Evaluating error of lidar derived dem interpolation for vegetation area

Abstract

Light Detection and Ranging or LiDAR data is a data source for deriving digital terrain model while Digital Elevation Model or DEM is usable within Geographical Information System or GIS. The aim of this study is to evaluate the accuracy of LiDAR derived DEM generated based on different interpolation methods and slope classes. Initially, the study area is divided into three slope classes: (a) slope class one (0° ' 5°), (b) slope class two (6° ' 10°) and (c) slope class three (11° ' 15°). Secondly, each slope class is tested using three distinctive interpolation methods: (a) Kriging, (b) Inverse Distance Weighting (IDW) and (c) Spline. Next, accuracy assessment is done based on field survey tachymetry data. The finding reveals that the overall Root Mean Square Error or RMSE for Kriging provided the lowest value of 0.727 m for both 0.5 m and 1 m spatial resolutions of oil palm area, followed by Spline with values of 0.734 m for 0.5 m spatial resolution and 0.747 m for spatial resolution of 1 m. Concurrently, IDW provided the highest RMSE value of 0.784 m for both spatial resolutions of 0.5 and 1 m. For rubber area, Spline provided the lowest RMSE value of 0.746 m for 0.5 m spatial resolution and 0.760 m for 1 m spatial resolution. The highest value of RMSE for rubber area is IDW with the value of 1.061 m for both spatial resolutions. Finally, Kriging gave the RMSE value of 0.790m for both spatial resolutions.