Significant effect of topographic normalization of airborne LiDAR data on the retrieval of plant area index profile in mountainous forests

As an important metric for describing vertical forest structure, the plant area index (PAI) profile is used for many applications including biomass estimation and wildlife habitat assessment. PAI profiles can be estimated with the vertically resolved gap fraction from airborne LiDAR data. Most research utilizes a height normalization algorithm to retrieve local or relative height by assuming the terrain to be flat. However, for many forests this assumption is not valid. In this research, the effect of topographic normalization of airborne LiDAR data on the retrieval of PAI profile was studied in a mountainous forest area in Germany. Results show that, although individual tree height may be retained after topographic normalization, the spatial arrangement of trees is changed. Specifically, topographic normalization vertically condenses and distorts the PAI profile, which consequently alters the distribution pattern of plant area density in space. This effect becomes more evident as the slope increases. Furthermore, topographic normalization may also undermine the complexity (i.e., canopy layer number and entropy) of the PAI profile. The decrease in PAI profile complexity is not solely determined by local topography, but is determined by the interaction between local topography and the spatial distribution of each tree. This research demonstrates that when calculating the PAI profile from airborne LiDAR data, local topography needs to be taken into account. We therefore suggest that for ecological applications, such as vertical forest structure analysis and modeling of biodiversity, topographic normalization should not be applied in non-flat areas when using LiDAR data.