.png){kind=icon}
.png){kind=icon}
Return
Seasonal effects of rainfall on characteristics and environment response of dissolved organic matter (DOM) in urban rivers: insights from spectroscopy and machine learning
.png){kind=icon}
OA
AI
K
J
Y
S
DOI：10.1016/j.watcyc.2025.06.005.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Urban river dissolved organic matter (DOM) is particularly vulnerable to monsoon climate rainfall impacts, but its mechanism remains unclear. The characteristics and environmental responses of DOM were analyzed by means of excitation-emission matrix combined with parallel factor analysis (EEM-PARAFAC) and machine learning. The results showed that the total nitrogen (TN) and chemical oxygen demand (COD) of downstream urban water (DUW) were significantly higher than those of upstream urban water (UUW) (P < 0.01), and the rivers were generally in a mesotrophic state (TSIM > 30). Parallel factor analysis PARAFAC identified two humic-like components (C1 and C3) and one protein-like component (C2), with the variation order being C3 -> C2 -> C1. In different seasons, the abundance of protein-like components in UUW (29.78%-86.90%) was significantly higher than that in DUW (9.06%-66.86%), while in DUW, humic-like components dominated (33.12%-90.94%). DOM overall exhibited low humification and strong autochthonous characteristics (HIX <4 and BIX >1). Machine learning indicated that nitrogen concentration, pH, and temperature (T) were the key environmental factors affecting DOM components. The DOM network of UUW exhibited more complex topological properties. Path analysis showed that the properties of DOM in UUW in summer were most affected by its source, while in other spatiotemporal distribution states, Fe had the greatest impact on DOM properties. Our study highlights the complexity of DOM dynamic changes and their environmental significance under the background of rainfall, urbanization, and climate change, providing a scientific basis for urban river water quality management and protection strategies.
Keywords:
Parallel factor analysis
Two-dimensional correlation spectroscopy
Self-organizing maps
Co-occurrence network
Partial least squares path model
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
W
IF:
8.7
Papers:
33
Citations:
630
Organization
Cited Papers