UBC Theses and Dissertations
Comparison of computer-based terrain storage methods with respect to the evaluation of certain geomorphometric measures Mark, David Michael
Topographic information can be digitized in several ways. Sampling may be surface-random (points selected according to partially or completely arbitrary criteria) or surface-specific (points selected according to their topographic significance). Surface-random sampling includes grids, contours, and randomly-located points. In this study, grid sampling and surface-specific sampling are compared. Surface behavior between sampled points is assumed to be linear. All aspects of surface form can be considered to reflect surface roughness. Horizontal variation includes the concepts of texture and grain, while vertical variation is discussed under relief. The relationships between these are embodied in slope and the dispersion of slope magnitude and orientation. The distribution of mass within the elevation range of a topographic surface is described under hypsometry. Parameters for investigation are selected from these categories. The variation of some selected geomorphometric parameters in southern British Columbia is examined via a stratified random sample consisting of forty-two 7 x 7 km areas. The values of some of these parameters are used to group the samples, and six are chosen for more detailed analysis. The relationships among the variables are examined using correlation analysis. For four geomorphometric measures (local relief, mean slope, roughness factor, and hypsometric integral), the theoretical errors involved in estimating the measures from the two selected terrain storage methods are discussed. The surface-specific point samples should produce better results than grids of reasonable densities. The latter, however, should require less digitization time and computer storage space per point. For at least local relief and hypsometric integral, grid error should be a linear function of grid spacing. Results of empirical comparison of the methods over the six selected areas are presented. The average surface-specific point data set is found to require some 2.6 times as much digitization time and 3.1 times as much computer storage space as the 15 by 15 grids used in the comparison. Computed estimates obtained from both of these data bases are presented for each of the four selected parameters, together with other estimates (obtained manually) in some cases. The average errors for the methods are found to differ significantly for local relief and mean slope but not for the hypsometric integral; for all three measures, the grids produce larger mean errors. The assumption of a linear relationship between grid spacing and grid error is used to estimate the grid spacing which would be required fo produce the same average error as the surface-specific points. For the three parameters used, these hypothetical grids are calculated to require more computer storage space and digitization time than the surface-specific point data sets. The influence of the density of surface-specific points and of base map scale appear to be related to the topographic texture. For a reasonably experienced terrain analyst, the reproduceability of these data sets appears to be good, although there remains a subjective element in point selection not present for grids. It is concluded that for a given amount of digitization time or computer storage space, better estimates of geomorphometric parameters can be obtained using sets of surface-specific points than using regular grids.
Item Citations and Data