Developing a forest-degradation index for forest ecosystem restoration using UAV-based RGB photography

Abstract

Forest degradation assessment is essential to plan restoration. This study was a first attempt to develop a forest-degradation index (FDI), based on data from unmanned aerial vehicles (UAVs). It aimed to find a practical solution to replace labor-intensive conventional ground surveys with complex multiple variables, to plan restoration projects. It explored correlations between UAV and ground data, to construct a FDI. Five forest-restoration trial plots, representing a wide range of degradation level, were surveyed, with ground sample plots and a UAV. Aerial photos were processed, to produce canopy-height models (CHMs) and orthophotos, used to measure six variables, related to degradation. Four were highly correlated between ground and UAV-derived measurements: tree stocking-density (TD, r = 0.84), per cent canopy cover (CC, r = 0.91), per cent ground vegetation (VEG, r = 0.84), and per cent exposed soil + rock (SOIL, r = 0.75). To construct the FDI, a highly intercorrelated variable (CC) was rejected, to prevent over-weighting of related factors. The three remaining criteria were weighted by experts and applied to the normalized values. The resultant FDI quantified degradation levels reasonably intuitively and ranked the sites in logical order of degradation. However, limitations of the technique included i) obscurement of tipping points, which define conventional degradation stages ii) use of 5 arbitrary categories in the FDI, and iii) exclusion of landscape criteria. Until these issues are resolved, a hybrid system, combining individual variables, with the UAV-derived FDI system, may be the best solution for planning restoration strategies.