| 筑坝河流浮游植物生态地球化学的BP人工神经网络模拟 | |
| 赵颜创 | |
| Thesis Advisor | 王宝利 |
| 2013 | |
| Degree Grantor | 中国科学院研究生院 |
| Place of Conferral | 北京 |
| Degree Name | 硕士 |
| Degree Discipline | 环境工程 |
| Keyword | Bp人工神经网络 河流-水库体系 敏感性分析 浮游植物动态变化 乌江 |
| Abstract | 筑坝是目前最为重要的影响河流水环境的人为活动之一。河流经大坝拦截后，水流减缓，营养负荷增加，易于浮游植物生长。经过梯级水电开发，天然河流演化为蓄水河流，同时浮游植物的生态环境也发生了相应的变化。因此，可以利用浮游植物对环境变化敏感性，来评估河流梯级水电开发的生态环境效应。自上世纪90年代始，一些学者利用人工神经网络建立模型来模拟河流或者湖泊中浮游植物的动态变化并取得成功；但这些研究多应用于单一的河流或湖泊研究，且主要用来模拟某一具体藻类的动态变化。因此，本论文尝试利用BP人工神经网络模型来模拟某一浮游植物类群在整个河流－水库系统中的生态动态变化。 乌江是我国西南典型梯级开发蓄水河流，且有一定研究成果积累。为此，本研究选取了以乌江干流的六冲河（未受筑坝影响的河流点代表）、洪家渡水库（库龄10年的年轻水库）、东风水库（库龄22年的水库）、六广（受筑坝影响的河流点代表）、乌江渡水库（库龄40年的老水库），及支流猫跳河上的红枫湖水库（库龄52年的老水库）为研究对象，自2011年5月至2012年5月期间对所有样点进行半月一次的密集采样，检测浮游植物的群落组成和细胞密度，及相关的物理和化学参数，并以此为基础，进行BP人工神经网络的模拟并验证。最后，利用模型对各个环境因子进行敏感性分析来识别蓄水河流中浮游植物动态变化的主要控制因子，为梯级水电开发对河流浮游植物变化的影响提供重要的科学依据。通过以上研究，获得如下主要结论： 1、研究区域浮游植物主要类群为蓝藻、绿藻和硅藻；浮游植物生物量和群落结构具有明显时空变化。在原始河流六冲河中，硅藻为优势类群，细胞密度占浮游植物总细胞密度（1.12×106cells/L）的67.2%，在八月份生物量最大。受筑坝影响的六广采样点，优势类群仍为硅藻，但细胞密度占浮游植物总细胞密度（0.99×106cells/L）比例下降为56.4%，在八月生物量最大；同六冲河相比，浮游植物群落结构已发生变化。蓝藻是水库浮游植物的主要优势类群，且因水库营养水平不同各有差异。浮游植物总细胞密度随着水库库龄的增加而越来越大。其中洪家渡水库浮游植物细胞密度为2.13×106cells/L，优势类群为硅藻（36%）和蓝藻（36%）；红枫湖水库的为34.88×106cells/L，优势类群为蓝藻（94%）。 2、相关性分析显示，原始河流点溶解硅和浮游植物生物量显著正相关（R=0.639；P<0.01），表明硅是浮游植物生长的限制性因子。在红枫湖水库浮游植物生物量和pH值显著相关（R=0.425；P<0.05）；在理想的碳酸盐岩溶解体系，而在其他各水库采样点中，温度与和浮游植物生物量均显著正相关，表明温度是浮游植物生长的控制因子。 3、本研究将水温、pH、溶解氧、溶解CO2、总氮、总磷、溶解硅作为输入变量，而浮游植物各类群的细胞密度作为输出变量，应用BP人工神经网络模型对浮游植物动态变化进行模拟。结果显示预测值和实测值在时间变化上显著相关，表明该模型可以成功地模拟各类群的时序（time-series）变化；此外该模型还成功的预测了优势浮游植物类群生物量峰值出现的时间。可见，尽管浮游植物存在不同种群间的竞争及河流和水库不同的生态环境，BP人工神经网络模型够模拟蓄水河流浮游植物生态变化这种复杂、非线性生态系统。 4、BP人工神经网络模型的敏感性分析，可以识别出不同系统中影响浮游植物动态变化的主要因子。敏感性分析结果显示：总的来说pH和溶解氧是蓄水河流浮游植物各类群的主要驱动因素；在河流点六冲河中还发现CO2、溶解硅是硅藻生态变化的主要驱动因素，而在水库中温度也是优势类群的主要驱动因素，六广处样点中浮游植物的主要控制因子与水库相似。敏感性分析得出的主要驱动因素和实测值相关性分析结果基本一致。模型显示在天然河流向蓄水河流转变过程中，水温对浮游植物动态变化的控制作用得到了加强。 |
| Other Abstract | Damming is one of the most important human activities affecting the river environment. The flow velocity of the river will slow down and nutrient loading will be greater due to the construction of the dam, which is suitable for the growth of phytoplankton. The original river will be an impounded river due to the construction of cascade hydropower and the living conditions of phytoplankton is changed. So we can evaluate the ecological effects of the cascade hydropower development by phytoplankton’s sensitivity to the change of conditions. Since the beginning of 1990s, some researchers use artificial neural networks (ANNs) to simulate the phytoplankton dynamics successfully in the rivers or lakes. But these researches always apply to simulate the dynamics of one or more particular species in single rivers or lakes. Therefore, this paper tries to simulate the dynamics of dominant phytoplankton groups in a whole river-reservoir system. The Wujiang River is a typical dammed river in south-western China and we have done works for years. Six sampling sites were selected in this study, they are the Liuchong River (respecting the original eiver ), the Hongjiadu Reservoir (10 years old), the Dongfeng Reservoir (22 years old), the Liuguang River ( the river affected by damming), the Wujiangdu Reservoir (40 years old) and the Hongfenghu