Multi-resolution modeling and simulation in natural hazards modeling

Donna Calhoun, Carsten Burstedde

Date : February 19, 2015

Chiba University, Chiba University, Chiba, Japan (Invited)

Abstract

Adaptive mesh refinement has been actively used since the early eighties as a way to save computational resources when solving partial differential equations (PDEs) on logically Cartesian meshes. By adaptively refining the solution in regons of interest, calculations can be done at much less computational expense then would be required by uniform meshes. While the idea of "adaptive mesh refinement" (AMR) is appealingly easy, however, developing numerical algorithms that can take advantage of AMR is another story. To spur development of AMR algorithms, we have developed ForestClaw (www.forestclaw.org), a parallel AMR code based on octree mesh refinement. Our target applications include the dispersion of volcanic ash in the aftermath of volcanic eruptions, tsunami modeling, and the study of paleo-tides. We will present results from prototypical examples demonstrating the capabilities of ForestClaw and discuss the underlying models used for our target applications.

Slides : chiba-2015.pdf