基于机载激光雷达数据的森林蓄积量模型研建

【目的】建立变量相同、结构稳定、具有普适性的基于机载激光雷达数据的森林蓄积量预估模型,为规范森林蓄积量建模与评价提供科学参考,为森林资源调查提供计量依据。【方法】利用东北林区落叶松林、红松林、杨树林和桦树林4种森林类型790块样地的激光雷达数据和地面实测蓄积量数据,首先采用多元线性回归和非线性回归方法,分别建立基于机载激光雷达数据的森林蓄积量回归模型,通过对比分析,确定具有相同变量和统一结构形成的普适性模型;然后采用哑变量建模方法,建立基于相同激光雷达变量的不同森林类型蓄积量模型。【结果】4种森林类型线性蓄积量回归模型的解释变量个数在2～6之间,确定系数(R~2)在0.701～0.827之间;非线性蓄积量回归模型的解释变量个数在2～4之间,R2在0.707～0.818之间。基于点云平均高度和平均强度的落叶松林、红松林、杨树林、桦树林非线性二元蓄积量模型,其R~2分别为0.679、0.814、0.698和0.703,平均预估误差分别为4.26%、2.90%、3.68%和3.83%,平均百分标准误差分别为24.44%、18.23%、21.47%和23.26%。【结论】基于机载激光雷达数据估计森林蓄积量,非线性模型优于线性模型;基于点云平均高度和平均强度的二元蓄积量模型具有普适性,可作为森林蓄积量估计的标准模型;本研究建立的4种森林类型蓄积量模型,其预估精度均达到森林资源调查相关技术规定要求,可在实践中推广应用。     

基于机载激光雷达数据估计林分蓄积量及平均高和断面积

基于东北林区191个红松林(Pinus koraiensis)样地的机载激光雷达数据和地面实测数据,首先,通过多元线性回归和非线性回归估计方法,确定林分蓄积量及平均高、断面积的基础回归模型;然后,利用误差变量联立方程组方法,建立基于激光雷达变量的林分蓄积量与平均高、断面积的模型系统。结果显示:建立的多元线性、多元和二元非线性林分蓄积量回归模型,其确定系数R~2分别为0.858,0.846和0.821,平均预估误差MPE分别为2.57%,2.66%和2.85%,平均百分标准误差MPSE分别为26.35%,16.35%和17.88%;利用模型系统对林分平均高、断面积和蓄积量进行估计,其R~2分别为0.597,0.750和0.822,MPE分别为1.90%,2.52%和2.84%,MPSE分别为10.85%,15.28%和17.73%。结果表明:基于机载激光雷达数据估计林分蓄积量、平均高等主要森林参数,非线性模型优于线性模型,而且基于点云高度变量(中位数)和强度变量(75%分位数)的二元非线性模型就能达到比较理想的预估效果;误差变量联立方程组方法,是建立林分蓄积量与平均高、断面积回归模型系统的一种可行方法;所建立的东北红松林平均高、断面积和蓄积量联立模型,其预估精度达到森林资源调查相关技术规定要求,可以在实践中推广应用。     

北京市主要森林类型蓄积量生物量碳储量模型研建

林分水平的蓄积量、生物量和碳储量模型,是开展森林资源规划设计调查的计量基础。基于北京市2016年森林资源连续清查的1 425个乔木林样地数据,分别利用非线性独立回归估计、误差变量联立方程组和含哑变量的误差变量联立方程组方法,建立了油松林、侧柏林、栎树林、桦木林、榆树林、刺槐林、杨树林、其他硬阔林、其他软阔林、乔木经济林等10种主要森林类型的林分蓄积量、生物量和碳储量模型。结果显示:10种主要森林类型的蓄积量、生物量和碳储量模型的确定系数(R~2)都在0.93以上,总体相对误差(TRE)和平均系统误差(ASE)都在±3%以内且多数趋近于0,平均预估误差(MPE)都在5%以内,平均百分标准误差(MPSE)都在15%以内。结果表明:不同森林类型的蓄积量主要取决于林分断面积和平均高,生物量主要取决于蓄积量和林分平均高;含哑变量的非线性误差变量联立方程组方法,是建立林分水平三储量(森林蓄积量、生物量和碳储量)模型系统的可行方法;所建北京市10种主要森林类型的蓄积量、生物量和碳储量模型,其预估精度达到相关技术规定要求,可以在实践中推广试用;为进一步提高模型的准确度,可采用基于二元模型计算的蓄积量和生物量样地数据对所建模型进行修正。     

