Supervised learning-based water quality prediction and ecological risk factor mining across China’s 12 major river basins

Jishu Guo

doi:10.1016/j.envsoft.2025.106840

Supervised learning-based water quality prediction and ecological risk factor mining across China’s 12 major river basins

2025-12-19

0

PRE

AI

J

Jishu Guo

Y

Yimin Huang

Y

YUN ZHANG *

DOI：10.1016/j.envsoft.2025.106840

Original

Original request for help

Save

Abstract

En 中文

• New method (spatio–temporal SANN). We propose a spatio-temporal aware neural network (SANN) for water quality prediction. • New perspective (interpretable drivers). We use a gradient-based attribution method to quantify the effects of indicators across river basins. • New application setting (12-basin evaluation). We construct a benchmark dataset covering China’s 12 major river basins and conduct water quality prediction tasks on this dataset. • Empirical gains (accuracy and robustness). Extensive water quality prediction experiments across 12 river basins show that SANN outperforms strong baselines.

Supervised learning-based water quality prediction and ecological risk factor mining across China’s 12 major river basins

Abstract

Journal

Organization

Cited Papers

Citing Papers