RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación
Artículo científico (JournalArticle)
Handheld sensing system for data-driven irrigation management in olive orchards
[EN] Proper water management is necessary to optimize the quanti-ty and quality of olive oil. The advent of monitoring tools capable of providing reliable and systematic information on tree water sta-tus, and thus vegetative growth, can be advantageous for growers of olive groves. Leaf temperature, environmental conditions, crop water stress index (CWSI), and other related vegetation indices (VI) measured non-invasively using proximal sensing tools are being increasingly employed for agriculture 4.0 applications. This study aimed to implement and evaluate a low-cost handheld sys-tem to determine the water conditions of olive trees under differ-ent irrigation treatments. Implementation of the data-driven sys-tem required the selection of the most efficient CWSI equation for the developed proximal sensing device. Specifically, five potential equations were evaluated, including two analytical models, one empirical equation derived from existing literature, a newly pro-posed empirical equation, and a hybrid model combining analyti-cal and empirical calculations. The sensing system was equipped with a Global Navigation Satellite System (GNSS), an infrared thermometer, a compact NDVI sensor with ambient light correc-tion, and an environmental measuring unit providing air tempera-ture and relative humidity. Additionally, the leaf water potential (LWP) was calculated in real time to better determine the actual hydric stress conditions of the trees. All data were acquired between 12:00 and 14:00 on both the sunny and shaded canopy sides. The experimental results showed that the handheld system eased the collection of field data to help growers schedule and manage irrigation for olive oil production through stress identifi-cation and precise GPS positioning. The best correlation between LWP and CWSI was found for the analytical formulas (R2 = 0.62), followed by the empirical formula (R2 = 0.55); however, both ana-lytical equations required a higher number of measurements when compared to the alternative models considered, which complicat-ed their practical implementation in the handheld prototype., This work was supported by the project "Piattaforma DIGItale per La difesa di precisione dell'OliVEto" - "DIGILOVE",- "National Research Centre for Agricultural Technologies (Agritech) " (MUR, PNRR-M4C2, CN00000022) , Spoke 2 'University degli Studi di Napoli Federico II', CUP: E63C22000920005. The authors are grateful to the Agricultural Robotics Laboratory research group of the Polytechnic University of Valencia for the scientific collaboration and personal support, in particular to: Dr. V. Saiz-Rubio, Prof. E. Ortiz, Mr. A. Cuenca Cuenca, Mr. Montano Perez, Prof. A. Torregrosa, and Prof. C. Ortiz.