【作者】 付甜；

【导师】 肖文发；

【作者基本信息】 中国林业科学研究院 ， 生态学， 2013， 博士

【摘要】 森林是陆地生态系统中最大的碳库,它对全球碳循环的贡献是不可比拟的。三峡库区地处长江流域中下游,生态地位特殊,是我国生态环境保护的重点地区,库区森林生态系统在保持水土、涵养水源、增加碳汇等方面均发挥了重要作用。采用模型模拟方法对库区森林生态系统进行碳储量和生产力的研究,一方面模型模拟的结果有利于更客观地评估三峡库区森林的生态功能；另一方面,通过对模型参数的本土化改进,有利于探索更精确的区域性碳计量的理论与技术方法。本文以长江三峡库区的马尾松林、杉木林等主要森林类型为研究对象,以加拿大碳收支模型CBM-CFS3为平台,先收集森林资源清查的一类调查样地的连续调查数据,运用若干种理论生长函数模型,通过比较得出各森林类型最优的林分水平的年龄—蓄积产量曲线,作为模型模拟森林碳储量和生产力的驱动力；然后通过收集大量文献样地资料,采用幂函数模型和多项logit模型分别估计出各森林类型的蓄积—生物量转换参数和林木器官(干、皮、枝、叶)生物量比例参数。最终,应用估算的产量曲线、生物量转换和比例参数以及收集到的生物量周转和凋落物分解参数,以1999年三峡库区森林资源清查二类调查数据中的林龄,面积等数据为基础,实现三峡库区的森林生态系统碳储量和生产力的模拟计算。模型参数估计的主要结论如下：(1)以Richards、Logistic和Korf生长函数为备选模型,运用非线性回归方法,分别建立林龄-蓄积量生长模型,并通过模型拟合统计量和抽样精度检验的结果比较,对各个模型进行评价,选出适合各森林类型的最优模型。结果表明：Richards模型表现出较强的适应性,其次为Logistic模型、Korf模型。各主要森林类型的最优曲线拟合结果表现良好,决定系数R2均达到0.5以上,均方根误差在10.99m3·hm"2～21.07m3·hm-2之间,精度检验的结果在44%～80%之间。各森林类型的蓄积预测值与实测值的残差呈正态分布,残差有随林龄的增大而增大的趋势。针叶混交林的生长潜力最大,其蓄积生长极限达到352.356m3·hm-2;柏木林长势较差,蓄积生长极值不超过80m3·hm-2。(2)蓄积-生物量转换模型的拟合结果除常绿阔叶林外均关系显著,预测残差随自变量的增大而升高,均方根误差均控制在6.520t·hm-2～23.123t·hm-2,精度检验结果在31.14%～91.79%之间,除常绿阔叶林以外,预测精度均达到70%以上。林木器官生物量比例模型拟合的结果除常绿阔叶林外均关系显著,预测残差随自变量的增大而减小。干、皮、枝和叶生物量比例模型的均方根误差均不超过0.1,分别达到0.031～0.085,0.005～0.041,0.029～0.103和0.016～0.083,其精度检验结果分别达到91.04%～96.14%,62.71%～94.48%,63.70%～94.47%和-8.86%～83.92%。各森林类型(除常绿阔叶林外)的乔木层地上生物量模型可行性较高,模型所得生物量参数也可为亚热带森林生物量参数研究提供参考。模型模拟主要结果如下：(1)三峡库区森林生态系统的死有机质(DOM)碳库的碳储量达到103.636×106t,其中地下DOM碳储量是地上DOM碳储量的3倍以上。对不同森林类型而言,马尾松林的DOM碳储量最高(占库区森林总DOM碳储量的26.1%),其后由大到小依次为落叶阔叶林(22.7%)、针阔混交林(20.2%)、针叶混交林(15.2%)、柏木林(5.5%)、常绿阔叶林(5.0%)、杉木林(3.4%)和温性松林(1.9%)。不同森林类型的DOM碳库的碳密度差异很大,表现为针叶林的DOM碳密度明显小于阔叶林和混交林,尤其是柏木林的各类DOM碳密度均为最低,针阔混交林的DOM碳密度最大。(2)三峡库区森林生态系统总有机碳储量达到151.018×106t,其中土壤有机质的碳储量高达80.163×106t,占总森林生态系统碳储量的53.1%,植被生物量的碳储量为47.382×106t(31.4%),枯落物的碳储量为23.472×106t(15.5%)。不同森林类型的生态系统碳储量差异较大,在3.144×106t到40.706×106t之间,分布面积最大的马尾松林的生态系统碳储量最高,生产力较低的柏木林总有机碳储量最低。三峡库区森林生态系统(植被、凋落物、土壤)的平均碳密度达107.353t·hm-2,低于全国平均水平258.83t·hm-2,远低于约275t.hm-2的世界平均水平(植被、土壤)。平均土壤有机质碳密度为56.984t·hm-2,植被碳密度为33.682t.hm-2,枯落物碳密度为16.685t·hm-2。针叶林的生态系统碳密度要小于阔叶林和混交林,针叶林中又属柏木林的碳密度最小；针阔混交林的碳密度最大；各森林类型的植被、枯落物和土壤有机质碳库的碳密度大小与其总碳密度大小正相关。(3)三峡库区森林生态系统净第一性生产力(NPP)和净生态系统生产力(NEP)的模拟结果表明：库区内森林植被的平均NPP为3.92t C·hm-2·a-1,其中超过三分之二的碳被分配给凋落物,造成2.71t C.hm-Z.a-1的碳排放,平均NEP为1.21t C·hm-2·a-1。库区植被年均固定大气中的C共5.513×106t,其中有1.694×106t C被固存在生态系统中,并因凋落物分解而释放3.819×106Mg C归还给大气。各森林类型的年均NPP在2.20tC·hm-2·a-1～5.08t C·hm-2·a-1之间,柏木林的碳固定能力最低,针阔混交林最高,其它森林类型的NPP由高至低依次为：针叶混交林>常绿阔叶林>落叶阔叶林>马尾松林>杉木林>温性松林。各森林类型的碳固存能力与其固定碳的能力并不一致,年均NEP在0.63t C·hm-2·a-1～-2.14t C·hm-2·a-1之间,由高至低分别为：针叶混交林>马尾松林>杉木林>常绿阔叶林>落叶阔叶林>针阔混交林>温性松>柏木林。通过与实测推算数据的对比验证,CBM-CFS3模型模拟出的森林植被碳密度结果合理；凋落物碳密度的估算结果大于实测推算数据,土壤有机碳密度则较低,远低于全国平均水平；模型死有机质中包含对死木的估算造成了凋落物碳密度的高估,而模型DOM碳库的初始化方法造成土壤有机碳密度的低估；NPP的估计结果基本与实测推算数据相一致。总的来说,模型模拟得出的三峡库区森林生态系统的平均碳密度低于全国平均水平,远低于世界平均水平,但其林龄结构以中幼龄林为主,未来库区森林的固碳潜力巨大。更多还原

【Abstract】 Forest resource is the largest carbon pool in terrestrial ecosystem, which contributes carbon to the global carbon cycle incomparably. The Three Gorges Reservoir Area locates in the upstream to midstream of Yangtze River under a special ecological status, where is the key area of ecological and environmental protection in China. The forest ecosystem in this reservoir plays a important role in soil and water conservation, and carbon sink increase, etc. Simulating the carbon storage and productivity of forest ecosystem using model method, on one hand, can make for a more objective assessment of forest ecological functions of Three Gorges Reservoir Area; on the other hand, is conducive to explore more precise regional carbon accounting theory and technical methods through the localization improvement of model parameters.This study considers the Masson Pine, Chinese Fir and other major forest types in Three Gorges reservoir area as