Applying a Robust Empirical Method for Comparing Repeated LiDAR Data with Different Point Density

A key aspect of vegetation monitoring from LiDAR is concerned with the use comparable data acquired multitemporal surveys and different sensors. Accurate digital elevation models (DEMs) to derive products, are required make comparisons among repeated data. Here, we aimed apply an improved empirical method based on DEMs difference, that adjust ground a low-density dataset high-density one for ensuring credible changes. The study areas collection six sites over Sierra de Gredos in Central Spain. methodology consisted producing “the best DEM difference” between low- (using classification filter, interpolation spatial resolution lowest vertical error) generate local “pseudo-geoid” (i.e., continuous surfaces differences) was used correct raw points. error estimated by 50th percentile (P50), normalized median absolute deviation (NMAD) root mean square (RMSE) differences. In addition, analyzed effects site-properties (elevation, slope, height distance nearest geoid point) accuracy. Finally, assessed if changes were related differences datasets. Before correction aggregating sites, ranged 0.02 ?2.09 m 0.39 0.85 (NMDA) 0.54 2.5 (RMSE). segmented-based filter algorithm (CSF) showed highest error, but there not significant methods or resolutions. After dropped significantly: ?0.004 ?0.016 0.10 0.06 0.28 0.46 (RMSE); CSF continued showing greatest error. terrain slope point most important variables explaining corrections, decoupled errors DEMs. This work using difference) better adjusted benchmark being adapted site-specific conditions. accuracy data, minimizing random nature decoupling those errors.