Tianci Zhu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Tianci Zhu has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
Fang Liu
3 shared publicationsQiong Guo
2 shared publications- Rui Deng
School of Smart City, Chongqing Jiaotong University, Chongqing, China
2 shared publications Yibo Li
2 shared publicationsRui Deng
School of Smart City, Chongqing Jiaotong University, No. 66 Xuefu Road, Nan'an District, Chongqing, 400074, China; Technology Innovation Center for Spatio-temporal Information and Equipment of Intelligent City, Ministry of Natural Resources, No. 6 Qingzhu East Road, Yubei District, Chongqing, 400021, China
1 shared publicationWeilin Zhou
School of Smart City, Chongqing Jiaotong University, No. 66 Xuefu Road, Nan'an District, Chongqing, 400074, China
1 shared publication
Latest Publications
- Open Access
- 9 June 2026
- Published by Elsevier in SSRN Electronic Journal
Differential Hydrological Responses of Turbidity, Total Phosphorus and Total Nitrogen in a Complex River Confluence Revealed by Remote Sensing and Machine Learning
River confluences are hydrologically dynamic environments where contaminant transport and water quality exhibit pronounced spatial and temporal variability. Taking the Chongqing section of the Jialing River and its confluence with the Yangtze River as the study area, this study integrated Sentinel-2 imagery with in-situ measurements of turbidity (NTU), total phosphorus (TP), and total nitrogen (TN) to construct machine learning-based inversion models for water quality assessment. Five machine learning algorithms, including Random Forest (RF), Extreme Gradient Boosting (XGB), Categorical Boosting (CatBoost), Convolutional Neural Network (CNN), and Multilayer Perceptron (MLP), were evaluated to identify the optimal inversion approach. The results showed that XGB achieved the highest accuracy for NTU and TN inversion (R² = 0.915 and 0.718, respectively), whereas RF performed best for TP (R² = 0.855).Pixel-level inversion from 2021 to May 2025 revealed pronounced spatiotemporal heterogeneity in water quality across the river confluence system. Across hydrological stages from low-flow to high-flow conditions, NTU and TP increased by 938% and 196%, respectively, whereas TN increased by only 9%. These contrasting responses indicate markedly different hydrological sensitivities between particulate-associated constituents (NTU and TP) and dissolved constituents (TN), highlighting the important role of hydrological variability in shaping contaminant dynamics within the confluence zone. By integrating remote sensing observations with machine learning techniques, this study provides a robust framework for large-scale water quality assessment and improves the understanding of hydrologically driven contaminant dynamics in complex river confluence systems.
- 9 June 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessDifferential Hydrological Responses of Turbidity, Total Phosphorus and Total Nitrogen in a Complex River Confluence Revealed by Remote Sensing and Machine Learning
River confluences are hydrologically dynamic environments where contaminant transport and water quality exhibit pronounced spatial and temporal variability. Taking the Chongqing section of the Jialing River and its confluence with the Yangtze River as the study area, this study integrated Sentinel-2 imagery with in-situ measurements of turbidity (NTU), total phosphorus (TP), and total nitrogen (TN) to construct machine learning-based inversion models for water quality assessment. Five machine learning algorithms, including Random Forest (RF), Extreme Gradient Boosting (XGB), Categorical Boosting (CatBoost), Convolutional Neural Network (CNN), and Multilayer Perceptron (MLP), were evaluated to identify the optimal inversion approach. The results showed that XGB achieved the highest accuracy for NTU and TN inversion (R² = 0.915 and 0.718, respectively), whereas RF performed best for TP (R² = 0.855).Pixel-level inversion from 2021 to May 2025 revealed pronounced spatiotemporal heterogeneity in water quality across the river confluence system. Across hydrological stages from low-flow to high-flow conditions, NTU and TP increased by 938% and 196%, respectively, whereas TN increased by only 9%. These contrasting responses indicate markedly different hydrological sensitivities between particulate-associated constituents (NTU and TP) and dissolved constituents (TN), highlighting the important role of hydrological variability in shaping contaminant dynamics within the confluence zone. By integrating remote sensing observations with machine learning techniques, this study provides a robust framework for large-scale water quality assessment and improves the understanding of hydrologically driven contaminant dynamics in complex river confluence systems.
- Open Access
- 2 May 2026
- Published by Elsevier in SSRN Electronic Journal
A Study on Water Quality Inversion and Spatiotemporal Characteristics of Converging Rivers Using Machine Learning
Taking the Chongqing section of the Jialing River and its confluence with the Yangtze River as the study area, this study developed multi-band combinations from Sentinel-2 imagery and integrated them with in-situ measurements of turbidity (NTU), total phosphorus (TP), and total nitrogen (TN). Based on the selected inputs, five machine learning models, including Random Forest (RF), Extreme Gradient Boosting (XGB), Categorical Boosting (CatBoost), Convolutional Neural Network (CNN), and Multilayer Perceptron (MLP), were developed and compared to determine the optimal inversion approach. The results show that XGB achieved the best performance for NTU and TN inversion (R2 = 0.915 and 0.718), whereas RF performed best for TP (R2 = 0.855). Pixel-level inversion revealed pronounced spatiotemporal heterogeneity in water quality from 2021 to May 2025. During the wet season, NTU and TP exhibited strong sensitivity to hydrological disturbances, with average increases of 938% and 196%, respectively, while TN showed a relatively weak response (9%). This pattern suggests that NTU and TP are primarily controlled by flow dynamics and associated sediment and nutrient transport, whereas TN remains comparatively stable. Through a multi-model comparison and optimization framework, this study enables high-precision inversion of multiple water quality parameters in a complex river confluence system and provides a reliable methodological reference for water quality monitoring and river ecosystem management in hydrologically dynamic watersheds.
- 2 May 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessA Study on Water Quality Inversion and Spatiotemporal Characteristics of Converging Rivers Using Machine Learning
Taking the Chongqing section of the Jialing River and its confluence with the Yangtze River as the study area, this study developed multi-band combinations from Sentinel-2 imagery and integrated them with in-situ measurements of turbidity (NTU), total phosphorus (TP), and total nitrogen (TN). Based on the selected inputs, five machine learning models, including Random Forest (RF), Extreme Gradient Boosting (XGB), Categorical Boosting (CatBoost), Convolutional Neural Network (CNN), and Multilayer Perceptron (MLP), were developed and compared to determine the optimal inversion approach. The results show that XGB achieved the best performance for NTU and TN inversion (R2 = 0.915 and 0.718), whereas RF performed best for TP (R2 = 0.855). Pixel-level inversion revealed pronounced spatiotemporal heterogeneity in water quality from 2021 to May 2025. During the wet season, NTU and TP exhibited strong sensitivity to hydrological disturbances, with average increases of 938% and 196%, respectively, while TN showed a relatively weak response (9%). This pattern suggests that NTU and TP are primarily controlled by flow dynamics and associated sediment and nutrient transport, whereas TN remains comparatively stable. Through a multi-model comparison and optimization framework, this study enables high-precision inversion of multiple water quality parameters in a complex river confluence system and provides a reliable methodological reference for water quality monitoring and river ecosystem management in hydrologically dynamic watersheds.
- 12 June 2025
- Published by Elsevier in Environmental Pollution
- Vol. 382, 126668
Machine learning based water quality evolution and pollution identification in reservoir type rivers
- 12 June 2025
- Published by Elsevier in Environmental Pollution
- Vol. 382, 126668
Machine learning based water quality evolution and pollution identification in reservoir type rivers