#O06 Nearshore component of MPAS-O

Abstract

The project will develop a new tidal and estuary system (TES) component in DOE’s MPAS-Ocean (‘MPAS-O’ hereafter) model to improve its skills in TES and to reduce uncertainties in the coastal biogeochemistry (BGC) simulation. The new TES model will be online, one-way coupled to MPAS-O at a prescribed boundary and driven by MPAS-O.

We have done preliminary work to narrow down the design of the TES model to be largely based on a co-volume approach proposed by Chen et al. (2013), where both velocity components are specified on cell edges, and the thickness variables evolve on both the primary and the dual cell centers. This approach can handle non-orthogonal grids although accuracy is best on orthogonal grids. We will explore both split explicit and implicit time stepping, using the original (i.e. not vector-invariant) form. This is because there is enough energy dissipation in TES and so energy conservation is not as important as in large scale. The spurious modes will be filtered using an appropriate advection and viscosity scheme. The implicit approach will take advantage of the disjoint TES domains in the world and assign different numbers of compute cores to these domains for efficiency. To enhance the strong scaling, sub-cycling method (in lieu of matrix solvers using global reduction) will be explored as well. A ghost-cell Immersed Boundary Method will be used to alleviate the staircase problem near the bottom, as the bathymetry is a major driver in TES.

The proposed approach strikes a best balance between accuracy, efficiency and usability. TES is strongly dominated by forcings like boundary condition and bathymetry. The complex geometry and bathymetry pose severe challenges for grid generation for orthogonal grids. Strong forcings and inundation also set a high bar for robustness, which may be better addressed with an implicit approach.

The development of the BGC component of the new model will be presented in a separate poster (Wang et al.).