Growth Modeling For Spruce-Fir Forest In Changbai Mountains

Forests are significant natural resources for human beings, and they have a variety of functions. However, the quality of Chinese forest is very low; thus, an urgent need is to increase their productivity. As natural forest is the dominant part of forest, its growth pattern and growth dynamic should be modelled, so as to make suitable management plan accordingly. This is a big project for Chinese forestry. In this dissertation, I took spruce-fir forest in Changbai Mountains as the object, used large dataset including clear-cutting plots, and numerous permanent re-measured plots, to develop a system of growth models for uneven-aged forest. Main conclusions are as following:(1) The Gompertz function is optimal for fitting diameter, height, and volume growth models, which taking age as the independent variable. According to the quantitative mature age, Korean pine, fir, spruce, tilia, acer, and Korean birch should be cut while their diameter reach 36.2cm,44.4cm,42.2cm, 59.4cm,25.8cm and 48.9cm. R2 of local height-diameter models are around 0.74, while it is between 0.86-0.91 for general height-diameter models. This result indicates that the general height-diameter models have higher precision. Among different types of volume functions, the single entry volume equations have the lowest precision, general volume equations which using diameter and height as the independent variables have the highest precision, the one entry volume function deduced from general volume equations and height-diameter models have the medium precision. Therefore, when height data was available, general equations should be used, otherwise, the deduced one entry volume equations should be chosen.(2)The competition group (including diameter, basal area in larger trees, relative diameter, basal area per ha, trees per ha, stand density index), stand attribute group (including basal area ratio of broadleaves species, basal area ration of the species itself, cutting intensity), and site index (including elevation, slope, aspect, humus depth, soil depth) were used as the candidate independent variables to develop individual diameter growth model for fir, spruce, and pine in spruce-fir forest in Changbai Mountains. Results indicated that stand attribute variables are very important in the model, and the cutting intensity may influence the diameter growth despite varying stand density. The mortality models for fir, spruce, and pine were developed using binary logistic model, and results indicated that the binary logistic can generate rational prediction.(3) Weibull function, beta function, and negative-exponential function were compared for fitting diameter distribution in uneven-aged forest. Results show that the weibull function is the best, while negative-exponential function is the worst. But the parameters of negative-exponential function are decreasing along time apparently. The ordinary linear regression was used to simulate the parameters variation, so as to develop diameter dynamic model. This model can be used to predict the diameter distribution and basal area in 10 years. Additionally, the variations of q value show that it will decrease in undisturbed stand, while increase after cutting.(4)Based on long-term permanent re-measure data, multinomial logistic models were used to develop matrix model for all species, fir, spruce, Korean pine, and other species group. Results indicated that this multinomial logistic method is very good, and can generate precise prediction. Using the matrix model developed to predict the dynamic for plot 4, as an example, and 13 different cutting scenarios were tested. From the dynamic of 50 years, I can make the conclusion that the optimal cutting strategy is to cut at 10% intensity with 10 years cutting cycle.The main innovation of this thesis are the three following aspect:(1) The models of uneven-aged forest was first developed systematiclly by using large sample trees and permanent re-measure plots data in spruce-fir forest in Changbai Mountains. (2) It is the first time in China to include cutting intensity as an independent variable in diameter growth models. (3) Multinomial logistic models were first be used in China to calculate the transition probability in matrix models.Further study could include competition variables into diameter-age, height-age, volume-age, volume-diameter, and volume-diameter & height functions. Better method for grouping species need be studied. As for the site index, height-diameter models and using past diameter increment could be tested. Spatial growth model is another new subject. The matrix models in this dissertation were only used to predict timber and size biodiversity, and it could be connected with more objectives. New methods such like error covariance model and mixed-effect models should be used to get higher precision.