Investigating Correlation among NDVI Index Derived by Unmanned Aerial Vehicle Photography and Grain Yield under Late Drought Stress Conditions

AbstractNormalized Difference Vegetation Index (NDVI) has been related with grain yield and other important agronomic traits of rainfed cereal crops grown under Mediterranean conditions. In most studies, spectral measurements were carried out using portable spectroradiometers; however, Unmanned Aerial Vehicles (UAVs) carrying conventional digital cameras equipped with spectral filters have been recently used as an alternative. The main scope of this work was to investigate whether grain yield and yield-related traits under late drought stress conditions are correlated with NDVI Index derived by low height combined photography obtained by UAVs.Experiments were sown at the Athalassa experimental station for two consecutive growing seasons. Twenty durum wheat varieties were sown in a Randomized Complete Block Design with four and six replications in the 1st and 2nd growing season, respectively. In the 1st growing season, plants received 270mm of rainfall and extra 30mm irrigation during booting. In the 2nd growing season, plants received 140mm of rainfall, 50mm irrigation during tillering and 50mm irrigation during booting. In both growing seasons, plants received no rainfall or irrigation from booting to maturity.Conventional and near infra red photography were taken by a CanonIS200 camera mounted on a UAV. Geo-processing of the imagery and orthorectification was applied on the data to become compatible. After that proper samples were masked from the orthoimagery and processed for NDVI index calculation. Two flights were carried out during the 1st growing season when most varieties were at booting and milk stage respectively. Four flights were carried out during the 2nd growing season when most varieties were at booting, heading, milk and dough stages respectively.In our study there was significant correlation between yield and NDVI in both growing seasons with the correlation been stronger during the 2nd season. Correlations were stronger when estimated during the milk and dough stages