A flux-form version of the conservative semi-Lagrangian multi-tracer transport scheme (CSLAM) on the cubed sphere grid

  • Authors:

  • Lucas M. Harris;Peter H. Lauritzen;Rashmi Mittal

  • Affiliations:

  • Department of Atmospheric Sciences, University of Washington, Seattle, WA, United States;Atmospheric Modeling and Predictability Section, Climate and Global Dynamics Division, NCAR Earth Systems Laboratory, National Center for Atmospheric Research, Boulder, CO, United States;Mesoscale and Microscale Meteorology Division, NCAR Earth Systems Laboratory, National Center for Atmospheric Research, Boulder, CO, United States

  • Venue:
  • Journal of Computational Physics
  • Year:
  • 2011

Quantified Score

Hi-index 31.48

Visualization

Abstract

A conservative semi-Lagrangian cell-integrated transport scheme (CSLAM) was recently introduced, which ensures global mass conservation and allows long timesteps, multi-tracer efficiency, and shape preservation through the use of reconstruction filtering. This method is fully two-dimensional so that it may be easily implemented on non-cartesian grids such as the cubed-sphere grid. We present a flux-form implementation, FF-CSLAM, which retains the advantages of CSLAM while also allowing the use of flux-limited monotonicity and positivity preservation and efficient tracer sub-cycling. The methods are equivalent in the absence of flux limiting or reconstruction filtering. FF-CSLAM was found to be third-order accurate when an appropriately smooth initial mass distribution and flow field (with at least a continuous second derivative) was used. This was true even when using highly deformational flows and when the distribution is advected over the singularities in the cubed sphere, the latter a consequence of the full two-dimensionality of the method. Flux-limited monotonicity preservation, which is only available in a flux-form method, was found to be both less diffusive and more efficient than the monotone reconstruction filtering available to CSLAM. Despite the additional overhead of computing fluxes compared to CSLAM's cell integrations, the non-monotone FF-CSLAM was found to be at most only 40% slower than CSLAM for Courant numbers less than one, with greater overhead for successively larger Courant numbers.