Small format aerial images to estimate the pine crown defoliation

Abstract

The paper discusses the opportunities of using small format aerial images to estimate the pine crown defoliation. Test area in eastern part of Lithuania was photographed using Canon EOS-1DsMark II digital camera, installed on-board SkyArrow ultra-light aircraft. Camera lenses were adopted to capture images corresponding to the conventional color-infrared photography. Ground sampling density of acquired images was 9 cm. Crown defoliation in the field was assessed for more than 500 trees located on 46 sample plots. 89% of field surveyed trees were identified on rectified aerial images. The individual crown defoliations were estimated using two non-parametric estimation techniques - most similar neighbor and k-nearest neighbor and small format aerial image characteristics as the auxiliary variables. The prediction root mean square errors were at two levels – around 9-11% for 170-years old trees and around 11-15% for the 65-years old trees. Accuracy of individual crown defoliation class prediction ranged from weak to moderate. Defoliation prediction root mean square errors at a sample plot level were 5.07% and 1.58% for 65-years and 170-years old stands respectively.