【作者】 黄广远；

【导师】 徐程扬；

【作者基本信息】 北京林业大学 ， 森林培育学， 2012， 博士

【摘要】 城市森林是城市生态系统的重要组成部分,结构和功能是其首要特征,了解城市森林的结构及其与功能的关系、协调城市森林多种功能的发挥是更好地建设城市森林的关键。而城市森林景观质量是其社会服务功能的重要方面。本研究以北京市城区城市森林为研究对象,以群落学调查为基础,对城市森林的植物组成结构特征及空间分布特点进行了研究；对景观质量评价的指标、方法进行了研究；对典型城市森林景观美学质量和主要树种的单体景观美学质量进行了评价、对景观美学质量预测检验进行了研究。以此阐述北京市城区城市森林的结构特征、探究结构与景观质量之间的关系、揭示影响城市森林景观美学质量的主要因子。以进一步为北京城市森林的景观建设和改造优化实践提供参考、为城市森林生态系统服务功能的提升提供理论依据。主要结论如下：(1)北京市城区城市森林依据群落外貌的优势种类型可划分为71个群落类型,分为针叶林、阔叶林、针阔混交林和竹林四种外貌类型和七种生活型类型。城区城市森林植物108科323属633种,分别占北京市植物总科、属、种数的63.12%、33.82%和29.62%,总体数量偏低；在木本植物中,松科、杨柳科、蔷薇科等构成了城区城市森林植物群落的优势科；乡土种仅占全北京乡土种的22.00%,比例偏低,而引进种占全北京引种植物的48.30%,国外引进种占20.68%,比例偏高；木本植物乡土种低于外来种；乡土乔木、灌木、草本种占北京山区野生乔木、灌木、草本种的62.22%、53.45%和23.56%。受地带性的影响同时在城市化建设的人为干扰下表现出一定的变异性；乡土植物利用不够,尤其灌木和草本植物乡土种有十分大的应用潜力；少数几个种应用频度高而多数种应用频度低,是造成植被景观单一的原因之一。(2)受人为因素干扰,乔、冠、草不同层次的多样性指数在不同城市森林绿地有一定的差异,总体多样性指数以草本层最高；乔、冠、草不同层次的Patrick指数为以草本层最高、灌木层最低、乔木层居中；Pielou指数表现为乔木层>灌木层>草本层。城区城市森林的建设受人为干扰影响较大,规律性较差,木本Simpson指数、Patrick指数、草本和乡土草本Patrick指数都呈高斯模型的变化趋势；草本Simpson指数呈指数模型变化趋势；乡土木本种Patrick指数呈球状模型变化趋势；半方差曲线中木本Patrick指数、乡土草本种Patrick指数和草本Patrick指数的基台值最高；木本Patrick指数和Simpson指数的结构方差比最高；木本Simpson指数各个方向异质性差；乡土木本种Patrick指数在不同方向上,基台值、块金值和变程均不相同,具有条带状各向异性的特征,表现出一定的不均衡性；木本Patrick指数在大斑块中间分布有值更高或更低的小斑块,南、北部有一定的差异；草本Patrick指数值除西北部较高、市中心区较低外,其余地方插值图斑块边界复杂、斑块破碎化较严重、斑块呈交错分布；乡土种Patrick指数值规律性明显,市中心区低、向市区四周延伸逐渐增加；北京城区城市森林植物种组成状况明显受到人为干扰,种类和分布都产生了复杂变化。(3)以实际调查查阅相关文献资料为基础初选评价因子,应用灰色系统理论、层次分析法、统计评判法等对影响城市森林景观美学质量的指标进行筛选,最后确定郁闭度、灌木盖度、草本盖度、胸径分布、林分密度、枯枝明显度、空间联系性、观赏特性多样性、灌草统一度、枯树倒木、枯枝落叶等23个指标是影响北京城区城市森林景观美学质量的重要指标,构建评价指标体系。(4)应用AHP法、SMPPC法、TOPSIS法和SBE法四种方法分别对城市森林景观样地进行评价,四种不同评价方法获得的景观值不存在显著性差异,都能够体现整体景观特征,可信度较强。心理物理学的SBE法以大众品味作为评判标准,评价结果具有数量基础,操作简单易行,且结果可信度高,SBE法仍然是当前行之有效的城市森林景观评价方法；其余三种方法评价客观性强,但过程比较复杂。实践中四种评价方法可以结合应用;随着计算机技术和地理信息技术的发展,新的评价方法不断涌现,对于城市森林景观建设具有很好的促进作用。(5)采用SBE法评价城市森林景观质量,不同类型城市森林景观美景度以近自然公园最高、郊野公园最低；不同植被类型景观质量以阔叶纯林、混交林和针阔混交林最高、常绿针叶纯林最低；采用多元数量化方法,构建了不同城市森林类型景观美景度模型；利用曲线回归构建了单因素模型,各模型拟合度高,可以反映不同城市森林绿地和各景观要素的景观美学质量特征,郁闭度、草本盖度、林分密度对景观的影响都呈二次曲线模型；调整林分结构使林分密度、郁闭度等处于中等级别以形成良好的景观空间是今后建设的重点；这为进一步城市森林的景观优化提供了理论基础依据。(6)景观美景度法和灰色关联法对城市森林树种单体景观的评价结果具有很好的一致性；单体景观评价显示,针叶树景观质量高于阔叶树、外来种高于乡土种；城区城市森林中Ⅰ、Ⅱ级景观乔木树种各占18.97%和63.79%,它们在城市森林中出现频率和单种数量均占绝对优势,体现了北京城市森林的整体景观面貌和风格特色；外来树种丰富了城市森林景观内容,但更要加大乡土种的应用力度,使其充分发挥景观效果。(7)以不同模型对城市森林景观美学质量进行预测评价,卡方统计检验说明实测值与五种模型的预测值之间没有显著差异,五种模型的预测值都是合理的。进一步分析,BP模型预测精度高、稳定性好、风险低,预测合格率达90%,同时其决定系数高,线性关系好；总体上BP模型的预测检验效果最好,建议使用BP模型进行城市森林景质量的预测评价。(8)典型公园城市森林景观引入POE评价,结果表明目前城市森林景观建设还不能很好地满足使用者的需求,景观质量的提升还有很大空间。要以多学科知识为指导、循序渐进,不断提高景观质量。更多还原

【Abstract】 Urban forest is the key components of urban ecosystem and structure and function is its primary characteristics. To understand the city forest structure and its relationship with function and to coordinate variety functions for the better construction is the key area. The landscape quality is the most important aspect of it’s social service function.This study taking the urban forest of Beijing city as the research object, the plants composition-structure characteristics and spatial distribution characteristics are studied by the community survey as the foundation, landscape quality evaluation index and method are studied, the typical urban forest landscape aesthetic quality and the main species of monomer landscape aesthetic quality are evaluated, landscape aesthetic quality prediction test was researched. Then elaborating the structural features of the urban forest, exploring the relationship between the structure and landscape quality, revealing the main factors to affect the aesthetic quality of urban forest landscape, providing a reference for landscape construction, renovation and optimization