Testing the performance of a forest characterization scheme using multiple dataset comparison

Characterising forest structure is an essential part of any comprehensive biodiversity assessment. In this study, the robustness of a previously proposed LiDAR based scheme for characterising the ecological structure of a dry Eucalypt forest landscape was examined by comparing two LiDAR datasets of the same area. This was validated using a network of field sites that recorded commonly used metrics of biodiversity. The eight category scheme, derived from LiDAR point density, proved resilient when recovering the openings above the ground and medium vegetation, canopy cover, and presence of mid-storey vegetation and high trees, however less so when recovering the opening above low vegetation, presence of understorey vegetation and vertical canopy density of high trees. These last categories were considered to be affected by the difference in the pulse repetition rate. Obtaining sufficient multiple returns by setting proper pulse repetition rate is the key to maintaining good performance of the scheme.