Digitalization of olive trees by using remote sensing techniques

Olive is one of the economic agricultural crops of Aegean Region. In order to be able to implement agricultural policies effectively and revise, an inventory of olive groves must be geographically known and recorded. Today, the olive planted area is in a greater extend can be determined more precisely in a short time through high resolution satellite images with low cost and labor. The olive tree inventory in Turkey is based on farmer's declarations and is kept in text files. Therefore, great difficulties are confronted in terms of updating the number of trees that are either planted or cut down. The aim of this study is determining new approach to determine precise olive trees by using new technologies such remote sensing and GTS. Tn this pilot study, a highly accurate digital olive tree inventory has been formed by making use of PLETADES (Spatial resolution=50 cm) and Worldviev-2 satellites images. As the study area, olive plantations in Gölmarmara district of Manisa province, where olive is intensively farmed, was chosen. Tn the study, olive trees have been marked on the maps and were grouped according to their canopy sizes. Also, this inventory has the quality to be based on in yield determination. Distinguishing characteristics of other trees, which might confuse the detections, were gathered through ground works. In doing so, reflection coefficients on VR, VG and NIR bands of these observed trees were determined. It was observed that pansharpened satellite images, which contain near infrared band, were quite effective in distinguishing olive trees from others. It was also experienced that the time of taking the image was another important factor in distinguishing olive trees. Winter months were found appropriate due to the fact that olive tree, an evergreen, is easier to distinguish then. In this study, all the parameters were determined geographically and inserted as a layer into the GIS to create database for the olive planted areas. Olive trees in the area were counted in seven groups that were categorized making use of indicator patterns prepared.