【作者】 王策；

【导师】 冯玉杰；

【作者基本信息】 哈尔滨工业大学 ， 环境科学与工程， 2013， 博士

【摘要】 地表水承载着各种有机污染物，尤其是我国的天然河流——它无时无刻不在见证着当地的工业发展速度和由此带来的工业废水、非点源污染、突发事件所排放的污水。这些有毒有机化合物进入河流后将对水生态系统产生直接或间接的影响，危害着人类健康安全，因此明晰有机污染物在水体中的归趋，对于有效改善水质状况及风险评估将具有重要意义。基于实验室分析有机污染物在环境中迁移转化规律常受到种种局限，譬如时间成本、人力成本、设备条件等等，而数学模型在这方面则扮演了重要角色。本文构建了两种数学模型对松花江中典型有机污染物的迁移转化规律进行了模拟和预测。本文研发了动态多介质模型，它耦合了水力学子模型和逸度Ⅳ子模型，用于模拟有机污染物的长期环境行为。首先采用一维运动波方程计算天然河流动态流量、水深和流速；其次逸度Ⅳ方程模拟污染物在多介质环境中的分布，其内部的所有参数具有时空变化性。动态多介质模型模拟了2007年松花江8种多环芳烃在多介质环境中的归趋变化，采用柯西不等式系数法(TIC)对松花江中8种多环芳烃的水相模拟浓度进行了验证，结果显示所有模拟结果与相应监测浓度匹配较好，TIC值均小于0.5。以污染物萘为代表，计算了它在每个环境介质的质量通量，结果显示环境系统的总输入和总输出分别为140.61×103Kg和140.26×103Kg，相对误差为0.25%，证明了模型质量守恒。从模型参数的灵敏度分析结果得知，点源源强和河水流量对污染物水中浓度影响最大，灵敏度系数分别为0.97和0.90。动态多介质模型模拟了2007年松花江中8种酚类污染物在多介质环境的行为变化。基于2007年不同水文季节的监测浓度，通过纳什效率系数(NSE)、百分比偏差系数(PBIAS)和均方根-测量标准偏差比(RSR)三种统计学测试方法评估模型的预测能力。整体而言，模拟浓度与监测浓度吻合度较好，大多数数据点同时满足NSR<0.5，RSR<=0.7和PBIAS∈[-70%，+70%]条件。虽然存在若干奇异点，但配对T检验结果说明当置信水平为0.05时，P>0.05接受零假设即模型所有模拟值与监测值无显著差异。以2，4-二氯苯酚为代表对污染物从河流系统到其它环境介质的分配路径和质量通量变化进行了评估，模型模拟期间水体平流的质量通量最大，净河流输出通量约为4500.1kg。采用蒙特卡洛方法考查模型输入数据对输出结果的影响程度。模型预测了2012年枯水期河流中酚类污染水平，与2007年同期进行比较，发现松花江的酚类污染得到了显著削弱。研究结果显示，动态多介质模型完全可以应用于松花江点源污染，为当地决策者制定水污染控制方案提供可靠的数据支持。针对突发水污染事件，本文研发了一维污染物迁移转化模型，它耦合了一维运动波方程和平流-扩散-反应方程。模型内部包括挥发、光解和生物氧化的动力学过程以及在悬浮颗粒物沉降和再悬浮作用下的水层和沉积层之间的扩散质量交换过程。模型应用在松花江硝基苯突发水污染事件中，模拟了污染物在水层和沉积层中的浓度变化。针对水层浓度采用NSE、PBIAS、RSR和TIC四种统计学方法评价模型表现力，结果显示模型的整体模拟效果令人满意。通过灵敏度分析得知，河流流速是影响水层硝基苯浓度最灵敏的输入参数，这意味着它对污染物的水中浓度分布和变化起到至关重要的作用。研究结果显示，污染物迁移转化模型完全有能力为松花江及与其水文特征相似的其它天然河流的水污染控制和预警提供可靠信息。通过建立数学模型可为水环境生态保护提供科学指导，从而预防、减少和修复水污染所造成的生态破坏，避免发生巨大的经济损失，切实维护人类健康和生态系统平衡，使百姓生活与国民经济的发展更加和谐、永续。更多还原

