A novel implementation of a decomposition-enhanced hybrid GWO–KELM model with LUBE for constructing prediction intervals of groundwater drought

Saman Shahnazi, Kiyoumars Roushangar, Armin Farshbaf, Kiyoumars Roushangar and Hossein Hashemi, (2026). A novel implementation of a decomposition-enhanced hybrid GWO–KELM model with LUBE for constructing prediction intervals of groundwater drought. Springer Nature Link, 19(31),

Document type:
Article
Metadata
Links
Versions
Statistics
  • Sub-type Journal article
    Author Saman Shahnazi
    Kiyoumars Roushangar
    Armin Farshbaf
    Kiyoumars Roushangar
    Hossein Hashemi
    Title A novel implementation of a decomposition-enhanced hybrid GWO–KELM model with LUBE for constructing prediction intervals of groundwater drought
    Appearing in Springer Nature Link
    Volume 19
    Issue No. 31
    Publication Date 2026-03-02
    Place of Publication London, United Kingdom
    Publisher Springer Nature Link
    Language eng
    Abstract Groundwater drought has become a serious concern for water security in arid and semi-arid climates, largely due to the overuse of aquifers combined with climate variability and anthropogenic activities. The complex dynamics of groundwater systems, which are influenced by a wide range of climatic, geological, and anthropogenic factors, necessitate the development of accurate and reliable predictive models. This study proposes a comprehensive point and interval prediction scheme that integrates the Grey Wolf Optimizer (GWO) and Kernel Extreme Learning Machine (KELM), enhanced by multiple signal decomposition techniques, to perform multi-horizon forecasting of the Standardized Groundwater Index (SGI). To address hydrogeological heterogeneity, wells were clustered based on their drought-response behaviors using Self-Organizing Maps (SOM), and representative wells from each cluster were subsequently selected for forecasting analysis. Pre-processing with Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN), Variational Mode Decomposition (VMD), Empirical Wavelet Transform (EWT), and Empirical Fourier Decomposition (EFD) significantly improved the input data quality by effectively isolating and extracting non-stationary features from the original time series. The point forecasting results demonstrated the consistent superiority of the VMD–GWO–KELM model, particularly for short-term horizons. Furthermore, uncertainty analysis using the Lower–Upper Bound Estimation (LUBE) framework highlighted the superiority of VMD–GWO–KELM in constructing reliable and sharp prediction intervals, with coverage probabilities approaching nominal confidence levels.
    Keyword Groundwater drought
    Machine learning
    Hybrid modeling
    Time series forecasting
    Iran
    Copyright Holder Springer Nature Link
    Copyright Year 2026
    Copyright type All rights reserved
    DOI 10.1007/s12145-026-02093-y
  • This website has been archived and its online forms are no longer available. Please contact us by email at contact@unu.edu.

  • Citation counts
    Google Scholar Search Google Scholar
    Access Statistics: 0 Abstract Views  -  Detailed Statistics
    Created: Fri, 12 Jun 2026, 02:27:44 JST by Luying Wang on behalf of UNU INWEH