【作者】 刘庆；

【导师】 吴志峰；

【作者基本信息】 中国科学院研究生院(广州地球化学研究所) ， 环境科学， 2016， 博士

【摘要】 流域水环境管理中,在点源污染得到较好控制的情况下,土地利用组成与空间布局的调整和优化可有效调控非点源污染。使探索河流水质对流域景观特征时空变异的响应及其机制,成为河流水质污染控制基础研究的热点之一,对流域景观优化设计与水环境保护具有重要意义。氮是陆地水体的重要污染物,相邻高地的氮经地表与地下径流通过河岸带排入水体。河岸带作为陆地和水体之间的过渡缓冲带,是截留转化陆地污染物防止其进入水生环境的最后一道屏障,在控制来自陆地非点源的氮素过程中发挥着重要的作用。因此,研究河岸带生态系统对污染物的去除作用,对改善河流水质和保护流域生态系统健康具有重要的理论和现实意义。本文以GIS技术为基础,结合实地采样调查,使用传统线性多元统计与非线性统计法分析了流域景观特征对河流水质的影响,并获得河流水质典型污染物种类,在此基础上,通过分析河岸带生态系统对该典型污染物的去除作用,来展示河岸带生态系统的重要作用。本文研究的主要内容及结论如下所示:本论文主要得到以下研究结果与结论:(1)以广州市流溪河流域为研究区,在水质外业断面采集的基础上,采用偏冗余分析(Partial redundancy analysis PRDA)和随机森林回归分析(Random forests regression RFR)等多种统计分析方法,分析不同景观变量组与河流水质间的关系,确定各景观因子对水质的影响和贡献程度。结果表明:流域景观特征从上游到下游表现为城市化递增的趋势,水质恶化程度与该趋势一致。其中,裸露地对水质的影响作用最强,其次是建设用地,两者间对水质变化起负效应,而林地与水质指标之间的关系表现为负相关,园地与水质指标关系不明显;在所有景观特征变量中,景观组成对水质的作用最明显,其对水质空间变异的解释贡献率占总解释方差的20.7%;其次是自然景观因子和景观格局指数,其值分别为8.1%和4.3%。与传统的线相关分析相比,偏冗余分析结合随机森林回归的非线性统计方法能够起到更高的降维作用,两者的结合是研究流域景观指标与水质关系的有效途径之一。(2)以广州市水源地流溪河为研究对象,在子流域尺度及30m、61m、100m、200m及300m河岸缓冲尺度中,通过冗余分析(RDA)筛选对水质空间变异解释力最佳的宽度,同时获得对水质影响最显著的景观变量组;利用CART模型进行的变化点分析筛选对单一水质指标影响最大的景观因子,确定沿流域景观梯度变化引起水质突变的景观阈值.结果表明:整体来看,河岸缓冲尺度对水质的影响要大于子流域尺度,且200m缓冲区尺度的景观因子对水质空间变异的解释力最大;在各尺度的景观特征中,裸露地面积比(bar%)是对水质影响最重要的景观因子.另一方面,在200m缓冲区尺度中,bar%、建设用地面积比(con%)及建设用地最大斑块指数(lpicon)是影响水质中tp、cod及nh4+空间变异的最主要景观变量;且当con%>15~23%、bar%>0.8~1.4%及lpico>5.0%~6.6%时会分别加重河流中cod、tp及nh4+的污染负荷。(3)以广州市水源地流溪河二级支流的河岸带为研究对象,分析流域多尺度环境因子对河岸带土壤理化性质的影响,确定调控河岸带土壤理化性质的不同尺度景观因子,为河岸带管理和修复提供参考和科学依据。结果表明:河岸带流域纵向尺度上的地形指数、坡度、海拔、土地利用类型等因子是影响土壤水分含量、土壤质地及ph等性质的主要因素;河岸带横向断面尺度上距河边远近不同的景观位置及植被覆盖度等因子是影响土壤可溶性有机碳、土壤硝态氮及土壤容重等性质的主要因素;河岸带垂直剖面尺度上人为活动对土壤干扰强度大小、剖面根系分布情况及地下水位高低等因子是影响土壤有机碳、总氮及氨氮等性质的主要因素。(4)以广州市水源地流溪河二级支流的河岸带为研究对象,分析嵌套结构的多尺度因素对河岸带土壤反硝化作用的影响,确定调节河岸带土壤反硝化酶活性的不同尺度控制因素。结果表明:以土壤反硝化酶活性表征的反硝化潜力,其在研究区内的剖面尺度空间变异性显著大于断面尺度和流域景观尺度;反硝化酶活性的空间变异受多尺度因子的作用,其中剖面尺度中土壤有机碳、全氮、硝态氮等因子对其空间分布起主要的直接控制作用,而断面尺度的景观位置、植被密度与流域景观尺度的地形指数、高程及土地利用类型等因子通过影响土壤性质而间接调控反硝化酶的空间分异性;研究区的反硝化酶活性在剖面尺度中表层最高,随着深度增加呈急剧减小的趋势;断面尺度中活性值最高位于河岸区中间的景观位置,因该位置具有良好的景观连接性和较高的植被密度;流域景观尺度中,除受人为扰动严重城市化地区外,河岸带土壤反硝化酶活性沿流域从上游至下游呈不断增加的趋势。(5)以广州市水源地流溪河二级支流的河岸带为研究对象,分析河岸带浅层地下水对氮素的去除作用,确定主要作用机理,探讨主要影响因素。本研究结果表明,所研究区的河岸带对来自高地地下水硝酸盐有很好的削减作用,可较好的起到对河流水质保护的作用,去除机理主要是反硝化作用,影响该机理的直接因素有地下水中氮源、碳源及氧气的供给情况,间接因素有纵向景观尺度的表征水文地质的含水层厚度、土地利用类等环境因子,及代表横向尺度因子的景观位置、河岸带宽度环境因子。三个断面的两个宽度的河岸带均显示了明显的对地下水硝酸盐的去除效果,其中,去除率最高的主要位于可提供丰富硝态氮的耕地用地类型的样地D2,其在23m与10m宽度的去除率均超过90%,显著高于其他样地;而出去率相对最低的位于林地背景的用地类型为菜地的样地D4,23m宽的去除率只有53%;在不考虑河岸区地下水硝氮输入浓度的差异时,23m宽的河岸带硝氮去除率要高于10m宽的。更多还原

【Abstract】 In the case of point source pollution getting better control, the optimization of land use composition and space layout can effectively control the non-point source pollution in water environment management. Thus, water quality response to the temporal and spatial variation of watershed landscape characteristics and its mechanism became the hotspot in controlling water pollution for basic research. Nitrogen is one of the important pollutants for terrestrial water, Nitrogen is from adjacent highlands going through the riparian zone into water by surface and underground runoff. Riparian zone, as an important buffer zone between land and water, is the last barrier to stop land pollutant getting into the water. It plays an important role in controlling nonpoint source. So reaching on riparian zone has great theoretic and practical significance in pollutants removal, water quality and watershed ecosystem health.Combining with the field sampling survey, this paper had analyzed the influence of watershed landscape characteristics on water quality by using GIS, linear multivariate and nonlinear statistical method, and got the typical pollutant of water quality. On this basis, another purpose is to show the important role of riparian zone ecosystem.Main