【Abstract】 Surface waters, especially natural rivers always witness the rapid developmentand advancement of local industries and act as receiving waters for various kinds oforganic contaminants from municipal and industrial wastewaters, organic chemicalsin use, non-point source pollutions. Organic compounds with toxicity dischargedinto river probably bring negative impact on aquatic ecosystem by direct andindirect effects on organisms, and put human’s life to much inconvenience. It is,therefore, of significance for us to clearly understand the transport and fate oforganic pollutants in aquatic environment in order to implement effectively waterquality management and risk assessment.The demand for numerical models to simulate and predict the transport and fateof organic pollutants in the environment is a reflection of certain restrictions ofexperimental research, in some cases by time limitations, laborious fieldobservations, and undeveloped laboratory determination equipment. Numericalmodeling plays an important role in predicting the behavior of pollutants in theenvironment. For this, two kinds of models are developed in this paper.Dynamic multimedia model has been developed in this study that coupleshydrodynamics submodel and fugacity Ⅳ submodel, to simulate the long-termenvironmental fate of organic pollutants. In the first, the one-dimensional networkkinematic wave equation is used to calculate varying water flow, depth and velocity.In the second, Fugacity Ⅳ equations in which all parameters are shown inspatio-temporal variability are implemented to predict contaminant distributions inmultimedia environments.The dynamic multimedia model has been applied to simulate multimediaenvironmental fate of eight PAHs in the Songhua River in2007. Theil’s inequalitycoefficient test (TIC) is used to implement model validation for modeledconcentrations of eight PAHs in water compartment. From results, all TIC values areless than0.5, indicating simulated concentrations are in acceptable agreement withmonitoring data. Taking Nap as representative, calculation of mass balance of Naphas effectively demonstrates that the model program is running properly, of whichthe result shows total input and output mass fluxes in the whole environment systemare140.61×103Kg and40.26×103Kg respectively with the0.25%of relative error.From sensitivity analysis, contaminant concentration in water column is highly sensitive to emission intensity and riverine flow parameters whose sensitivitycoefficients are0.97and0.90, respectively.The resulting model is also applied to describe the multimedia environmentalbehavior of eight phenol compounds in the Songhua River in2007. Based on fieldobservations during different hydrological seasons of the year2007, the predictivepower of this model is evaluated by three different kinds of quantitative statisticalmethods, namely Nash-Sutcliffe efficiency (NSE), percent bias (PBIAS) and ratio ofthe root-mean-square to the standard deviation of measured data (RSR). Thegoodness-of-fit of model prediction for phenolic contaminants is, in general,agreeable. The conditions of NSR<0.5，RSR<=0.7and PBIAS∈[-70%，+70%] arefulfilled together for most of data points. The paired t-test result (P>0.05) indicatesthat there is no significant difference between simulated and observed data at thechosen significance level of0.05, although some outliers are present. For one focalcompound-2,4-DCP, the pathways and mass fluxes from the river system to thesurrounding environment are also evaluated. During simulation, mass flux ofriverine advection is the most remarkable and net output mass flux is around4500.1kg. Uncertainty analysis is performed by Monte Carlo stimulation to judgethe influence of variability of input parameters on modeled results. Modelsimulation indicates that phenolic pollution in the river during the low-flow periodof2012is remarkably reduced when compared to the same period of2007. Theresearch results indicate that this model can be applied in point-source pollutioncases in the river and has the ability to provide decision makers with valuablereference data for consideration of water pollution controls.For water pollution emergency, contaminant transport and fate model has beendeveloped in this study, which couples kinematic wave flow option withadvection-dispersion-reaction equation. The model includes kinetic processes suchas volatilization, photolysis and biodegradation, and diffusive mass exchangebetween water column and sediment layer as a function of particles settling andresuspension. It has been applied to simulate Nitrobenzene pollution emergency inthe Songhua River and concentration variances of Nitrobenzene in overlying watercolumn and underneath sediment bottom are obtained. For aqueous concentration,four kinds of quantitative statistical tests, namely NSE, PBIAS, RSR and TIC, areadopted to evaluate model performance. The results generally show that the modeledand detected concentrations exhibit good consistency. Flow velocity in the river ismost sensitive parameter to Nitrobenzene concentration in water column based onsensitivity analysis. It indicates flow velocity has important impact on bothdistribution and variance of contaminant concentration. The model performs satisfactory for prediction of organic pollutant fate in the river, with the ability tosupply necessary information for pollution event control and early warning, whichcould be applied to similar long natural rivers.We expect to establish numerical model to give a scientific guide for aquaticecosystem protection so that prevention, reduction and recovery of ecologicaldamage by water pollution can be realized with the aim of free-economic loss,maintenance of human health and ecosystem balance, harmony and sustainability ofpeople’s life and economic development.更多还原