基于地面高光谱遥感的科尔沁沙地植被盖度估算及生物量模拟

发布时间：2017-12-31 21:21

本文关键词：基于地面高光谱遥感的科尔沁沙地植被盖度估算及生物量模拟　出处：《内蒙古农业大学》2016年博士论文　论文类型：学位论文

【摘要】：植被的光谱参数和定量模型是对其生长实时监测与精确诊断研究的关键内容。随着人们对生态环境的持续关注,生态环境监测、评估、治理等方面的发展迫切需要低成本、低耗时、高精度、高密度和高稳定性的信息获取技术。而基于地面高光谱遥感技术的植被生态应用使得快速监测植被长势以及准确估测植被理化性质成为可能。本文以我国内蒙古自治区通辽市科尔沁左翼后旗东南方向的阿古拉生态水文试验区为研究区域,利用美国Analytical Spectral Device (ASD, Inc., Boulder, CO,USA)公司的FieldSpec HandHeld便携式地物光谱仪开展了一系列典型植被野外观测试验,并在此基础上设计控制试验,评估了异向反射率效应(Anisotropic Reflectance Effect, ARE)对基于光谱混合分析(Spectral Mixture Analysis, SMA)的植被盖度反演估算的影响,最后对试验区两个天然草地试验点进行了高光谱植被指数模拟地上干、鲜生物量的研究。取得的主要成果如下：1.由于植被叶片各种色素、液态水分、内部结构的差异,不同植被类型的反射率光谱存在差异性,但也会有“异物同谱”现象的发生；外界环境对植被生长的影响是十分显著的,植被生长情况的差异可以反映在其冠层反射率光谱曲线上,从而引发了“同物异谱”的现象。2.在不考虑异向反射率效应的情况下,运用线性光谱混合分析(分别使用可见光波段和近红外波段)对三个试验寸苔草盖度的估算效果并不理想,可见光波段和近红外波段的均方根误差RMSE分别高达0.160和0.164。考虑异向反射率效应后,利用近红外波段对寸苔草盖度估算的精度提升明显,均方根误差RMSE由0.164降低到了0.079,降幅高达52%。未来的植被盖度估算研究应该重视异向反射率效应,因为它可以认为是同种端元类型光谱差异性的另一重要来源。3.天然条件下,即使对于具有同一植被类型的区域,其范围内的植被参数也会有较大的空间异质性,尤其在土壤、水分以及地形等局部环境因子存在较大差异的条件下,此现象更为明显。因此通常研究所做的关于特定植被类型叶面积指数和地上生物量等植被参数在一定空间上具有较小异质性的假设是不正确的。4.试验区两个天然草地地上干、鲜生物量与其单波段原始高光谱冠层反射率相关关系很弱,C4试验点最优决定系数R20.202,而C3试验点受其植被群落结构空间上分层的影响,最优决定系数R20.062。5.基于原表达式波段范围构建的四个植被指数(简单比值植被指数、归一化植被指数、土壤调节植被指数和增强型植被指数)与地上干、鲜生物量的最佳相关系数明显低于基于全波段范围构建的植被指数：结合多元线性逐步回归分析对构建出来的四个最佳植被指数进行筛选,对于C3试验点地上干生物量,挑选出归一化植被指数和增强型植被指数,决定系数R2为0.732,均方根误差RMSE为57.5g/m2,而对于C3试验点地上鲜生物量以及C4试验点地上干、鲜生物量,选出的植被指数都是土壤调节指数和增强型植被指数,其决定系数R2分别为0.769,0.676,0.693,对应均方根误差RMSE分别为184 g/m2、65.8 g/m2、134g/m2。
[Abstract]:The spectral parameters and the quantitative model of vegetation growth is the key content of real-time monitoring and accurate diagnosis of its study. As people continue to focus on the ecological environment, ecological environment monitoring and assessment, development, treatment and other aspects of the urgent need for low cost, low consumption, high precision, high density and high information acquisition technology and ecological stability. The application of ground vegetation based on hyperspectral remote sensing technology makes rapid monitoring vegetation growth and accurate estimation of vegetation physicochemical properties become possible. The eco hydrological agula test area based on China's Tongliao the Inner Mongolia Autonomous Region city Kerqinzuoyihouqi southeast as the study area, using the United States Analytical Spectral Device (ASD, Inc., Boulder, CO, USA) FieldSpec HandHeld portable spectrometer, carried out a series of typical vegetation field test, and on the basis of the design of control test, evaluation Estimation of anisotropic reflectance effect (Anisotropic Reflectance Effect, ARE) of spectral mixture analysis (based on Spectral Mixture Analysis, SMA) to evaluate the effect of vegetation coverage inversion, at the end of the two natural grassland experiment test areas were Hyperspectral Vegetation Index simulation on dry, fresh biomass. The main achievements are as follows: 1. the vegetation leaf pigment, liquid water, the difference of internal structure, the reflectance spectra of different vegetation types are different, but there will also be "objects with the same spectrum" phenomenon; effect of external environment on the growth of vegetation is very significant, the differences can be reflected in the growth of vegetation canopy reflectance spectra the curve, which led to the "synonyms spectrum" phenomenon of.2. without considering the anisotropic reflectance effect, using linear spectral mixture analysis (respectively using visible light And the near infrared band) to estimate the effect of three inch test coverage of Carex is not ideal, the root mean square error of RMSE visible and near infrared bands were as high as 0.160 and 0.164. considering the anisotropic reflectance effect, estimation of coverage by near infrared band Carex inch accuracy increase significantly, root mean square error by RMSE 0.164 reduced to 0.079, a decline of up to 52%. estimation of vegetation coverage in future should pay attention to the different reflection effect, because it can be considered the same end another important source of.3. natural condition difference spectral element types, even to have the same vegetation area, vegetation parameters within its range will be large spatial heterogeneity, especially in soil, moisture and terrain of local environmental factors of different conditions, this phenomenon is more obvious. So usually the research done on specific vegetation types Leaf area index and aboveground biomass of vegetation parameters with smaller heterogeneity in a certain space on the assumption that the.4. is not the correct test area of two natural grassland aboveground dry biomass, fresh and original single band hyperspectral canopy reflectance correlation is very weak, C4 decided to test the optimal coefficient of R20.202 and C3. The test point affected by the spatial structure of vegetation community on the stratification, optimal decision coefficient R20.062.5. four vegetation index based on the scope of the original expression band (ratio vegetation index, vegetation index, soil adjusted vegetation index and enhanced vegetation index) with dry, the best correlation coefficient of fresh biomass was significantly lower than that of vegetation full wave band based on the index: the combination of multiple linear stepwise regression analysis of the four best vegetation index constructed were screened for the C3 test on dry biomass, selection The normalized difference vegetation index and enhanced vegetation index, the coefficient of determination R2 was 0.732, the root mean square error of RMSE 57.5g/m2, and C3 test for aboveground biomass and C4 test on dry, fresh biomass and vegetation index are selected vegetation index and enhanced vegetation index, the coefficient of determination R2 were 0.769,0.676,0.693, corresponding to the root mean square error of RMSE were 184 g/m2,65.8 g/m2134g/m2.

【学位授予单位】：内蒙古农业大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：Q948

【相似文献】