基于地理加权回归模型的大兴安岭中部天然次生林更新分布

以黑龙江省大兴安岭新林林业局翠岗林场为研究区,基于2018年7—8月在研究区建立的45块固定样地数据,以林分因子、地形因子、林分空间结构、土壤厚度和物种多样性5方面的9个因子为自变量,建立全局泊松模型和以地理加权回归模型为基础的4种尺度(5、10、15和20 km)地理加权泊松模型(GWPR)对该地区天然次生林更新状况进行模拟,利用全局MoranI和局域MoranI分别对模型残差的全局空间自相关性和空间分布状况进行描述,评价全局模型和各尺度局域模型的拟合效果,对尺度效应下各局域模型之间的差异进行说明,采用5 km尺度局域模型绘制研究区森林更新的空间分布,对研究区森林更新状况进行评价和分析。

Taking Cuigang Forest Farm, Xinlin Forestry Bureau, Daxing'anling Mountains, Heilongjiang Province as the study area, this research was based on the data of 45 permanent sample plots established in the study area from July to August 2018. Taking 9 factors from 5 aspects including stand factors, topographic factors, stand spatial structure, soil thickness and species diversity as independent variables, a global Poisson model and four geographically weighted Poisson regression models (GWPR) at scales of 5, 10, 15 and 20 km based on the geographically weighted regression model were established to simulate the natural secondary forest regeneration status in this region. Global Moran's I and local Moran's I were used to describe the global spatial autocorrelation and spatial distribution of model residuals respectively. The fitting performance of the global model and each local model at different scales was evaluated, and the differences among the local models under scale effects were explained. Finally, the 5 km scale local model was adopted to map the spatial distribution of forest regeneration in the study area, and the forest regeneration status of the study area was evaluated and analyzed.