植被特征尺度与尺度优化研究
发布时间：2019-06-26 08:46
【摘要】：尺度是研究事物或事物现象的空间维和时间维的度量大小。地球空间表面复杂,人们在某一尺度上所建立的模型或总结出的规律,在另一尺度上未必有效或需要修正,不同区域所需空间观测尺度不一致,用单一的观测尺度去衡量整个复杂的地表显然具有一定的局限性;其次不同的研究方法和研究目的,影像的观测尺度也各异,因此,最优观测尺度的选择是有必要的。不同地物有其自身最适宜的观测尺度,并不是越细微越好,只有在最优观测尺度下,才能进行最全面的数据挖掘与探索,数据分析也可以得到事半功倍的效果。本文首先深入剖析了遥感影像中尺度选择的重要性,并简要介绍了几种常用的最优观测尺度选择方法,并以植被冠层像元值和植被遥感反演参量叶面积指数为研究对象,利用高分辨率Google Earth影像数据、1:10万全国土地利用资料和Landsat TM数据为研究数据,对该三种适宜尺度选择方法进行了相应的改进,从而分析了不同区域下各类植被冠层和植被景观最优空间观测尺度。本文的主要工作和结论有:(1)以亚米级高分辨率森林遥感图像为研究对象,基于局部方差,进行植被冠层最优观测尺度的选择。首先基于冠层特征尺度的物理定义,引入局部方差和倒置的指数拟合模型,建立了冠层特征尺度计算模型,并利用美国北部的洛克研究区和南部的梅肯研究区的高分辨率影像进行模型验证,对论文提出的冠层特征尺度模型进行了定量验证分析,发现冠层特征尺度模型值与树林株行距实测值存在密切联系,线性复相关系数达0.95以上。研究结果表明,冠层特征尺度模型是具有普适性与合理性的,论文建立的冠层特征尺度模型为茂盛植被冠层特征尺度定量计算提供了一种新方法。(2)以亚米级高分辨率森林遥感图像为研究对象,基于半方差,进行植被冠层最优观测尺度的选择。首先基于冠层特征尺度的物理定义,引入半方差函数变程参数,计算了不同影像的半方差变程值,并利用美国北部的洛克研究区和南部的梅肯研究区的高分辨率影像进行验证,发现半方差函数计算的变程值与树林株行距存在密切联系,线性复相关系数达0.91。研究结果表明,论文提出的半方差函数的变程值为茂盛植被冠层观测尺度的选择是可行的,但在普适性上没有局部方差好。(3)在植被叶面积指数尺度效应研究基础上,基于景观指数和LAI尺度效应,进行植被景观最优观测尺度的选择。首先选用2000年全国1:10万土地利用资料和Landsat TM数据,建立了聚合指数与LAI尺度效应的经验统计模型,设置LAI尺度效应阈值,根据该统计模型,即确定聚合指数值,从而计算不同LAI尺度效应条件下中国区域的最优空间观测尺度,结果表明不同地表区域所需的空间观测尺度存在巨大差异,并建立了中国区域的最优空间观测尺度先验知识,为智能对地观测的植被景观观测尺度优化提供了一种新的方法。
[Abstract]:The dimension is the measure of the dimension of a space or time dimension that studies the phenomenon of things or things. the spatial surface of the earth is complex, the model or the summarized rule of people on a certain scale is not necessarily effective or needs to be corrected on another scale, and the space observation scale required for different regions is not consistent, It is obvious to measure the whole complex surface with a single observation scale; secondly, the different research methods and the research object, the observation scale of the image is also different, and therefore, the choice of the optimal observation scale is necessary. The most suitable observation scale of the different figures is not the finer the better, and the most comprehensive data mining and exploration can be carried out only under the optimal observation scale, and the data analysis can be obtained with less effort. In this paper, the importance of the mesoscale selection of remote sensing images is deeply analyzed, and several common optimal observation scale selection methods are briefly introduced, and the leaf area index of the vegetation canopy image element value and the vegetation remote sensing inversion parameter is the research object. Using the high-resolution Google Earth image data and 1: 100,000 national land-use data and Landsat TM data as the research data, the three suitable scale selection methods were improved, and the optimal spatial observation scale of the vegetation canopy and vegetation landscape under different areas was analyzed. The main work and conclusion of this paper are as follows: (1) The selection of the optimal observation scale of the vegetation canopy is carried out based on the local variance based on the local variance of the sub-rice-level high-resolution forest remote sensing image. based on the physical definition of the scale of the canopy characteristic, the local variance and the inverted index fitting model are introduced, a canopy characteristic scale calculation model is established, and the model verification is carried out by using the high-resolution image of the Locke research area in the north and the Meiken research area in the south, Based on the quantitative analysis of the canopy characteristic scale model proposed in this paper, it was found that the model value of the canopy characteristic was closely related to the measured value of the line spacing of the forest, and the linear complex correlation coefficient was above 0.95. The research results show that the canopy characteristic scale model is universal and reasonable, and the canopy characteristic scale model established by the paper provides a new method for the quantitative calculation of the canopy characteristic scale of the luxuriant vegetation. (2) The optimal observation scale of the vegetation canopy is selected based on the semi-variance based on the semi-variance of the sub-variance. First, based on the physical definition of the scale of the canopy characteristic, the half-variance function variable-range parameter is introduced, the half-variance value of the different images is calculated, and the high-resolution image of the Mekken study area in the northern part of the United States and the southern region of Meken is used for verification, It was found that the value of the half-variance function is closely related to the line spacing of the forest, and the linear complex correlation coefficient is 0.91. The results show that the variation value of the semi-variance function proposed by the paper is feasible for the selection of the canopy observation scale of the luxuriant vegetation, but there is no local variance in universality. (3) On the basis of the scale effect of the vegetation leaf area index, the selection of the optimal observation scale of the vegetation landscape is carried out based on the landscape index and the LAI scale effect. firstly, using the 2000-year national 1: 100,000 land-use data and Landsat TM data, a empirical statistical model of the aggregation index and the LAI scale effect is established, the LAI scale effect threshold is set, and the aggregation index value is determined according to the statistical model, so as to calculate the optimal spatial observation scale of the Chinese region under different LAI scale effect conditions, the results show that the spatial observation scale required for different surface areas is great difference, and the optimal space observation scale prior knowledge of the Chinese region is established, And provides a new method for the intelligent peer-to-earth observation scale optimization of the vegetation landscape.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：Q948;P237
