EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Leveraging level data for accurate downstream flow prediction in the Narmada River Basin with advanced machine learning models

Vijendra Kumar, Namal Rathnayake, S. Hariprasad Reddy, Naresh Kedam, Upaka Rathnayake, Yukinobu Hoshino

Journal of Hydroinformatics · 01.02.2025 · 27 (2) · 141-158

DOI: 10.2166/hydro.2025.131 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Hydroinformatics, online erschienen am 29.01.2025. Verfasst von Vijendra Kumar und 5 weiteren Autor:innen. Die Arbeit wurde bisher 10-mal zitiert (Crossref) und nennt 41 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate river flow prediction is essential for effective water resource management and flood forecasting. This study investigates advanced machine learning (ML) models CatBoost, LGBM, Random Forest, and XGBoost for predicting downstream flow at ‘Mandleshwar' using level data from key stations in the Narmada River basin. The dataset, spanning 1978–2020 with over 15,000 daily observations, was obtained from India’s Water Resources Information System (WRIS). Rigorous preprocessing ensured high-quality input data for modeling. During training, XGBoost achieved an MAE of 0.0031, MSE of 0.00002, RMSE of 0.0048, and R2 of 0.99, effectively capturing flow dynamics. In validation, LGBM outperformed with an R2 of 0.95, MAE of 0.0646, and RMSE of 0.2262, while CatBoost, XGBoost, and Random Forest followed closely. In testing, XGBoost led with an R2 of 0.96, RMSE of 0.2326, and MAE of 0.0680, while Random Forest and other models also demonstrated strong predictive capability. The findings emphasize the utility of level data for downstream flow prediction and the effectiveness of ML models in hydrological applications. This study provides actionable insights for optimizing reservoir operations, flood mitigation, and sustainable water management, supporting policymakers in developing adaptive strategies for the Narmada River basin.

Zitieren

APA

Kumar, V., Rathnayake, N., Reddy, S. H., Kedam, N., Rathnayake, U., & Hoshino, Y. (2025). Leveraging level data for accurate downstream flow prediction in the Narmada River Basin with advanced machine learning models. Journal of Hydroinformatics, 27(2), 141-158. https://doi.org/10.2166/hydro.2025.131

Vancouver

Kumar V, Rathnayake N, Reddy SH, Kedam N, Rathnayake U, Hoshino Y. Leveraging level data for accurate downstream flow prediction in the Narmada River Basin with advanced machine learning models. Journal of Hydroinformatics. 2025;27(2):141-158. doi:10.2166/hydro.2025.131

BibTeX
@article{Kumar2025,
  author = {Kumar, Vijendra and Rathnayake, Namal and Reddy, S. Hariprasad and Kedam, Naresh and Rathnayake, Upaka and Hoshino, Yukinobu},
  title = {Leveraging level data for accurate downstream flow prediction in the Narmada River Basin with advanced machine learning models},
  journal = {Journal of Hydroinformatics},
  year = {2025},
  volume = {27},
  number = {2},
  pages = {141--158},
  doi = {10.2166/hydro.2025.131}
}
Kostenlos starten →