Spatio-Temporal Analysis of Yield Estimation of Corn and Sunflower Using Single-Dated Satellite Imagery Between 2014-2016

Multi-temporal satellite images are generally used to calculate yield estimation. However, investigating the feasibility of yield estimation using a single-dated satellite image still needs to be solved today. The benefits of using a single-dated satellite image in yield estimation and optimum timing are among the issues to be considered in future studies. This study developed a parcel-based yield model using a single satellite image for the 2014, 2015, and 2016 years and meteorological data in Kadirli-Osmaniye, Türkiye. The SPOT 6&7 satellite images were acquired at an appropriate time of the plant cover (the greenest period of the plant). For four months, photographs and values of plant height are provided weekly from fieldwork to determine crop cover fraction and monitor vegetation development. Regression analysis was performed with 12 vegetation indexes obtained from satellite images and crop cover fractions. In all calculated yield models, the relationship between I.-II.-III. soil classes selected, considering soil maps and yield, are compared. The relative errors of the low-yield fields were calculated between 10% and 35% (according to the yields obtained at harvest time). As a result, it was determined that a single satellite image could model yield estimation of corn and sunflower for each year on fields with an average of 0.10 relative errors.