江西省主要树种不同立地等级的地上生物量与不确定性估计

发布时间：2018-01-11 01:02

本文关键词：江西省主要树种不同立地等级的地上生物量与不确定性估计　出处：《中国林业科学研究院》2017年博士论文　论文类型：学位论文

【摘要】：森林立地生产力是森林植被的潜在生产能力,是指导森林经营管理、制定经营决策方案的重要指标,对森林可持续经营有着重要的意义。森林生物量是森林群落在其生命过程中所产生的干物质的积累量,可以作为反映森林立地生产力的指标,并与立地类型和质量息息相关。由于立地条件的多样性或差异性,同一树种在不同区域的生物量估计结果以及生物量估计误差也会随立地质量的不同而变化,忽略立地质量差异引起生物量估计结果以及生物量估计误差不同的结果必然是粗略而不精准的。为了精准获得不同立地条件下的森林生物量及立地生产力的估计,本文将我国江西省第六次和第七次国家森林资源连续清查数据主要树种分为杉木、马尾松、硬阔和软阔四类,对主要树种立地条件进行了样地水平的立地分级;采用异速生物量模型和生物量扩展转换因子(BCEF)模型,估计了区域不同立地条件下森林生物量及其不确定性;探究了林分起源、龄组、密度等因素对森林地上生物量及其各组分估计误差的影响;建立了基于气候和立地因子的江西省主要树种的生物量增量模型,为降低森林生物量估计的不确定性、改进生物量估计模型精度提供技术和方法基础。得到的主要结论如下:(1)采用树高分级法,建立了江西省杉木林、马尾松林、硬阔林、软阔林的优势木树高—胸径模型,用优势木树高等级代表立地等级,将江西省杉木林分立地质量为7个等级,马尾松林分立地质量分为5个等级,硬阔林分立地质量分为5个等级,软阔林分立地质量分为3个等级。通过计算杉木林和马尾松林分在不同立地等级下的生物量均值和误差估计,发现不同立地等级的区域马尾松优势林分地上生物量均值随着立地水平提高而增大,即立地质量越高,森林生物量密度越大。证明了优势木树高分级用作划分立地等级依据的可行性。(2)采用三种异速生物量模型、两种生物量扩展转换因子(BCEF)模型,结合蒙特卡洛模拟法估计了江西省杉木和马尾松在不同立地等级下的地上生物量均值及误差。以地上生物量均值的相对均方根误差的大小作为评价指标,比较并确定了三种异速生物量模型、两种BCEF模型在估计杉木和马尾松不同立地等级下地上生物量均值的最优模型形式。A.三种异速地上生物量模型形式下的杉木和马尾松地上生物量估计结果相比,加入树高变量的单木生物量模型比单独用胸径做变量的模型能够提高估计精度。在同时拥有胸径和树高两种变量存在的时候,马尾松拥有两个参数的模型较有三个参数的模型获得了更好的估计效果。而杉木则是三个参数的模型好于两参数的模型,具体应用时应根据树种选模型形式。异速模型的生物量估计误差在立地平均水平——也就是中间立地等级的误差最小;立地质量越接近平均水平,生物量估计的相对误差越小。B.两种BCEF模型相比较,经验(回归)模型法估计的各立地等级的生物量均值误差相差不大,而用连续函数法估计的各立地等级的生物量均值误差相差较大。异速生物量模型与BCEF模型相比较,在估计区域尺度的生物量均值时,BCEF模型相对误差较异速模型估计误差要小。在估计不同立地等级的生物量时经验(回归)模型要优于异速生物量模型,异速生物量模型优于BCEF连续函数模型,对于包含多种不同立地条件的大尺度区域,BCEF经验(回归)模型法的估计结果更可靠。对于江西省杉木和马尾松而言,BCEF均值在不同立地等级下的差异不大。(3)选择合适的预测变量是建立生物量增长模型的关键,比较各生物地理气候因子的重要性后发现,林木竞争是江西省森林生物量增长的主要影响因素。对于针叶林而言,次要影响因素是气候因子(无霜期天数、年平均温度)、龄组,最后是地形因子(海拔)。对于阔叶林来说,立地等级是次要的影响因子,最后是气候因子。立地等级对于针叶林生物量增量估计影响不大,影响江西省森林(主要树种)生物量增量的气候因子是温度。地形因子对阔叶林影响无显著性,对针叶林影响较小。因此,在用生物量增长估计江西省针叶林立地生产力时,气候因子和龄组可以作为立地指数替代变量;而在估计阔叶林立地生产力时,立地等级似乎是更好的选择。(4)不同起源分类对于杉木林异速生物量模型估计地上及各组分生物量均值影响不大,对于马尾松林人工林生物量均值模型误差要高于天然林分,对马尾松各组分中树枝、树叶影响较大。起源对于异速生物量模型估计误差的影响因树种而异。随着龄组增大,杉木地上生物量及各组分生物量的模型误差估计值是随着龄组增大而降低的。龄组对马尾松异速生物量模型估计误差的影响为幼龄林模型误差最大,其次是中龄林、成熟林、近熟林和过熟林。密度对杉木异速生物量模型地上生物量估计误差影响不大,对于树枝和树叶而言,低密度林分中的生物量均值估计模型误差相较高密度林分及区域水平更大。密度对马尾松异速生物量模型地上生物量估计误差影响不大,对于各组分而言,除树叶在区分密度后生物量均值模型误差较不分类时升高而外,在树干、树皮及树枝中,低密度和高密度林分的生物量均值模型误差较不分类时均降低。在分组分计算马尾松生物量均值时,使用林分密度分类是降低模型误差的好方法。(5)杉木和马尾松在调查数据原始抽样设计和三个抽样间距、三种多起点系统抽样设计下的地上生物量均值及误差值估计的差异不大,每一种抽样方式均能很好的反映江西省杉木和马尾松总体的地上生物量均值。随着抽样间距的增大,抽样单元增多,抽取样地数的减少,相对均方根总误差的绝对值和相对值在上升。这是由于抽样误差的绝对值和相对值均在上升,而模型误差差异不大。相应的,抽样误差在总误差中的占有率也在上升。但是,综合考虑抽样难度和成本,三种不同抽样间距设计及三种多起点系统抽样设计均能很好的反映调查总体的地上生物量均值水平,可以为其他的大尺度区域调查时的系统抽样设计提供参考。
[Abstract]:Forest site productivity is the potential productivity of forest vegetation, is the guidance of forest management, an important indicator of making management decisions, is of great significance to the sustainable management of forest. Forest biomass is the dry matter accumulation of forest community generated in the process of life, can reflect the productivity index of forest site. It is closely related to the site type and quality. Because the site conditions of diversity or differences, the same species and biomass estimation with site quality varied in the estimation of biomass in different regions, ignoring the site quality differences caused biomass estimation results and biomass estimation results of different error is inevitable rough and not precise. In order to obtain accurate forest biomass under different site conditions and estimation of site productivity, this will be China's Jiangxi Province, Sixth And the seventh national forest resources inventory data for Chinese fir trees, pine, hardwood and softwood four kinds of main tree species in the site conditions of site classification sample level; the allometric biomass model and biomass expansion factor conversion (BCEF) model to estimate the regional forest biomass in different site under the condition of uncertainty; explores the stand origin, age group, density factors estimation error influence on biomass and forest biomass increment were established; the model of main tree species in Jiangxi Province Based on the climate and site factors, in order to reduce the uncertainty in the estimation of forest biomass estimation, and provide basic technology methods the accuracy of the model improved biomass. The main conclusions are as follows: (1) the tree classification method, Jiangxi Province set up Chinese fir forest, masson pine forest, hardwood forest, softwood forest the dominant height and DBH. Type, with dominant height level representative site grade in Jiangxi Province, the site quality of Chinese fir forests into 7 grades, masson pine forest site quality is divided into 5 grades, hardwood forest site quality is divided into 5 grades, softwood forest site quality is divided into 3 grades. The mean biomass estimation and error calculation of Chinese fir and Masson Pine Forest in different site level, different site grade regional advantage Forest Aboveground Biomass of Pinus massoniana with mean site level increases, the site quality is high, the forest biomass density is higher. It is proved that the advantages of high grade trees as feasibility of dividing site grade basis. (2) using three kinds of allometric biomass models, two kinds of biomass expansion factor conversion (BCEF) model, Monte Carlo simulation method to estimate the Jiangxi Province, Chinese fir and Masson Pine in different site grade under ground biomass and mean Error. With biomass average relative root mean square error of the size as the evaluation index, and identified three allometric biomass models, two kinds of BCEF model in the estimation of Chinese fir and Masson pine of different site grade under ground biomass model.A. optimal biomass models of mean three different speed on the ground under Chinese fir and Masson pine aboveground biomass estimation results compared to single tree biomass model with tree height variables compared with DBH variable model can improve the estimation accuracy. When there have both tree height and DBH of two kinds of variables, Ma omatsu has two parameter model with three parameters the model get better estimation effect. Chinese fir is the model parameters of the three models are better than the two parameters, the application should choose according to the tree model. The biomass allometric model estimation error in average site Which is in the middle level of site grade error; site quality is close to the average level, the relative error estimation of biomass is smaller.B. two BCEF model comparison, experience (regression) biomass average estimation error model of each site grade difference, the biomass and mean error in continuous function estimation the site grade is larger. The allometric biomass model and the BCEF model are compared in the estimation of biomass average regional scale, BCEF model of relative error is allometric model estimation error is smaller. Experience in the estimation of biomass in different site grade when (regression) model of allometric biomass is better than the model, different the speed of biomass model is better than the BCEF continuous function model for large scale region contains a variety of different site conditions, BCEF (regression) estimation model more reliable results. For Chinese fir and Masson Pine and Jiangxi Province BCEF, mean difference in site grade under little. (3) choose the suitable forecasting variable is key to establish