关键词: 材积表, 混合模型, 哑变量模型, 检验, 落叶松

Abstract: [Objective] This paper aimed at accurately assessing whether there are deviations or not and grasping the differences among tree volume tables of larch (Larix spp.) in different regions, so as to provide the basis for revising and improving the two-variable tree volume tables.[Method] Based on the newly collected volume data of 480 sample trees of larch from four regions, i.e., northeastern, northern, northwestern and southwestern regions, the current two-variable tree volume tables for different regions were examined using adaptability test method of regression equations, the differences of two-variable tree volume equations were analyzed using mixed model approach, and the two-variable tree volume equations with region-specific parameters were developed using dummy-variable model method.[Result] The validation results of 4 two-variable tree volume tables published as ministerial standards showed that the estimation errors of three tree volume tables exceeded allowable range (±3%) and the largest error reached to about 12%, indicating significantly systematic bias. From the results of difference significance test of two-variable tree volume equations for different regions, it was found that the four regions ranked by volume estimate from large to small as northeast, northwest, southwest and north; the difference of volume estimates between northeastern and northern regions was highly significant (P<0.01), that between northwestern and northern regions was significant (0.01≤ P< 0.05), that between northeastern and southwestern regions was lowly significant (0.05≤ P< 0.10), and those between any other two regions were not significant. The developed tree volume equations with region-specific parameters showed that little differences were existed among three modeling alternatives, i.e., one population only, two populations and four populations for four regions. From the comparison of two alternatives, i.e., one generalized volume equation for the whole country and four volume equations for four regions, the largest relative error of volume estimates in different regions was only about 3%.[Conclusion] Most of the two-variable tree volume tables published as ministerial standards are probably biased, it is recommended that all two-variable tree volume tables should be systematically examined, and the biased volume tables should be updated or revised. The differences of two-variable tree volume equations among different regions are not so large, and it is suggested that the population classification for developing two-variable tree volume equations for major tree species should be considered by State Forestry Administration at national level. It is necessary to develop gradually a system of forestry tables covering the whole country to promote the standardization for establishing forestry tables.

Key words: volume table, mixed model, dummy-variable model, validation, Larix