Forewarning model for water pollution risk based on Bayes theory

Jun Zhao, Juliang Jin, Qizhong Guo, Yaqian Chen, Mengxiong Lu, Luis Tinoco

Research output: Contribution to journalArticlepeer-review

10 Scopus citations


In order to reduce the losses by water pollution, forewarning model for water pollution risk based on Bayes theory was studied. This model is built upon risk indexes in complex systems, proceeding from the whole structure and its components. In this study, the principal components analysis is used to screen out index systems. Hydrological model is employed to simulate index value according to the prediction principle. Bayes theory is adopted to obtain posterior distribution by prior distribution with sample information which can make samples' features preferably reflect and represent the totals to some extent. Forewarning level is judged on the maximum probability rule, and then local conditions for proposing management strategies that will have the effect of transforming heavy warnings to a lesser degree. This study takes Taihu Basin as an example. After forewarning model application and vertification for water pollution risk from 2000 to 2009 between the actual and simulated data, forewarning level in 2010 is given as a severe warning, which is well coincide with logistic curve. It is shown that the model is rigorous in theory with flexible method, reasonable in result with simple structure, and it has strong logic superiority and regional adaptability, providing a new way for warning water pollution risk.

Original languageEnglish (US)
Pages (from-to)3073-3081
Number of pages9
JournalEnvironmental Science and Pollution Research
Issue number4
StatePublished - Feb 2014

All Science Journal Classification (ASJC) codes

  • Environmental Chemistry
  • Pollution
  • Health, Toxicology and Mutagenesis


  • Bayes theory
  • Forewarning
  • Hydrological model
  • Prediction
  • Principal components analysis
  • Water pollution risk


Dive into the research topics of 'Forewarning model for water pollution risk based on Bayes theory'. Together they form a unique fingerprint.

Cite this