基于BP人工神经网络和模糊理论的太湖蓝藻水华发生风险评价

INTEGRATION OF BP ARTIFICIAL NEURAL NETWORK AND FUZZY THEORY ON EVALUATING CYANOBACTERIA-DOMINANT BLOOM OCCURRENCE RISK OF LAKE TAIHU

目前,湖泊蓝藻水华是我国乃至世界的重大环境问题之一。蓝藻水华的暴发机制复杂,具有明显的不确定性。以太湖为例,根据近5 a水环境和水华发生的实测数据,结合BP(Back Propagation)人工神经网络和模糊理论,建立了蓝藻水华发生风险的模糊风险评价方法。对太湖9个水环境功能区的评价结果表明:西部沿岸区、梅梁湖蓝藻水华发生风险最大,为重度蓝藻水华风险区;竺山湖、五里湖次之,为中度蓝藻水华风险区;南部沿岸区、贡湖、湖心区为轻度蓝藻水华风险区,东太湖和东部沿岸区水华发生风险最小,为轻微蓝藻水华风险区。建立的评价方法和评价结果,可为蓝藻水华的预测、预警以及风险管理提供参考和依据。

Cyanobacteria-dominant bloom is one of the most serious environmental problems both in China and all the world.However,the algal bloom outbreak mechanisms are complex with obvious uncertainty.This study aimed to develop an approach to evaluate algal bloom occurrence risk in Lake Taihu.Firstly,BP artificial neural network was applied to reveal the relations of algal bloom and the impact factors in order to evaluate the status of algal bloom.Then,according to these results,the fuzzy theory was integrated to deeply evaluate algal bloom occurrence probability with different possibility,which would greatly quantify the uncertainty of algal bloom outbreak.Finally,the composite algal bloom occurrence risk of Lake Taihu was assessed based on the recent fine years＇ monitoring data with 65 sites.The results showed that the west coast area and MeiLiang Bay are of the highest risk,Lake Zhushan and Lake Wuli are of moderate risk,the southern coast area,Lake Gonghu and the central region of Lake Taihu are of low risk,Middle East area and East Lake are of little risk.The methodology developed and results can be utilized to support further study on algal bloom forecast and water resource management.