A combined contour lines iteration algorithm and Delaunay triangulation for terrain modeling enhancement

Abstract

Digital Elevation Models (DEMs) play a crucial role in civil and environmental applications, such as hydrologic and geologic analyses, hazard monitoring, natural resources exploration, etc. Generally, DEMs can be generated from various data sources, such as ground surveys, photogrammetric stereo methods, satellite images, laser scanning, and digitized contour lines. Compared with other data sources, contour lines are still the cheapest and more common data source becausethey cover all areas, at different scales, in most countries. Although there are different algorithms and technologies for interpolation in between contour lines, DEMs extracted solely from contours still suffer from poor terrain quality representation, which in turn negatively affects the quality of analytical applications results. In this paper, an approach for improving the digital terrain modeling based on contour line densification and Delaunay triangulation is presented to acquire a more suitable DEM for hydrographic modeling and its applications. The proposed methodology was tested using a variety of terrain patterns in terms of intensity: hilly, undulated, and plain (1:25,000 topographic map, 5 m contour interval). The precision of the extracted GRID model increases as the number of added contours increases. Adding four contour lines, the Root Mean Square Error (RMSE) of examining points were 0.26 m, 0.29 m, and 0.05 m for hilly, undulated, and plain samples, respectively, and the Mean Absolute Error (MAE) were 0.50 m, 0.48 m, and 0.17 m. The convergence probabilities between extracted and original flow lines for the same regions were 96.91%, 94.93%, and 84.03%. Applying the methodology, experimental results indicate that the developed approach provides a significant advantage in terrain modeling enhancement, generates DEMs smoothly and effectively from contours, mitigates problems and reduces uncertainties.