Rendimiento en arrozales a partir de información textural y espectral (drone) bajo riego con alternancia de humedecimiento y secado
Loading...
Código QR
Authors
Solis Paucar, César Augusto
Contact Email
Abstract
La creciente escasez de agua a nivel mundial, agravada por la crisis del cambio climático, afecta considerablemente a los cultivos de arroz, especialmente a aquellos que emplean el método de riego por inundación continua, el más utilizado globalmente. En este contexto, la presente investigación tiene como propósito evaluar el rendimiento del arroz en pozas manejadas con dos sistemas de riego: inundación continua (IC) y alternancia de humedecimiento y secado (AWD), este último aplicado con niveles de agua de 5 cm, 10 cm y 20 cm por debajo de la superficie del suelo (AWD5, AWD10 y AWD20). Además, se buscó predecir el rendimiento del cultivo empleando índices de vegetación (IV) y medidas texturales (MT). El estudio se desarrolló en la zona experimental del INIA-Vista Florida, situada en el distrito de Picsi, departamento de Lambayeque. Para ello, se emplearon imágenes multiespectrales y RGB obtenidas mediante vuelos realizados con un UAV. Los resultados revelaron que los modelos predictivos más precisos se lograron en las etapas fenológicas de macollamiento y floración, utilizando hasta tres variables dependientes (IV y/o MT). Entre las variables más destacadas se identificaron los IV NDVI y SP y las MT VAR y SA. Los modelos con validación cruzada alcanzaron coeficientes de determinación (R²) de 0,674 y 0,643, con errores cuadráticos medios (RMSE) de 0,781 y 0,848 tn ha⁻¹, respectivamente. Estos resultados evidencian que los modelos que combinan IV y MT presentan una mayor precisión, como lo demuestra su bajo RMSE. En conclusión, esta investigación proporciona herramientas para una gestión más eficiente del agua en sistemas agrícolas y fomenta el desarrollo de modelos predictivos más precisos y adaptados a los desafíos impuestos por el cambio climático.
The increasing global water scarcity, exacerbated by the climate change crisis, significantly impacts rice crops, particularly those using continuous flooding irrigation, the most widely used method worldwide. In this context, the present study aims to evaluate rice yield in plots managed under two irrigation systems: continuous flooding (CF) and alternate wetting and drying (AWD). The latter was applied with water levels of 5 cm, 10 cm, and 20 cm below the soil surface (AWD5, AWD10, and AWD20). Additionally, this research sought to predict yield using vegetation indices (VI) and textural measures (TM). The study was conducted in the experimental area of INIA-Vista Florida, located in the district of Picsi, Lambayeque department. Multispectral and RGB images captured through UAV flights were used for analysis. The results showed that the most accurate predictive models were achieved during the phenological stages of Tillering and Flowering, using up to three independent variables (VIs and/or TMs). Among the most significant variables identified were the VIs NDVI and SP, and the TMs VAR and SA. The models with cross-validation achieved coefficients of determination (R²) of 0,674 and 0,643, with root mean square errors (RMSE) of 0,781 and 0,848 t ha⁻¹, respectively. These results demonstrate that models combining VIs and TMs provide greater accuracy, as evidenced by their low RMSE. In conclusion, this research provides tools for more efficient water management in agricultural systems and promotes the development of more accurate predictive models adapted to the challenges imposed by climate change.
The increasing global water scarcity, exacerbated by the climate change crisis, significantly impacts rice crops, particularly those using continuous flooding irrigation, the most widely used method worldwide. In this context, the present study aims to evaluate rice yield in plots managed under two irrigation systems: continuous flooding (CF) and alternate wetting and drying (AWD). The latter was applied with water levels of 5 cm, 10 cm, and 20 cm below the soil surface (AWD5, AWD10, and AWD20). Additionally, this research sought to predict yield using vegetation indices (VI) and textural measures (TM). The study was conducted in the experimental area of INIA-Vista Florida, located in the district of Picsi, Lambayeque department. Multispectral and RGB images captured through UAV flights were used for analysis. The results showed that the most accurate predictive models were achieved during the phenological stages of Tillering and Flowering, using up to three independent variables (VIs and/or TMs). Among the most significant variables identified were the VIs NDVI and SP, and the TMs VAR and SA. The models with cross-validation achieved coefficients of determination (R²) of 0,674 and 0,643, with root mean square errors (RMSE) of 0,781 and 0,848 t ha⁻¹, respectively. These results demonstrate that models combining VIs and TMs provide greater accuracy, as evidenced by their low RMSE. In conclusion, this research provides tools for more efficient water management in agricultural systems and promotes the development of more accurate predictive models adapted to the challenges imposed by climate change.
Description
Universidad Nacional Agraria La Molina. Facultad de Ciencias. Departamento Académico de Ingeniería Ambiental, Física y Meteorología
Keywords
BY NC SA
Citation
Date
2025
Collections
Seleccionar año de consulta:
Licencia de uso

Excepto si se señala otra cosa, la licencia del ítem se describe como info:eu-repo/semantics/openAccess