A comparative study of the accuracy of digital terrain models (DTMs) based on various data models

This paper describes a comparative study of the accuracies of digital terrain models (DTMs) derived from four different data models, namely, contour data only, contour data with additional feature-specific (F-S) data (peaks, pits, points along ridges, points along ravines, and points along break lines, etc.), square-grid data only, and square-grid data with additional F-S data. It has been found that: 1. (1) The accuracy of DTMs derived from (photogrammetrically measured) contour data, in terms of RMSE (root mean square error) or σ (standard deviation), is about 1/3 to 1/5 of the contour interval, depending on the characteristics of the terrain topography. However, if additional F-S data are included, these figures can be reduced by 40% to 60%. 2. (2) The accuracy of DTMs derived from gridded data plus F-S data is linearly related to the grid-interval. However, if F-S data are not included, the relationship becomes parabolic. 3. (3) The accuracy of a DTM derived from contour data with a vertical interval CI matches that from grid data with an interval of (K x CIcotα), where α is the mean slope angle of the area and K is a constant ranging from 1.5 to 2.0. However, if additional F-S data are included, then the value of K can be reduced. In this case, the value of K ranges from 1.0 to 1.5, depending on the characteristics of the terrain topography.