【作者】 舒娱琴；

【导师】 祝国瑞；

【作者基本信息】 武汉大学 ， 地图制图学与地理信息工程， 2004， 硕士

【摘要】 由于森林主体的乔木个体大、寿命长，其动态变化具有两个显著的特性，一是时间跨度大，二是空间尺度大。正是这两个因素使得林业研究面临着非常大的难题。而虚拟森林环境应用计算机图形学技术和林业科学知识，构建复杂的森林对象，表达及其分析复杂的林分生长规律，其努力超越时间与空间的限制的特性，为林业科学研究提供了一个新型的平台。 构建虚拟森林环境的目的就是为了让森林管理者和科研者们更方便、更高效地经营和研究森林。林业管理者关心的是树木的最终采伐量，而树木采伐量是与树高、胸高直径等基本参数有关的；他们在空间结构上关心的是树木之间的竞争关系，而竞争关系又主要体现在树木的位置、树高、胸径、冠幅、冠长等几何特征量上。林分是森林区划的最小地域单位，其内部特征相同且与四周相邻部分有显著区别的小块森林。而林分内这些几何特征信息在本文中特指为森林空间数据。森林空间数据是树木三维建模的参数来源之一，是虚拟森林环境构造的基础数据。纵观森林资源调查监测技术的研究，单依靠遥感和地面测量技术还是无法方便有效地得到我们需要的森林空间数据。用计算机生成树木的图形模型一直是国内外学者的研究热点。目前在植物模拟方法上主要有L系统、迭代函数系统、扩散受限、粒子系统、交互式建模，这些方法的模拟计算过程复杂、计算量大。对于需要完成大数据量场景交互的可视化系统来说，上述模型的引入，势必会影响到系统的运行速度。目前整体地分析森林中有哪些形态不同的树木，建好的三维几何树模型又是怎样放置在森林中，如何生成一个实实在在的森林环境，若干年后，森林是如何变化的，特别是如何将现实的林分生长规律应用到虚拟森林环境中，如何模拟虚拟森林生长全过程等问题的研究却少之又少。 因此，基于林分生长规律的虚拟森林环境的构建需要解决两个大问题：一是森林空间数据的分析与仿真的问题，不仅要描述现时的林分内树木的位置信息及其对应的树木几何特征量，而且还要研究这些森林空间数据随着时间的流逝而发生的具体变化。第二是在森林空间数据给定的情况下，如何应用计算机图形学的建模和绘制技术来有效地表现树木及其绘制森林，最终建立一个可以反演过去、再现现实、预测未来的虚拟森林。本文的研究工作是在福建省空间信息工程研究中心承担的国家重大基础研究前期研究专项“分布式虚拟地理环境与森林遥感建模研究”的资助下，在现有南方森林资源和森林生长模型的基础上，论述了基于林分生长规律的虚拟森林环境的构建思路，提出了具体实现的技术和方法。本文的研究工作主要由以下几个部分展开： 1 森林空间数据的现时性分析与仿真。利用现有的森林资源二类调查的林分平均信息和林相图等整体信息和少量野外采样的数据，采取基于林分结构规律的整体到个体的仿真方法反演到个体，即获得森林空间数据。解决了目前的森林资源监测技术还无法得到森林空间数据的问题。常见的二类调查中就是小班调查。小班是以林分为基础划分的，其中小班面积、树种、林分密度、平均胸径、平均树高、立地质量等调查因子是森林空间数据的仿真中主要用到的林分平均信息。林相图是森林资源信息的主要表达形式，其小班边界图是用来计算树木位置的。不论是人工林还是天然林，在未遭受到严重干扰的情况下，林分内部许多特征因子，如胸径、树高、树冠等都具有一定的分布状态，而且表现出较为稳定的结构规律性，在林学中称为林分结构规律，林分结构内涵着这些反映林分特征因子的变化规律，以及这些因子之间的相关规律。林分直径结构是最基本的林分结构，可以采用正态分布、Weibun分布、p分布及r分布的方法对林分直径的分配状态进行拟合。整体到到个体的具体实现方法就是在林分结构规律的基础上提出来的。 2森林空间数据的时变性分析与仿真。在己获得现时的森林空间数据后，通过自然稀疏规律和林木竞争来分析林分生长规律，提出了以单木竞争生长模型为基础的生长预测和林木枯死概率为标准的淘汰机制的时变性仿真的实现方法。其研究结果能够为树木、森林的动态过程模拟建模提供足够详细的可靠的动态数据源，从而实现整个虚拟森林生长的动态变化。森林的资源总量是有限的，林木之间为了竞争阳光和养分，必然存在“适者生存”的自然进化规律。随着树木的平均大小的增大，密度过于拥挤从而导致一些被压木死亡，林分密度随着时间的推移必将呈现某种变化规律，称之为森林自然稀疏规律。森林自然稀疏规律可以用自疏现象的3/2自疏法则、张大勇等提出的自疏模型模型和BP一MSM混合人工神经网络模型来描述。如果用整个林分密度因子来解释林分中单株林木个体的生长仍然是不大可能的。因为林分中具体某一株林木的生长不是受整个林分密度因子的控制，而是受其周围的小环境的控制。在同一生长环境内，林木之间展开生长空间的竞争，致使一些林木给另一些林木造成间接危害的现象，称为林木竞争。林木竞争的模拟关键在选择竞争指数、组建单木生长模型和枯死模型。 3树木的交互式参数建模技术。基于森林经营管理的主要应用目的，林业管理者或其它用户需要?更多还原

