基于地理分区及神经网络的湖泊水库富营养化研究

An Integrated Eutrophication Assessment for Lakes and Reservoirs

构建了一个基于地理分区及神经网络的湖泊水库富营养化综合评价体系.以美国环境保护署Nutrient Criteria Database数据库为参照对象,有针对性地研究我国湖泊水库的情况,首次提出了基于地理分区的简易评价标准,对不同地理特征的水体富营养化临界值进行了定量研究.同时本研究还建立了基于神经网络的富营养化评价模型,可反映富营养化过程中的非线性特征,对以氮为控制因子的湖泊水库具有较高的评价准确性.二者有机结合而成的综合评价体系具有通用性强的优点;运用30组湖泊水库的实测数据进行评价验证,表明其不但准确可靠,且操作简便,可作为湖泊水库水环境管理的参考工具.

An integrated eutrophication assessment framework is developed for lakes and reservoirs based on an ecogeographical classification method and an artificial neural network model.Using the USEPA Nutrient Criteria Database as the basic reference and considering the ecogeographical characteristics of Chinese lakes and reservoirs,a simple eutrophication assessment criterion considering the ecogeographical characteristics is proposed for the first time.This criterion places the emphasis on the determination of critical values of key parameters for various regions.Moreover,an artificial neural network（ANN） assessment model is developed,considering the complexity and nonlinearity of eutrophication process.It is found that this ANN assessment model offers the advantage to assess with more accuracy the trophic status in nitrogen-limited water bodies.Integrating such two assessment methods can establish a simple but general eutrophication assessment framework;verification with 30 lakes and reservoirs shows that it can be served as a reliable and cost-effective tool for aquatic environmental management.