Effect of LiDAR derived DEM resolution on terrain attributes, stream characterization and watershed delineation

A raster grid with 1m spacing was created using simple kriging from LiDAR point clouds. The 2, 3, 4, 5, 8, and 10 m resolution were created from the original 1 m DEM using nearest neighbor resampling method. The derived DEM were used to study the sensitivity of terrain attributes while changing DEM resolution and possibly introduce a minimum required resolution for the study area. A variety of terrain attributes were calculated to analyze their rate of changes in different resolutions. Using a coarser DEM resolution (8 and 10 m) instead of a finer one (1, 2 and 3 m), tends to ignore a variety of valuable features, and changes spatial distribution of terrain attributes. It also changes minimum, maximum and standard deviation of elevation within each unit leading to smoother output images. Large differences were observed between 1 and 10 m on all topographical and hydrological attributes, implying that the 10 m DEM is too coarse for this landscape to extract streamlines, and landform attributes. Landform elements were the most sensitive among other tested attributes. The plan curvature trends indicated that a 4 m DEM is capable to capture the grain and the longest significant relief wavelength across slopes, at least in this landscape; this implies that 4 m grid size can be the minimum DEM resolution to study the relief changes accurately, in this particular landscape.