A Catchment-Based Hierarchical Spatial Tessellation Approach to a Better Representation of Land Heterogeneity for Hyper-Resolution Land Surface Modeling

To represent the physical processes at hillslope scales for hyper-resolution land surface modeling, we propose a hierarchical, catchment-based spatial tessellation method. The land surface is divided into a hierarchical structure: catchments, height bands along hillslopes within a catchment, and land cover patches within a height band. This catchment-based structure explicitly represents hillslope drainage networks and can be applied at various resolutions determined by a pre-defined maximum height band size. The proposed tessellation method is superior to the conventional grid-based structure in representing land surface heterogeneity, resulting in a higher aggregation skill through the height band representation. The spatial variations in air temperature, leaf area index, saturated soil hydraulic conductivity, and soil porosity are generally lower within a height band than those in a conventional rectangular grid, reflecting the nature of topographic control on climate, vegetation, and soil distribution. The improvement in aggregation skill depends on resolutions and terrain slope angle, more pronounced at 1/6 degrees model resolution and over steeper terrains. Finally, we demonstrate that our proposed catchment-based structure performs better than the grid-based structure through modeling tests over the Columbia River basin at resolutions of 1/2 degrees, 1/6 degrees, and 1/20 degrees and a global test at 1/2 degrees using the ILAMB model evaluation metrics.