水库消落带湿地植被的时空演替模式及其适生机制研究

【作者】 林川；

【导师】 赵文吉；

【作者基本信息】 首都师范大学 ， 地图学与地理信息系统， 2014， 博士

【摘要】 水库消落带(reservoir hydro-fluctuation belt)具有陆地和水域的双重属性,是一种较为特殊的季节性湿地生态系统,近年来已经成为研究热点之一。水库消落带植被是水库消落带生态系统的重要组成部分,消落带植物群落未来演替方向将会给生态环境带来巨大影响。因此,从内在机理和宏观表征两个方面,研究水库消落带湿地植被的适生机制和时空演替模式,对于保持水库消落带生态系统的生态平衡,以及维持水库生态系统的生态功能,保护水库生态环境具有十分重要的意义和价值。本文以湿地生态学、生态水文学等为理论依据,以遥感和GIS为技术支撑,选择消落带生境退化严重的官厅水库为研究区,在收集整理研究区1984-2013年间中、高分辨率遥感影像的基础上(中分辨率的Landsat系列影像13景,高分辨率的SPOT-5影像2景,高分辨率的ZY-3影像1景),再结合地面实测数据,首先利用长时间序列的Landsat系列卫星数据和官厅水库历史水文统计资料,基于遥感变化探测技术和GIS空间分析技术,划分了官厅水库消落带的边岸类型,合理确定了官厅水库消落带的范围及其分区；其次,以中、高分辨率的遥感影像和地面实测高光谱数据为数据源,并针对不同空间分辨率的数据源制定了与其对应的湿地植被分类体系,利用面向对象分类、反向传播人工神经网络(BP-ANN)和Fisher线性判别分析,对消落带内的湿地植被进行了分类提取；然后,利用从5期中分辨率Landsat影像上提取的消落带湿地植被空间分布格局图,同时结合消落带的分区数据,基于景观格局和CA-Markov(元胞-马尔可夫)的分析方法,揭示了消落带不同分区内湿地植被的时空演替规律,并对未来消落带不同分区内湿地植被的空间分布格局进行了预测；最后,利用野外采集的湿地植被地面实测高光谱数据、水质数据和室内测定的湿地植被生化参数数据,利用数理统计方法和物种分布统计模型-GAM(结合空间分布预测模块GRASP),以水生植物为例,对湿地植被的适生机制进行了分析。本文的主要研究结论如下：(1)获得了三种官厅水库的边岸类型,分别为高坡度稳定型边岸、舒缓坡度稳定型边岸和舒缓坡度淤积型边岸；定量地确定了官厅水库消落带的分区,分为长期出露区和淹水频繁区,二者的范围总和即为官厅水库消落带的总范围,其中,长期出露区的面积为42.06km2,淹水频繁区的面积为46.19km2,官厅水库消落带的总面积为88.25km2。(2)5景Landsat系列影像的平均总体分类精度达到84.86%,Kappa系数达到0.81,3景高分辨率影像的平均总体分类精度达到86.67%,Kappa系数达到0.86,提取结果均较为理想；利用地面实测高光谱数据,根据选定的8个光谱特征变量进行典型湿地植被识别,BP-ANN和Fisher线性判别的总分类精度分别达到85.5%和87.98%,识别精度也较为理想。(3)在近30年的时间里,长期出露区内,随着水体面积的不断萎缩,沉水植物、挺水植物、湿生植物以及中生植物在空间上呈现较为明显的退化趋势,而盐生植物和耕地则呈现明显的扩张趋势,其中耕地景观成为长期出露区内的优势景观类型,在空间上连续分布,面积较大,盐生植物成为长期出露区内仅次于耕地的第二优势景观类型；淹水频繁区内由于水分相对充足,沉水植物、挺水植物、湿生植物和中生植物未呈现退化趋势,反而在整体上有一定程度的扩张,1987-2013年间沉水植物、挺水植物、湿生植物和中生植物的增加幅度分别为102.17%、172.80%、160.20%和256.32%,同时,充足的水分一定程度上抑制了盐生植物和耕地的扩张,表现为盐生植物在淹水频繁区内的平均比例仅为1.58%,耕地在2007-2013年随着水域面积的增大而减少；采用CA-Markov模型,以2007年和2013年的数据为基础,对2019年的官厅水库消落带湿地植被空间分布格局进行了预测,预测结果表明：长期出露区内的湿地植被整体上仍然呈现进一步的退化趋势,生态环境进一步恶化；淹水频繁区内的湿地植被仍然没有表现出明显的退化趋势,生态环境仍然较为良好。基于水库消落带不同分区湿地植被的时空变化和演替分析结果以及基于CA-Markov模型的水库消落带湿地植被预测结果,再结合水库消落带不同分区湿地植被的转移方向分布图,得到长期出露区内湿地植被的时空演替模式为抑制性演替,而淹水频繁区内则存在两种湿地植被时空演替模式,抑制性演替和促进性演替。(4)沉水植物由于受到水体和水中悬浮物等因素的影响,其反射光谱特征较为特别,其他5种植物生态类型的反射光谱则具有一定的相似性,6种植物生态类型的WP r、Dr WP g、Rg以及510nm和675nm附近的吸收特征参数(吸收深度DEP和吸收面积AREA)均存在不同水平的差异；分别建立了含水量和叶绿素含量对典型湿地植被光谱特征参数的响应模型,含水量的响应模型为y=-9.462x2-2.671x+0.608(x为黄边面积SDy)和y=0.219e1.010x(x为SRWI),叶绿素含量的响应模型为y=20.89x-18.45(x为ND(565,735)),经过交叉检验,响应模型均取得了较好的测试效果；最后以水生植物为例,分析了物种分布对环境因子的适应性机制,在预先选定的6个潜在影响沉水植物和挺水植物空间分布的环境因子中,CHLa、TP和distowater是影响二者空间分布的主要因素,但二者的空间分布对于CHLa、 TP和distowater这3个影响因子的响应有所差异,沉水植物一般分布在CHLa浓度和TP含量较低,distowater较大的区域,而挺水植物一般分布在CHLa浓度和TP含量较高,distowater较小的区域。更多还原

