Examination of adaptive and static mesh refinement for idealized tropical cyclone simulations in a global shallow water model

Currently, it is not possible to resolve the intricate details of deep moist convection in the tropics with global non-hydrostatic models, since the computational expense would prohibit these models from being run a very fine horizontal scales across the entire globe (order of 1 km). A way to bridge this scale discrepancy is through the use of variable resolution meshes. Ongoing work is being done to understand the utility of such variable resolution meshes for idealized tropical cyclone simulations. In this work, the utility of static and adaptive mesh refinement (SMR and AMR, respectively) are examined for idealized tropical cyclone (TC) simulations in a global shallow water model based on elemental discontinuous and continuous Galerkin methods. The SMR simulations have varying sizes of the statically refined meshes (geometry-based) while the AMR simulations dynamically refine and de-refine based on some threshold. We examine two cases in detail: (i) a TC-like vortex moving in the environmental flow on the sphere, and (ii) aspects of the intertropical convergence zone (ITCZ) development and breakdown into TC-like vortices on the sphere. Sensitivity tests are conducted based on the number of fully refined elements in the vicinity of the TC. It is demonstrated that the feature that feature of interest (the TC) can be resolved as nearly as well as the truth runs with significantly less computational expense. More realistic TC simulations on the sphere are conducted using the forced shallow water equations (mimicking the effects of deep moist convection).