Sun Peiyan^1,2, Gao Zhenhui¹, Zhang Youchi¹, Wang Xiulin² Shiqiang¹ , Yang Jianqian¹, Xie li¹, and Ma Zhaodang¹, ¹The North China Sea Monitoring Cente, SOA, China; ²Ocean University of Qingdao

Abstract

In this paper, based on the monitoring data of HAB (harmful algal blooms) in Bohai Sea area near Bayuquan in June 30 ~ Sep. 16, 2001, a neural network model is built to recognize different samples in no HAB period, HAB prior period and HAB occurrence period. The input parameters are the total quantity and ratio of inorganic nitrogen and phosphate and the accuracy rate of recognition is about 82%. Then by designing several groups of data, the relation between the total quantity and ratio (N and P) and HAB occurrence is discussed. The results indicate: HAB occurrence needs a reasonable range of total quantity and ratio of inorganic nitrogen and phosphate. Also, their general scope which HAB may happen in this sea area is calculated. This is helpful for HAB predication and mechanism research.