Classification of Defoliated Trees Using Tree-Level Airborne Laser Scanning Data Combined with Aerial Images

Abstract

Climate change and rising temperatures have been observed to be related to the increase of forest insect damage in the boreal zone. The common pine sawfly (Diprion pini L.) (Hymenoptera, Diprionidae) is regarded as a significant threat to boreal pine forests. Defoliation by D. pini can cause severe growth loss and tree mortality in Scots pine (Pinus sylvestris L.) (Pinaceae). In this study, logistic LASSO regression, Random Forest (RF) and Most Similar Neighbor method (MSN) were investigated for predicting the defoliation level of individual Scots pines using the features derived from airborne laser scanning (ALS) data and aerial images. Classification accuracies from 83.7% (kappa 0.67) to 88.1% (kappa 0.76) were obtained depending on the method. The most accurate result was produced using RF with a combination of data from the two sensors, while the accuracies when using ALS and image features separately were 80.7% and 87.4%, respectively. Evidently, the combination of ALS and aerial images in detecting needle losses is capable of providing satisfactory estimates for individual trees.

Remote Monitoring of Forest Insect Defoliation -A Review-

“…Remote detection of defoliation in sparse cover vegetation is still much more difficult to assess (Dennison et al, 2009). Nevertheless, Kantola et al (2010) found that combining (in a fusion approach) different sensor data, like spectral features from digital photographs with those from high density ALS data, can enhance mapping accuracy for two defoliation classes to higher values (88.1%) than obtained by each method separately (80.7% and 87.4% respectively). Correct assessment of defoliation by detection of changes much depends on obtaining data relatively free of exogenous noise, so an adequate pre-processing of image data is critical.…”

“…The difficulties for classifying damage severity using traditional methods may be underlined by the fact that less than half of the studies in this review were able to map defoliation above two severity classes in a continuous manner (see Table 1), and in two of these significant accuracy was attained only for two classes (Ilvesniemi 2009;Kantola et al, 2010). Nevertheless, accuracy of 80% or above and Kappa values ≥ 0.6 were achieved in one third of the cases using a dichotomous classification.…”

Influence of soil and topography on defoliation intensity during an extended outbreak of the common pine sawfly (Diprion pini L.)

“…At its most extensive range (beginning of 2000's), the outbreak area covered ca. 10 000 ha (Kantola et al 2010, Talvitie et al 2011. Changes in the D. pini population and resulting tree damage intensity within the study area have been monitored annually since 2000 (De Somviele et al 2007).…”

“…The radius of the plots varied between 8.5 and 13 m, resulting in an average of 24 trees on each plot. As part of a larger project (Kantola et al 2010, 2013, Talvitie et al 2011, eleven of the plots were established in 2002 and 17 in 2007. The locations of the plots were chosen subjectively to ensure that the range in defoliation intensity levels within the study area was covered.…”

“…The LiDAR point cloud was acquired from the Palokangas area in July 2008 using an ALS50-II SN058 ® laser scanner (Leica Geosystem AG, Heerbrugg, Switzerland) as described by Kantola et al (2010). With the standard TerraScan approach (Axelsson 2000), the data set was classified into ground or non-ground points and a high resolution (1 m) digital elevation model (DEM) was generated using classified ground points.…”