Computational Instability arising from YIN-YANG Boundary

“Yin-Yang Grid” (Kageyama and Sato 2004) is a composite mesh that is convenient when expanding a regional dynamical model to create a global one. In 2009, Japan Meteorological Agency (JMA) initiated a fundamental research work to develop a global non-Hydrostatic model, and a Yin-Yang Grid model has been one of candidates for a further development. Shallow-water test cases proposed by Williamson et al. (1992) were examined, and a three-dimensional sample model adopting a regional non-hydrostatic model ASUCA (Kawano et al. 2013) has been developed. There are a lot of ways to adopt Yin-Yang Grid (Qaddouri and Lee 2011, Baba et al. 2010, Peng et al. 2006). These are very depending on discretization methods and advection (or flux) calculation schemes. So far, the finite-volume method with the positive finite difference advection scheme proposed by Hundsdorfer et al. (1995) is examined using the boundary exchanging method (Sugimura et al. 2006) to form a Yin-Yang Grid model. On the way to develop shallow-water models and a three-dimensional model, we have encountered many types of computational instability problems. We have analyzed results to overcome the problems, and found that some of the problems are closely related to discretization methods and advection schemes. For instance, the bilateral boundary exchanging method for Yin-Yang Grid seems to have a limitation in spatial resolution relative to the grid size when using the finite volume method, just as advection schemes have such limitations. A low-pass filter or a higher order advection scheme seems effective to overcome such kind of instabilities. In this study, we will show some instability examples stemming from composite mesh formation related to Yin-Yang Grid, and will discuss ways to overcome issues.