The Use of UAV-Acquired Multiband Images for Detecting Rockfall-Induced Injuries at Tree Crown Level

In this paper, we present an identification of rockfall-injured trees based on multiband images obtained by unmanned aerial vehicle (UAV). A survey with a multispectral camera was performed three rockfall sites versatile tree species (Fagus sylvatica L., Larix decidua Mill., Pinus sylvestris Picea abies (L.) Karsten, and Abies alba Mill.) different characterizations rockfalls rockfall-induced injuries. At one site, injuries were induced in the same year as survey. second they after initial injuries, at third six years first surveys row. Multiband used to extract vegetation indices (VIs) crown level further studied see which VIs can identify injured how successfully. total 14 considered, including individual bands (green, red, red edge, near-infrared) using regression models differentiate between uninjured groups for single consecutive years. The model also VI differentiations among recorded injury size possible where least before UAV survey, could still be identifiable site only four months not possible. that explain largest variability (R2 > 0.3) were: inverse ratio index (IRVI), green–red (GRVI), normalized difference (NDVI), (NRVI), (RVI). RVI most successful, explaining 40% variance two sites. R2 values increased few percentages (up 10%) when observed over period mostly did change significantly, thus indicating if vitality or decreased. Differentiation show promising results, while, other hand, there strong correlation (RVI) according basal area (so-called index). Both case broadleaves conifers sites, achieved value 0.82. presented results indicate UAV-acquired surveying protection forests order monitor their vitality, is crucial maintaining protective effect through time space.