Knowledge Resource Center for Ecological Environment in Arid Area
| 中国草地类型、覆盖度遥感划分及时空格局研究 | |
| 其他题名 | CLASSIFYING GRASSLAND TYPE AND FRACTIONAL COVERAGE AND STUDYING THEIR SPATIO-TEMPORAL PATTERN IN CHINA BASED ON REMOTE SENSING |
| 温庆可 | |
| 出版年 | 2009 |
| 学位类型 | 博士 |
| 导师 | 张增祥 |
| 学位授予单位 | 中国科学院遥感与数字地球研究所 |
| 中文摘要 | 草地生态系统是陆地生态系统中最重要、分布最广的生态系统类型之一，与森林和海洋并列为地球的三大碳库，在全球变化中占有重要地位。草地类型和草地覆盖度作为草地的基本属性和直观量化指标，在水文、气象、生态等研究领域的过程分析及建模中，都作为重要的控制因子或基本的输入变量而存在，是反映草地资源状况最基本的两个指标。然而在草地资源研究领域，有关草地生物量、退化状况等更具现实意义的草地研究较多，系统的进行草地类型和覆盖度的研究并不多见。对草地基础性指标的深入研究，是提高我国草地资源利用、保护精细程度的关键。因此，论文以全中国为研究区域，以中国草地类型和草地覆盖度为研究对象，基于中分辨率成像光谱仪(Moderate Resolution Imaging Spectroradiometer ，MODIS)的增强型植被指数(Enhanced Vegetation Index，EVI)时序遥感数据，提出了中国草地二级类型的分类方法，实现了基于草地类型的草地覆盖度的遥感估算。在此基础上，进一步对中国草地类型和草地覆盖度的时空格局展开分析。\n论文的主要成果和创新点包括：\n（1）基于MODIS_EVI时序数据，构建草地生长季EVI特征曲线。提取曲线的傅立叶组分特征并计算相似性。以反映各类型草地物候特征的草地生长季EVI特征曲线相似性为参量，同时辅以影响草地类型形成的自然因子特征，建立草地二级类型分类决策树。以涵盖所有草地二级类型的甘肃省为试验区，利用1：500 000中国草地资源类型图对分类方法进行验证和评价。与经典的最大似然法监督分类结果相比，总体精度提高了7.93％，kappa系数提高了0.06。进而在全国范围内以省级行政区域分别建立决策树，实现2005年中国草地二级类型分类。各省级区域的总体分类精度均在60％以上，其中，西藏、新疆、内蒙古、青海、甘肃等几个草地资源大省，总体分类精度在70％以上。\n（2）针对不同的草地类型的植被覆盖差异，建立分区、分草地类型的等密度和非等密度混合像元分解草地覆盖度估算模型，按照同一思想按省级行政单位分别选取模型参数，实现了区域可比性强的全国范围的草地覆盖度估算。利用野外实测点和现有研究区域成果，分别在北方地区、南方地区和西部地区进行精度验证，估算结果平均相对误差值均在10.20％以下，相关系数在0.75－0.88之间。\n（3）利用空间统计分析的方法，揭示中国各类型草地及草地覆盖度的数量特征和空间分布特征。按面积从大到小排列，高寒草甸>典型草地>荒漠草地>高寒草原>草甸草地>灌丛草地。高寒草甸、高寒草原、荒漠草地在空间上呈聚集分布，草甸草地、典型草地在全国范围呈分散分布。灌丛草地在华北、华中及南方地区分散分布。2005年中国草地覆盖全国平均值为29.60％。八个草地资源大省及自治区均排在低草地覆盖度的前列。通过绘制箱须图，揭示草地类型的覆盖度特征。六种类型草地的平均覆盖度由小到大依次为荒漠草地<高寒草原<典型草地<高寒草甸<草甸草地<灌丛草地，覆盖度的四分位间距分别为6－11％、20－25％、19－38％、27－49％、34－55％及46－59％。\n（4）通过测算不同等级覆盖度草地动态度、综合草地动态度指数及转移矩阵，揭示中国20世纪80年代－2005年的草地覆盖度动态特征。20世纪80年代－2005年期间，草地覆盖度呈现出先减少、再好转、继而保持平稳状态的发展趋势。各等级覆盖度草地之间的转换规律也显示，在2000年以后草地的质量相比2000年以前发生好转，整体覆盖度变高。各覆盖度等级草地的增加，主要转变来源是耕地、林地、未利用地的转变。2000年后，我国耕地向草地转换、草地向林地转换的程度都有所增加，肯定了我国“退耕还林、还草”工程在生态环境保护方面发挥的作用。高、中、低覆盖度草地最主要的被占用原因是耕地开垦。\n（5）选取全面反映景观特征的8项景观指数，分析草地景观的空间异质性特征。在全国草地景观中占有明显的优势地位的是高寒草原和高寒草甸，很大程度上控制着中国草地生态环境的状况，且边界褶皱程度高，相对受人为因素干扰较少，较好的保持了自然状态。最不占优势的草地类型是灌丛草地，破碎度最高、空间聚集度最低，几何形状较为规则，受人为因素干扰最为严重。\n（6）对草地类型与气温、降水、海拔、坡度和土壤类型为生境因子进行空间叠加分析，定量的揭示各类型草地的生境特征。各类型草地的水热组合特征、地形特征、土壤类型分异明显，加之各类型草地覆盖度的四分位间距的特征分析，对基于遥感的草地类型深入研究具有指导意义。 |
| 英文摘要 | Grassland ecosystem is the most important and most widely distributed type of ecosystem in terrestrial ecosystems. Together with forests and oceans, they are considered as the Earth's three major carbon pools. So grassland plays very important role in global changes. As the basic properties of grass and intuitive quantitative indicators, the grassland type and fractional coverage are important control factors or basic input variables in the process of analysis and modeling in hydrology, meteorology, ecology research. They are two basic indicators to reflect the status of grassland resources. However, in research fields of grassland resources, more practical significance studies such as the grass biomass or the grass degradation are concerned more. Systematic study on the grassland type and fractional coverage are rare in China. However, further study on the basic indicators of grassland is significant to realize precision utilization and protection of grassland resource. Therefore, the paper takes the grassland type and fractional coverage in China as the study object. Basing on the time-series images of Moderate Resolution Imaging Spectroradiometer( MODIS) Enhanced Vegetation Index( EVI), six types of grassland are classified and the fractional coverage of grassland is estimated. On this basis, spatio-temporal pattern of the grassland type and fractional coverage in China are analyzed. \nThe main results and initiatives are as follows: \n(1) Basing on time-series MODIS_EVI data, the growth characteristic EVI curve are constructed. Extract the Fourier components of the curve