【作者】 罗仙仙；

【导师】 亢新刚； 杨华；

【作者基本信息】 北京林业大学 ， 森林经理， 2010， 博士

【摘要】 中国林业正处在全面实施以生态建设为主的林业发展战略,加速推进传统林业向现代林业转变,着力构建林业生态和产业体系,促进林业又快又好发展的重要时期。我国林业监测体系中不同的监测项目是在一定的社会经济和需求条件下产生和发展壮大的,各自具有其自身的特点,每种监测都有自已明确的目标、监测内容、监测方法、监测周期。根据社会形势对监测工作的需求,在不同尺度上,如区域(省)级及经营单位级层次上,如何整合现有的监测项目,如何构建统一的森林资源综合监测指标,建立与监测对象相适应的地面样地布设方案,其中抽样设计是关键。同时,针对不同指标匹配抽样调查方法,对改进和完善现有的监测体系有重要理论研究价值与实践意义。本论文是在国家“十一五”科技支撑课题“森林资源综合监测指标与技术体系研究(编号2006BAD23B01)”资助下进行的,其工作在于：(1)从核心概念出发,比较国内外森林资源清查体系特点,同时分析了世界主要林业发达国家森林资源综合监测现状。在分析我国森林资源监测现状与存在问题的基础上,针对我国森林资源监测的实际情况与社会形势对监测工作的需求,指出我国森林资源综合监测的发展趋势和建设目标为监测内容多样化、监测周期年度化、监测技术标准化、监测手段一体化、监测信息共享化。(2)抽样调查技术是森林资源综合监测关键技术。简略阐述抽样调查的历史背景。详细进行抽样理论的分类与比较。在分析国内外森林资源综合监测抽样调查体系的基础上,根据Michael et al.的抽样分类系统,重点综述了不同抽样方法在我国森林资源综合监测当中的应用现状。为实现我国现阶段监测的新需求,对如何改进和完善我国现有的抽样调查体系进行探讨。(3)提出了综合监测指标体系框架,总结了用于遥感监测分类方法的最近邻域分类法(k-NN分类法)和改进最近邻域分类法(ik-NN分类法),同时,提出了三阶段分层抽样设计,为森林资源综合监测提供抽样理论基础。(4)在介绍临界距离、角规常数、样地半径系数等概念基础上,对点抽样的总量估计、胸高断面积估计、总体大小估计、均值估计、方差和方差估计以及乘积估计进行系统建模。同时,对双重点抽样方法进行系统推导,可为建立经营单位级森林资源监测提供抽样技术指导。(5)为实现我国现阶段监测的新需求,对不同类型线截抽样(包括直线型、L型、Y型)的总体总值、密度、均值的条件和无条件估计,以及方差、估计区间进行系统推导,为森林资源综合监测提供抽样技术指导。(6)建立了直线型线截抽样的倒木蓄积总值、密度、均值、长度,以及估计精度、方差、估计区间等计算公式。在检查法的基础上,根据现有森林生态系统监测需求,应用直线型线截抽样原理对Ⅰ大区4小区进行了倒木调查的实验研究。分析表明：截线长度是影响精度的主要因子,不能把所有截线长度累加成用一条等长度的截线进行调查,这样严重影响样本选取的概率,对统计估计不利。(7)分析了倒木调查因子与截线长度的关系。分析表明：倒木蓄积总量的估计精度随截线的长度增加而增加；倒木总条数估计值和倒木蓄积密度随截线的长度增加先增大后减少,当截线长度区间为[50,100]m时,倒木总条数估计值和倒木蓄积密度相对稳定变化。从而确定直线型线截抽样的最优截线长度为50 m,此时,实验区倒木蓄积总量为536.85 m3,蓄积密度为27.53 m3·hm-2,倒木总条数为5887条,抽样精度达90.05%(可靠性为95%),为森林资源综合监测地面样地设计提供理论与实践依据。(8)对世界主要国家的森林资源清查地面样地设计进行了述评,包括样本单元的构成、样地大小、形状、样本单元中树木的选择方法、调查因子和监测周期。发现样本单元多采用群团样地或样地群,圆形样地应用较多,调查因子除森林资源外,还包括森林健康、生物多样性、碳和土壤等。建议在现有清查体系样地的基础上,布设小样方或样圆、样线等,扩展新的调查因子。(9)在假定没有样地边缘效应基础上,对基于固定面积嵌套样地抽样的资源现状进行理论探讨,在检查法的基础上,对Ⅰ大区4小区的系统样地内嵌套布设单个天然更新小样地(10 m×10 m),并进行幼苗幼树更新调查。利用所建数学模型,对研究区域的天然更新株数进行统计估计,其结果为：区域内云杉、冷杉、其它树种及总体幼苗幼树更新公顷株数无偏估计分别为420株·hm2、2251株·hm2、4501株·hm2、7178株·hm2,从而为天然林数量化经营提供了一个新的视角。同时,在假定全部固定样地的基础上,对净增量一般化表达、方差及其估计,分子总体净增量的估计、方差及方差估计进行推导,为森林资源综合监测提供理论依据。总之,所建立的三阶段分层抽样方法、点抽样方法、线截抽样方法以及基于固定面积嵌套样地的资源动态估计对改进我国森林资源监测体系提供科学理论依据。更多还原

