Resumen

SANTOYO-DE LA CRUZ, M.F. et al. Soil salinity levels, using unmanned aerial vehicles imagery, neural networks and decision trees. RIIIT. Rev. int. investig. innov. tecnol. [online]. 2023, vol.11, n.64, pp.49-65.  Epub 26-Ene-2026. ISSN 2007-9753.

Soil salinity is a global problem that threatens crop growth, and yields and impedes the and sustainable development of modern agriculture with major staple crops unable to complete their life cycle when soil NaCl concentrations exceed 200 mM. More than one third of the world's irrigated plots are affected by salinization. Unmanned aerial vehicles offer a viable alternative for acquiring remote sensing data. One way to classify and detect soil salinity variables are neural networks and decision trees. The objective of this work is the detection and classification of soil salinity through the use of multispectral images captured by drones as well as the use of decision tree models and neural networks, to estimate salinity variables: Exchangeable Sodium Percentage (ESP), Adjusted Sodium Adsorption Ratio (SARaj) and sodium concentration (mEq L-1). Agricultural soil with salinity problems was sampled, processed in the laboratory, the results were interpreted and classified according to salinity levels. A drone overflight was conducted to captured images and extract reflectances of the four multispectral bands (Green, red, red edge and near infrared). The salinity indexes were calculated to classify soil salinity using neural networks and decision trees. The neural networks classified 82.6% of the samples in the medium level category of ESP estimation. The model also presented an accuracy of 71.88% when determining sodium ion concentration in the neural networks. The Research work showed that it is possible to classify soil salinity with acceptable precision from images captured with unmanned aerial vehicles. Neural network models showed better accuracy in the estimation of soil salinity indicators with respect to decision trees. These technological procedures can be applied in soil salinity mapping to know the affected areas in an easy and fast way.

Palabras llave : drones; multispectral images; precision agriculture; reflectances; salinity index.

        · resumen en Español     · texto en Español     · Español ( pdf )