TY - JOUR
T1 - Modeling periphyton biomass in a flow-reduced river based on a least squares support vector machines model
T2 - Implications for managing the risk of nuisance periphyton
AU - Huang, Wei
AU - Wu, Leixiang
AU - Wang, Zhuowei
AU - Yano, Shirichiro
AU - Li, Jiake
AU - Hao, Gairui
AU - Zhang, Jianmin
N1 - Funding Information:
The authors thank the anonymous reviewers for their valuable comments. This work was jointly supported by the National Key Project R & D of China (2016YFC0401709), National Natural Science Foundation of China (Grant No. 51879278, 51879279), IWHR Research & Development Support Program (grant numbers WE0145B782017, WE0145B342016), National Major Science and Technology Program for Water Pollution Control and Treatment of China (2018ZX07111003), the Open Research Fund of the State Key Laboratory of Hydraulics and Mountain River Engineering (Grant No. Skhl 1825), and the State Key Laboratory of Eco-hydraulics in Northwest Arid Region, Xi'an University of Technology (Grant No.2018KFKT-8).
Funding Information:
The authors thank the anonymous reviewers for their valuable comments. This work was jointly supported by the National Key Project R & D of China ( 2016YFC0401709 ), National Natural Science Foundation of China (Grant No. 51879278 , 51879279 ), IWHR Research & Development Support Program (grant numbers WE0145B782017 , WE0145B342016 ), National Major Science and Technology Program for Water Pollution Control and Treatment of China ( 2018ZX07111003 ), the Open Research Fund of the State Key Laboratory of Hydraulics and Mountain River Engineering (Grant No. Skhl 1825 ), and the State Key Laboratory of Eco-hydraulics in Northwest Arid Region, Xi’an University of Technology (Grant No.2018KFKT-8).
Publisher Copyright:
© 2020 Elsevier Ltd
PY - 2021/3/1
Y1 - 2021/3/1
N2 - Predicting periphyton biomass is essential for controlling algal biomass in regulated rivers; however, accurate modeling remains a challenge. Many traditional models have been developed to predict periphyton biomass under different conditions, but they are only useful under specific conditions and cannot accurately model the wide range of factors that co-control the periphyton biomass. This prompts us to look at alternate systems, including artificial intelligence methods that have advantages in dealing with non-linear relationships in a complex system. In this study, we calculated the periphyton biomass based on three well-established models, using field data from the Ohyama River in Japan. The field data indicates that the average chlorophyll a concentration was 21.04 μg/cm2, which is above the ‘nuisance’ threshold. Among the three models, the performance of the least squares-support vector machine (LS-SVM) model was found to be superior to those of the artificial neural network (ANN) and multiple linear regression (MLR) models. The relationships between water temperature, light intensity, TN, TP, discharge, and chlorophyll a were derived based on the LS-SVM model. Two scenarios for flushing flow were identified (i.e., for ‘degraded’ and ‘common’ conditions) and optimal flow rates of ∼10 m3/s and 9 m3/s, respectively, were recorded. The findings can improve river management, optimize dam operations, and reduce excess periphyton growth in flow-reduced downstream reaches of rivers.
AB - Predicting periphyton biomass is essential for controlling algal biomass in regulated rivers; however, accurate modeling remains a challenge. Many traditional models have been developed to predict periphyton biomass under different conditions, but they are only useful under specific conditions and cannot accurately model the wide range of factors that co-control the periphyton biomass. This prompts us to look at alternate systems, including artificial intelligence methods that have advantages in dealing with non-linear relationships in a complex system. In this study, we calculated the periphyton biomass based on three well-established models, using field data from the Ohyama River in Japan. The field data indicates that the average chlorophyll a concentration was 21.04 μg/cm2, which is above the ‘nuisance’ threshold. Among the three models, the performance of the least squares-support vector machine (LS-SVM) model was found to be superior to those of the artificial neural network (ANN) and multiple linear regression (MLR) models. The relationships between water temperature, light intensity, TN, TP, discharge, and chlorophyll a were derived based on the LS-SVM model. Two scenarios for flushing flow were identified (i.e., for ‘degraded’ and ‘common’ conditions) and optimal flow rates of ∼10 m3/s and 9 m3/s, respectively, were recorded. The findings can improve river management, optimize dam operations, and reduce excess periphyton growth in flow-reduced downstream reaches of rivers.
UR - http://www.scopus.com/inward/record.url?scp=85098462634&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85098462634&partnerID=8YFLogxK
U2 - 10.1016/j.jclepro.2020.124884
DO - 10.1016/j.jclepro.2020.124884
M3 - Article
AN - SCOPUS:85098462634
VL - 286
JO - Journal of Cleaner Production
JF - Journal of Cleaner Production
SN - 0959-6526
M1 - 124884
ER -