北京市二类调查小班蓄积量预估模型研究

为解决当前北京市二类调查通过角规绕测技术预估林分蓄积量存在的问题,基于北京市二类调查数据,根据优势树种(组)的不同,将北京市森林划分成10个不同的树种组。在此基础上,利用一类清查数据,以林分蓄积量为因变量,林分参数及立地参数为自变量构建非线性蓄积量预估模型,计算确定系数(R~2)、总相对误差(TRE)、估计值的标准差(SEE)、平均系统误差(MSE)、平均预估误差(MPE)和平均百分标准误差(MPSE),并对模型拟合效果进行评价。结果表明:构建的蓄积量预估模型拟合效果较好,各树种组蓄积量预估模型的确定系数(R~2)均大于0.94,MPE均小于5%,MPSE基本在10%以下,可以应用于北京市二类调查中蓄积量的预估。     

湖南省主要树种单木和林分生长率模型研建

以湖南省1999年、2004年、2009年和2014年4期的森林资源连续清查为基础，利用大量、连续、系统的固定样地和样木数据，根据胸径和生长率的一般分布规律，选取常用的生长率回归方程作为基础模型式，采用非线性回归估计方法，构建了11个树种组的单木胸径生长率和材积生长率模型，以及9个树种组的林分蓄积量生长率模型。结果表明：各模型确定系数R²均在0.88以上，单木生长率模型的总体相对误差和平均预估误差均在4%以内，胸径生长率模型的平均预估误差大部分在10%以内；林分蓄积量生长率模型的平均预估误差和总体相对误差基本在4%以内，蓄积量生长率模型的平均预估误差在20%以内。各项指标表明，拟合模型能满足精度要求，具有较高的实用性，可为湖南省森林资源年度更新和森林经营管理提供技术支撑。     

基于无人机机载激光雷达构建森林蓄积量模型反演及精度分析

基于机载激光雷达点云数据提取的嵩县测区森林资源数据，结合样地调查数据，采用多元线性回归模型，重点分析无人机载激光雷达获取的点云数据在森林蓄积量模型反演方面的精度分析，为河南省森林蓄积量的测算提供参考依据。结果显示：山区栎类蓄积量调整决定系数AdjR2=0.890m³·hm^-2，平均相对误差MSE=0.237 m³·hm^-2，均方根误差RMSE=0.478 m³·hm^-2，结合分层地面样地调查数据对山区栎类蓄积量数据进行多元线性回归模型反演，模型精度为96.01%。无人机机载激光雷达能够自动获取大面积栎类标准地的激光点云数据，可以提取森林的垂直结构信息(高度参数)和水平结构信息(郁闭度)具备三维结构参数提取能力，通过全覆盖的激光雷达数据反演结果以及地面验证两个部分的数据验证，得到的精度测算结果较好，为计算森林蓄积量提供新的方法。     

基于森林资源连续清查数据的西南地区云南松林分生长模型研究

为掌握西南地区云南松林分生长规律,以森林资源连续清查568块固定样地的云南松林分调查数据为基础,系统研究了云南松林分平均树高、林分断面积、林分立地指数、林分蓄积量因子与林分平均年龄、林分优势木平均高、林分公顷株数等的关系,建立了云南松林分断面积生长和林分收获预估模型。模型拟合结果表明,模型具有较高的精确度和稳定性,可应用于西南地区云南松林的经营管理和收获预估。     