the research objects. Firstly, the Age-Volume yield curves are created by comparing sevaral growth functions based on forest resource inventory of continuous sample survey data for each forest type, as the driving force for model to simulate forest carbon stocks and productivity; Secondly, volume-to-biomass conversion parameters and forest biomass component (stem, bark, branch, foliage) proportion parameters of every main forest type were estimated using the power function model and multiple logit model and plots by collecting a large number plots from literature data in various forest types.Ultimately, the model simulation of forest ecosystem carbon stocks and productivity in Three Gorges Reservoir finished, using the estimated yield curve, volume-to-biomass conversion parameters and forest biomass component proportion parameters parameters, biomass turnover and litter decomposition parameters searched from literature,adding the base information as age, area, etc. in Three Gorges Forest Inventory data.Model parameters estimation results are as follows: (1) The Age-Volume growth models of main forest types in Three Gorges Reservoir Area were estimated using the mathematical statistics method of nonlinear regression, based on the optimal model of Richards, Logistic and Korf growth function. Finally the most suitable models for every forest type were chosen through comparing the results of model fit statistics and sampling accuracy test. It shows that the Richards growth function appears its strong adaptability, the Logistic and Korf growth function. All the fitting results of best curves for every forest type are good, their determine coefficient R2are all above0.5, and the RMSE varied from10.99m3·hm-2to21.07m3·hm-2, the results of accuracy test are between44%-80%. The residuals between predicted and observed volume which show normal distribution, increased with the age rising. Coniferous mixed forest has the greatest potential to growth, its accumulation growth limit is352.356m3·hm-2, and the cypress forest is poorest, with the growth extreme value less than80m3·hm-2. All the growth models can provide reference and basis to stand growth prediction and forest management for this region.(2) The model fitting results of volume-to-biomass conversion for every forest type show good relations except evergreen broad-leaved forest, and the predicting residuals increased with the increment of the independent variable value. The root mean square errors of them are all controlled within6.520t·hm-2-23.123t·hm-2. The results of accuracy test are between31.14%-91.79%, all the forest types of it except evergreen broad-leaved forest are above70%. On the other hand, The results of proportion models for biomass components show significant relationship in addition to evergreen broad-leaved forest, and the predicting residuals decreased with the independent variable value increased. The root mean square error of stem, bark, branch and leaf biomass proportion of total tree biomass were all less than0.1, as0.031～0.085,0.005～0.041,0.029～0.103and0.016～0.083respectively. Their results of accuracy test reach91.04%～96.14%,62.71%～94.48%,63.70%-94.47%and-8.86%～83.92%. The models obtained (except the models for evergreen broad-leaved forest) in this paper are reasonable, and the model parameters also can offer reference for the study of the subtropical forest biomass parameters. Model simulation results are as follows:(1) The carbon storage of dead organic matter (DOM) pool of forest ecosystems reaches103.636×106t in Three Gorges Reservoir Area, and the carbon storage of underground DOM is3times more than the aboveground. For different forest types, Masson Pine forest has the highest carbon stocks in DOM (accounts for26.1%to total DOM carbon stocks), follow in descending order are deciduous broad-leaved (22.7%), coniferous and deciduous mixed (20.2%), coniferous mixed (15.2%), Cypress(5.5%), evergreen broadleaved (5.0%), Chinese Fir (3.4%) and temperate pine forest (1.9%). The carbon stocks in DOM carbon density of different types of forest varied greatly, shows that the coniferous forest is significantly less than broadleaved and mixed forests, especially the Cypress forest types which has the lowest DOM carbon intensity.