COMPARING UNIFORM AND RANDOM DATA REDUCTION METHODS FOR DTM ACCURACY

The digital cartographic representation of the elevation of the earth's surface created from discrete elevation points is defined as a digital terrain model (DTM). DTMs have been used in a wide range of applications, such as civil planning, flood control, transportation design, navigation, natural hazard risk assessment, hydraulic simulation, visibility analysis of the terrain, topographic change quantification, and forest characterization. Remote sensing, laser scanning, and radar interferometry become efficient sources for constructing high-accuracy DTMs by the developments in data processing technologies. The accuracy, the density, and the spatial distribution of elevation points, the terrain surface characteristics, and the interpolation methods have an influence on the accuracy of DTMs. In this study, uniform and random data reduction methods are compared for DTMs generated from airborne Light Detection and Ranging (LiDAR) data. The airborne LiDAR data set is reduced to subsets by using uniform and random methods, representing the 75%, 50%, and 25% of the original data set. Over the Mount St. Helens in southwest Washington State as the test area, DTM constructed from the original airborne LiDAR data set is compared with DTMs interpolated from reduced data sets by Kriging interpolation method. The results show that uniform data reduction method can be used to reduce the LiDAR datasets to 50% density level while still maintaining the quality of DTM.