Development of tropical grassland biomass prediction model based on UAV RGB images
¹University State Southwestern Bahia, Postgraduate Program in Animal Science,
BR 415, Itapetinga – BA, Brazil
²University State Southwestern Bahia, Department of Plant Science and Animal Science, Road of Good Will, km 04, Vitória da Conquista – BA, Brazil
³University State Southwestern Bahia, Department of Agricultural and Soil Engineering, Road of Good Will, km 04, Vitória da Conquista – BA, Brazil
⁴University State of Feira de Santana, Department of Biological Sciences, Av. Transnordestina, Feira de Santana – BA, Brazil
*Correspondence: agro.barbara@outlook.com
Abstract:
The objective of this study is to assess the predictive potential of indices derived from RGB images captured by a camera mounted on a remotely piloted vehicle (RPV) to estimate the fresh and dry forage yield of grasses from the Urochloa genus. The experiment was conducted between December 2021 and January 2023, involving four cultivars of the Urochloa genus (U. brizantha cv. Braúna, U. brizantha cv. Paiaguás, U. hybrid cv. Camello, and U. decumbens cv. Basilisk), with flights conducted at two heights (20 and 100 metres). The values of the Green Leaf Index (GLI) and Digital Vegetation Model (DVM) extracted were correlated with the yields of fresh (FFY), dry forage yield (DFY), dry matter content (DM), and crude protein (CP). The results showed that DVM exhibited greater efficiency in estimating DM and CP at a flight altitude of 20 m. In contrast, GLI proved more efficient in estimating FFY and DFY at 100 m altitude, suggesting the potential for combining DVM and GLI to develop predictive models. The RGB images obtained via RPV have potential for estimating forage productivity and quality, expanding the possibilities of pasture management techniques.
Key words:
digital vegetation model, drone, green leaf index, pasture, production and consumption charts