龙景湖叶绿素a浓度预测模型敏感性分析

发布时间：2018-01-06 22:32

本文关键词：龙景湖叶绿素a浓度预测模型敏感性分析　出处：《重庆大学》2015年硕士论文　论文类型：学位论文

【摘要】：龙景湖为封闭深水型湖泊,水动力条件差,且湖内营养物质丰富,因此湖体极容易在条件适宜的情况下导致藻类大量生长,影响湖体的景观作用。论文致力于环境因子对藻类生长影响机制的研究,探索对藻类生长影响较大的环境因子,为防治龙景湖藻华爆发提供一定指导。在对龙景湖水质特征进行监测的基础之上,针对龙景湖的不同区域,建立了叶绿素a浓度预测模型,并利用此模型,对各个模型的输入参数进行了局部敏感性分析和全局敏感性分析,主要研究内容和结论如下:①利用相关性分析,对各环境因子与叶绿素a浓度的相关性和各环境因子间的相关性进行了考察,在建立叶绿素a浓度预测模型时,利用此分析结果筛选模型输入变量。剔除了与叶绿素a浓度无关(R0.3)的环境因子,并对高度相关(R0.8)的环境因子进行了合并。②通过对模型预测精度及决策系数的比较,确定了预测模型隐含层转移函数为tansig,输出层转移函数为purelin。露天剧场湖湾、湖心和秋亭桥湖湾三个监测点的隐含层神经元数量分别为8、13、13。模型的平均预测误差分别为2.23%、2.01%和2.37%,达到了良好的预测水平。③利用基于偏导的局部敏感性分析,分析得出三个监测点对各环境因子的敏感性,发现叶绿素a浓度对水温、ORP、TP和CODMn表现出了较高的敏感性。叶绿素a对各个因子的敏感性都是随时间的变化而变化的,在十一月中旬之前,各因子的敏感性系数处于较高水平,且各因子的敏感性系数差异较大;十一月中旬之后,各因子的敏感性系数降低,且相互间差异较小,但水温在都表现出了最高的敏感性系数。④利用基于二阶偏导的全局敏感性分析,找出了对叶绿素a浓度有较高影响的双因子组。其中,以水温和TP、水温和CODMn以及真光层深度和CODMn为敏感性系数最高的双因子组。另外,通过与局部敏感性分析结果进行比较,发现某些局部敏感性系数不高的因子在与其他因子结合后其敏感性系数得到了提高,二者表现出协同作用。⑤通过比较TN和TP在局部敏感性分析和全局敏感性分析中的表现,发现叶绿素a浓度对TN的敏感性要明显低于对TP的敏感性,说明就营养盐限制而言,龙景湖表现为磷限制型的特征。
[Abstract]:Longjing Lake is a closed deep-water lake with poor hydrodynamic conditions and abundant nutrients in the lake. Therefore, the lake body is easy to cause algal growth under suitable conditions. The effect of environmental factors on algae growth is studied in this paper, and the environmental factors that have a great impact on algae growth are explored. On the basis of monitoring the water quality characteristics of Longjing Lake, a prediction model of chlorophyll a concentration was established for different areas of Longjing Lake. The input parameters of each model are analyzed by local sensitivity analysis and global sensitivity analysis. The main contents and conclusions are as follows: 1 using correlation analysis. The correlation between the concentration of chlorophyll a and the concentration of chlorophyll a and the correlation between the concentration of chlorophyll a and the concentration of chlorophyll a were investigated, and the prediction model of the concentration of chlorophyll a was established. The input variables of the model were screened and the environmental factors independent of chlorophyll a concentration (R0.3) were eliminated. By comparing the prediction accuracy and the decision coefficient of the model, the implicit layer transfer function of the prediction model is determined to be tansig. The transfer function of the output layer is purelin.The number of hidden layer neurons in the lake bay, the center of the lake and Qiuting Bridge Lake Bay is 813, respectively. 13. The average prediction error of the model is 2.23% and 2.37%, respectively, which achieves a good prediction level. 3. The local sensitivity analysis based on partial derivation is used. The sensitivity of three monitoring sites to various environmental factors was obtained, and the effect of chlorophyll a concentration on water temperature and ORP was found. TP and CODMn showed high sensitivity. The sensitivity of chlorophyll a to each factor varied with time, and before middle of November, the sensitivity coefficient of each factor was at a higher level. The sensitivity coefficient of each factor is different. After middle of November, the sensitivity coefficient of each factor decreased, and the difference between each factor was small, but the water temperature showed the highest sensitivity coefficient 4. 4 using the global sensitivity analysis based on second order partial derivative. The double factor groups with high effect on chlorophyll a concentration were found. Among them, water temperature and TP, water temperature and CODMn, and the depth of the true light layer and CODMn were the most sensitive groups. By comparing with the results of local sensitivity analysis, it is found that some factors with low local sensitivity coefficient are enhanced by combining with other factors. By comparing TN and TP in local sensitivity analysis and global sensitivity analysis, it was found that the sensitivity of chlorophyll a concentration to TN was significantly lower than that to TP. The results showed that Longjing Lake was characterized by phosphorus limitation as far as nutrient limitation was concerned.
【学位授予单位】：重庆大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：X832;X524

【参考文献】