We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Numerical Analysis

Title: Conservation with moving meshes over orography

Abstract: Adaptive meshes have the potential to improve the accuracy and efficiency of atmospheric modelling by increasing resolution where it is most needed. Mesh re-distribution, or r-adaptivity, adapts by moving the mesh without changing the connectivity. This avoids some of the challenges with h-adaptivity (adding and removing points): the solution does not need to be mapped between meshes, which can be expensive and introduces errors, and there are no load balancing problems on parallel computers. A long standing problem with both forms of adaptivity has been changes in volume of the domain as resolution changes at an uneven boundary. We propose a solution to exact local conservation and maintenance of uniform fields while the mesh changes volume as it moves over orography. This is solved by introducing a volume adjustment parameter which tracks the true cell volumes without using expensive conservative mapping.
A finite volume solution of the advection equation over orography on moving meshes is described and results are presented demonstrating improved accuracy for cost using moving meshes. Exact local conservation and maintenance of uniform fields is demonstrated and the corrected mesh volume is preserved.
We use optimal transport to generate meshes which are guaranteed not to tangle and are equidistributed with respect to a monitor function. This leads to a Monge-Amp\`{e}re equation which is solved with a Newton solver. The superiority of the Newton solver over other techniques is demonstrated in the appendix. However the Newton solver is only efficient if it is applied to the left hand side of the Monge-Amp\`{e}re equation with fixed point iterations for the right hand side.
Subjects: Numerical Analysis (math.NA)
DOI: 10.1016/j.jcp.2022.111217
Cite as: arXiv:2108.00805 [math.NA]
  (or arXiv:2108.00805v1 [math.NA] for this version)

Submission history

From: Hiroe Yamazaki [view email]
[v1] Mon, 2 Aug 2021 12:21:10 GMT (5972kb,D)

Link back to: arXiv, form interface, contact.