[Prediction Models of Sapwood Density, Heartwood Density, and Bark Density in Larix Olgensis Plantation].
Yu-Xin Peng,Feng-Ri Li,Fu Liu,Li-Hu Dong
DOI: https://doi.org/10.13287/j.1001-9332.202004.007
2020-01-01
Abstract:In this study, the Beta regression models of sapwood, heartwood, and bark density of Larix olgensis were constructed. A total of 35 trees were destructively sampled from plantations in three different sites, Linkou Forestry Bureau of Heilongjiang Province, Dongjingcheng Forestry Bureau, and Maoershan Experimental Forest Farm of Northeast Forestry University. AIC, R2, BIAS, RMSE and LRT were used as the goodness-of-fit statistics to compare and select the most optimal models for sapwood, heartwood, and bark density. The jackknife resampling technique was used to verify and evaluate the developed models. The results showed that the independent variables of the optimal sapwood, heartwood, and bark density model were not identical. Sapwood density had a good relationship with tree age, tree height, relative height, and the square of relative height. The independent variables of the optimal heartwood density model were annual growth, relative height, and the square of relative height. The independent variables of the optimal bark density model were tree age, annual growth, relative height, and the square of relative height. The analysis of the optimal model showed that from the base to the tip of the trunk, sapwood density decreased gradually, heartwood density initially decreased and then increased regularly, bark density initially increased and then decreased gradually. The established Beta regression models could predict sapwood, heartwood, and bark density of L. olgensis at any position in the research area and be an essential basis for the study of trunk average density and biomass.