Estimación de biomasa disponible mediante imágenes multiespectrales adquiridas con un dron

In Uruguay, pastures are essential for livestock and dairy farming, allowing for competitive and differentiated production. However, pasture monitoring is not a common practice in establishments, as it requires a methodology to manage and improve the efficiency of the system through decision-making. Normally, this is done using direct and indirect methods, which require greater dedication and higher costs. The objective of this work is to estimate the available biomass using multispectral images acquired with a drone. Subsequently, achieve its application for decision-making through the monitoring of state variables (Stock), in the grazing area of ​​two intensive pastoral-based milk production systems, for the winter season. The work was carried out at the Southern Regional Center (CRS) of the Faculty of Agronomy (FAgro) in 2023 in the months between June and September. A DJI Phantom 4 Multispectral drone was used in a second-year Dactylis pasture mixed with Alfalfa. The work consisted of two experiments called “drone calibration” and “AP flights”. In the first, weekly flights were carried out for 3 months in two 6x4m test plots, in which 22 50x30cm cuts were made each week in which the biomass within the quadrants (KgMS/ha), height (cm) and NDVI were determined, thus generating data for 7, 14, 21, 28 and 35 days of growth. The data were analyzed using a 2nd degree polynomial regression model correlating the variables available biomass and height with the NDVI to determine its goodness of fit using the regression coefficient. The correlation between NDVI and available biomass (kgDM/ha) showed a correlation of 0.41 (R²), although with variations in the different growth stages. On the other hand, the correlation between NDVI and height presents an R² of 0.29. Regarding the “AP flights” experiment, using the model from the first experiment, flights were made in the CRS Grazing Area (AP) with the purpose of estimating the forage stock and comparing it with another method (C-Dax). The flights were made with overlaps of 20%. The average stock values ​​in the AP were 669 to 831 KgDM/ha and a difference with the values ​​estimated with the C-Dax of +/-1700 KgDM/ha throughout the study period. In conclusion, for this work, and based on the experiments carried out, the drone reflected average correlation data for a Dactylis and Alfalfa pasture in winter conditions, but it provides a first approximation to what could be an important technological advance and the first precedent in terms of biomass estimation with this methodology in the AP.