Use of a UAV for statistical-spectral analysis of vegetation indices in sugarcane plants in the Eastern Amazon.

The use of unmanned aerial vehicles is increasingly present in agricultural activities, representing an important innovation tool, being advantageous, for example, for the management of difficult-to-access areas and continuous monitoring of plantations, producing, with a relative low cost, high quality images. In this sense, the present work aimed to use a remotely piloted aircraft to carry out a statistical-spectral analysis of vegetation indices in sugarcane plantations in the Brazilian Eastern Amazon. To achieve this objective, an EbeeSQ model aircraft was used, which has a multispectral camera with a Sequoia sensor, which allowed aerial photographs to be obtained in four distinct spectral bands: green, red, red-edge and near-infrared. For the spectral analyses, thirteen study zones were considered, where four vegetation indices were calculated: NDVI, SAVI, GNDVI and NDRE. After generating the indices, it was possible to perform descriptive statistics for each one of them, as well as for the pixels of the original photograph. Tests were also carried out to verify the normality of the data, as well as correlation and regression tests between the vegetation indices and the digital numbers of the aerial photograph. Through the analyses, it was noticed that the GNDVI was the index that best adjusted to the data, presenting the highest mean, range, magnitude of correlation and R-square. While, NDVI and SAVI presented the least satisfactory results. Finally, from the GNDVI, it was possible to perform an image classification and area calculation of the low, medium and high portions of the GNDVI. [ABSTRACT FROM AUTHOR]