基于人工神经网络的夏季辽东湾北部河口区富营养化评价

发布时间：2018-11-18 06:41

【摘要】：为掌握辽东湾北部海域营养盐污染状况及富营养化水平,本文基于2014—2016年夏季现场调查资料,分析了该海域无机氮、活性磷酸盐、化学需氧量和叶绿素a的时空分布特征,并建立误差逆向传播算法(Back Propagation)的人工神经网络模型对该海域富营养化水平进行评价。结果表明:受2015年夏季辽宁省干旱和2016年夏季东北地区大规模降雨的影响,辽东湾北部海域营养盐、化学需氧量和叶绿素a浓度均呈现2014年较高,2015年降低,2016年又上升的趋势。调查海域营养盐结构在2014年和2015年主要表现为磷限制,2016年调查海域北部为磷限制,南部为氮限制。BP人工神经网络的评价结果表明,调查海域夏季富营养化水平呈现2014年较为严重,2015年降低,2016年又升高的"U"型特征。富营养化程度严重的区域主要出现在大辽河口、辽河口及其近岸海域。大辽河口富营养化程度在3年夏季均保持较高水平,其中2014年和2016年与辽河口富营养化水平相当,而2015年远高于辽东湾北部其他海域。BP人工神经网络在进行富营养化评价过程中,能够综合考虑各评价指标对海水富营养化的贡献率,避免对营养盐的过度依赖,同时减少评价过程中的主观误差,是一种更加客观合理的富营养化评价方法。
[Abstract]:In order to understand the status of nutrient pollution and eutrophication in the northern part of Liaodong Bay, the temporal and spatial distribution characteristics of inorganic nitrogen, active phosphate, chemical oxygen demand and chlorophyll a were analyzed based on the field investigation data of 2014-2016. An artificial neural network model of the error reverse propagation algorithm (Back Propagation) was established to evaluate the eutrophication level of the sea area. The results showed that under the influence of drought in Liaoning Province in 2015 and heavy rainfall in Northeast China in 2016, the concentrations of nutrients, chemical oxygen demand and chlorophyll a in the northern part of Liaodong Bay were higher in 2014 and decreased in 2015. The trend is rising again in 2016. In 2014 and 2015, the nutrient structure of the investigated area was mainly characterized by phosphorus limitation. In 2016, phosphorus limitation was observed in the northern part of the survey area and nitrogen limitation in the southern part of the survey area. The evaluation results of the BP artificial neural network show that, The level of eutrophication in summer was more serious in 2014, decreased in 2015, and increased in 2016. The areas with serious eutrophication mainly occur in the mouth of Daliao River, Liaohe Estuary and its inshore waters. The eutrophication level of Daliao Estuary maintained a high level in summer of three years, and the eutrophication level in 2014 and 2016 was similar to that in Liaohe Estuary. In the process of eutrophication evaluation, BP artificial neural network can comprehensively consider the contribution rate of each evaluation index to seawater eutrophication and avoid excessive dependence on nutrients. At the same time, it is a more objective and reasonable method to evaluate eutrophication by reducing the subjective error in the evaluation process.
【作者单位】：上海海洋大学海洋科学学院;国家海洋环境监测中心;国家海洋局第一海洋研究所;
【基金】：海洋公益性行业科研专项(201305043) 环保部环境基准与风险评估国家重点实验室基金“基于神经网络的河口近岸海域富营养化评价技术研究”资助
【分类号】：X824

【相似文献】