收藏切换
Prediction of DOM photochemical properties in surface water based on interpretable machine learning
收藏切换
Bing LIU1, *, Kai-ge SHI1, Yan-min CHEN1, Xin LU1, Shuai-jun HAN1, Li GU2
China Environmental Science | 2025, 45(7) : 3925 - 3934
Less
收藏切换
China Environmental Science | 2025, 45(7): 3925-3934
• Environmental Ecology •
Prediction of DOM photochemical properties in surface water based on interpretable machine learning
Full
Bing LIU1, *, Kai-ge SHI1, Yan-min CHEN1, Xin LU1, Shuai-jun HAN1, Li GU2
Affiliations
  • 1.School of Chemistry and Chemical Engineering, Zhengzhou Normal University, Zhengzhou 450044, China
  • 2.College of Environment and Ecology, Chongqing University, Chongqing 400044, China
Published: 2025-07-20
Outline
收藏切换

Dissolved organic matter (DOM) in surface water, which serves as a significant component within the carbon cycle, plays a vital role in the biogeochemical cycle of aquatic ecosystems. The triplet excited state of DOM (3DOM*), singlet oxygen (1O2), along with hydroxyl radicals (·OH), which are the photochemical products generated from DOM, were taken as the research targets. Three regression algorithms were employed to constructed the machine learning (ML) models predictive models for evaluating importance of features including environmental factors, ultraviolet-visible light parameters and fluorescence parameters. The ML model based on the categorical boosting algorithm exhibited better performance for the prediction targets in the test set, with R2 values of 0.9344, 0.8091, and 0.7775, respectively. Temperature, DOC, and pH were identified as the most important features for predicting 3DOM*, 1O2 and ·OH, respectively. When applying the chained ML model incorporating with the categorical boosting algorithm, the prediction performance for 1O2 and ·OH was further optimized. For 1O2 the R2 value in the test set increased from 0.8091 to 0.8233, and the RMSE decreased from 1.322 to 0.418. For ·OH the R2 value in the test set rose from 0.7775 to 0.8094, and the RMSE dropped from 2.982 to 0.312. Additionally, the ML models perform better than the partial least squares regression model in predicting 3DOM*, 1O2 and ·OH, indicating that the ML models possessed strong capability to handle non-linear relationships. The feature importance analysis based on ML successfully distinguished the key features in the DOM photochemical reaction. This finding can provide direct evidence for the intensive management of numerous influencing factors.

dissolved organic matter  /  explainable  /  machine learning  /  photochemistry  /  prediction
刘冰, 时凯歌, 陈燕敏, 卢鑫, 韩帅军, 古励. 基于可解释机器学习的地表水DOM光化学性质预测. 中国环境科学, 2025 , 45 (7) : 3925 -3934 .
Bing LIU, Kai-ge SHI, Yan-min CHEN, Xin LU, Shuai-jun HAN, Li GU. Prediction of DOM photochemical properties in surface water based on interpretable machine learning[J]. China Environmental Science, 2025 , 45 (7) : 3925 -3934 .
Year 2025 volume 45 Issue 7
54
17
Cite this Article
BibTeX
Article Info
  • Receive Date:2024-11-21
  • Online Date:2026-10-08
  • Published:2025-07-20
Article Data
Affiliations
History
  • Received:2024-11-21
Funding
Affiliations
    1.School of Chemistry and Chemical Engineering, Zhengzhou Normal University, Zhengzhou 450044, China
    2.College of Environment and Ecology, Chongqing University, Chongqing 400044, China
References
Share
https://castjournals.cast.org.cn/joweb/zghjkx/EN/
Share to
QR

Scan QR to access full text

Cite this article
BibTeX
Citations
表12种不同金属材料的力学参数
科
Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)
属
Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏
  • BibTeX
  • EndNote
  • RefWorks
  • TxT