【Abstract】 The forest dynamic changes have two distinct characters because of arbor’s big shape and its long life-span. One is long time scale, the other is large space scale. The two characters just make forestry research more difficult. In Virtual Forest Environments, complex forest objects are constructed, the laws of stand growth are presented and analyzed with computer graphics technology and forestry knowledge. So Virtual Forest Environments can provide a new way for forestry research due to its character of trying to exceed the limit of time and space. Generating Virtual Forest Environments aims to let forest governors and researchers manage and study forestry more conveniently and efficiently. What forest governors concern about is the fall of arboreal which is relevant to the basic parameters such as diameter at breast height, height of tree and so on. Moreover, they also concern about the competition relationship of trees in spatial configuration which relates to trees’ position, height, diameter at breast height, crown width, crown length. Forest stand is the least unit of forest partitions, which inner characteristics are same and distinguished markedly from neighboring sections. In this dissertation these characteristic information in forest stand refer to forest spatial data that is one of parameters of modeling three-dimensional tree model and basic data source of generating Virtual Forest Environments. Reviewing the techniques of surveying forest resource inventory, those forest spatial data can not be obtained conveniently and efficiently just with the remote sensing and ground measurement. Modeling tree with computer has been the popular research issue the domestic and international scholars are interested in. At present, in modeling three-dimensional trees there are L system, iterated function system, diffusion limited aggregation, particle system and interactive modeling. The computing process is complex and amount of calculation is great as trees are modeled trees with these methods. As for visualization system which needs to interact with scene that includes great amount of data, the running speed of system will be influenced when tree’s models are imported in scene. Nowadays, there are just few researches about the whole simulating process of the virtual forest growth such as which trees in different shapes are in forest, how to place three-dimensional geometric model of tree into forest scene, how to render realistic Virtual Forest Environments, what will happen in forest after a few years, especially how to apply the practical law of stand growth to Virtual Forest Environments.Therefore, two key problems about generating Virtual Forest Environments which are true of the law of stand growth should be resolved in this dissertation. One is to analyze and simulate forest spatial data, namely not only to describe the present characters of tree in stand, but also to research on how forest spatial data change with time. The other is to construct three-dimensional geometric model of tree and to render forest with modeling and rendering technology of computer graphics ifthe forest spatial data are given, finally to generate a virtual forest which can recur to the past, represent the present, and forecast the future. Research in this dissertation is subsidized by national important foundation research project named "Research On Distributed Virtual Geographic Environments And Modeling Forest By Remote Sensing " that Spatial Information Research Center of Fujian province takes on. Based on existing forest resource and forest growth model in south, the idea of constructing Virtual Forest Environments which are true of the law of stand growth are expounded, and the detailed implementing techniques and methods are put forward. The main research contents of the dissertation are as below:1. Analysis and simulation of the present forest spatial data. The whole-to-individual method based on the law of stand structure is introduced and discussed in order that the individual, namely the present forest spatial da更多还原