【作者】 王利；

【导师】 丰震；

【作者基本信息】 山东农业大学 ， 森林培育， 2003， 硕士

【摘要】 采用带状样地法在山东药乡林场的麻栎林中共布设样地10块，测得样木共计406株，并且测定了这些树木的树高、胸径、地径、枝下高、平均冠幅五项指标，利用多元线性回归分析、逐步回归分析等方法，研究了树高曲线模型以及胸径与地径、冠幅与胸径之间的相关关系，并且利用华北地区阔叶树的二元立木材积表导算了适于本地区的麻栎一元立木材积表，其研究结果如下： 1. 通过多元线性回归分析，首次建立了计算树高的多元线性回归方程，经检验达到极显著水平，平均精度为98.09%，最低精度为95.51%，其结果表明树高可利用地径、胸径、枝下高、平均冠幅进行计算，该方法精度高，结果可靠。 2 通过逐步回归分析，在显著水平0.15的情况下，引入变量地径、胸径、枝下高，首次建立了计算树高的逐步回归方程，经检验回归方程达到极显著水平，平均精度为98.06%，最低精度93.99%，与多元线性回归相比较，调查的工作量减少，但精度稍低。3 研究了树高与枝下高、平均冠幅的相关关系，选出的各分组非线性回归方程平均精度为94.86%，而线性回归方程平均精度为90.84%，最低精度为78.84%，由此可见，其非线性化回归方程好于线性回归方程。4首次研究了树高与胸径、平均冠幅的相关关系，非线性回归方程（平均精度为98.43%）好于线性回归方程（平均精度为97.71%，最低精度为94.04%）。 5 通过对树高与胸径相关关系进行研究，选出了适用于不同范围的回归<WP=5>方程，线性回归方程复相关系数为0.1525-0.4454，平均精度为97.45%，非线性回归方程平均精度为96.21%，最低精度为89.01%，可见，其线性回归方程好于非线性回归方程。6 研究了树高与枝下高的相关关系，选出的各分组回归方程的复相关系数为0.119-0.3829，平均精度为96.35%，最低精度为67.39%，其结果可靠程度低于树高与胸径的回归方程。 7 研究了树高与平均冠幅的相关关系，选出了适用于不同范围的回归方程，其相关系数为0.03-0.29，平均精度为93.34%，最低精度为69.13%，其结果可靠程度低于树高与胸径的回归方程。 通过以上研究单因子和多因子树高曲线模型的结果表明，在单因子模型中，以胸径建立的树高回归方程（平均精度为96%以上）为最好。在多因子模型中，以多元线性回归、逐步回归和双因子(胸径和平均冠幅)建立的回归方程为最好。从精度上讲，可采用多元线性回归方程（y =2.25037 - 0.02336x1 + 0.29419x2 + 0.47206x3 + 0.11931x4）计算树高；从效率上讲，可采用单因子（胸径）回归方程（适用于胸径14cm以下树木的最佳回归方程为适用于胸径14-24cm树木的最佳回归方程为y=-2.74871+1.03083x2-0.01816x22，适用于胸径24cm树木的最佳回归方程为y=2.0901+0.52963x2-0.00526x22。）、双因子（胸径与平均冠幅）回归方程（y1=5.15252+0.5225x1+0.00019x24.66289，y2=2.23591+0.44189x1+3.27044x20.36849，y3=10.72627+0.39608x1+0.01832x23.6925。）计算树高。8研究了胸径和地径的相关关系，选出了适用于不同范围的回归方程x2=a+bx1，个别方程精度较低仅为72.37%，其余精度均在93%以上，估计<WP=6>结果比较可靠，可用相应方程利用地径计算树木的胸径。 9 研究了树木冠幅与胸径的相关关系，选出的各分组回归方程估计精度在91%以上，估计结果较可靠，可以用胸径计算树木的冠幅，以确定树木的营养面积。 10 先利用树高与胸径的回归方程y=2.0901+0.52963x2-0.00526x22计算各径阶平均高，然后借助于华北地区的阔叶树二元立木材积表导算出了麻栎的一元立木材积表。更多还原

【Abstract】 Height curve models and corelations between diameter at breast height(DBH) and collar diameter(CD)、crown breadth and diameter of Sawtooth Oak (Quercus acutissima) have been studied by establishing 10 belt sample plots ,measuring tree height(H)、 DBH、 CD、 height blow branch(HBB)、CB of 406 trees,employing multifactor linear regression analysis、stepwise regression analysis and other functions.Using hardwood standard volume table of Huabei region and height-DBH equations,one-way volume table has been established.The results are as follows:Multifactor linear regression equations are established to calculate the height by multifactor linear regression analysis . The significance levels of equations are very high, The average precision is 98.09%, the lowest precision is 95.51%,which indicate that there tree height can be estimated through CD,DBH,HBB and CB.At the 0.15 significance level,the stepwise regression equations for tree height are established by using the stepwise regression analysis.The average precision is 98.01%,the lowest precision is 93.99%.Compared with multifactor linear regression analysis method,this method enhances efficiency of investigation ,although its precision is lower a litter than that of the multifactor linear regression analysis method.The corelations are studied among height、HBB、CB,different regression equations for different DBH groups are established.The average accuracy of non-linear regression equation is 94.86%,hower,the average accuracy of linear regression equation is 90.84%,and the lowest accuracy is 78.84%,the results indicate that non-linear regression equations are better than linear regression equations.<WP=8>The corelations are studied among tree height、DBH、CD,also the different regression equations for different DBH groups are established,,the results indicate that non-linear regression equations (the average precision 98.43%) are better than linear and the changeable linear regression equations(the average precision 97.71%, the lowest precision 94.04%).Regression equations for different DBH groups are establisheded by studying the relations between tree height and DBH.The results indicate that the changeable linear regression equations are better than non-linear regression equations ( the average precision 96.21%, the lowest precision 89.01%). The corelations between tree height and HBB are studied ,different equations are established for different diameter groups. Their relative coefficient squares 0.119-0.3829, the average precision is 96.35%, the lowest precision is 67.39%.The reliability of the equations is lower than that of height-diameter regression equations The corelations between tree height and CB are studied, and the different equations for different DBH groups are established, their relative coefficient squares 0.03-0.29, the average precision is 93.34%, the lowest precision is 69.13%,the reliability of the equations is lower than that of height-DBH regression equations. Through above studies, the results indicate that height-DBH equations are best among the single factor models. In the multifactor models, the better equations are multifactor linear regression equation、 stepwise regression equation and binary factor equation. As to equation precision, multifactor linear regression equation can be used to calculate tree height; as to efficiency of investigation, tree height can be calculated by using the single factor (DBH)equation、binary factor (DBH、CB)equation.The corelations between DBH and CB are studied, and the equations for different diameter groups are established.Particular equation precision(72.37%) is slower ,however,precisions of other equations are high up to 93%.Their regression results are more reliable,and the equations can be used to calculate DBH through CB.Different DBH group regression equations are established by studying the relations between CB and DBH. The precision of the equations is above 91%, the results are more reliable.The CB can be calculated through DBH and<WP=9更多还原