【Abstract】 Forestry in China is implementing its developing strategy which focuses on ecological construction in an all-around way, accelerating the transform from the traditional forestry to a modern one. It is experiencing an important period when efforts are concentrated on the construction of forestry ecology and industrial system for a rapid and steady progress. Various monitoring projects belonging to the forestry monitoring system have come into being and developed under certain social economic and demanding conditions, having their own characteristics. Every monitoring project has its own clear aim, content, method and period. According to the demand of social situation for the monitoring work, at different scale (e.g. the scale of an area or a province and its management unit), sampling design is crucial for the unification of the present monitoring projects, the construction of a universal and comprehensive monitoring indicators of forest resources and the creation of the ground plot scheme appropriate to the monitoring objects. At the same time, matching different indicators with their corresponding sampling methods has important research value and practical significance as well. This paper is aimed at:(1) From the core concept, the inventory and monitoring methods were introduced, and characteristics of national forest inventory at home and abroad was compared. At the same time, the current condition of forest resources monitoring in mainly developed forestry countries was analyzed. Based on the current condition and existing problem in China, according to the actual condition in forest resources monitoring and the monitoring need from the social practice, the development trend and construction goal of forest resources comprehensive monitoring in China was pointed out, that is, the diversification of monitoring content, annually monitoring period, the standardization of monitoring technology, the integration of monitoring measures and the sharing of monitoring information.(2) Sampling survey technique is key to forest resources comprehensive monitoring. The historical of sampling survey was briefly introduced. Classification and comparison of sampling theory were carried out in detail. Based on the analysis of forest resources comprehensive monitoring at home and abroad, according to the sampling classification of Michael Kohl, et al., the application condition of different sampling method on forest resources comprehensive monitoring was reviewed. In order to meet the new need of the current monitoring, improving the sampling survey system in China was discussed.(3) The indicators system framework of comprehensive monitoring was put forward. The classification method(k-NN, k-Nearest Neighbor and ik-NN, improved k-Nearest Neighbor) of remote sensing was summarized. At the same time, three-stage sampling for stratification was put forward, it would provide the theoretical basis for comprehensive forest resources monitoring.(4) Based on the some concepts, such as limiting distance, basal area factor, plot radius factor, estimation of population total, basal area, population size, mean, variances and product of point sampling was systematically built. At the same time, the method of double point sampling was systematically deduced, it would provide the sampling guidance for setting up the forest resources monitoring at the management sacle.(5) To meet the new need at present, the conditional estimation and unconditional estimation of population total, density and mean, at the same time, variance, estimation interval and precision of various kinds of line intersect sampling(including straight-line, Ell-shaped transects and Y-shaped transects) was systematically deduced so that it can provide the scientific proof for forest resources comprehensive monitoring.(6) The formula of calculation of total volume, density, mean and total number of log, estimation precision, variance and estimation interval was systematically deduced. Based on the method control, according to the need of the forest ecological system at present, the straight-line intersect sampling was applied in the inventory of log in the fourth small section of I section. The results show that the lengthen of transect is a main factor impacting the precision, A inventory could’t be carried out by one transect with the length equivalent to the total length of all transects, because the probility of selecting sample will be affected too much, and it was not good for statistical estimation.(7) Relationship between the inventory factors of logs and the length of transect was analysed. The analysis shows that:the precision of the total volume of logs increased with the increase length of transect; Both the total number of logs and the density of volume firstly increased and then decreased with the increase length of transect. When the interval of the length of transect arrived at [50,100] m, they relatively stable. In the end, the best transect length was determined 50 meter, then the total volume of logs in the research area was 536.85 m3, the density of volume was 27.53 m3·hm-2, the total number of logs was 5887 and the sampling precision gained 90.05%(reliability was 95%), it can provide the scientific and practical proof of ground plot design for forest resources comprehensive monitoring.(8) Ground Plot Design techniques for National Forest Inventory of major countries in the world are reviewed. They cover sample unit, plot configuration, plot size, selection of trees in sample unit, and investigation variables and interval. Clustered circular plots are commonly adapted. Forest health, biomass, biodiversity and soil carbon are also included in the investigation variables besides normal ones. Based on the plots of national forest inventory, Small plots including rectangle, circle, line, and so on should be configurated to get more information.(9) On the hypotheses of no edge effect of plots, estimation following fixed area plot sampling was carried out. Singular and small plots(10 m x 10 m) were systematically nested in the plots of the method control, and regeneration survey was carried out. According to the builded model, the number of seedings and small trees was estimated. The unbiased estimation of the number per hectare of spruce, fir, other tree species and total seedlings was 420,2252,4501,7178 respectively.At the same time, on the basis of all the fixed plots, the general expression of increment, variance and its estimation were built, as well as the increment of different domains. It has provided theoretical evidences for the comprehensive monitoring of forest resource.In summary, three-stage sampling for stratification, point sampling, line intersect sampling and dynamical estimation of resources based on fixed and nested plot was buildt, they will benefit to improve the forest resources monitoring system.更多还原