EUVIMEDEUVIMED entdecken
Crossref

Interpretable AI-driven prediction of surface water quality using temporal fusion transformers and hydrochemical index integration

K. Vidhya, Sneha Gautam, B. Nagarajan, D. Rasi, D. Sharmila

Environmental Engineering Research · 15.09.2026 · 32 (2) · 260424-0

DOI: 10.4491/eer.2026.424 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Environmental Engineering Research, online erschienen am 15.09.2026. Verfasst von K. Vidhya und 4 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Seasonal variability and nonlinear hydrochemical relationships complicate water-quality forecasting in monsoon-influenced reservoirs. This study aimed to predict water quality one month ahead and interpret seasonal patterns in the Bhavani River–Bhavani Sagar system, Tamil Nadu, by integrating composite indices, a Temporal Fusion Transformer (TFT), principal component analysis (PCA), and SHapley Additive exPlanations (SHAP). Monthly CPCB monitoring records from January 2018 to December 2024 comprised 252 station-month observations across three locations and 14 physicochemical and microbiological parameters. Measurements at month t were used to forecast the weighted arithmetic water quality index (WA-WQI) at month t+1. Eight models were evaluated using five-fold chronological cross-validation, with validation periods following their corresponding training periods. TFT yielded the lowest prediction errors, with a mean absolute error of 1.94, root mean square error of 2.97, coefficient of determination of 0.98, and WQI class accuracy of 94.6%. SHAP ranked dissolved oxygen and biochemical oxygen demand as the strongest predictors. Summer showed greater oxygen-related stress, whereas monsoon conditions were associated with higher turbidity and coliform counts. These findings support basin-specific forecasting and seasonal monitoring priorities; application to other reservoirs requires independent validation.

Zitieren

APA

Vidhya, K., Gautam, S., Nagarajan, B., Rasi, D., & Sharmila, D. (2026). Interpretable AI-driven prediction of surface water quality using temporal fusion transformers and hydrochemical index integration. Environmental Engineering Research, 32(2), 260424-0. https://doi.org/10.4491/eer.2026.424

Vancouver

Vidhya K, Gautam S, Nagarajan B, Rasi D, Sharmila D. Interpretable AI-driven prediction of surface water quality using temporal fusion transformers and hydrochemical index integration. Environmental Engineering Research. 2026;32(2):260424-0. doi:10.4491/eer.2026.424

BibTeX
@article{Vidhya2026,
  author = {Vidhya, K. and Gautam, Sneha and Nagarajan, B. and Rasi, D. and Sharmila, D.},
  title = {Interpretable AI-driven prediction of surface water quality using temporal fusion transformers and hydrochemical index integration},
  journal = {Environmental Engineering Research},
  year = {2026},
  volume = {32},
  number = {2},
  pages = {260424--0},
  doi = {10.4491/eer.2026.424}
}
Kostenlos starten →