EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

An explainable Bi-LSTM model for winter wheat yield prediction

Abhasha Joshi, Biswajeet Pradhan, Subrata Chakraborty, Renuganth Varatharajoo, Abdullah Alamri, Shilpa Gite, Chang-Wook Lee

Frontiers in Plant Science · 17.01.2025 · 15

DOI: 10.3389/fpls.2024.1491493 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Frontiers in Plant Science, online erschienen am 17.01.2025. Verfasst von Abhasha Joshi und 6 weiteren Autor:innen. Beteiligte Einrichtung: Centre for Advanced Modeling and Geospatial Information Systems (CAMGIS), School of Civil and Environmental Engineering, Faculty of Engineering & IT,…. Die Arbeit wurde bisher 39-mal zitiert (Crossref) und nennt 63 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: University of Technology Sydney, King Saud University, Kangwon National University.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate, reliable and transparent crop yield prediction is crucial for informed decision-making by governments, farmers, and businesses regarding food security as well as agricultural business and management. Deep learning (DL) methods, particularly Long Short-Term Memory networks, have emerged as one of the most widely used architectures in yield prediction studies, providing promising results. Although other sequential DL methods like 1D Convolutional Neural Networks (1D-CNN) and Bidirectional long short-term memory (Bi-LSTM) have shown high accuracy for various tasks, including crop yield prediction, their application in regional scale crop yield prediction remains largely unexplored. Interpretability is another pressing and challenging issue in DL-based crop yield prediction, a factor that ensures the reliability of the model. Thus, this study aims to develop and implement an explainable DL model capable of accurately predicting crop yield and providing explanations for the predictions. To achieve this, we developed three state-of-the-art sequential DL models: LSTM, 1D CNN, and Bi-LSTM. We then employed three popular interpretability techniques: Local interpretable model-agnostic explanations (LIME), Integrated Gradient (IG) and Shapley Additive Explanation (SHAP) to understand the decision-making process of the models. The Bi-LSTM model outperformed other models in terms of predictive performance (R 2 up to 0.88) and generalizability across locations and ranges of yield data. Explainability analysis reveals that enhanced vegetation index (EVI), temperature and precipitation at later stages of crop growth are most important in determining Winter wheat yield. Further, we demonstrated that XAI methods can also be used to understand the decision-making process of the models, to understand instances such as high- and low-yield samples, to find possible explanations for erroneous predictions, and to identify regions impacted by particular stress. By employing advanced DL techniques along with an innovative approach to explainability, this study achieves highly accurate yield prediction while providing intuitive insights into the model’s decision-making process.

Schlagwörter

Stichwörter der Autor:innen

bidirectional LSTMcrop yieldexplainabilityintegrated gradientslimeshap

Autor:innen und Institutionen

  • Centre for Advanced Modeling and Geospatial Information Systems (CAMGIS), School of Civil and Environmental Engineering, Faculty of Engineering & IT, University of Technology Sydney, Ultimo, NSW, Australia.
  • School of Science and Technology, Faculty of Science, Agriculture, Business and Law, University of New England, Armidale, NSW, Australia.
  • Department of Aerospace Engineering, University Putra Malaysia, UPM, Serdang, Malaysia.
  • Department of Geology and Geophysics, College of Science, King Saud University, Riyadh, Saudi Arabia.
  • Symbiosis Centre for Applied Artificial Intelligence, Symbiosis Institute of Technology, Symbiosis International Deemed University, Pune, India.
  • Department of Science Education, Kangwon National University, Chuncheon-si, Republic of Korea.

Förderung

  • University of Technology Sydney
  • King Saud University
  • Kangwon National University

Zitieren

APA

Joshi, A., Pradhan, B., Chakraborty, S., Varatharajoo, R., Alamri, A., Gite, S., & Lee, C. W. (2025). An explainable Bi-LSTM model for winter wheat yield prediction. Frontiers in Plant Science, 15. https://doi.org/10.3389/fpls.2024.1491493

Vancouver

Joshi A, Pradhan B, Chakraborty S, Varatharajoo R, Alamri A, Gite S, et al. An explainable Bi-LSTM model for winter wheat yield prediction. Frontiers in Plant Science. 2025;15. doi:10.3389/fpls.2024.1491493

BibTeX
@article{Joshi2025,
  author = {Joshi, Abhasha and Pradhan, Biswajeet and Chakraborty, Subrata and Varatharajoo, Renuganth and Alamri, Abdullah and Gite, Shilpa and Lee, Chang-Wook},
  title = {An explainable Bi-LSTM model for winter wheat yield prediction},
  journal = {Frontiers in Plant Science},
  year = {2025},
  volume = {15},
  doi = {10.3389/fpls.2024.1491493}
}
Kostenlos starten →