基于无人机LiDAR特征变量的南方集体林区蓄积量估测

基于广西壮族自治区森林资源年度监测评价成果数据,采用逐步回归选择机载激光雷达特征变量,建立多元线性回归、Logistic回归和随机森林模型,预测南方集体林区桉树、杉木和天然阔叶林样地的蓄积量。结果表明:1)桉树和杉木样地的逐步回归特征变量多为高度和强度变量,而天然阔叶林样地则是间隙率、覆盖度、叶面积指数等综合变量;2)桉树和天然阔叶林样地,随机森林模型的蓄积量估测精度(桉树R~2=0.97,RMSE=12.60m~3/hm~2;天然阔叶林:R~2=0.90,RMSE=18.45m~3/hm~2)高于多元线性回归和Logistic回归模型,而杉木样地在多元线性回归模型中得到了最优的蓄积量估测结果(R~2=0.91,RMSE=24.30m~3/hm~2);3)在3种模型估测精度中,人工桉树和杉木样地均优于天然阔叶林样地。可见,高密度的激光雷达点云可以获取更优的特征变量,针对复杂的样地条件需要灵活选择估测模型实现蓄积量调查,以便为林草部门进行森林资源调查、监测和经营管理提供科学依据。     

黑龙江人工落叶松林分蓄积量生物量和碳储量建模

林分水平的计量数表是开展森林资源二类调查必不可少的度量衡。利用黑龙江省国家森林资源清查的53个人工落叶松(Larix spp.)样地实测数据,结合已经颁布实施的《落叶松立木生物量模型及碳计量参数》行业标准,建立了林分水平的基于断面积和平均高的蓄积量、生物量和碳储量模型。结果表明:所建模型的确定系数(R2)在0.91以上,平均预估误差(MPE)在4%左右,可以满足森林资源二类调查的精度要求;提出的建模方法,也适用于建立其它地区或其它树种的林分蓄积量、生物量和碳储量模型。     

哑变量在森林蓄积量模型估测中的应用

【目的】建立含哑变量的林分蓄积量估测模型，分析哑变量在香格里拉高山松林分蓄积量模型中的意义与作用。【方法】以香格里拉为研究区，基于2008—2009年3幅TM遥感影像与2008年抽样控制样地数据，对香格里拉高山松林分神经网络模型与考虑龄组构造的哑变量神经网络模型两种类型建立蓄积量遥感估测模型，并进行精度评价。对比模型的估测值与实测值，计算模型残差，检验各龄组残差均值与0之间的差异性；同时对模型的预测值结果进行组间均值的差异性检验，以此作为确定龄组分类形式构建哑变量的标准与依据。【结果】2个模型的独立样本检验结果表明，引入哑变量的神经网络估测模型比神经网络模型拟合效果要好，其决定系数要高于神经网络模型，决定系数从0.516提高到0.783。模型预估精度从神经网络模型的66.3%提高至哑变量模型的74.8%，估算误差优于神经网络模型。【结论】根据模型的残差差异性结果得出，哑变量模型可以在一定程度上解决在估测幼龄林、中龄林蓄积量低值高估的问题；可见引入哑变量估测森林蓄积量的方法是相对有效的。     

Predicting stand damage and tree survival in burned forests in Catalonia (North-East Spain)

The study developed models for predicting the post-fire tree survival in Catalonia. The models are appropriate for forest planning purposes. Two types of models were developed: a stand-level model to predict the degree of damage caused by a forest fire, and tree-level models to predict the probability of a tree to survive a forest fire. The models were based on forest inventory and fire data. The inventory data on forest stands were obtained from the second (1989–1990) and third (2000–2001) Spanish national forest inventories, and the fire data consisted of the perimeters of forest fires larger than 20 ha that occurred in Catalonia between the 2nd and 3rd measurement of the inventory plots. The models were based on easily measurable forest characteristics, and they permit the forest manager to predict the effect of stand structure and species composition on the expected damage. According to the stand level fire damage model, the relative damage decreases when the stand basal area or mean tree diameter increases. Conversely, the relative stand damage increases when there is a large variation in tree size, when the stand is located on a steep slope, and when it is dominated by pine. According to the tree level survival models, trees in stands with a high basal area, a large mean tree size and a small variability in tree diameters have a high survival probability. Large trees in dominant positions have the highest probability of surviving a fire. Another result of the study is the exceptionally good post-fire survival ability of Pinus pinea and Quercus suber.     