本文编号：2506069
[Abstract]:The dimension is the measure of the dimension of a space or time dimension that studies the phenomenon of things or things. the spatial surface of the earth is complex, the model or the summarized rule of people on a certain scale is not necessarily effective or needs to be corrected on another scale, and the space observation scale required for different regions is not consistent, It is obvious to measure the whole complex surface with a single observation scale; secondly, the different research methods and the research object, the observation scale of the image is also different, and therefore, the choice of the optimal observation scale is necessary. The most suitable observation scale of the different figures is not the finer the better, and the most comprehensive data mining and exploration can be carried out only under the optimal observation scale, and the data analysis can be obtained with less effort. In this paper, the importance of the mesoscale selection of remote sensing images is deeply analyzed, and several common optimal observation scale selection methods are briefly introduced, and the leaf area index of the vegetation canopy image element value and the vegetation remote sensing inversion parameter is the research object. Using the high-resolution Google Earth image data and 1: 100,000 national land-use data and Landsat TM data as the research data, the three suitable scale selection methods were improved, and the optimal spatial observation scale of the vegetation canopy and vegetation landscape under different areas was analyzed. The main work and conclusion of this paper are as follows: (1) The selection of the optimal observation scale of the vegetation canopy is carried out based on the local variance based on the local variance of the sub-rice-level high-resolution forest remote sensing image. based on the physical definition of the scale of the canopy characteristic, the local variance and the inverted index fitting model are introduced, a canopy characteristic scale calculation model is established, and the model verification is carried out by using the high-resolution image of the Locke research area in the north and the Meiken research area in the south, Based on the quantitative analysis of the canopy characteristic scale model proposed in this paper, it was found that the model value of the canopy characteristic was closely related to the measured value of the line spacing of the forest, and the linear complex correlation coefficient was above 0.95. The research results show that the canopy characteristic scale model is universal and reasonable, and the canopy characteristic scale model established by the paper provides a new method for the quantitative calculation of the canopy characteristic scale of the luxuriant vegetation. (2) The optimal observation scale of the vegetation canopy is selected based on the semi-variance based on the semi-variance of the sub-variance. First, based on the physical definition of the scale of the canopy characteristic, the half-variance function variable-range parameter is introduced, the half-variance value of the different images is calculated, and the high-resolution image of the Mekken study area in the northern part of the United States and the southern region of Meken is used for verification, It was found that the value of the half-variance function is closely related to the line spacing of the forest, and the linear complex correlation coefficient is 0.91. The results show that the variation value of the semi-variance function proposed by the paper is feasible for the selection of the canopy observation scale of the luxuriant vegetation, but there is no local variance in universality. (3) On the basis of the scale effect of the vegetation leaf area index, the selection of the optimal observation scale of the vegetation landscape is carried out based on the landscape index and the LAI scale effect. firstly, using the 2000-year national 1: 100,000 land-use data and Landsat TM data, a empirical statistical model of the aggregation index and the LAI scale effect is established, the LAI scale effect threshold is set, and the aggregation index value is determined according to the statistical model, so as to calculate the optimal spatial observation scale of the Chinese region under different LAI scale effect conditions, the results show that the spatial observation scale required for different surface areas is great difference, and the optimal space observation scale prior knowledge of the Chinese region is established, And provides a new method for the intelligent peer-to-earth observation scale optimization of the vegetation landscape.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：Q948;P237
【参考文献】
相关期刊论文 前1条
1 舒波;;An enhanced landscape aggregation index[J];Journal of Chongqing University(English Edition);2011年04期
，本文编号：2506069
本文链接：https://www.wllwen.com/wenyilunwen/huanjingshejilunwen/2506069.html
