum

Abstract. Mathematical models serve as invaluable tool for comprehending marine ecosystems. The performance of these models is often highly dependent on their parameters. Traditionally, refining these models involved a time-intensive trial-and-error approach to identify model parameter values that are able to reproduce observations well. However, as ecosystem models grow in complexity, this approach becomes impractical. With advancements in computing power, optimization techniques have emerged as a viable alternative. Yet, these techniques often exhibit model-specific tailoring, limiting their broader application. In this study, we introduce a parameterisation toolbox founded on a Particle Swarm Optimizer (PSO) implemented in the Framework for Aquatic Biogeochemical Models (FABM), which allows its reuse between numerous existing models in FABM, and thus makes the optimizer more accessible to the community. The PSO toolbox's effectiveness is demonstrated through its implementation on a 1D physical-biogeochemical model (GOTM-ECOSMO E2E), which successfully parameterised the Sylt-Romo Bight ecosystem. The toolbox was able to identify most of the tuned parameters and to suggest potential ranges for poorly constrained parameters. In addition, the toolbox uncovers a number of parameter sets with notable differences in some parameter values, but resulting in not much difference in biomass and fluxes. Furthermore, by experimenting with optimisation models of varying complexity, the toolbox was able to define an optimal model for the Sylt-Romo Bight.

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.
Share
Download
Short summary
Parameterisation is key in modeling to reproduce observations well but is often done manually....
Share