Abstract

In this study, the trajectory of water droplets in a sprinkler irrigation system was studied using the optical tracking method. The analysis was carried out on the basis of 1920×1080 pixel, 60 frame/s video recordings taken in real field conditions. Image segmentation and trajectory detection algorithms were developed on the OpenCV and Python platforms to determine the trajectory of water droplets. The main focus of the study was to determine the dynamics of the movement of water droplets with a diameter of 0.002 meters after exiting a sprinkler head rotating at an angle of 360°. The following parameters were taken into account in the mathematical modeling process: exit velocity - 11 m/s, exit angle - 30°. Based on these data, the trajectory of the droplet was calculated using the equations of ballistic motion and air resistance (drag). The Random Forest model was used to assess the influence of factors. The results showed that the factors that have the greatest impact on the water spray trajectory are wind (31.2%) and terrain slope (25.6%). This means that small changes in wind speed and slope significantly reduce the water spray radius and cause uneven water distribution. The Convolutional Neural Network (CNN) model was used to spatially analyze and classify areas, achieving 93% accuracy in flat terrain and 74–79% accuracy in windy and uneven areas. This result indicates that the modeled system works with high reliability even in real field conditions. At the end of the study, the sprinkler exit angle and installation spacing were optimized, and the drift zones of water due to wind were reduced from 21.8% to 7.1%. This change has increased the stability of water distribution and allowed for a significant reduction in water consumption in crop production.