THE COMPARE of OLIVE TREE COUNTING METHODS

In this project, which is aimed to determine the presence of olive trees, the template method which is applied manually (screen digitize) and OLICOUNT software which can make semiautomatic counting, eCognition software which is an object-based approach, and new M-OLICOUNT methods created by modifying the OLICOUNT were compared. In this study, 60 cm resolution images of Quickbird satellite including NIR wavelength energy were used. In order to prevent the decrease of accuracy by other trees mixing with the olive trees, the image of the winter months when the trees shed their leaves was used. As the study area, Karaot Village and its environs in the northeast of Torbali district of izmir province, villages in the center and surrounding of Urla District and lands in Mordogan Town of Karaburun District where the olive cover is widespread were chosen. As a result, it was determined that the accuracy rate of manual counting method was higher in comparison with other methods in the areas where olive trees were found mixed with natural vegetation where the topography was sloped under the conditions of Aegean Region and therefore the shadow effect was high. The other three methods were found to provide better performance in terms of time, labor and cost in areas which are flat and with a slope that is close to being flat and where the olive tree cultivation is made as an orchard production. © 2020 Parlar Scientific Publications. All rights reserved