| 摘要: |
| 为了探究杉木 Cunninghamia lanceolata 的碳储量生长规律, 科学指导碳汇导向型森林的经
营, 最大化实现杉木人工林的碳汇服务功能和价值, 研究基于 2019 年广东省二类调查的杉木人工林样
地数据, 通过条件筛选和异常值处理筛选出建模数据 17 680 条, 以林龄为解释变量, 碳储量为目标变
量, 采用加权非线性回归方法求解模型参数, 在 Logistics、 Mitscherlich、 Gompertz 和 Schumacher 4 种
基础模型中筛选最优基础模型, 在最优基础模型中引入以杉木适生区分布区域为特征的哑变量, 在不
同参数引入哑变量以确定最佳的引入参数位置, 建立可兼容不同区域的杉木人工林碳储量生长模型。
结果表明, 在 4 种候选理论生长模型中, Schumacher 模型的拟合效果最佳, 各项评价指标在候选模型
中均为最优, 确定系数为 0. 814 3, 标准误差为 8. 526 1, 平均百分比标准误差为 19. 03%, 平均百分比
误差为 4. 66%。 在参数 b 中引入哑变量效果较好, 模型的评价指标得到显著优化, 最终模型的确定系
数为 0. 857 6, 标准误差为 7. 576 9, 平均百分比标准误差为 14. 63%, 平均百分比误差为 4. 14%, 说明
引入哑变量后有效提高模型的拟合效果和稳定性, 包含哑变量的碳储量生长模型可用于预测广东省不
同区域杉木人工林碳储量。 |
| 关键词: 碳储量 生长模型 哑变量 杉木人工林 |
| DOI：10. 20221 / j. cnki. 2096-2053. 202504005 |
| 分类号: |
| 基金项目:广东省林业科技创新专项资金项目 (2023KJCX001)。 |
|
| Construction of A Growth Model for Carbon Storage in Cunninghamia lanceolata Plantations in Guangdong Province Based on Dummy Variables |
|
GU Luowei, HE Yuheng, HUANG Huanqiang, FENG Mingchun, LIN Na, CHEN Shiqing
|
|
South China Agricultural University,College of Forestry and Landscape Architecture
|
| Abstract: |
| To explore the growth patterns of carbon storage in Cunninghamia lanceolata and scientifically
guide carbon-sink-oriented forest management practices, aiming to maximize carbon sequestration services and
value of C. lanceolata plantations, this study utilized data from the 2019 second - class forest inventory of
C. lanceolata plantations in Guangdong Province. After conditional screening and outlier removal, 17 680 modeling data entries were selected. Using stand age as the explanatory variable and carbon storage as the response
variable, model parameters were solved for using weighted nonlinear regression methods. The optimal basic
model was selected from four candidates: Logistic, Mitscherlich, Gompertz, and Schumacher. Dummy varia-bles representing suitable growth regions of C. lanceolata were introduced into the optimal basic model. The optimal parameter position for incorporating dummy variables was identified to establish a region-compatible carbon
storage growth model. The results showed that among the four candidate theoretical growth models, the Schumacher model exhibited the best fit with an R
2
of 0. 814 3, standard error (SE) of 8. 526 1, mean percentage
standard error (MPSE) of 19. 03%, and mean percentage error (MPE) of 4. 66%. After introducing dummy
variables at parameter b, the model′s performance significantly improved: R
2
increased to 0. 857 6, SE decreased to 7. 576 9, MPSE decreased to 14. 63%, and MPE decreased to 4. 14%. This demonstrates that dummy variable incorporation effectively enhanced the model′s fitting accuracy and stability. The final carbon storage growth model with dummy variables can be used to predict C. lanceolata plantations across different regions
of Guangdong Province. |
| Key words: carbon storage growth model dummy variable Cunninghamia lanceolata plantation |
