EUVIMEDEUVIMED entdecken
Crossref

Machine-learning-enhanced geostatistical assessment of drinking- and irrigation-water quality in the Oued Souf aquifer, El Oued, Algeria

Issam Zaiz, Lakhdar Sek, Mohammed Seghir Ammari, Nadjet Zair, Abdelmonem Miloudi, Nabil Mega, Hamad Khelaifa

Water Supply · 01.10.2026 · 26 (10) · 1083-1111

DOI: 10.2166/ws.2026.189 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Water Supply, online erschienen am 18.08.2026. Verfasst von Issam Zaiz und 6 weiteren Autor:innen. Die Arbeit nennt 53 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Groundwater degradation in arid regions threatens drinking-water supply and agricultural development. This study proposes an integrated framework combining hydrochemical indices, geostatistics, and machine learning to evaluate groundwater suitability in the phreatic aquifer of the Oued Souf (El Oued, southeastern Algeria). Forty-one samples from a single field campaign (August 2022)were analysed for major physicochemical parameters. The Drinking Water Quality Index (DWQI) ranged from 131 to 444 (mean 254), with 90% classified as Doubtful-to-Unsuitable, driven by evaporite dissolution and agricultural contamination; the Irrigation Water Quality Index (IWQI) ranged from 43 to 127 (mean 80), with 83% as Excellent-to-Good.Ordinary Kriging (OK) outperformed Inverse Distance Weighting (IDW) reducing RMSE by 5.3% for DWQI (53.60 vs. 56.61) and 2.8% for IWQI (17.13 vs. 17.62). Random Forest (RF) and Artificial Neural Network (ANN) models were trained on an 80/20 split with 5-fold cross-validation; RF achieved high predictive accuracy (R2 ≈ 0.96) while ANN exhibited severe overfitting, confirming RF for data-scarce contexts. Electrical conductivity emerged as the dominant quality driver (74% DWQI feature importance), highlighting its value as a cost-effective monitoring proxy. The ML-geostatistical workflow – is transferable to analogous arid aquifer systems, while site-specific components require recalibration. Single-season sampling design is acknowledged as a limitation; multi-season validation is recommended.

Zitieren

APA

Zaiz, I., Sek, L., Ammari, M. S., Zair, N., Miloudi, A., Mega, N., & Khelaifa, H. (2026). Machine-learning-enhanced geostatistical assessment of drinking- and irrigation-water quality in the Oued Souf aquifer, El Oued, Algeria. Water Supply, 26(10), 1083-1111. https://doi.org/10.2166/ws.2026.189

Vancouver

Zaiz I, Sek L, Ammari MS, Zair N, Miloudi A, Mega N, et al. Machine-learning-enhanced geostatistical assessment of drinking- and irrigation-water quality in the Oued Souf aquifer, El Oued, Algeria. Water Supply. 2026;26(10):1083-1111. doi:10.2166/ws.2026.189

BibTeX
@article{Zaiz2026,
  author = {Zaiz, Issam and Sek, Lakhdar and Ammari, Mohammed Seghir and Zair, Nadjet and Miloudi, Abdelmonem and Mega, Nabil and Khelaifa, Hamad},
  title = {Machine-learning-enhanced geostatistical assessment of drinking- and irrigation-water quality in the Oued Souf aquifer, El Oued, Algeria},
  journal = {Water Supply},
  year = {2026},
  volume = {26},
  number = {10},
  pages = {1083--1111},
  doi = {10.2166/ws.2026.189}
}
Kostenlos starten →