基于连清样地数据的全国杉木人工林平均木树高-胸径模型

目的 基于我国森林资源连续清查（简称“连清”）样地数据，分省区研建全国杉木人工林平均木树高-胸径的最优基础模型，以期为全国各省区杉木人工林的树高预测提供基础模型。 方法 研究范围为杉木人工林分布的15个省份，数据来自第六次、第七次连清样地数据的树高调查表，总样本数为23 239个。选取18种基础生长方程作为候选模型，分别拟合各省区杉木平均木树高与胸径的关系，根据模型的决定系数（R2）、平均绝对误差(MAE)、平均相对误差（MRE）、均方根误差(RMSE)和平均预估误差（MPE），并结合模型残差分布图，确定各省区最优模型，同时采用5折法验证各省区最优模型的预测能力，最终决定各省区最优树高-胸径模型。 结果 15个省区的杉木最优树高-胸径模型并不相同，四川、云南、重庆、陕西、浙江、江西、湖南、广西的最优模型为模型18（Mitscherlich方程），江苏、安徽、河南和福建的最优模型为模型16（Hossfeld方程），广东、湖北、贵州的最优模型分别为模型10（双曲线方程）、模型11（Logistic方程）和模型13（Gompertz方程），R2分布在0.602～0.807之间，MAE分布在0.94～1.53 m之间，MRE分布在−2.93%～−4.72%之间，RMSE分布在1.23～2.00 m之间，MPE分布在0.50%～2.77%之间。模型拟合效果较好，满足精度要求，且参数具有生物学意义，可作为全国各省区杉木人工林平均木树高-胸径基础模型。 结论 本研究构建全国杉木人工林分布的15个省区的最优树高-胸径基础模型，能较好的模拟各省区的杉木平均木树高随胸径的变化规律，可以作为全国各省区基本的杉木人工林平均木树高-胸径模型，为各省区杉木人工林的树高预测提供依据。

Average Tree Height-Diameter Models of Cunninghamia lanceolata in China Based on Continuous Forest Inventory Plot Data

Objective Based on the national permanent forest plots, the basic average tree height - diameter models of Cunninghamia lanceolate was established by province for predicting tree height . Methods There was a total of 23 239 samples distributed in 15 provinces in this study. Eighteen candidate base growth equations were used to fit height-diameter relationship in each province. The coefficient of determination (R2), mean absolute error (MAE), mean relative error (MRE) and root mean square error (RMSE) combing with the residual plots were used for model evaluation. In addition, the 5-fold method was used to test the optimal model in each province. Result The best height-diameter model in each province was not the same. The best model for Chinese fir in Sichuan, Yunnan, Chongqing, Shanxi, Zhejiang, Jiangxi, Hunan and Guangxi was the Mitscherlich equation. For Jiangsu, Anhui, Henan and Fujian, the best model was the Hossfeld equation. And for Guangdong, Hubei and Guizhou, the best model was hyperbolic equation, Logistic equation and Gompertz equation, respectively. The R2 of the model ranged from 0.602 to 0.807, MAE ranged from 0.94 to 1.53 m, MRE ranged from −4.72 to −2.93%, RMSE ranged from 1.23 to 2.00 m, and MPE ranged from 0.50 to 2.77. These models performed well and had biological significance, which indicated that these models could be used as the basic height-diameter models of C. lanceolate plantation in each province Conclusion height-diameter models of C. lanceolate distributed in 15 provinces are developed in this study, which simulates well the average tree height of C. lanceolate in each province, and can be used as basic models of height-diameter for C. lanceolate plantation in each province in China.