【Abstract】 The reservoir hydro-fluctuation belt is a kind of special seasonal wetland ecological system and has the dual attributes of land and water, which has become the focus of research in recent years. The vegetation of reservoir hydro-fluctuation belt is an important part of the reservoir hydro-fluctuation belt ecosystem and its succession will take a dramatic ecological impact. Therefore, study the suitable mechanism and temporal and spatial succession pattern of the reservoir hydro-fluctuation belt wetland vegetation from two aspects of internal mechanism and macroscopic characterization, which has the important meaning and value for keep the balance of the reservoir hydro-fluctuation belt ecosystem, maintaining the ecological function of the reservoir hydro-fluctuation belt ecosystem, and protecting the ecological environment of the reservoir hydro-fluctuation belt ecosystem.In this paper, taking the wetland ecology and ecological hydrology as theoretical basis, using remote sensing and GIS technology, selecting Guanting Reservoir where the hydro-fluctuation belt habitat degradation seriously as the study area, collecting the remote sensing images during1987-2013in the study area which include3high resolution remote sensing images (SPOT-5and ZY-3) and medium-resolution remote sensing images (Landsat ETM+, Landsat TM and Landsat8OLI), and combined with the ground measured data. Firstly, based on the long time series of Landsat satellite data and the historical hydrological data of Guanting Reservoir; the remote sensing change detection and GIS spatial analysis technology were used to discriminate the stability of the Guanting Reservoir shore, the coverage and partition of Guanting Reservoir hydro-fluctuation belt reasonably. Secondly, based on the moderate, high resolution satellite images and the ground measured hyperspectral data; the different classification systems were made for the different spatial resolution data; and the object-oriented classification, back propagation artificial neural network (BP-ANN) and Fisher linear discriminant analysis were made use of to extract the wetland vegetation of the hydro-fluctuation belt. Then, based on the5wetland vegetation spatial distribution maps of hydro-fluctuation belt extracted from the medium-resolution Landsat images and the partition data of hydro-fluctuation belt; the analysis methods of landscape pattern and CA-Markov were used to reveal the temporal and spatial succession pattern of wetland vegetation with different partitions of hydro-fluctuation belt, and predicted the future spatial distribution pattern of wetland vegetation with different partition of hydro-fluctuation belt. Finally, based on the ground measured hyperspectral data, water quality data and laboratory biochemical parameter data; the mathematical statistics method and the species distribution statistical model (GAM) combined with spatial distribution prediction module (GRASP) were used to analyze the suitable mechanism of wetland vegetation. The principal conclusions of this paper are as follows:a) The high slope stability shore, soothing slope stability shore and soothing slope deposit shore were mainly3shore types of the Guanting Reservoir. The water area of Guanting Reservoir first increased and then decreased during1987-2013, which had the maximum water area in1996(113.12km2) and minimum water area in2007(32.49km2), the overall atrophy of water body in Guanting Reservoir was more serious. According to overlay analysis, the hydro-fluctuation belt of Guanting Reservoir was divided into long-term outcrop region and frequent flooding region. The total coverage of Guanting Reservoir hydro-fluctuation belt was also determined reasonably. The area of long-term outcrop region, frequent flooding region and Guanting Reservoir hydro-fluctuation belt were42.06km2,46.19km2and88.25km2, respectively.b) The average overall classification accuracy of5Landsat series images was84.86%and the Kappa coefficient was0.81. The average overall classification accuracy of3high