Assessing a novel modelling approach with high resolution UAV imagery for monitoring health status in priority riparian forests

Black alder (Alnus glutinosa) forests are in severe decline across their area of distribution due to a disease caused by the soil-borne pathogenic Phytophthora alni species complex (class Oomycetes), “alder Phytopththora”. Mapping different types damages is challenging high density ecosystems which spectral variability canopy heterogeneity. Data obtained unmanned aerial vehicles (UAVs) may be particularly useful for such tasks resolution, flexibility acquisition and cost efficiency this type data. In study, A. glutinosa was assessed considering four categories tree health status field: asymptomatic, dead defoliation above below 50% threshold. A combination multispectral Parrot Sequoia UAV -red green blue (RGB) data were analysed using classical random forest (RF) simple robust three-step logistic modelling approaches identify most important indicators while adhering principle parsimony. total 34 remote sensing variables considered, including set vegetation indices, texture features from normalized difference index (NDVI) digital surface model (DSM), topographic photogrammetry-derived structural DSM at crown level. The identified RF yielded an overall accuracy 67%, aggregation legend three classes (asymptomatic, defoliated, dead) two (alive, improved 72% 91% respectively. On other hand, confusion matrix, computed models leave-out cross-validation method accuracies 75%, 80% 94% four-, three- two-level classifications, study findings provide managers with alternative classification rapid, effective assessment areas affected non-affected disease, thus enabling them hotspots conservation plan control restoration measures aimed preserving black forests.