Spectral data modelling for tropical soil fertility analysis: association of vis-NIR and XRF techniques for the modernization of the traditional methods of analysis

Abstract

The modernization of soil fertility diagnostics using sensor systems is a current and prominent topic of research in the context of Soil Science and Precision Agriculture. The application of non-destructive techniques in this diagnosis makes it possible to reduce the use of chemical reagents and reduces the number of operations involved in the analytical procedure, making it faster. This has a positive impact on the environment and has the potential to optimize the management of fertilizers for crop production. Our research proposal aims to develop a method for using X-Ray Fluorescence (XRF) spectrometry and its integration with visible and Near-Infrared (vis-NIR) diffuse reflectance spectroscopy as an analysis tool for soil fertility diagnosis. For this, three steps with well-defined objectives will be conducted. In the first one, the temporal stability of XRF models for predicting fertility attributes will be assessed, as well as the influence of the change in fertilization management on this temporal stability. In the second step, the ability of vis-NIR and XRF spectra to group samples with similar performance for predicting fertility attributes via XRF sensor will be assessed by multivariate statistics. Finally, the third step aims to evaluate the potential and applicability of data fusion techniques to exploit the synergy between vis-NIR and XRF sensors to improve the accuracy of soil fertility prediction. We also highlight that the evaluation of the second and third steps will be conducted using a massive group of data, which will give a comprehensive coverage and confidence in the results. With the execution of this project it is expected to evolve the Technology Readiness Level (TRL) of the application of XRF and vis-NIR sensors as an analytical method of soil fertility, subsidizing the integration of these sensors in agricultural analysis laboratories or in on-board systems for on-site analysis. (AU)