北温带落叶阔叶森林沼泽遥感识别及其群落特征分析

发布时间：2018-02-04 05:48

本文关键词： 落叶阔叶森林沼泽雷达影像遥感识别群落特征　出处：《东北师范大学》2014年硕士论文　论文类型：学位论文

【摘要】：森林沼泽群落位于生态交错带,属于隐域性植被,其空间分布格局和特殊的群落特征一直是湿地生态学领域的研究难点和热点。本论文旨在为高效获取森林沼泽资源分布信息及认识群落特征而服务,研究了北温带长白山地区的落叶阔叶森林沼泽的遥感识别研究过程,并对森林沼泽的群落特征进行了初步分析。根据森林沼泽的环境特征,构建了整合地形、植被、水文、物候响应4个特征量的森林沼泽遥感识别决策树模型,利用实地调查数据,对模型识别结果进行了验证,并对遥感分类误差进行了深入分析。在群落水平上,沿着沼泽到森林的环境梯度,选择草本沼泽、森林沼泽及典型森林群落,对森林沼泽的群落特征进行归纳总结和初步分析。主要结果如下： (1)所构建森林沼泽决策树模型具有很好的遥感识别效果,森林沼泽和平缓森林的遥感分类总体精度为83.6%,森林沼泽制图精度为85.5%,森林制图精度为82.3%。基于雷达影像提取的林下水文特征量对整个遥感分类过程的贡献率最大,物候响应特征能有效改善分类效果。整合更多时相的雷达影像,获取更准确全面的林下水文特征,是森林沼泽遥感识别研究的关键技术。 (2)处于生态过渡带的森林沼泽群落,地表环境要素的异质性影响雷达影像遥感分类的效果。地表草本覆盖为优势特征的沼泽区和被枯落物覆盖的森林区,二者雷达影像散射特征的分离性较差,是遥感分类误差的主要原因。物候响应特征有助于提高平缓森林的遥感识别。面向对象分类是解决环境异质性和像元异质性的重要方法,有利于湿地边界的有效识别和分类效果的改善,是后续研究关注的重点内容。 (3)林下水文特征及物候响应特征量的理想阈值范围,位于森林沼泽和平地森林相应特征量散射系数分布曲线的交点附近,其中交点值是较为理想的取值之一。若研究区拥有分类要素充足的训练区,借助本论文构建的模型,有助于快速开展森林沼泽遥感识别工作。 (4)研究区内的落叶阔叶森林沼泽为水曲柳林,群落各层次物种数的关系为乔木层灌木层草本层。较森林群落相比,森林沼泽群落木本植物的物种数较少,均匀度较低。 (5)在森林沼泽群落中,先锋物种白桦的喜光性最强,优势度显著的水曲柳耐湿性最高,适应性较强的怀槐稍耐阴性。对比森林群落,森林沼泽乔木优势种的喜光喜湿润性更显著。 (6)对比草本沼泽、森林沼泽和森林3类群落,群落各层次间的生物多样性表现为：森林沼泽群落草本层的生物多样性和物种丰富度均先后高于草本沼泽和森林群落；乔木层和灌木层的生物多样性和物种丰富度均低于森林群落。森林、森林沼泽、草本沼泽3类群落生物多样性的总体趋势为：森林群落森林沼泽群落草本沼泽群落。
[Abstract]:The forest swamp community is located in the ecotone and belongs to the hidden vegetation. Its spatial distribution pattern and special community characteristics have been the research difficulties and hot spots in wetland ecology field. This paper aims to efficiently obtain the information of forest and swamp resources distribution and recognize the community characteristics. The process of remote sensing recognition of deciduous broadleaved forest swamps in Changbai Mountain area of northern temperate zone was studied and the community characteristics of forest swamps were analyzed. According to the environmental characteristics of forest swamps, a decision tree model of remote sensing recognition of forest swamps is constructed, which integrates the four characteristics of terrain, vegetation, hydrology and phenology, and makes use of field survey data. The model recognition results were verified, and the classification error of remote sensing was analyzed in depth. On the community level, the herbaceous swamp, forest swamp and typical forest community were selected along the environmental gradient from swamp to forest. The community characteristics of forest swamp are summarized and analyzed. The main results are as follows: 1) the forest swamp decision tree model has good remote sensing recognition effect. The overall accuracy of remote sensing classification of forest swamp and flat forest is 83.6, and the precision of forest swamp mapping is 85.5%. The accuracy of forest mapping is 82.3. The hydrological characteristics extracted from radar images have the largest contribution to the whole process of remote sensing classification. Phenological response features can effectively improve the classification effect. Integrating more temporal radar images and obtaining more accurate and comprehensive understory hydrological features is the key technology of forest and swamp remote sensing recognition research. (2) in the forest and swamp communities in the ecological transition zone, the heterogeneity of surface environmental factors affects the effect of remote sensing classification of radar images. The surface herbaceous cover is the dominant marsh area and the forest area covered by litter. The separation of the scattering characteristics of the two radar images is poor. Phenological response is helpful to improve remote sensing recognition of flat forest. Object-oriented classification is an important method to solve environmental heterogeneity and pixel heterogeneity. It is beneficial to the effective identification of wetland boundary and the improvement of classification effect, which is the focus of further research. 3) the ideal threshold range of hydrological characteristics and phenological response characteristics is located near the intersection of the scattering coefficient distribution curve of the corresponding forest characteristics in the forest swamp and peaceful land. The intersection value is one of the more ideal values. If the study area has sufficient classification elements of the training area, with the help of the model constructed in this paper, it is helpful to quickly develop the forest and swamp remote sensing identification work. (4) the deciduous broadleaved forest swamps are mandshurian forests, and the relationship of species number at each level is the herbaceous layer of Arbor layer and shrub layer. Compared with the forest community, the species number of woody plants in the forest swamp community is less than that in the forest community. The uniformity is low. Among the forest swamp communities, Bai Hua, a pioneer species, had the highest photoicity, the highest moisture tolerance of Fraxinus mandshurica with significant dominance, and the slightly negative tolerance of Huai Huai with strong adaptability. The dominant species of forest swamp trees are more favorable to light and moisture. Comparison of herbaceous swamps, forest swamps and forest communities. The biodiversity of the forest swamp community was higher than that of the herbaceous layer and the species richness of the herbaceous layer was higher than that of the herbaceous swamp and forest community. The biodiversity and species richness of Arbor layer and shrub layer are lower than that of forest community, forest, forest swamp. The general trend of biodiversity of herbaceous swamp community is forest forest swamp herbaceous swamp community.
【学位授予单位】：东北师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P237;P941.78

【参考文献】