Exploring a Multiresolution Approach Using AMIP Simulations

Abstract This study presents a diagnosis of a multiresolution approach using the Model for Prediction Across Scales–Atmosphere (MPAS-A) for simulating regional climate. Four Atmospheric Model Intercomparison Project (AMIP) experiments were conducted for 1999–2009. In the first two experiments, MPAS-A was configured using global quasi-uniform grids at 120- and 30-km grid spacing. In the other two experiments, MPAS-A was configured using variable-resolution (VR) mesh with local refinement at 30 km over North America and South America and embedded in a quasi-uniform domain at 120 km elsewhere. Precipitation and related fields in the four simulations are examined to determine how well the VRs reproduce the features simulated by the globally high-resolution model in the refined domain. In previous analyses of idealized aquaplanet simulations, characteristics of the global high-resolution simulation in moist processes developed only near the boundary of the refined region. In contrast, AMIP simulations with VR grids can reproduce high-resolution characteristics across the refined domain, particularly in South America. This finding indicates the importance of finely resolved lower boundary forcings such as topography and surface heterogeneity for regional climate and demonstrates the ability of the MPAS-A VR to replicate the large-scale moisture transport as simulated in the quasi-uniform high-resolution model. Upscale effects from the high-resolution regions on a large-scale circulation outside the refined domain are observed, but the effects are mainly limited to northeastern Asia during the warm season. Together, the results support the multiresolution approach as a computationally efficient and physically consistent method for modeling regional climate.