The Research On An Integrated Simulation-diagnosis Water Quality Model Of The Second Songhua River Based On The System Dynamic And Bayesian Network

With the rapid development of society and technology, it is increasingly important for human to explore and utilize water resources reasonably. In order to solve the core issues of water resources management, it is urgent to introduce a water quality model with high efficiency and good adaptability. The purpose of this model is to assess water pollution based on the collected data. Further, it can also be used to predict the tendency of water quality variation in the future. On the other hand, in order to make full use of the monitoring data of water pollution, it calls for an integrated system which can assess and diagnose the water quality. The purpose of water quality diagnosis is to clarify the relationship between main pollution sources and relevant parameters, aiming at solving the uncertainty within water resources management.In this research, we explored the application of an integrated water quality simulation-diagnosis model based on the collected data of The Second Songhua River. The simulation model and diagnosis model were established based on the system dynamics and Bayesian network respectively. The purposes of this model are: to analyze water quality variation tendency; to confirm the key pollution sources and the sensitivity responses of changes in water quality to relevant parameters; to infer various possible relationships between the water quality of point pollution sources and Songlin section. The key pollution sources were decided based on the maximum possibility.In the part of water quality simulation, we adopted system dynamics to establish water quality model. In order to improve the accuracy, the model not only considered point pollution sources, but also involved in non-point pollution sources. As the simulation result shows, the average error between monitoring data and simulation data is lower than 10%, the comparing results indicate great consistency. Through sensitivity analysis, the upstream, Dacheng Corn, Hadashan Reservoir and Yinma River were selected as key pollution sources. Further, it could be concluded by Bayesian network diagnosis that upstream water quality and outlet of Hadashan Reservoir are the most important impact factors of Songlin Section. The water quality of Songlin Section and upstream presents obvious positive correlation. So it is the best way to reduce the risk of water pollution accident by controlling the water quality of upstream. The simulation result of the integrated model showed the likelihood of more reasonable and reliable way for the management of outlet to avoid further water environment deterioration. The research results will provide necessary technique support for the government and basin managers. The methods providing scientific basis for the solution to basin pollution control and trans-regional water environment issues.