(2) The carbon storage of forest ecosystem reaches151.018×106t in Three Gorges Reservoir Area, where soil organic carbon stock is up to80.163×106t, accounts for53.1%to the total forest ecosystem carbon stocks, and with the biomass carbon stock of47.382×106t (31.4%), litter carbon reserves of23.472×106t (15.5%). The forest ecosystem carbon stocks of different forest are quite different from3.144x106t to40.706×106t. The Masson Pine forest which has the largest distribution area gets the highest ecosystem carbon stocks, while the Cypress forst with lower productivity reserves a minimum. The average forest ecosystems carbon density (vegetation, litters and soil) of107.353t·hm-2of Three Gorges Reservoir Area is below the national average level of258.83t·hm-2, which is far below the world average level of275t·hm-2(vegitation and soil). The average soil organic carbon density is56.984t·hm-2, with the vegetation of33.682t·hm-2, litter of16.685t·hm-2. For different forest types, the ecosystem carbon density of coniferous forest is less than broadleaved and mixed forests, in which the Cypress forest owns the minimal carbon density, the conifer-deciduous mixed forest gets the maximum carbon density.(3) The simulation results of Net Primary Productivity (NPP) and Net Ecosystem Productivity (NEP) for forest ecosystem in Three Gorges Reservoir show that:the average vegetation NPP is3.92t C·hm-2·a-1, of which more than two thirds of the carbon is assigned to litters, results in the emissions of2.71t C·hm-2·a-1. The average NEP is1.21t C·hm-2·a-1. The vegetation fixes5.513×106t atmospheric C every year, in which1.694×106t C sequestrate in ecosystems, and release3.819x106t C returned to the atmosphere because of litter decomposition.. The average annual NPP of every forest types varies from2.20-5.08t C·hm-2·a-1. The carbon fixation capacity of forest in descending order as follows:coniferous and deciduous mixed> coniferous mixed> evergreen broad-leaved> deciduous broad-leaved> Masson Pine> Chinese Fir> temperate pine> Cypress forest. The carbon sequestration capacity are not consistent with its fixation ability fixed carbon, with average annual NEP at0.63t C·hm-2·a-1～2.14t C·hm-2·a-1. The carbon sequestration capacity of forest in descending order as follows:coniferous mixed> Masson Pine> Chinese Fir> evergreen broad-leaved> deciduous broad-leaved> coniferous and deciduous mixed> temperate pine> Cypress forest.By the comparison with estimated results used measured data, CBM-CFS3model simulation results of vegetation carbon density are reasonable, but the litter and soil organic carbon density express the overestimation and underestimation respectively, the dead organic matter including the deadwood and the DOM initialization methods may be the main reasons. NPP estimation results are basically consistent with the measured projection data. Overall, the average forest ecosystems carbon density of Three Gorges Reservoir Area derived from model is lower than the national, and far below the world level. But benefit from its young-middle age structure, the carbon sequestration potential of forest in this area will appear quickly in the future.更多还原