Assessing impacts of climate change on water resources in regions with complex terrain requires precipitation simulations at sufficiently fine resolutions that capture orographic effects, which is not practical for a global model. To address this challenge, a topography-based subgrid dataset is being developed for the ACME Atmospheric Model, where each quadrilateral grid is discretized into subgrid units based on land surface elevation classes derived from high resolution global elevation data. For this purpose, a more consistent 90 meter resolution global surface elevation data is being developed by combining elevation data obtained from various sources. Each subgrid unit represents the fractional area within a range of surface elevation. To determine the intervals of the surface elevation, a local elevation classification method, which uses an elevation-area profile relationship to capture topographic patterns, is employed at each atmospheric quadrilateral grid to derive the subgrid units. The local elevation classification method results in more number of subgrid units over mountainous regions as compared to flat regions. For consistency, the same high resolution surface elevation data and local elevation classification method is used to define subgrid topographic landunits for the land model. In this presentation, we describe the methods employed and show results for some continents.