results from study are as follows:(1) To explore the relationship between water quality and landscape characteristic at the Liuxi River watershed, based on collected water quality data, Partial redundancy analysis(PRDA) and Random forests regression(RFR) analysis were integrated to identify the key landscape factors and to determine the important landscape characteristic variable group at the watershed scale. The results showed that between water quality and landscape characteristic variables there was a significant correlation. On the whole, landscape characteristics from upstream to downstream enhanced the gradient of urbanization along with the water quality deterioration. Thereinto, percentage of bare land area had a most important negative effect on water quality, then was percentage of construction land area. However, percentage of forest land area had a certain positive effect on water quality. There were no obvious relations between the percentage of watershed area, orchard area and water quality. Out of the explained variation of water quality, landscape composition(20.7% of 0.657; that is 0.136/0.657) was the major contributions at the watershed scale. The result indicated the importance of the landscape composition for the water quality of the watershed. Specifically, the percentage of bare land area and construction land area were the most important factors to the concentration of total phosphorus and ammonia nitrogen, COD in the river; Meanwhile, because redundancy analysis combined with random forest regression, they could have higher dimension reduction effect, it proved that a combination of both was one of the effective ways to study the relationship between landscape characteristic and water quality.(2) Combing the scale dependence and change-point analysis to discuss the relation between the landscape and the water quality is important to provide valuable information for the restoration of riparian buffer and river management. Liu Xi River watershed in Guangzhou city was as a typical case study area, at the scales of sub-basin and 30 m, 61 m, 100 m, 200 m, 300 m buffer zones, redundancy analysis(RDA) was used to screen the best buffer scale in that the landscape variables had more explaining power for water quality spatial variation, and to gain the best landscape variable group. Change-point analysis was applied to determine the optimum landscape factor for explaining every water quality indicator and to estimate the speci?c locations along a gradient of landscape metric that result in a sudden change in the water quality variable by the Classification and regression tree(CART) model. The results showed that: the water quality might to be better correlated with the buffer landscape metrics than with the sub-basin scale landscape indices, the landscape pattern at the200 m buffer was believed to have the greatest impact on the water quality; The area percentage of bare land(BAR%), the area percentage of construction land(CON%) and the large patch index of construction land(LPIcon) were recognized respectively as the dominant variable influencing the TP, COD and NH4+ for a 200 m buffer zone. The result of change-point analysis indicated the key interval values of the tree landscape variables within the 200 m buffer zone. When the BAR% was >0.8~1.4%, CON% was >15~23% and LPIcon was >5.0%~6.6%, the water quality might increase opportunity for pollutant reduction.