基于混合效应模型及生存分析方法的森林天然更新模拟方法研究

As the best method for forest restoration, natural regeneration has a profound impact on the structure and biodiversity of future forest. The model for natural regeneration can simulate the reality and future situation of natural regeneration and provide accurate forest plan for forest managers. In constructing the model for natural regeneration, the mixed effects model and survival analysis method have obvious advantages. But there are still many problems with these two methods. In zero-inflated model count model, multi-level random intercept effect or random coefficient effect is not considered simultaneously on two distributions. In constructing the survival model of sapling using survival analysis method, the influences of covariates and random effects were not taken into account. The method of height distribution model and total seedling estimation based on mixed effect model is still blank. In the study, a seedling counting model for natural regeneration was constructed based on multivariate (covariates) multilevel generalized linear mixed effect model; Based on the multivariate multilevel mixed effect model, logistic model and survival analysis method were used to construct the survival and development for the sapling of natural regeneration. The generalized distribution model for sapling height of natural regeneration is constructed based on the multivariate multilevel mixed effect model. The statistics of total seedling count of natural regeneration were fitted based on sampling and hybrid estimators methods.

天然更新是森林恢复最好的办法，对未来林分结构和生物多样性具有深远的影响。更新模型能够模拟天然更新的现实和未来状况，为森林经营者提供准确的森林计划。混合效应模型和生存分析方法在构建天然更新模型上具有明显的优势。但这两种方法目前还存在着诸多问题，在构建零膨胀计数模型时，没有同时在两个分布上考虑多水平随机效应；在构建幼苗-幼树生存模型时，基于生存分析方法的研究刚刚起步，并没有考虑协变量及随机效应对更新的影响；基于混合效应模型的幼树高度分布模型和幼苗总体估计量的方法研究还是空白。本研究基于多元（协变量）多水平广义线性混合效应模型方法，构建天然更新幼苗计数模型；基于多元多水平混合效应模型方法，分别采用logistic模型和生存分析方法，构建天然更新幼苗-幼树生存和发展模型；基于多元多水平混合效应模型方法，构建天然更新广义的幼树高度分布模型；基于混合估计量方法模拟天然更新幼苗总体株数统计量。

在森林经营中，在有条件的情况下，天然更新是森林恢复最好的办法，不仅能够节约时间和财力。更新模型能够模拟天然更新的现实和未来状况，为森林经营者提供准确的森林计划。混合效应模型和生存分析方法在构建天然更新模型上具有明显的优势。但这两种方法目前还存在着诸多问题，在构建零膨胀计数模型时，没有同时在两个分布上考虑多水平随机效应；在构建幼苗-幼树生存模型时，基于生存分析方法的研究刚刚起步，并没有考虑协变量及随机效应对更新的影响；基于混合效应模型的幼树高度分布模型和幼苗总体估计量的方法研究还是空白。本研究基于多元（协变量）多水平广义线性混合效应模型方法，构建了天然更新幼苗计数模型；基于多元多水平混合效应模型方法，分别采用logistic模型和生存分析方法，构建了天然更新幼苗-幼树生存和发展模型；基于多元多水平混合效应模型方法，构建了天然更新广义的幼树高度分布模型；基于混合估计量方法模拟了天然更新幼苗总体株数统计量。截止目前为止，在本基金的支持下，利用云冷杉阔叶混交林数据，完成上述全部研究内容，发表论文2篇，后续还将发表一些本课题资助的成果。这些成果能够为未来林分天然更新方法提供科学依据。