Early Detection of Rice Blast Disease Using Satellite Imagery and Machine Learning on Large Intrafield Datasets
[EN] This study explores the use of remote sensing and machine learning (ML) for early detection of Pyricularia oryzae (rice blast) in 'Bomba' rice. Conducted in Spain's Albufera Natural Park over four seasons (2021-2024), 94 fields were monitored using Sentinel-2 imagery...
| Autores: | , , , , , , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2025 |
| País: | España |
| Institución: | Universitat Politècnica de València (UPV) |
| Repositorio: | RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia |
| Idioma: | inglés |
| OAI Identifier: | oai:riunet.upv.es:10251/233262 |
| Acceso en línea: | https://riunet.upv.es/handle/10251/233262 |
| Access Level: | acceso abierto |
| Palabra clave: | Remote sensing Sentinel-2 Rice Blast Machine learning |
| Sumario: | [EN] This study explores the use of remote sensing and machine learning (ML) for early detection of Pyricularia oryzae (rice blast) in 'Bomba' rice. Conducted in Spain's Albufera Natural Park over four seasons (2021-2024), 94 fields were monitored using Sentinel-2 imagery and Topcon Yield Trakk data. Principal Component Analysis (PCA) identified key spectral bands (B03, B04, B05, B07, B08, B11) at early stages (35 and 55 DAS). Three ML classifiers-K-Nearest Neighbors (KNN), Random Forest (RF) and Support Vector Machines (SVMs)-were tested to categorize fields by yield-based infection levels. RF achieved the best performance (up to 94% Accuracy), showing high robustness across band combinations and dates. KNN was more input-sensitive, and SVM performed weakest. Integrating multispectral and multitemporal data enhanced accuracy. Overall, RF and remote sensing proved reliable tools for early disease detection, supporting Precision Agriculture and real-time pest management. |
|---|