Regional Modeling Applications of a Global Climate Model Using Local Mesh Refinement

Traditional regional climate models (RCMs) are limited area models that require lateral boundary conditions to be specified from an outside source, such as a global climate model (GCM). Information flow at the lateral boundaries is inward only. Events simulated within the RCM cannot propagate outside to the GCM domain, and thus cannot impact global flow patterns which in turn feed back into the RCM through the lateral boundaries. This restriction limits the applicability of scientific problems that can be investigated, for example, assessing the impact of local changes in climate forcing on the local climate. The Ocean-Land-Atmosphere Model (OLAM) is a new numerical simulation model that is based on the RAMS regional model but encompasses a global domain and uses a geodesic triangular mesh. Local mesh refinement enables OLAM to function simultaneously as a GCM and RCM. Selected geographic areas are represented with very high resolution typical of RCMs, while the remainder of the globe is covered with a lower resolution GCM-like grid. This configuration enables two-way information transfer between low and high resolution regions, eliminates problematical lateral boundary conditions, and allows simulations of comprehensive cause and effect relationships between proximate or widely separated events within a single model system. It is common for the refined mesh regions to contain more than half the grid cells in the model domain and to require most of the computational steps. Therefore, compared with an RCM, the OLAM coarse global grid adds a modest increase in computational expense while providing major benefits. Different mesh structures that provide local refinement are compared. Standard global simulation tests are used to evaluate performance of different refinement structures and as a guide for optimizing them, particularly in the transitional region between lower and higher resolution. Techniques are discussed to enable physical parameterizations to adapt to a wide range of grid resolution. Simulations of shallow and deep convective systems in highly refined regions of the grid test model performance at the finest scales.