富营养化数值模拟预测模型比较分析  

Comparison Analysis for Numerical Simulation Prediction Models of Eutrophication

作　　者：郑月蓉[1] 姜世中[1] 任平[1] 宁翔[1] 

机构地区：[1]四川师范大学地理与资源科学学院,四川成都610066

出　　处：《四川师范大学学报（自然科学版）》2010年第5期684-687,共4页Journal of Sichuan Normal University（Natural Science）

基　　金：国家重点基础研究发展规划(2003CB415202)资助项目

摘　　要：分析三岔湖水库水文特征,发现水库冬季温度几乎没有分层现象,库表水温与库底水温相差不到0.1℃;夏季温度分层明显,最大温差达12.7℃,春秋季也有分层现象,一般温差为1.4～9.0℃.选用磷和溶解氧作为三岔湖水库水质的研究对象,运用人工神经网络模型对水质进行模拟分析,对比箱式模型结果,箱式模型作为一种传统的、成熟的研究手段,具有直观、明了等特点,但箱式模型的误差较大,相对误差为-20%～55%;而人工神经网络作为一种较新的研究方法,误差相对较小,精度较高,相对误差为-1.15%～0.838%.By means of the phosphorus and dissolved oxygen in Sancha Lake reservoir,this paper analyzes the hydrological features,in winter,there is no temperature stratification phenomenon in Sancha Lake Reservoir,the temperature difference in water between the outsurface and underside is less than 0.1 ℃;in summer,thermal stratification is obvious,the maximum temperature difference reaches 12.7 ℃,there are also stratification phenomena in spring and autumn,generally the temperature difference is between 1.4 ～ 9.0 ℃.On this basis,the authors use the neural network to analyze water quality,compare the results with those of box model.The box model is a traditional,sophisticated research tool,simple and clear.But its relative error is larger,between the range of-20% ～ 55%.As a relatively new research method,the error of neural network is relatively small,the relative error is between-1.15% ～ 0.838%.The artifical neurat net work has a higher accuracy.

关 键 词：三岔湖水库 富营养化 箱式模型 人工神经网络模型 

分 类 号：X820.2[环境科学与工程—环境工程]