苏州太湖三山岛国家湿地公园景观健康评价

发布时间：2018-04-16 11:29

本文选题：湿地公园 + 景观健康　；参考：《湿地科学》2017年05期

【摘要】：苏州太湖三山岛国家湿地公园位于太湖之中,对其进行景观健康评价可以有效评估湿地公园建设绩效,对指导湿地公园的科学性规划与建设具有重要意义。从研究区的景观数据中,提取土地利用类型面积所占比例、道路长度、不同类型坡岸长度等景观评价指标,采用F-score方法和随机森林法,对这些指标进行特征选择,并作为输入特征向量;采用主成分分析方法,分析水质评价指标,将其分为优质、良好、一般和较差4类,并作为输出分类结果。在此基础上,建立支持向量机分类评价模型;通过选择合适的核函数和核参数,实现景观健康评价。研究结果表明,苏州太湖三山岛国家湿地公园景观健康水平具有明显的空间差异性,研究区西部几乎未受到人为干扰,景观健康等级为优质;主岛的东北部位于旅游核心区域,人为干扰程度较重,景观健康等级大多属于一般和较差;主岛东南部建有人工修复湿地,明显改善了该区域的景观健康等级。在模型训练过程中,对比多项式核函数、Sigmoid核函数和高斯核函数三者的分类结果发现,高斯核函数的分类精度最高,训练集精度为90.0%,检核集精度为85.7%。在影响湿地公园景观健康因素众多、野外监测数据较少的条件下,本研究采用的支持向量机方法可以快速取得合理的评价结果,揭示湿地公园景观健康的空间差异。
[Abstract]:Suzhou Taihu Sanshan Island National Wetland Park is located in Taihu Lake. The evaluation of landscape health can effectively evaluate the construction performance of the Wetland Park and it is of great significance to guide the scientific planning and construction of Wetland Park.From the landscape data of the study area, the landscape evaluation indexes, such as the proportion of land use type area, the length of road, the length of different types of slope and bank, were extracted. The F-score method and the random forest method were used to select the characteristics of these indicators.The method of principal component analysis is used to analyze the water quality evaluation indexes, which are classified into 4 categories: high quality, good quality, general and poor, and as the result of output classification.On this basis, the classification evaluation model of support vector machine is established, and the landscape health evaluation is realized by selecting appropriate kernel function and kernel parameters.The results show that the landscape health level of Sanshan Island National Wetland Park in Taihu Lake Suzhou has obvious spatial difference, the western part of the study area is almost undisturbed by human beings, the landscape health grade is of high quality, and the northeast of the main island lies in the core area of tourism.The degree of human disturbance is serious, the landscape health grade is mostly general and poor, and the artificial restoration wetland is built in the southeast of the main island, which obviously improves the landscape health grade in this area.In the course of model training, by comparing the results of polynomial kernel function Sigmoid kernel function and Gao Si kernel function, we find that Gao Si kernel function has the highest classification accuracy, the accuracy of training set is 90.0 and the accuracy of check set is 85.7.Under the condition that there are many factors affecting wetland park landscape health and less field monitoring data, the support vector machine method used in this study can quickly obtain reasonable evaluation results and reveal the spatial difference of wetland park landscape health.
【作者单位】：苏州科技大学环境科学与工程学院;
【基金】：国家自然科学基金项目(51278523) 江苏省“六大人才高峰”第十一批高层次人才选拔培养资助项目(2014-NY-020) 苏州市科技计划项目(SYN201510)资助
【分类号】：TU986.5;X820.4

【相似文献】