LEGOS team: P. Marchesiello (PI task I), F. Lyard (PI task VI), G. Cambon, Y. Soufflet. Coordination: L. Debreu (LJK, Grenoble)
This project focuses on the first thematic axis of the call for proposal by the ANR "Numerical Models": complex systems modeling. It addresses particularly the thematic on modeling of environmental sciences, specifically oceanography.
COMODO: Ocean Modeling Community
The French ocean modeling community has been recently assembled under the group name COMODO (COmmunauté de Modélisation Océanique). This community is diverse and offers a variety of applications and numerical approaches for ocean modeling; it also relies at various degrees on the international community. For the first time, a global effort is proposed to strengthen interactions between community members. The effort is directed towards two main objectives: improving existing models and their numerical methods; and providing guidelines for the development of future generation ocean models. Existing ocean models suffer from a number of well-identified issues that are addressed in this project.
ANR COMODO project
To improve on those issues, COMODO proposes an innovative evaluation of effective resolution, i.e. numerical dissipation mechanisms, especially in the context of submesoscale modeling; and improvement of advection-diffusion schemes for reduction of spurious diapycnal mixing and accurate representation of active and passive tracers. The second part of the proposal is based on recent advances of our community on vertical coordinate systems, unstructured meshes and non-hydrostatic modelling. The objective is here both to continue fundamental research in these topics and to contribute to the design of future generation models involving their system of equations and numerical methods. The proposed developments will be evaluated thanks to a benchmark suite that covers both idealized test cases design to assess basic important properties of numerical schemes and more complex test cases that will be set-up for a thorough evaluation of progresses made during this project. This benchmark suite, accompanied with the results of the different models, will be made publicly available so as to provide elements for future model developments as well as an opportunity for more theoretical work on numerical schemes to be evaluated in the context of ocean modeling.
Effective resolution (task I)
The IRD/LEGOS team will address the effective resolution problem. The forward cascade of kinetic energy in the ocean surface at submesoscale implies that numerical closure can be made more consistent with physical closure. Nevertheless, dissipation in submesoscale models remains dominated by numerical constraints rather than physical ones. In Marchesiello et al. (2011), model convergence at submesoscale is controlled by numerical dissipation, which overpowers submesoscale energy production and transfer. Effective resolution can thus be defined as the dissipation wavelength marking the start of dissipation range, below which model dynamics become unphysical. This dissipation range is a function of the model numerical filters. Going beyond our present knowledge requires a better understanding of numerical dispersion/dissipation ranges and their connection to the submesoscale range in models. This is the objective of TASK 1 of the ANR-COMODO project.
FD and FE models examination and improvements (task VI)
It has been noticed for a while that connexions can be established between structured and unstructured methods on one hand, finite elements (FE) and finite differences (FD) discretization on the other hand (Lin, 1997, Adcroft al., 2008). Still, strong distinctive characteristics remain. However, thanks to some recent theoretical progress in unifying continuous and discontinuous Galerkin methods (Lyard et al., 2012), FD can be treated in a unified theoretical framework based on an extended FE formalism. We propose to scrutinize the design and numerical characteristics of existing structured and unstructured models, with a methodology based on the extended FE formalism. In a second step, we will examin the unstructured mesh modeling computational efficiency and accuracy, in comparison to FD models.