Knowledge Resource Center for Ecological Environment in Arid Area
| 稀疏高寒草原植被盖度遥感反演方法研究 | |
| 其他题名 | The Method Research About Vegetation Coverage of Sparse Alpine Grassland By Remote Sensing Inversion |
| 李磊磊 | |
| 出版年 | 2015 |
| 学位类型 | 硕士 |
| 导师 | 范建容 |
| 学位授予单位 | 中国科学院大学 |
| 中文摘要 | 植被覆盖度是刻画地表植被覆盖的重要参数，被用于水文、气候、植被产量、土地资源管理等领域，因此准确获取植被覆盖度意义重大。目前针对干旱区的稀疏植被研究存在一定进展，而在高海拔寒冷地区的稀疏高寒草原的覆盖度研究相对匮乏。日喀则地区地处青藏高原地带，受低温环境条件的影响，高寒草原的生长季节较短，草群稀疏、低矮，生物产量明显较低，覆盖度低。低覆盖度地区是潜在荒漠化区域。因此对该区域的植被覆盖度进行研究极具有科学价值。由于针对低覆盖度的遥感提取的研究较少，本研究诣在探索稀疏高寒草原的覆盖度提取，主要聚焦在两个方面，一是如何准确的提取高寒草原类型；二是如何选取模型和数据准确估算出植被覆盖度。本研究取得的主要成果如下： （1）通过野外多光谱相机实测和数码照相法估算，提取样方覆盖度，建立实测覆盖度和实测植被指数的回归模型。野外进行大量分散采集，实测不同盖度的高光谱数据，证实低盖度的样方的光谱仍具有正常植被所具有的明显特征但光谱较弱，进行植被指数特征应用具有意义。 （2）基于地表覆被分类的植被覆盖度提取方法，可有效剔除了非植被区域，较传统方法反演精度有很大提高。本研究选取Landsat8和Landsat7影像，通过决策树和神经网络分类法对研究区进行分类。其中决策树分类法是采用面向对象分类方法进行切割，基于影像的光谱特征，影像植被指数和高程等信息建立决策树。神经网络分类是基于ENVI软件选取样本，采用三层神经网络结构进行遥感分类，提取稀疏高寒草原。其中决策树法分类精度较高，达到85.3%。 （3）基于提取的高寒草原类型，选取基于野外实测盖度的像元二分模型、改进的三波段最大梯度差法和基于NDVI和SAVI指数的回归模型法，估算稀疏高寒草原的覆盖度。同时本文对比野外实测的样方覆盖度数据进行检验，评价最优提取方法。根据结果分析，研究区内稀疏高寒草原覆盖度的总体精度约为82%，像元二分法效果较好。稀疏高寒草原的平均覆盖度为37.6%。 |
| 英文摘要 | Vegetation coverage, an important parameter to characterize the surface vegetation, was usually used in the field of hydrology, climate, vegetation and land resource management research. Therefore it had a great significant to obtain accurate vegetation coverage. The middle Brahmaputra river were in the area of the Qinghai Tibet Plateau, which owns a special climatic environment. Influenced by the low temperature, alpine grassland had a short growing season, which leaded the grass sparser, the biomass less and the coverage lower than the usual. Meanwhile, low coverage area is the potential desertification area. So the research of vegetation coverage is of great scientific value. Due to the little research of low coverage by remote sensing before, the purpose of this study is to explore the extraction of sparse grassland coverage. The paper mainly focused on two aspects, one is how to accurately extract the type of alpine grassland; the other is how to select appropriate models and data which can estimate vegetation coverage accurately. The main results were as following: Firstly, through the multi spectral camera and digital photography measurement method, the research estimated the sample coverage, make the regression model between the measured coverage and the measured vegetation index. Meanwhile, it proves that low coverage vegetation owned apparent spectral characteristics although weaker than normal vegetation spectrum. Secondly, the vegetation coverage extraction method based on land classification, can effectively eliminate the non-vegetation area and make higher accuracy than the traditional method. This study used Landsat8 and Landsat7 images to make classification through the decision tree and neural network method. The decision tree classification method mainly used the object-oriented classification method to cut image in ecognition software. Then we analyzed spectral characteristic, vegetation index and elevation information from the selected sample to build and apply the decision rule tree. We selected samples from raster in ENVI software, builded the three-layer neural network to make supervised classification and extracted the alpine grassland. The total classification accuracy was 85.3%. After the extraction of alpine grassland, we selected three models to estimate the coverage of sparse grassland including binary pixel model, three-band maximum difference gradient model and regression model. Then we compared with the measured coverage of alpine grassland sample to check up and selected the optimal method. The total extraction accuracy is 82 percent. The effect of binary pixel model was best. The average coverage of alpine grassland was 37.6 percent. |
| 中文关键词 | 植被覆盖度 ; 遥感分类 ; 像元二分模型 ; 回归模型 |
| 英文关键词 | vegetation coverage raster classification two-divided pixel model |
| 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 | 中国科学院成都山地灾害与环境研究所 |
| 资源类型 | 学位论文 |
| 条目标识符 | http://119.78.100.177/qdio/handle/2XILL650/287473 |
| 推荐引用方式 GB/T 7714 | 李磊磊. 稀疏高寒草原植被盖度遥感反演方法研究[D]. 中国科学院大学,2015. |
| 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 |
| [李磊磊]的文章 |
| 百度学术 |
| 百度学术中相似的文章 |
| [李磊磊]的文章 |
| 必应学术 |
| 必应学术中相似的文章 |
| [李磊磊]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。
