Bazı Ekmeklik ve Makarnalık Buğday Çeşitlerinde Sarı Pas Hastalığının Fenolojik Dönemlere göre Hastalık-Verim İlişkisinin Çok Bantlı (Hiperspektral) Veriler Kullanılarak Araştırılması

Abstract

The aim of this study was to evaluate the severity of yellow rust in different phenological periods by 
 subjecting bread (Bayraktar 2000, Demir 2000, Eser and Kenanbey) and durum (Çeşit-1252, Eminbey, Kızıltan 91 and Mirzabey 2000) wheat varieties to different spore doses (0%, 25%, 50% and 100%) under controlled epidemic conditions. The research was conducted in Yenimahalle, Ankara, Turkey during the 2018-2019 growing season. In the study, the morphological changes in yellow rust severity were determined at different phenological developmental stages of the test materials with the reflectance values obtained by using handheld spectroradiometer in different spore dose applications during the period from tillering to stalk emergence. These reflectance values were converted into vegetation index values expressed by mathematical formulae and used in determining yield estimates. Considering the results obtained, it was determined that the spectral indices calculated especially in the early flowering period (25 May 2019, Feekes 10.5.1) were effective in yield estimation for all bread varieties except Kenanbey variety (15 June 2019, Feekes 10.5.4). It was determined that the spectral band region of 25 May 2019 (Feekes 10.5.1), which includes all indices determined to predict yield in all bread and durum varieties and which is the beginning of flowering, was effective. In grain yield estimation, it was determined that there was a decrease in the correlation values of the spectral indices starting from the early flowering period (Feekes 10.5.1) towards the grain setting period (Feekes 10.5.3) and milk maturity period 
 (Feekes 10.5.4). When the correlations between these index values and yield values were examined, it was 
 determined that prominent phenological periods and high correlation indices could be calculated for these periods. Nowadays, with the use of optical sensor technology instead of traditional disease surveillance methods, the way has paved the way for the development of new approaches for early, fast and accurate yield estimation as a result of the verification of images taken by unmanned aerial vehicles on which multispectral and hyperspectral cameras are located with ground data using artificial intelligence and deep learning techniques.