基于3-PGmix过程生长模型和优化算法的长白落叶松人工林潜在生产力估计OA
Estimation of Potential Productivity of Larix olgensis Plantations Based on 3-PGmix Process-Based Model and Optimization Algorithm
[目的]提出一种基于过程生长模型和优化算法估计林分潜在生产力的新方法,为立地质量评价和森林经营提供科学依据.[方法]基于长白落叶松纯林连续观测样地数据校准 3-PGmix 模型,应用决定系数(R2)、平均绝对误差(MAE)、均方根误差(RMSE)和相对均方根误差(rRMSE)评价模型性能;设置蓄积连年生长量和 5年定期平均生长量最大 2种目标情景,采用粒子群优化算法求解不同立地等级下林分蓄积生产力最大值和对应的最优林分密度,在此基础上通过 Weibull分布函数的尺度和形状参数模型得到对应的最优直径结构.[结果]1)在校准数据和验证数据中,3-PGmix 模型模拟的长白落叶松人工林的平均胸径、株数密度、蓄积量和地上生物量与样地实测值一致性较好,决定系数均大于 0.86,相对均方根误差均小于 16%.2)基准年龄 30年时,长白落叶松人工林连年生长量和 5年定期平均生长量最大时 5个立地等级的蓄积潜在生产力分别为 4.50～8.11和 4.53～8.18 m3·hm-2a-1.3)达到蓄积潜在生产力时的径阶数量随林龄增加单调递增,径阶香农指数在生长前期经波动变化后也呈现出增加趋势,且立地等级越高二者数值越大.[结论]结合过程模型 3-PGmix 与粒子群优化算法可以实现蓄积潜在生产力的估计,弥补传统方法未考虑林分自稀疏和气候因素影响的不足,为林分潜在生产力估计提供了一种新的方法.
[Objective]This study aims to develop a novel method for estimating stand potential productivity based on a process-based model and optimization algorithm,thereby providing a scientific basis for site quality assessment and forest management.[Method]The 3-PGmix model was calibrated using continuous observation data from pure stands of Larix olgensis obtaied from China's national forest inventory.The model was evaluated by the coefficient of determination(R2),mean absolute error(MAE),root mean square error(RMSE)and relative root mean square error(rRMSE).Two target scenarios were set:maximization of annual volume increment and 5-year periodic mean increment.A particle swarm optimization(PSO)algorithm was applied to estimate the maximum stand volume productivity and corresponding optimal forest density under five different site quality grades.Based on these results,the optimal diameter structure was derived using the scale and shape parameter models of the Weibull distribution.[Result]1)In the calibrated and validated data,the mean diameter at breast height(DBH),stem density,stand volume,and stem biomass of L.olgensis plantations simulated by the 3-PGmix model were consistent with the measured values in the plot,with R2 values greater than 0.86 and rRMSE less than 16%.2)At the base age of 30 years,the potential volume productivity of L.olgensis plantations at the maximum annual and periodic average growth increment for the five site grades ranged from 4.50 to 8.11 m3·hm-2a-1 and from 4.53 to 8.18 m3·hm-2a-1,respectively.3)The number of diameter classes,which reached the potential productivity of accumulation,increased monotonically with stand age,and the Shannon index of diameter classes exhibited an increasing trend after a fluctuating pattern at the early growth stage.The two indicators showed higher values under better site quality conditions.[Conclusion]The integration of the process-based 3-PGmix model and the PSO algorithm enables effective estimation of volume potential productivity,compensating for the limitations of traditional methods that do not account for self-thinning and climatic factors,and thereby provides a new approach for potential productivity estimation of forest stands.
黄宏超;雷相东;国红;罗光成;何潇
林木资源高效生产全国重点实验室 国家林业和草原局森林经营与生长模拟实验室 中国林业科学研究院资源信息研究所 北京 100091林木资源高效生产全国重点实验室 国家林业和草原局森林经营与生长模拟实验室 中国林业科学研究院资源信息研究所 北京 100091林木资源高效生产全国重点实验室 国家林业和草原局森林经营与生长模拟实验室 中国林业科学研究院资源信息研究所 北京 100091林木资源高效生产全国重点实验室 国家林业和草原局森林经营与生长模拟实验室 中国林业科学研究院资源信息研究所 北京 100091||北京林业大学林学院 北京 100083林木资源高效生产全国重点实验室 国家林业和草原局森林经营与生长模拟实验室 中国林业科学研究院资源信息研究所 北京 100091
农业科技
立地质量评价3-PGmix模型潜在生产力粒子群优化
site quality assessment3-PGmix modelpotential productivityparticle swarm optimization
《林业科学》 2026 (7)
38-49,12
国家重点研发计划课题"典型人工林立地质量评价与生产力提升技术(2022YFD2200501).
评论