Using high resolution UAV imagery to estimate tree variables in Pinus pinea plantation in Portugal

Abstract

Aim of study: The study aims to analyse the potential use of lowcost unmanned aerial vehicle (UAV) imagery for the estimation of Pinus pinea L. variables at the individual tree level (position, tree height and crown diameter). Area of study: This study was conducted under the PINEA project focused on 16 ha of umbrella pine afforestation (Portugal) subjected to different treatments. Material and methods: The workflow involved: a) image acquisition with consumer-grade cameras on board an UAV; b) orthomosaic and digital surface model (DSM) generation using structure-from-motion (SfM) image reconstruction; and c) automatic individual tree segmentation by using a mixed pixel‑and region‑based algorithm. Main results: The results of individual tree segmentation (position, height and crown diameter) were validated using field measurements from 3 inventory plots in the study area. All the trees of the plots were correctly detected. The RMSE values for the predicted heights and crown widths were 0.45 m and 0.63 m, respectively. Research highlights: The results demonstrate that tree variables can be automatically extracted from high resolution imagery. We highlight the use of UAV as a fast, reliable and cost-effective technique for small scale applications