practice of the urban forest and providing a theoretical basis for the enhancement of the urban forest ecosystem services in Beijing. The main conclusions are as follows:(1) It can be divided into72community types based on community appearance of dominant species in urban forest of Beijing city, and it can be divided into coniferous forest, broad-leaved forest, broad-leaved and coniferous mixed forest and bamboo forest four physiognomy types and seven kinds of life types. According to the urban floral category, there are633plant species in total which belong to108families and323genus in the green space in urban forest of Beijing city, which occupy63.12%,33.82%and29.62%of the total plant families, genus and species, the overall population is low and focus on Compositae, Gramineae, Leguminosae, Rosaceae, Cyperaceae such worldwide families. Pinaceae. Salicaceae and Rosaceae constituted dominant families in plant community of woody plants. Native species have a low proportion and occupy22%of the total native plant species in Beijing. Exotic species have a high proportion and occupy48.30%of exotic plants in Beijing, and native arbors, shrubs, herbs occupy62.22%,53.45%and23.56%of total species of Beijing mountainous area. It showed a certain amount of variability by influence of zonal effects and urbanization of the human disturbance. Native plants use is not enough, especially shrub and herb plants have very great application potential. A few kinds of application of high frequency application of low frequency and which was one of the reasons the vegetation landscape single.(2) Arbor, shrub, herb diversity index have certain differences in different urban forest green space from human interference. Patrick index is arranged in the herbs layer, arbor layer and shrub layer, and Pielou index is arranged in the arbor layer, shrub layer and herbs layer. Wood Simpson index, Patrick index and herb Patrick index were Gaussian model trend, herb Simpson index was exponential model trend, native wood Patrick index showed a trend of spherical model. Semivariance curve of wood Patrick index and herb (local herb) Patrick index have high Still value, woody Patrick index and Simpson index have high structural variance ratio value, native wood Patrick index are not the same trends in different directions on Sill, Nugget value and the Range process and banded anisotropic characteristics, showing a certain degree of imbalance. Wood Patrick index in the big patches distribution among have value higher or lower small patches, there is a certain difference from South to north. Herb Patrick index was lower in city center and higher in the northwestern, the rest place where the boundary complex, patch fragmentation more serious, patch is crisscross distribution. Native Patrick index regularity is obvious, the downtown area low, toward the city around extensions gradually increase. Urban forest plant species composition condition clearly influenced by the human interference, types and distribution has had a complex changes.(3) Based on the actual investigation and access relevant literature material as the basic primary evaluation factors, the indicators of the aesthetic quality of urban forest landscape were screened through grey system theory, analytic hierarchy process and statistics evaluation method. Then Identified density of canopy, coverage of herbage, stock density, diameter distribution, noticeability of dead branch, spatial relation, ornamental characteristics diversity, unified degree of shrub and grass et cl23indicators that affect the aesthetic quality of urban forest landscape of Beijing city, then evaluation index system were established.