Reservoir (52 years old). Water samples were semimonthly collected from May 2011 to May 2012. Phytoplankton community composition, cell density and relevant physical, chemical parameters were analyzed. Then BPANNs models were constructed and validated base on the data. Finally, the sensitivity analyses were done to discern factors influencing phytoplankton dynamics. Several important conclusions, which can provide a scientific basis for effects on phytoplankton dynamics of the construction of cascade hydropower, have been drawn as follows: 1. In the study areas there were three main algal groups: Chlorophyta, Bacillariophyta, and Cyanophyta. In the Liuchong River, the dominant group was Bacillariophyta, which accounted for 67.2% of phytoplankton total cell density (1.12×106cells/L), and the cell density reached maximum value in August. In the Liuguang River, Bacillariophyta was still the dominant group, but proportion fell to 56.4% of phytoplankton total cell density (0.99×106cells/L). Phytoplankton community composition in the Liuguang River has been changed compared to the original river. Chlorophyta was the dominant group in the reservoirs and that in each reservoir was different due to the ages of the reservoirs. The phytoplankton total cell density was larger with the reservoirs got elder. The phytoplankton total cell density in Hongjiadu Reservoir was 2.13×106cells/L and the dominant group were Chlorophyta (36%) and Bacillariophyta (36%), while that in Hongfenghu Reservoir was 34.88×106cells/L and the dominant group was Chlorophyta (94%) 2. Correlation analysis showed the phytoplankton cell density was significantly and positively correlated with the dissolved silicon (R = 0.639; P < 0.001) in the original Liuchong River. It indicated that dissolved silicon was the restrictive factor for the growth of phytoplankton. In the Hongfenghu Reservoir, phytoplankton cell density was significantly and positively correlated with pH (R = 0.639; P < 0.001). In the other reservoirs, phytoplankton cell density was significantly and positively correlated with the water temperature, which showed that water temperature was the restrictive factor for the growth of phytoplankton. 3. In this study, water temperature, dissolved oxygen, dissolved CO2, total nitrogen, total phosphorus and dissolved silicon were selected as the input parameters, and the phytoplankton cell density was selected as the output parameter. Then the BPANNs models were constructed to simulate the phytoplankton dynamics. The results that the predicted values has well correlation with the measured values showed the models can predict time-series variation of phytoplankton abundance in such complex and nonlinear phenomena in the river-rese |
| Subject Area | 环境地球化学 |
| Language | 中文 |
| Document Type | 学位论文 |
| Identifier | http://ir.gyig.ac.cn/handle/352002/5869 |
| Collection | 研究生_研究生_学位论文 |
| Recommended Citation GB/T 7714 | 赵颜创. 筑坝河流浮游植物生态地球化学的BP人工神经网络模拟[D]. 北京. 中国科学院研究生院,2013. |
| Files in This Item: | ||||||
| File Name/Size | DocType | Version | Access | License | ||
| 筑坝河流浮游植物生态地球化学的BP人工神（10595KB） | 暂不开放 | License | Application Full Text | |||
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.
Edit Comment