The representation of spatially heterogeneous land surface states and fluxes is essential for modeling processes that are nonlinearly related to the land surface states, such as the partitioning of sensible and latent heat fluxes. The contrast between the fine spatial scales at which this heterogeneity is manifest and the coarser scales at which most land surface models are run remains a challenging problem in land surface hydrology. An effort is being made to perform long-term land surface simulations at 1 km resolution over regional scales and 4-km resolutions over continental scales. The effects of sub-grid variability of inputs and parameters below these scales will be represented by introducing adjustment measures. In this study, the effects of sub-grid heterogeneity are investigated over aggregated grid areas of 250 m to 8 km. Land surface model simulations were performed over the Southern Great Plains region to quantify the effects of aggregation on water and energy states and fluxes. The land surface model, TOPLATS, combines a detailed representation of surface water and energy balance processes while capturing the topographically induced horizontal redistribution of subsurface water. The land model was run with forcings and model parameters at varying levels of aggregation. The differences among the fluxes produced by forcings and parameters at various resolutions provide a measure of the non-linearity in the land surface responses. The impact of factors such as wet/dry conditions and topography is evaluated. Alternative strategies that include running the averaged grids with a distribution of variables (e.g. soil-topographic index) are evaluated; how the sub-grid gets partitioned will be explored further. The results would lead to a clearer understanding of how to develop a sub-grid parameterization. Procedures to adjust states and fluxes computed at coarse resolutions will be developed. This simulation-based approach to sub-grid parameterization will be compared to statistical approaches.