Automating site assessments using data from UAVs
ABSTRACT: Much progress has been made in remotely detecting forest loss, particularly by using satellite imagery. However, quantification of different stages of forest degradation continues to be challenging. Compared with satellites, UAVs (or drones) can deliver images of much higher spatial resolution and enable estimation of forest characteristics with greater accuracy. Hence, such data from UAVs may enable the quantification of different levels of forest degradation in greater detail than ever before.
In this paper, we discuss the potential of data from UAVs to i) assess forest degradation at the site level, ii) determine the conditions of reference (or target) forest ecosystems and iii) detect the extent of forest regeneration. Additionally, we quantify and compare several forest stand-level variables, measured in the field (observed) and from UAVs (detected) in a Chilean temperate forest.
Detected values from UAV data were 27-100% of the observed values for species richness, 25-61% for counts of trees and 67-81% for basal areas. Observed vs detected basal area measurements were highly correlated (R2=0.9). Results, using a canopy structure metric, to predict tree species richness (R2=0.42) and number of trees (R2=0.45), were promising.
We conclude that data from UAVs may be useful to detect gradients in vegetation structure, to determine degradation stages of restoration sites and consequently, to establish restoration goals and thus derive the most appro-priate methods to achieve them.