Monitoramento do nitrogênio foliar e da cobertura vegetal de cafeeiros em sistema de cultivo orgânico com aeronave remotamente pilotada

Abstract

High-resolution images, obtained in aerial surveys with remotely piloted aircraft, have been incorporated into agricultural studies to assist in the planning of land use and agricultural inputs, and thus collaborate with the development of more sustainable agricultural practices. In this context, the objective of this study was to evaluate the potential of vegetation indices, coming from images of the visible region, in monitoring the status and estimating the leaf Nitrogen content of coffee trees in transition for organic farming system. The research was carried out in two phases: the first phase was carried out in May 2019, and consisted of a single flight for imaging, in addition to readings with the SPAD chlorophyll meter and leaf analysis (Kjeldahl method for determining the Nitrogen content) in an uncontrolled area of 1 ha. In this phase, Pearson's correlation coefficient, analysis of variance, and classification and regression models with the Random Forest algorithm were used as methods of analysis and estimation. In the second phase, an area of 0.23 ha was divided into 16 plots, in which 4 doses of organic fertilizer (0, 87.5, 175 and 262 kg N ha-1) were applied, and flights were carried out in October 2019, and January and March 2020, in addition to readings with the SPAD chlorophyll meter and leaf analysis (Kjeldahl method) in October and January. In addition, the variation in the vegetation cover index over time was also used as a variable of analysis. For statistical analysis, Pearson's correlation coefficient and analysis of variance were also used, along with logistic regression models. In the first phase of the study, the regression and classification models failed to monitor Nitrogen variability in the area, and its correlations with vegetation indices and SPAD values were not statistically significant. However, the images were very efficient for the supervised classification of land use. In the second phase, no significant correlations were found between the vegetation indices and leaf nitrogen at any time during the research. The logistic regression models, elaborated with the indices individually and combined, were also not able to identify the nutritional variation of the coffee trees. On the other hand, the readings with the SPAD were strongly related to the leaf nitrogen content in January 2020. The vegetation cover index was positively related to the different treatments, although the analysis of variance showed that the differences between the means were not significant. Despite this, the use of visible images in monitoring Nitrogen is still incipient, and further research is needed for a better understanding of the factors that may interfere with the results. In addition, these images proved to be efficient in monitoring the fraction of vegetation cover, very important information for the management of agricultural practices, not only in coffee growing.