SCALING LIDAR TERRAIN UNCERTAINTY USING LANDSCAPE PROPERTIES

The examination of landscape surface processes is often accomplished using a digital elevation model interpolated from a lidar point cloud. The resulting DEM often does not include the inherent uncertainties of the individual lidar points, and, even if provided, those uncertainties do not include consideration of the DEM interpolation method used, or the effect of terrain surface roughness and raw point density. Here, we present a single comprehensive model of the vertical terrain uncertainty and the characteristics of the terrain (normalized surface roughness), based on a representative number of datasets from a range of landscapes with various surface roughness and vegetation cover. To show the utility of properly scaled vertical covariance, we then use this modeled relationship to scale per-point estimates of lidar source data uncertainty and consider the estimated point accuracy in a normalized cross-correlation for change detection to show that the result is a more representative measure of change detection uncertainty.