and calculate the similarity of the growth characteristics. Take the similarity as the parameters, assistant with the natural factors impacting the formation of vegetation types, the grassland type classifying decision tree is constructed. Take Gansu Province, which contains all types of grassland of China, as study case to verify and evaluate the classification precision, under the assistant of the 1:500 000 scale grassland resources map in China. Comparing with the classical Max likelihood classification, decision tree classification improves the overall accuracy by 7.93%, and improves kappa coefficient by 0.06. And then establish decision tree across the country in provincial-level administrative regions, respectively, and realize the classification of grassland types in China. The overall classification accuracy in each province is above 60%.\n(2) Considering the coverage differences among different types of grassland, the density and non-density sub-pixel coverage estimation model are established via different grass types. As the model parameters in each province were selected under the same thinking, this national estimation of grass fractional coverage has strong regional comparability. Accuracy verification carried out in the north, south and western China by the measured field and the existing results show that the average estimation error is under 10.20％, with correlation coefficient between 0.75-0.88.\n(3) Spatial statistical analysis reveals the number and spatial distribution characteristics of each type and fractional coverage of the grassland in China. According to area from large to small, each type of grassland can be arranged as high-cold meadow steppe> typical steppe> desert steppe> high-cold typical steppe>meadow steppe> shrub herbosa in turn. High-cold meadow steppe, high-cold typical steppe and desert steppe are cohesion distributed in space, while meadow steppe and typical steppe distributed sparsely nationwide, shrub herbosa distributed sparsely in the north, central and south China. The average fractional coverage of grassland in China is 29.60％ in 2005. Eight provinces and autonomous regions which owe the largest number of grassland are all with the lowest coverage. Through drawing box-plot map, fractional coverage quartile range characteristics of each grassland types are revealed as desert steppe: 6-11％, high-cold typical steppe: 20-25％, typical steppe: 19-38％, high-cold meadow steppe: 27-49％, meadow steppe: 34-55％ and shrub herbosa: 46-59％.\n(4) Through calculating the dynamic degree of grassland fractional coverage, comprehensive grassland dynamic index and transfer matrix, the dynamic characteristics of the grass fractional coverage in China from 1980s to 2005 are exposed. During this period, grassland fractional coverage decreased at first, and then turned for the better around the year 2000, and then maintained a steady state. The main source of the grassland increase is cultivated land, forest land and unused land. After the year 2000, the conversion extents of both cultivated land to grassland and grassland to forest land conversion increased, which affirms the effect of \ |
| 中文关键词 | 中国 ; 草地类型 ; 覆盖度 ; 遥感 ; 决策树 ; 时空格局 ; 生境 |
| 英文关键词 | China grassland type fractional coverage remote sensing decision tree spatio-temporal pattern habita |
| 语种 | 中文 |
| 国家 | 中国 |
| 来源学科分类 | 地图学与地理信息系统 |
| 来源机构 | 中国科学院遥感应用研究所 |
| 资源类型 | 学位论文 |
| 条目标识符 | http://119.78.100.177/qdio/handle/2XILL650/286808 |
| 推荐引用方式 GB/T 7714 | 温庆可. 中国草地类型、覆盖度遥感划分及时空格局研究[D]. 中国科学院遥感与数字地球研究所,2009. |
| 条目包含的文件 | 条目无相关文件。 | |||||
| 个性服务 |
| 推荐该条目 |
| 保存到收藏夹 |
| 导出为Endnote文件 |
| 谷歌学术 |
| 谷歌学术中相似的文章 |
| [温庆可]的文章 |
| 百度学术 |
| 百度学术中相似的文章 |
| [温庆可]的文章 |
| 必应学术 |
| 必应学术中相似的文章 |
| [温庆可]的文章 |
| 相关权益政策 |
| 暂无数据 |
| 收藏/分享 |
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。
