Pre-selection of monitoring stations for marine water quality using affinity propagation: A case study of Xincun Lagoon, hainan, China. (1st January 2023)

• Record Type: Journal Article
• Authors: Fang, Xin; Luo, Chengshu; Zhang, Dongrong; Zhang, Haifeng; Qian, Jian; Zhao, Canghai; Hou, Zonghao; Zhang, Yifei
• Is Part Of: Journal of environmental management. Volume 325:Part B(2023)
• Publication Date: 2023-01-01
• Subjects: Environmental investigation station -- Clustering algorithm -- Layout optimization -- Bias parameter selection -- Coastal waters; Environmental policy -- Periodicals; Environmental management -- Periodicals; Environment -- Periodicals; Ecology -- Periodicals; 363.705
• DOI: 10.1016/j.jenvman.2022.116666
• Languages: English
• ISSNs: 0301-4797

Abstract

Abstract: The development, protection, and restoration of bays require works in scientific research and applications, and the success of which depends on a well deployment of monitoring stations for marine water quality. However, for bays without historical data, it is difficult to carry out related research on deployment of the monitoring stations, resulting in very few research works. This paper has introduced the affinity propagation (AP) clustering algorithm and achieved good results by correcting the preferences. The results show that under the given preference, that is, when the value of M is −6800, the number of monitoring stations in the Xincun lagoon area is 24. Simultaneous the sensitivity analysis of preferences shows that the number of exemplars decreases with lower preferences, that is, when M decreased from −4000 to −12000, the number also decreased from 70 to 36. However, some exemplars remain unchanged or being changed to adjacent positioning. This shows the stability of computation results and the rationality of AP. The research results can be well applied to other bays, even open waters. Highlights: AP clustering can be used to layout seawater monitoring stations. Multi-dimensional feature parameters is used to modify the bias parameters of the algorithm. Sensitivity analysis shows that AP clustering can be applied in bays lacking historical data.