【作者】 张志；

【导师】 孙玉军；

【作者基本信息】 北京林业大学 ， 森林经理学， 2004， 硕士

【摘要】 在对研究地区基本情况进行广泛调查的基础上，以1997年金沟岭林场1：5万的地形林相图、研究区1997年的森林经理小班调查资料和1997年30m×30m分辨率的LANSAT5TM影像图为基本资料，应用地理信息系统技术，在GIS软件的支持下，应用景观生态学的有关原理、方法和指标体系，研究该区的森林景观格局，得出了如下结论。 (1) 景观多样指数为2.608538，而聚集度达到0.82，景观异质性高，异质景观要素斑块交错镶嵌，存在着受某些环境因子和干扰状况综合控制形成的景观空间格局。 (2) 各森林类型之间的比较分析研究表明：处于演替早期阶段的森林类型，斑块规模小，中小斑块比例高，斑块形状简单规整；而处于演替后期的森林类型，斑块规模较大，大型斑块比例高，斑块形状复杂。 (3) 在景观要素各个森林类型中，混交林的各个森林类型的斑块联系度较高，而纯林、相对纯林的各个森林类型的斑块间的联系度较小；混交林各森林类型的相邻度大于它们与纯林、相对纯林的相邻度，而纯林、相对纯林的各个森林类型的相邻度极低，但与混交林各个森林类型的相邻度较高，说明在整个景观范围内，混交林与纯林、相对纯林是交错分布的，是景观的主体，而纯林、相对纯林总体是以分散的小斑块分布于混交林斑块间。 (4) 通过趋势面分析发现，水平位置的变化对景观格局没有显著的系统影响，海拨、坡度、坡向及其组合状况影响着森林群落类型的空间分布。 (5) 根据金沟岭林场的实际情况，采用人工神经网络方法建立了基于人工神经网络的混交林可变密度蓄积量BP网络模型，采用回归方程适应性检验法分别对模型的拟合效果进行适应性检验，检验结果表明所有模型是实用的，不存在系统偏差，表明用人工神经网络的方法对森林资源的进行预测是可行的，可以满足林业生产经营的精度要求。更多还原

【Abstract】 Choosing the Jingouling forest farm as the study object, which located at the wangqing county in the jilin province, the forest landsca pattern of this area is studied. In this study many data are used such as the 1:50000 topographic and forest phase map, the forest management Information though Detailed investigation in 1997 and the 30mx30m TM image in 1997. Supported by the GIS software and the principles, methods and Index systems of landscape ecology, the following conclusions are arrived in this study.(1) The landscape index is 2. 608538 and the contagion index is 0. 82, so the landscape spatial heterogeneity is more higher and heterogeneity patches is mosaic. There are landscape spatial pattern controlled by some environmental factors and status of disturbance at Jingouling.(2) Comparing all kinds of forest types, It shows that the patches of the forest types in the early succession are the scales of small and middle and have regular shapes;the patches of the forest types in the later succession are the scales of large and have compound shapes.(3) In the all kinds of forest types, The relation index between the forest types of the mixed stand are larger than the relation index between the forest types of the pure stand and the relative pure stand;The neighour index between the forest types of the mixed stand are larger than the neighour index between the forest types of the mixed stand and the forest types of the pure stand and the relative pure stand and The neighour index between the forest types of the pure stand and the relative pure stand are small. It shows that within the landscape the mixed stand is the main body and the pure stand and the relative pure stand are distributed in the mixed stand.(4) According to trend surface analysis, the change of horizontal location has not systematic effect on the landscape pattern. The altitude, slope, aspect and the combining of them effect the distribution of the forest types.(5) According to the practical condition in Jinggouling forest farm, variable density stand volume for BP models are developed with the method of the artificial neural network. The test method of regression equation adaptability is used to test the simulative effects of models. The testing result indicates the model is practicable and no system deviation exist and shows The BP model can satisfy the precision of forestry production.更多还原