森林生长是个复杂的自然过程，受到立地条件、气候变化、森林经营及本身特性的影响。随着人们对森林生长过程的认识，模型结构日趋复杂，表现出林分生长与收获模型的不确定性。林分生长与收获模型的不确定性主要表现在模型输入变量和对应参数的不确定性以及模型结构的不确定性。然而，以往构建的林分生长与收获模型，并没有考虑模型的不确定性，也没有对模型的不确定性进行分析评价。贝叶斯推断法是分析模型不确定性一种常用有效的方法。本研究以江西、福建杉木（Cunninghamialanceolata（Lamb.））为例，基于贝叶斯推断法分析杉木生长模型和枯损模型的不确定性，并分析评价两种不同来源不确定性对生长和枯损模型不确定性的贡献；通过贝叶斯法分析直径分布模型的不确定性，并分析评价两种不同来源不确定性对直径分布模型不确定性的贡献。

Forest growth is a complicated nature process influenced by site conditions, climate change, forest management and tree characteristics. With the understanding of the processes of forest growth, structures of forest growth and yield models are becoming more complex, resulting in model uncertainty. Two main sources of model uncertainty are input variables and corresponding parameters,and model structure. However, the traditional forest growth and yield models did not consider the model uncertainty, nor evaluate the model uncertainty derived from the two sources. Bayesian inference is a commonly used and effective method for dealing with model uncertainty. Based on the Chinese fir (Cunninghamialanceolata(Lamb.)) in Jiangxi and Fujian province, the objectives of the project were: quantify uncertainty of growth models and mortality models derived from input variables and correspongding parameters and model structure, and evaluate these two uncertainty; quantify uncertainty of diameter distribution model derived from two sources, and evaluate these sources uncertainty.