Spatial Variability Measuring Information Content

Abstract

When building a fully distributed grid-based distributed hydrologic model from geospatial data, a fundamental question is: What resolution is adequate for capturing the spatial variability of a parameter or input? The model computational elements, whether finite element or difference, require parameter values that are representative of the grid cell. The composite for the model should then be representative of the spatial variation found in the watershed. When solving conservation of mass, momentum, or energy, a representative parameter value is required for that element size. The correspondence of the computational element size and the sampled resolution of the digital map affect how well the model will represent the spatial variation of parameters and modeled rainfall runoff. Within each grid cell, the conservation of mass and momentum is controlled by the parameters taken from special purpose maps whose values are hydraulic roughness, rainfall intensity, slope and infiltration rate for each computational element. The grid cell map supplies parameter values within the model grid. To take advantage of distributed parameter hydrologic models, the spatial distribution of inputs (e.g., rainfall) and parameters (e.g., hydraulic roughness) should be sampled at a sufficiently fine resolution to capture the spatial variability.