(4) Landscape sample evaluation were conducted with AHP method SM-PPC method, TOPSIS method and SBE method. Four different method for the evaluation of the landscape value does not exist significant differences, to be able to reflect the overall landscape features, credibility is stronger. Psychological paradigm’s SBE method are easy operation with mass taste as evaluation standards and Evaluation results are reliable, SBE method is still effective landscape evaluation method current. The objectivity of the other three kinds of methods is strong, but the process is more complex. Four evaluation methods can be combined with practice applications. With the development of computer technology and geographic information technologies, new evaluation methods continue to emerge, has a good role in promoting the construction of urban forest landscape.(5) Landscape quality evaluation were conducted using SBE method in urban forest, and nearly nature park is the highest value and country park is the lowest value of landscape beauty in different types of urban forest, and broad-leaved and coniferous mixed forest has the highest value and deciduous broad-leaved pure forest has the lowest value of landscape beauty in different vegetation types. Different types of urban forest landscape beauty models were built using multiple quantitative method, the single factor models were constructed using curve regression method. The model fitting high, each model fitting degree is high, can reflect the characteristics of different urban forest green space and the landscape elements of landscape aesthetic quality. Crown density, herb coverage and stand density are two times curve model in test. Adjust forest structure so that the stand and canopy density are in medium level so as to form a good landscape is the key construction aspect in the future. Which provides a theoretical basis for further optimization of urban forest landscape.(6) Landscape beauty value method and gray association analysis in good agreement with the evaluation results has the very good consistency with scenic beauty estmation method and grey correlation analysis method in urban forest species monomer landscape. Monomer landscape evaluation shows that conifers landscape quality higher than broadleaf trees, exotic species higher than native species. Ⅰ,Ⅱ grade landscape tree species each accounted for18.97%and63.79%, frequency and single-number account for an absolute advantage in the urban forest. It embodies the overall appearance and style of urban forest landscape in Beijing. We should make fully exerting landscape effect to increase the application of native species. (7) Urban forest scenery aesthetic quality was predicted and tested in five different model, and chi-square statistical test shows no significant differences between the measured value and the predictive value of the five models. The BP model has high forecast accuracy, good stability, low risk, higher decision coefficient, good linear relationship and predict pass rate up to90%. Generally, the BP model prediction effect is the best, we suggest using BP model in the prediction of urban forest landscape quality.(8) POE method was used in typical parks, the results show that the current urban forest landscape construction still cannot meet the needs of users. It has a large space in increasing the quality of urban forest landscape and improving the quality of urban forest landscape constantly based on multi-disciplinary knowledge is a very hard work.更多还原