Utilización de drones en la estimación de parámetros para la gestión de sistemas pastoriles

In Uruguay, meat and milk production primarily comes from pastoral systems, whose productivity improves with efficient forage management. This requires understanding of forage availability and planning its use. This study aimed to estimate forage availability, height, and the number of leaves per tiller using a DJI Phantom 4 Multispectral drone and the Normalized Difference Vegetation Index (NDVI). The experiment, conducted in winter on perennial ryegrass and ryegrass with plantain pastures at the Southern Regional Center of the Faculty of Agronomy (Udelar), included two stages: calibration and practical application. During the calibration stage, forage height was measured using a rising plate meter (RPM) and a graduated ruler, alongside recording the number of leaves per tiller and biomass. Sampling was carried out weekly. For sampling, thirty 0.17 m² quadrats were placed using the double sampling technique. A total of 420 quadrats were sampled over 14 sessions. Simultaneously, drone flights were conducted, and the average NDVI of each quadrat was calculated using Agisoft Metashape Professional and QGis software. Simple linear regressions revealed the following correlations: R² between height measured with RPM and biomass was 0.69; R² between height measured with the graduated ruler and biomass was 0.75; R² between biomass and NDVI was 0.48; R² between height measured with RPM and NDVI was 0.37; and R² between height measured with the graduated ruler and NDVI was 0.46. These regressions enabled the calibration of the drone to estimate both biomass and forage height through NDVI. However, no relationship was found between the number of leaves per tiller and biomass, nor between the number of leaves per tiller and NDVI. With the calibration established, biomass estimated through the drone's weekly flights was compared to biomass estimated by an indirect determination device (C-Dax). This comparison revealed an average overestimation of 58% by the drone relative to the C-Dax. Nonetheless, the drone accurately represented the growth behavior of the pastures, which was encouraging. Additionally, forage availability before grazing and residuals after grazing were estimated using the drone, the C-Dax, and the RPM. The results were promising, demonstrating that the drone can be incorporated into the estimation of disappeared forage in plots, which would be useful in pastoral systems. The drone's ability to estimate forage height before and after grazing was also evaluated against the RPM. Results showed that the drone was a good estimator of height, with fewer overestimations than the RPM. Finally, spatial maps of biomass and height estimated by the drone were generated, offering visual representations of biomass and serving as valuable tools for monitoring and decision-making.