Eigenvector Analysis of Digital Elevation Models in a GIS: Geomorphometry and Quality Control

摘要

Digital elevation models (DEMs) cover a wide range of scales, and allow statistical analysis of geomorphometric parameters. At global or continental scale, DEMs covering rectangular quadrangles can be considered random samples. Average quadrangle values can be compared at different DEM scales, and for different physiographic provinces. Three independent variables provide valuable descriptions of terrain: average elevation, average slope, and the degree of terrain organization. DEMs with spacings of 30 inches (global coverage), 3 inches (continental United States), and 30 m (local United States) provide almost perfect correlations for average quadrangle elevation and slope, although the slope values increase as the DEM spacing decreases. The degree of terrain organization also correlates across DEM scales, but with lower correlation coefficients, especially in areas of lower relief. Terrain variables computed from 10 m and 30 m USGS Level 2 DEMs are essentially identical. Slope algorithms perform differently in different physiographic provinces, and the aspect algorithm performs poorly in low relief areas. Geomorphometric analysis can provide a rapid and effective assessment of DEM quality control, and should be integrated into the DEM production process.