基于不同遥感数据源的毛竹林地上部分生物量反演

发布时间：2018-03-05 15:54

本文选题：毛竹　切入点：生物量　出处：《安徽农业大学》2016年硕士论文　论文类型：学位论文

【摘要】：毛竹在我国是分布最广、面积最大的竹种,具有速生丰产、用途广泛、再生能力强、经济价值高和可持续更新等特点。据不完全统计,我国毛竹林面积386.83万ha,约占竹林面积的70%,占全世界竹林面积20%,在维护生态平衡方面发挥明显的作用。借助遥感技术对毛竹生物量的研究将为毛竹固碳能力的研究提供基础数据。LiDAR遥感技术可以获得植被高精度、高密度的三维坐标数据,并可构建植被的三维立体模型,进而反演植被生物量。将LiDAR技术应用在毛竹生物量遥感估测上将为今后毛竹林生物量估测提供更多手段。本文立足于研究机载LiDAR数据与机载高光谱数据分别反演毛竹林地上部分生物量的可行性,并且比较反演精度。以安徽省黄山市作为飞行区域,获取机载LiDAR数据与机载高光谱数据;在飞行航迹内调查50块(有效44块)毛竹林样地并计算生物量。分别提取样地范围内不同遥感数据特征变量作为自变量,样地生物量作为因变量,建立基于不同遥感数据源的反演模型。比较分析两种反演模型精度的原因。主要研究结论如下:(1)机载LiDAR数据经过归一化处理消除了地形因子的影响;点云分类使用软件分类与手工编辑的方法,区分开了地面点,植被点以及噪声点;点云统计定义高于地面2m的点为毛竹林反射点,避免了杂灌等植被的影响。因此用于提取变量的点全部是毛竹林的反射点。(2)机载LiDAR数据经过预处理在ENVI IDL模块下编程统计点云信息作为自变量,地面调查获取的毛竹林生物量作为因变量,使用SPSS 22软件进行多元线性回归分析可以建立反演模型f1:InW=5.024+0.101×Inh50+0.226×Inhmax-0.318×Ind15+0.582×Inc,该模型可解释生物量64.13%的变动。这表明借助机载LiDAR技术反演毛竹林地上部分生物量可行。(3)机载高光谱数据经过处理,使用ENVI软件提取位置变量、面积变量、植被指数、原始波段以及地面调查样地平均高作为自变量,地面调查获取的生物量作为因变量,使用SPSS 22软件多元线性回归分析的方法建立反演模型f2:W2=1.514+0.765×Dr+4.324×SDr-1.602×VI2+0.937 hmean,,该模型可解释毛竹林生物量58.3%的变动。表明借助机载高光谱技术反演毛竹林地上部分生物量是可行的。(4)机载LiDAR数据与机载高光谱数据提取特征变量配合地面调查数据建立毛竹林地上部分生物量反演模型是可行的,通过比较各自模型的决定系数(R2)、复决定系数(Ra2)、绝对均方根误差(RMSE)以及自相关性检验(DW)发现机载LiDAR数据建立的反演模型精度优于机载高光谱数据建立的反演。
[Abstract]:Phyllostachys pubescens is the most widely distributed and the largest bamboo species in China. It has the characteristics of fast growing and high yield, wide use, strong regeneration ability, high economic value and sustainable renewal, etc. According to incomplete statistics, The area of Phyllostachys pubescens forest in China is three million eight hundred and sixty-eight thousand and three hundred ha. about 70% of the total area of bamboo forest, accounting for 20% of the world's bamboo forest area, which plays an obvious role in maintaining ecological balance. Provide basic data. LiDAR remote sensing technology to obtain high accuracy vegetation, High-density three-dimensional coordinate data, and can build three-dimensional vegetation model, The application of LiDAR technology to the remote sensing estimation of bamboo biomass will provide more means for estimating the biomass of Phyllostachys pubescens in the future. This paper is based on the study of airborne LiDAR data and airborne hyperspectral data respectively. Feasibility of aboveground biomass of Phyllostachys pubescens forest, Taking Huangshan City, Anhui Province as flight area, the airborne LiDAR data and airborne hyperspectral data are obtained. In flight track, 50 plots (44 effective plots) were investigated and biomass was calculated. The characteristic variables of different remote sensing data were extracted as independent variables and biomass of sample plots as dependent variables. The inversion models based on different remote sensing data sources are established, and the reasons for the accuracy of the two inversion models are compared and analyzed. The main conclusions are as follows: 1) the airborne LiDAR data are normalized to eliminate the influence of terrain factors; Point cloud classification uses the method of software classification and manual editing to distinguish ground points, vegetation points and noise points, and point cloud statistical definition above 2 m above the ground is the reflection point of Phyllostachys pubescens forest. Therefore, all the points used to extract the variables are the reflection points of bamboo forest. The airborne LiDAR data are pre-processed and programmed under the ENVI IDL module to calculate the point cloud information as independent variables. The biomass of Phyllostachys pubescens forest obtained by ground survey was taken as dependent variable. The inversion model f1: Inw5.024 0.101 脳 Inh50 0.226 脳 Inhmax-0.318 脳 Ind15 0.582 脳 Inc can be established by using SPSS 22 software for multivariate linear regression analysis. This model can explain the variation of biomass 64.13%. This indicates that it is feasible to retrieve aboveground biomass of Phyllostachys pubescens forest by using airborne LiDAR technique. The spectral data are processed, Using ENVI software to extract location variables, area variables, vegetation index, original wave band and average height of ground survey samples as independent variables, and biomass obtained from ground survey as dependent variables. The inversion model f2: W2t1. 514 1. 765 脳 Dr 4. 324 脳 SDr-1.602 脳 VI2 0. 937 hMeV was established by using SPSS 22 software multivariate linear regression analysis. The model can explain the variation of biomass 58.3% of Phyllostachys pubescens forest. It shows that the airborne hyperspectral technique can be used to estimate the aboveground biomass of Phyllostachys pubescens forest. Row. 4) it is feasible to establish a model of aboveground biomass inversion of Phyllostachys pubescens forest by extracting characteristic variables from airborne LiDAR data and airborne hyperspectral data combined with ground survey data. By comparing the determination coefficients of their respective models, the complex determination coefficients, the absolute root mean square error (RMSE) and the autocorrelation test (DW), it is found that the inversion model built by airborne LiDAR data is more accurate than that established by airborne hyperspectral data.
【学位授予单位】：安徽农业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：S795.7

【参考文献】