Robert L. Higdon, Oregon State Univ.
This talk will describe some recent developments and numerical tests related to a two-level time-stepping method that I previously developed for a barotropic-baroclinic splitting for layered models. The algorithmic improvements involve the relation between velocity and momentum density as layer thicknesses tend to zero, limiting mass and momentum fluxes between thick and thin cells near variable bottom topography, solving the barotropic equations explicitly versus implicitly, and conserving mass within individual layers. Numerical tests involve external and internal Rossby waves in a two-layer fluid in a channel, an upwelling/downwelling problem that involves fluid interfaces moving upward and downward along sloping bottom topography, and a double-gyre circulation that displays meanders and eddies. If the explicit viscosity is set to zero, the model runs stably and with no grid noise.