Katipoğlu et al. (2026) Robust Monthly Lake Water Level Forecasting Using an Ensemble of Singular Spectrum Analysis Based Deep Learning and Categorical Boosting Models

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Water Resources Management
• Year: 2026
• Date: 2026-06-26
• Authors: Okan Mert Katipoğlu, Veysi Kartal, Türker Tuğrul, Veysel Süleyman Yavuz, Sema Arıman
• DOI: 10.1007/s11269-026-04806-w

Research Groups

• Department of Hydrology, Middle East Technical University (Ankara, Türkiye)
• Institute of Water Resources Management, Turkish Ministry of Agriculture and Forestry (Ankara, Türkiye)

Short Summary

This study presents a hybrid modelling approach for predicting one-month-ahead water levels in two major lakes in Türkiye, utilizing Singular Spectrum Analysis and machine learning algorithms to optimize sustainable water resources management. The developed methodology demonstrates superior performance over traditional models, particularly when incorporating CNN-based architectures.

Objective

• To develop an innovative hybrid modelling approach for predicting one-month-ahead water levels in complex hydrological systems

Study Configuration

• Spatial Scale: Two major lakes (Eber and Eğirdir) in Türkiye
• Temporal Scale: One-month-ahead predictions, with a focus on short-term forecasting

Methodology and Data

• Models used: Hybrid approach combining Singular Spectrum Analysis (SSA), Mutual Information (MI), Partial Autocorrelation Function (PACF), Deep Neural Networks (DNN), Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), and CatBoost algorithms
• Data sources: Hydrometeorological time series from in-situ measurements and satellite observations

Main Results

• SSA-based data preprocessing significantly improves the performance of prediction models
• CNN-based models demonstrate superior performance for Lake Eber, while DNN architecture excels for Lake Eğirdir
• The developed hybrid approach outperforms traditional models in terms of deterministic and probabilistic metrics

Contributions

• Original methodology integrating SSA, MI, PACF, and machine learning algorithms for predicting water levels in complex hydrological systems
• Contribution to the development of sustainable water management strategies and adaptability to similar hydrological systems

Funding

• This research was funded by the Turkish Ministry of Agriculture and Forestry (Project Code: 2020/1) and the Middle East Technical University Research Fund (Grant No. METU-2022-01).

Citation

@article{Katipoğlu2026Robust,
  author = {Katipoğlu, Okan Mert and Kartal, Veysi and Tuğrul, Türker and Yavuz, Veysel Süleyman and Arıman, Sema},
  title = {Robust Monthly Lake Water Level Forecasting Using an Ensemble of Singular Spectrum Analysis Based Deep Learning and Categorical Boosting Models},
  journal = {Water Resources Management},
  year = {2026},
  doi = {10.1007/s11269-026-04806-w},
  url = {https://doi.org/10.1007/s11269-026-04806-w}
}