运用林分密度和平均高估测思茅松人工林地上生物量

以65块云南省普洱地区思茅松人工林圆形样地数据和sentinel-2多光谱影像数据为研究对象,利用林分平均高与林分密度(每公顷株数、林分疏密度、植被覆盖度、叶面积指数)估测思茅松人工林林分地上生物量。分析思茅松人工林林分地上生物量与林分密度指标的相关性;采用参数模型(不变参数模型和可变参数模型)和非参数模型(包括支持向量机、随机森林和BP神经网络)探索平均高和林分密度等变量估测林分思茅松人工林地上生物量。结果表明:思茅松人工林林分地上生物量与每公顷株树、林分疏密度、植被覆盖度、叶面积指数呈显著正相关(r>0.5);在构建思茅松人工林地上生物量的所有模型中,每公顷株数-林分平均高构建的可变参数模型(R2=0.966 0,RMSE=10.05 t·hm^-2)效果最优,林分平均高-林分疏密度构建的RF模型(R2=0.901 7,RMSE=19.37 t·hm^-2)次之,林分平均高-植被覆盖度构建的RF模型(R2=0.748 4,RMSE=33.36 t·hm^-2)最差;林分密度-平均高的地上生物量模型与实测地上生物量的相关性较高(R2=0.966 0),反演误差值较低(RMSE=10.05 t·hm^-2);叶面积指数比植被覆盖度对林分地上生物量变动有更好的解释能力,每公顷株数对林分地上生物量变动的解释能力好于林分疏密度。

Aboveground Biomass Estimation of Simao pinewith Stand Average Height and Density of Plantation

With the investigation data on 65 circular sample plots of Pinuskesiya var. langbianensis plantation in Pu’er City Yunnan Province, the average height and stand density, including number of per hectare, stand density, fraction of vegetation cover(fcover) and leaf area index(LAI), were explored to estimate the aboveground biomass. The correlation between aboveground biomass and stand density in P. kesiya var. langbianensis pine plantation was analyzed. Parametric models(including invariant model and variable-parameter model) and non-parametric models(including SVM, RF, and BP models) were used to estimate the aboveground biomass of P. kesiya var. langbianensis plantation with mean height and stand density. Stand density indicators, including number of per hectare, stand density, fcover and LAI, were positively correlated with the aboveground biomass(r>0.5) of P. kesiya var. langbianensis plantation. Among all aboveground biomass models of P. kesiya var. langbianensis plantation, the variable parameter model based on number of per hectare and average height had the best effect(R2=0.966 0, RMSE=10.05 t·hm^-2), the RF model based on stand density and average height had the second effect(R2=0.901 7, RMSE=19.37 t·hm^-2), the RF model based on LAI and average height had the third effect(R2=0.855 1, RMSE=24.59 t·hm^-2), the RF model based on average height and fcover was at the bottom(R2=0.748 4, RMSE=33.36 t·hm^-2). The correlation between the models and the aboveground biomass of P. kesiya var. langbianensis plantation were well(R2=0.966 0, RMSE=10.05 t·hm^-2). LAI was better than fcover in explaining the change of aboveground biomass. Compared with stand density, the number of per hectare had a better ability to explain changes in aboveground biomass.