A Decision Tree-based Classification of Diseased Pine and Oak Trees Using Satellite Imagery

Abstract

Tree diseases contribute to the reduction of forest areas over the years and early detection of these diseases is essential to prevent its rapid spread and eventually provide immediate cure. In this study, the Japanese pine wilt (JPW) and the Japanese oak wilt (JOW) diseases were used. These two tree diseases were detected using high-resolution satellite imagery. JPW is a lethal disease that brought damagr and devastation to the greater number of pine trees in Japan which is primarily brought by the pinewood nematode (Bursaphelenchus xylophilus). JOW, on the other hand, is a vector-borne disease caused by a symbiotic fungus spreaded by the flying ambrosia beetle (Platypus quercivorus) that serves as a vector. A machine learning (ML) algorithm based on decision tree (DT) was implemented and programmed using the ML repository dataset obtained from the University of California, Irvine (UCI). The data will be used to classify image segments into two types: diseased or wilted trees, and others. The trained algorithm was able to classify the image segments with a high accuracy of 98.14%.