Speaker
Description
The determination of physic-chemical parameters in soil and plants are of fundamental importance for agriculture and livestock farming because they describe soil fertility and plant nutrition and have direct impact in production costs. The use of proximal soil sensing technology based on spectroscopy techniques is an innovative tendency in soil and plant analysis[1]. Energy Dispersive X-ray Fluorescence (EDXRF) with benchtop and portable instrumentation have been increasingly used in agriculture purpose. We are conducting a project which objective is to optimize the employment of EDXRF combined with machine learning methods for the rapid determination of soil fertility attributes and plant nutrition parameters. The goal is to offer versatility and speed for application in precision agriculture. Thus, it is proposed to optimize the operating conditions of the spectrometers as well as the processing parameters of the equipment output data. The main focus is on the generation of robust prediction models accompanied by a physical and chemical interpretation of the variables that allow the generation of such models. Therefore, pre-processing and selection of variables in the data have been studied, as well as the figures of merit of the models. The realization of this proposal will allow obtaining alternative methodologies for soil analysis and foliar analysis, significantly reducing time, instrumentation and the quantity of reagents. Moreover, datasets from the same place monitored over the years, and datasets samples from different places are being modelled and evaluated. The temporal and spatial validation of models is a complex task which requires different approaches. Finally, the use of spectrometric sensors will allow innovations in precision agriculture and as consequence a feasible possibility of analysis with high point density, serving as a basis for variable and optimized soil correction or foliar fertilization. Case of studies considering soil fertility analysis from three areas with
different soil types [2,3] and the evaluation of some plants like soybean, coffee and grass will be presented.
[1] T.R. Tavares, et al., Agronomy 11 (6), 2021, 1028
[2] F.R. dos Santos, J.F. de Oliveira, E. Bona, G.M.C. Barbosa, F.L. Melquiades, Microchem. J. 191, 2023, 108813
[3] J.V. Ribeiro, F.R. dos Santos, J.F. de Oliveira, G.M.C. Barbosa, F.L. Melquiades. Spectrochim Acta Part B At Spectrosc 211, 2024, 106835
