Increasing the use of expert opinion in forest characterisation approaches based on LiDAR data

Forest structure characterisation approaches using LiDAR data and object-based image analysis remain scarce to forestry agencies as these automated procedures usually require the use of expensive software and highly skilled analysts. The integration of forest expert opinion into semi-automated approaches would simplify the access of forest managers to new technologies and would allow the incorporation of personal experience and the introduction of specific forest management criteria. The aim of this study is to explore new alternatives to a previously published automated approach based on LiDAR data and object-based image analysis. We compare four approaches, ranging from null to high incorporation of expert opinion and from fully automated to fully manual. These four approaches consist of three stages: (1) forest stand identification from LiDAR models, (2) forest stand classification into forest structure classes (manual and based on cluster analysis), and (3) validation. Quantitative attributes for validation (i.e. hypsographs and percentiles) provided slightly lower degree of separability for forest structure classes, in the mixed procedures with increasing incorporation of expert opinion than for the fully automated approach. The new mixed approaches proposed are comparable to the automated procedures for the characterisation of forest structure in heterogeneous pine forest stands. They also offer additional advantages: (1) they make it possible to give a specific management focus and (2) they provide accessibility by the forest managers to the source of LiDAR information.