resolution images was86.67%and the Kappa coefficient was0.86. The extraction results were ideal. Making use of the selected8spectral characteristic variables to typical wetland vegetation identification, the total classification accuracy of BP-ANN and Fisher linear discriminant analysis were85.5%and87.98%respectively. The classification results were also ideal.c) With the water area shrinking, The submerged plant, emerged plant, hygrophilous plant and mesophyte plant presented degeneration obviously in long-term outcrop region from the recent30years; while the halophilous plant and cultivated land presented expansion obviously. The cultivated land landscape became the dominant landscape type of long-term outcrop region; it had continuous spatial distribution and large area. The halophilous plant became the second dominant landscape type of long-term outcrop region. Compared with the long-term outcrop region, water in the frequent flooding region was relative enough, so submerged plant, emerged plant, hygrophilous plant and mesophyte plant did not present the trend of degeneration; instead, they presented a certain degree of expansion. In addition, the enough water greatly inhibited the growth and distribution of the halophilous plant and cultivated land, the halophilous plant and cultivated land did not become the dominant landscape types of the frequent flooding region. The CA-Markov model was used to simulate the spatial distribution pattern of wetland vegetation in the Guanting Reservoir hydro-fluctuation belt based on the2007and2013data. The predicted results show that the wetland vegetation will still present degradation trend in the long-term outcrop region and the ecological environment will further deteriorated; the wetland vegetation will not present degradation trend obviously in the frequent flooding region and the ecological environment will still good. Based on the temporal and spatial analysis results with different partition of Guanting Reservoir hydro-fluctuation belt, succession analysis results with different partition of Guanting Reservoir hydro-fluctuation belt and the CA-Markov prediction result, combined with the transfer direction distribution map of wetland vegetation in different partition of Guanting Reservoir hydro-fluctuation belt, the temporal and spatial succession pattern of long-term outcrop region was inhibitory succession, the temporal and spatial succession pattern of frequent flooding region was inhibitory succession and promotive succession.d) The reflectance spectrum of submerged plant was special because of the influence of water body, suspended solid in water, etc. The reflectance spectra of the other five plant ecological types were similar. The WP_r, Dr, WP_g, Rg,510nm and675nm absorption feature parameters had certain differences. The response models of water content and chlorophyll content to spectral feature parameters of typical wetland vegetation were established respectively. The response model of water content was y=-9.462х2-2.671х+0.608(х was the yellow edge area, SDy) and y=0.219e1.010x (х was SRWI) respectively, the response model of chlorophyll content was y=20.89х-18.45(x was ND (565,735)). According to Cross Validation examination, the response models have achieved the satisfactory test results. Finally, take the aquatic plant for example to analyze the suitable mechanism between the species distribution and environmental factors. The6environmental factors which had the effect to the spatial distribution of the submerged plant and emerged plant were selected in advance; CHLa, TP and distowater were the main factors that influence the spatial distribution of the submerged plant and emerged plant, but the response between the spatial distribution and the environmental factors (CHLa, TP and distowater) had certain differences. The submerged plant were generally distributed where the CHLa and TP were lower and distowater was larger, the emerged plant were generally distributed where the CHLa and TP were higher and distowater was smaller.更多还原