Detection of Red Crown Rot of Soybean in Illinois Fields Using High-Resolution Satellite Imagery and Machine Learning
Remote Sensing · 08.11.2026 · 18 (21) · 3839
DOI: 10.3390/rs18213839 ↗Kurz gefasst
Fachartikel in der Zeitschrift Remote Sensing, online erschienen am 08.11.2026. Verfasst von Bruno D. Pugliese und 6 weiteren Autor:innen. Die Arbeit nennt 65 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Illinois Soybean Association.
Abstract (Originalsprache)
Red crown rot (RCR), caused by Calonectria ilicicola, is an emerging soybean disease in the U.S. Midwest. While field monitoring relies on labor-intensive scouting, satellite-based detection remains unexplored, and scalable approaches to characterize within-field distribution are lacking. This study evaluated high-resolution PlanetScope satellite imagery for mapping RCR-affected soybean canopies across 15 commercial fields in Illinois (2024–2025). A total of 2921 georeferenced canopy plots were classified as asymptomatic or RCR-affected and paired with six multispectral bands and seven vegetation indices. Model transferability to unseen fields was assessed using spatially independent leave-one-field-out cross-validation. Seven machine learning classifiers representing five model families were benchmarked to determine whether model complexity improves generalization. All classifiers showed strong and near-equivalent discrimination (ROC-AUC: 0.963–0.982), indicating that increased model complexity provided no meaningful advantage. Regularized logistic regression achieved the highest performance (accuracy: 0.945, balanced accuracy: 0.945, F1-score: 0.948, ROC-AUC: 0.982) at the optimized decision threshold. Permutation analysis identified EVI, NDVI, and red reflectance as the most influential predictors across representative architectures. Satellite-derived classification maps generally corresponded with symptomatic patterns in high-resolution UAV imagery, although mixed pixels reduced precision near patch boundaries. To our knowledge, this is the first study to evaluate satellite-based machine learning classification of RCR across independent commercial fields, and the results show that a simple, interpretable regularized model provides a transferable framework for within-field mapping of RCR-associated symptoms.
Förderung
- Illinois Soybean Association
Zitieren
APA
Pugliese, B. D., Paredes, J. A., Ruiz, A. F., Bowman, N. D., Ersoz, E., Martin, N. F., & Camiletti, B. X. (2026). Detection of Red Crown Rot of Soybean in Illinois Fields Using High-Resolution Satellite Imagery and Machine Learning. Remote Sensing, 18(21), 3839. https://doi.org/10.3390/rs18213839
Vancouver
Pugliese BD, Paredes JA, Ruiz AF, Bowman ND, Ersoz E, Martin NF, et al. Detection of Red Crown Rot of Soybean in Illinois Fields Using High-Resolution Satellite Imagery and Machine Learning. Remote Sensing. 2026;18(21):3839. doi:10.3390/rs18213839
BibTeX
@article{Pugliese2026,
author = {Pugliese, Bruno D. and Paredes, Juan A. and Ruiz, Andres F. and Bowman, Norman Dennis and Ersoz, Elhan and Martin, Nicolas F. and Camiletti, Boris X.},
title = {Detection of Red Crown Rot of Soybean in Illinois Fields Using High-Resolution Satellite Imagery and Machine Learning},
journal = {Remote Sensing},
year = {2026},
volume = {18},
number = {21},
pages = {3839},
doi = {10.3390/rs18213839}
}
EUVIMED