Open AccessDetermination of tree height according to data of raster images different resolution
Author(s) -
Petro Diachuk
Publication year - 2020
Publication title -
ukrainian journal of forest and wood science
Language(s) - English
Resource type - Journals
eISSN - 2664-4460
pISSN - 2664-4452
DOI - 10.31548/forest2020.03.002
Subject(s) - scots pine , raster graphics , aerial photography , digital elevation model , raster data , elevation (ballistics) , forest inventory , remote sensing , tree (set theory) , standard deviation , spatial analysis , geography , mathematics , forestry , environmental science , statistics , computer science , forest management , pinus <genus> , computer vision , mathematical analysis , botany , geometry , biology
The collection of information on the growth and development of trees is the basis for planning forestry and horticulture, while the relevance and reliability of such data defines the quality of forest and park inventory outputs. Currently in Ukraine, the height of growing trees and shrubs is measured mostly by clinometer. The enhancement of unmanned aerial vehicles (UAVs) and methods of processing the collected information allow to amplify the level of quality and accuracy of the collected data. Our goal was to consider the possibility of determining the height of trees based on aerial photography materials obtained by UAVs and to assess the accuracy of measured indicators modeling crown height at different spatial resolutions. Here we used methods of creating a digital canopy height model (CHM) from aerial photographs obtained by UAVs. We produced 8 digital elevation and terrain models for the calculation of CHM. Raster image analysis was performed using the ArcGIS software and Spatial Analyst toolkit using the Focal statistics filter. We have confirmed the possibility of CHM utilization to measure the height of trees in structurally homogeneous stands. Here we have shown the change of height values of Scots pine trees (Pinus sylvestris L.) and the deviation of the arithmetic mean value of the height for model trees applying the raster images with different spatial resolutions. Predicted tree heights were compared with the empirical values, which were obtained directly measuring the felled sample trees with the addition of stump height. CHM analysis with a cell size of 0.04 m2 and 0.1 m2 shows the smallest height deviations for model trees in the stand. The deviations relative to arithmetic mean were 2.3 % and 2.6 %. Raster images with a more coarse resolution (more than 1 m2) are not recommended in forest practice, since their utilizing entails a measurement error of 17% and higher, and thus exceeding the permissible deviations in tree height measurements according to the forest inventory guidelines currently applied in Ukraine.