biomass growth model, the importance of bio geographical climate factors found that forest competition is the main influence factors of Jiangxi forest biomass growth. For coniferous forest, the secondary factor is climatic factors (frost free days, the annual average temperature), age group, and finally terrain factors (elevation). For the broad-leaved forest, site grade is a minor factor, finally is climate factor. Site grade for coniferous forest biomass increment estimation has little effect on the impact of Jiangxi province forest (main species) bio climatic factors the temperature increment is. The influence of terrain factors on the broad-leaved forest had no significant effect on small coniferous forest. Therefore, with the growth in biomass estimation in Jiangxi province coniferous forest productivity, climatic factors and age groups Can be used as substitute variables and site index; in the estimation of broad-leaved forest productivity, site level appears to be a better choice. (4) different origin classification for Chinese fir forest biomass allometric models to estimate ground biomass and mean little effect for Pinus massoniana forest plantation biomass average model error is higher than that of natural stand on the branches of Pinus massoniana components in leaves of influence. As for the origin of allometric biomass models, the effect of the estimation error varies among species. With the increasing age of Chinese fir, aboveground biomass and biomass of the model error estimate is reduced with the increase of age. The age group of Masson Pine allometry biomass model estimation error for maximum error model of young forest, followed by forest, mature forest, near mature forest and over mature forest. The density of Chinese fir model of allometric biomass biomass estimation error influence No, the branches and leaves, biomass average low stand density estimation in high density and phase error model of regional level greater density on the model. The allometric biomass of Pinus massoniana biomass estimation error has little effect on the components, in addition to the leaves increased. The distinction between density biomass average model the error is not classified in the trunk, bark and branches, the mean biomass model error of low density and high density stand is not decreased. The classification calculation of Pinus massoniana biomass in the mean grouping, using the stand density is a good way to reduce the error of classification model. (5) in the original survey data of Chinese fir and masson pine the design of sampling and three sampling interval, sample design three different starting point system under ground biomass and mean value estimation error has little difference, each sampling method can well reflect the Jiangxi The average biomass of Chinese fir and Masson pine, overall land. With the increase of the sampling interval, the sampling unit increased, reduce the number of sampling plots, the absolute value of relative root mean square error of the total and relative value is on the rise. This is due to the absolute value of sampling error and relative values are on the rise, while the model error is insignificant. Accordingly, the sampling error share in total error is also on the rise. However, considering the difficulty and cost of sampling, the sampling design of three different sampling space design and three kinds of starting point system can well reflect the overall investigation on the biomass level, can provide a reference for the design of sampling system the investigation of other regions.

【学位授予单位】：中国林业科学研究院
【学位级别】：博士
【学位授予年份】：2017
【分类号】：S718.5

【参考文献】