(3) Taking the second-order tributaries of Liu Xi River in Guangzhou city as an example, the effect of multi-scale catchment landscape features on the soil physical and chemical properties of riparian zone soils were analyzed. The objectivity of this research is to explore and identify the main factors that influence soil properties at different spatial scales and provide scientific data for riparian zone management and restoration. The results suggest that along the longitudinal direction of the riparian zone, topographic index(TI), land slope, land elevation and land use types are the governing factors on the spatial heterogeneity of soil water content(SWC), soil textures and p H; along the transversal direction, the distance of a landscape from riverside and the vegetation coverage are the factors affecting the spatial heterogeneity of soil dissolved organic carbon(DOC), nitrate nitrogen(NO3--N) and bulk density(BD); along the vertical profile of the riparian zone, the intensity of human disturbance on soil, the root system distribution in soil profile and the groundwater tables are the major factors affecting the spatial heterogeneity of TOC, TN and NH4-N in riparian zones.(4)Soil denitrification enzyme activity was measured by a hierarchical sampling approach within the riparian zone of the Liu Xi River tributary of Guangzhou city, and explored and identified multi-scale factors that influenced soil denitrification. The results showed that soil denitrification potential—represented by soil denitrification enzyme activity(DEA)—showed more heterogeneity on a profile-scale than on a crosssectional scale and landscape-scale in riparian zones; multi-scale factors controlled the degree of spatial variation in soil denitrification. The profile-scale factors, including soil TOC(total organic carbon), TN(total nitrogen) and NO 3--N(nitrate nitrogen), were the major direct regulators of the spatial distribution of DEA. The cross-sectional scale factors, including landscape position and vegetation density, and the landscape-scale factors, including TI, elevation and land-use types, indirectly regulated the distribution of DEA. At the profile scale, the highest DEA occurred on the upper soil, but dramatically declined from the soil surface to the deeper layers. At the cross-sectional scale, the highest DEA was observed at the landscape position, with good landscape connectivity and high-density vegetation within the middle part of the riparian zone. At the watershed scale, DEA showed an increasing trend from upstream to downstream except for the sites in urbanized areas.(5)We examined associations between shalow groundwater of the riparian zone and nitrogen removal efficiency in the secondary tributary of the Liuxi River to determine the main mechanism and main influencing factors. Results showed that the riparian zone had a great cutting effect on nitrate from the high groundwater, so it had better protective effect on river water quality. Denitrification was the main remove mechanism, which directly affected by the nitrogen in the groundwater, carbon source, or the supply of oxygen. And some factors had indirect effects on this mechanism, such as aquifer thickness, land use, et al in vertical landscape scale and landscape position, riparian zone width, et al in horizontal scale. The three sections of both the width of riparian zone showed apparent effect on removing nitrate of the groundwater. D2 had the highest removal rate. The land use type of this sample area is cultivated land, which can provide a rich nitrate. The removal rate of 23 m and 10 m width in D2 were over 90%, significantly higher than other samples.The lowest removal rate was happened in vegetable sample area-D4 which located in the woodland. The removal rate of 23 m riparian zone was only 53%. Without considering the input concentration difference of Nitrate nitrogen, nitrate nitrogen removal rate of 23 m riparian zone was higher than 10 m.更多还原