Sensor multiespectral embarcado em sistema de aeronave remotamente pilotada para manejo de nitrogênio em milho e feijão-comum

Abstract

This comprehensive study aims to advance the field of rapid phenotyping by leveraging unmanned aerial vehicles (UAVs) and multispectral imagery to enhance the assessment of critical plant traits, with a particular focus on nitrogen content and yield optimization. Building upon a systematic review of 41 peer-reviewed papers from 13 countries, on the first chapter, we synthesized existing knowledge on the association between vegetation indices (VIs) and plant traits across different growth stages of 11 major crop species. Drawing insights from experiments conducted over two consecutive years, we investigated the correlation between VIs, leaf nitrogen content (LNC), and yield at key growth stages of corn and common bean. The second chapter examined the efficacy of VIs such as Green Normalized Difference Vegetation Index (GNDVI), Green to Near–Infrared Band Ratio (GN) and Transformed Chlorophyll Absorption in Reflectance Index (TCARI) in predicting nitrogen rates and grain yield in corn at V6, V11, and R1 stages. The third chapter focused on common bean growth, integrating selected VIs with texture data derived from UAV-based multispectral images to accurately estimate LNC across critical growth stages (V4, R5, and R7). Notably, the Green Normalized Difference Vegetation Index (GNDVI) and Modified Chlorophyll Absorption in Reflectance Index (MCARI) emerged as robust predictors of LNC, with the addition of texture metrics enhancing accuracy, especially when combined with machine learning models like random forest (RF) and support vector machine (SVM). Our findings underscore the potential of UAV-based multispectral imagery coupled with advanced analytical techniques to revolutionize phenotyping efforts, facilitating more precise monitoring of key plant traits and optimizing crop management strategies for better use of inputs and maintenance of yield over time.