MODELING CHLOROPHYLL A OF BOHAI BAY BASED ON STRUCTURAL EQUATION MODEL AND BAYESIAN NETWORKS

In this paper, a SEM-BN model is established to model chlorophyll a in Bohai Bay. The structural equation model (SEM) method has the advantage to investigate complex networks of relationships and has the disadvantage to predict with new data. On the contrary, the Bayesian networks (BN) method can produce good prediction accuracy with incomplete data sets, but it is limited in identifying the true and false relationships among factors. In our study, the two emerging tools are combined to model chlorophyll a in Bohai Bay. By using SEM, the relationship between the phytoplankton dynamics and coastal environment variables during the summer period in Bohai Bay is hypothesized and tested. And by using the BN, which is built based on the structure of SEM, the concentration of chlorophyll a is simulated. The result of simulation shows that the SEM-BN model has great performance, and shows good promise for future environmental modeling efforts in other regions.