The article of Kim Calders, John Armston, Glenn Newnham, Martin Herold and Nicholas Goodwin: Implications of sensor configuration and topography on vertical plantprofiles derived from terrestrial LiDAR, has been published in Agricultural and Forest Meteorology, Volume 194, Pages 104-117
AbstractThe vertical distribution of plant constituents is a key parameter to describe vegetation structure andinfluences several processes, such as radiation interception, growth and habitat. Terrestrial laser scanning(TLS), also referred to as terrestrial LiDAR, has the potential to measure the canopy structure with highspatial detail and accuracy. Vertical plant profiles, which describe the plant area per unit volume (PAVD)as a function of height, are often used to quantify the vertical structure. However, most studies do notaccount for topography, use registered multiple TLS scans or use a detailed airborne LiDAR digital terrainmodel to account for this variation in ground height. Airborne LiDAR is often not available or expensiveto acquire. Here, we present an approach that facilitates rapid, robust and automated assessment ofthe vertical structure of vegetation. We use single scans and local plane fitting to correct for topographiceffects in vertical plant profiles and test our approach in five different Australian forest types with differenttopography and understorey. We validate our approach with topography-corrected vertical plant profileswith digital terrain models derived from airborne LiDAR. Our results demonstrate that not correcting fortopography can lead to significant errors in the vertical distribution of plant constituents (CV(RMSE) upto 66.2%, typically ranging from 4.2% to 13.8%). This error decreases significantly when topography isaccounted for with TLS plane fitting (CV(RMSE) up to 20.6%, typically ranging from 1.5% to 12.6%). Wedemonstrate that height metrics from vertical plant profiles that are not corrected for topography departsignificantly from those that are inferred from the reference profile. The effect is most noticeable forcanopy top height and the peak PAVD height. Correcting topography with a TLS plane fitting approachreduces the error in canopy top height by at least 77% and up to 100%, and reduces the error in peak PAVDheight by 83.3% and up to 100%. We also show the advantage of a multiple return over a first return TLSinstrument. The definition of the ground returns with a first return instrument might be problematic inenvironments with dense herbaceous understorey and there is an overall trend of lower height metricscompared to multiple return instruments. We present a data-driven approach that is based on singlescan TLS data. The latter is of importance for large area sampling as it allows more sites to be sampledfrom existing resources and facilitate consistent processing of archived TLS data, which is often singlescan data with no survey control.
Keywords: Terrestrial LiDAR; Vertical plant profiles; Vegetation structure; Topography; Canopy height