Probability models for pine twisting rust (Melampsora pinitorqua) damage in Scots pine (Pinus sylvestris) stands in Finland

Factors affecting the probability that pine twisting rust (Melampsora pinitorqua) damage occur in a Scots pine (Pinus sylvestris) stand were analysed using the 7th Finnish National Forest Inventory data (NFI7) from southern Finland in 1977–1983. The inventory was based on systematic sampling. The NFI7 data was measured in clusters, each of which consisted of 21 sample plots. In addition to the stand and site characteristics measured for forest management planning purposes, the data included records of damage by pine twisting rust and occurrence of aspens (Populus tremula, the other host plant of the pathogen) in the stands. Two multilevel logit models were developed for predicting the overall probability of pine twisting rust damage and the probability of severe pine twisting rust damage. Site and stand characteristics were used as explanatory variables in the models. Residual variance in the models was studied on the inventory crew, cluster and year levels. The occurrence of aspens and site fertility were the most important factors increasing the probability that pine twisting rust damage will occur in a stand. The damage probability also decreased with increasing effective temperature sum calculated for the location. The overall damage probability was equally high on peatlands and on mineral soil if there were aspens in the stand. If, however, there were no aspens in the stand, the probability of damage was higher on mineral soils than on peatlands. In addition, the overall probability was lower in naturally regenerated stands than in planted or sown stands, and it decreased with increasing mean age of pines. In both models, the residual variance was significant on the both the inventory crew and the cluster levels.     

基于国家森林资源清查数据的不同生物量和碳储量估计方法的对比分析

[目的]通过对不同生物量和碳储量的估计方法进行对比分析,为确定在国家森林资源清查中生物量和碳储量的具体估计方法提供依据。[方法]以广东省2012年森林资源清查的100个杉木林和80个马尾松林的实测样地资料为基础,利用近年来我国建立的主要树种立木生物量模型,对改进IPCC法、生物量模型法和转换因子连续函数法(即方精云法)3种方法按一元和二元模型共6种方案进行了对比;同时,基于改进IPCC法一元和二元模型的生物量估计值,用平均含碳系数法、组分含碳系数法和固定含碳系数(0.5或0.47)法分别对碳储量进行估计。[结果]用二元生物量模型法得到的杉木林和马尾松林样地的总生物量分别为320 Mg和331 Mg,一元生物量模型法的结果分别相差0.9%和6.2%;改进IPCC法的估计结果,采用二元和一元模型时杉木林分别相差-3.6%和-11.9%,马尾松林分别相差-8.5%和-19.6%;而方精云法的估计结果,采用二元和一元模型时杉木林分别相差6.65倍和6.60倍,马尾松林分别相差-14.3%和-18.0%。平均含碳系数法和组分含碳系数法的碳储量估计结果,杉木林仅相差0.2%,马尾松林相差约0.4%;固定含碳系数法的估计结果因树种而异,对杉木林要低估0.6%5.4%,对马尾松林要低估3.3%9.1%。[结论]对生物量的估计,采用生物量模型法准确性最高,而林木水平的生物量模型其预估精度要高于林分水平的模型;IPCC法是基于材积源的通用方法,将其中的缺省参数改进为可变参数模型,可大大提高方法的适应性;方精云法只是基于IPCC法所建立的林分水平模型在大尺度上的一种具体应用方法,其精度要低于林木水平的生物量模型法,不适于中小尺度应用。对碳储量的估计,采用平均含碳系数法与组分含碳系数法差异很小,但采用固定含碳系数法则误差较大。     

Compatible volume and taper models for economically important tree species of Turkey

• Introduction  

The accurate estimation of stem taper and volume are crucial for the efficient management of the forest resources. Compatible segmented polynomial taper and volume equations were developed for Brutian pine (Pinus brutia Ten.), Lebanon cedar (Cedrus libani A. Rich.), Cilicica fir (Abies cilicica Carr.), Scots pine (Pinus sylvestris L.), and Black pine (Pinus nigra Arnold.).     

基于RS、GIS的马尾松林分蓄积量判读模型研究

以从RS、GIS可提取的因子为自变量 ,通过数量化、逐步聚类、逐步回归等方法建立马尾松林分蓄积量判读模型 ,并用不同时相的遥感数据、连续清查第三次复查外业调查数据对模型进行适合性检验、精度计算。结果表明 ,所建立的马尾松林分蓄积量判读模型线性关系显著 ,估算结果与外业调查数据无显著差异 ,且估算精度达到连续清查规程的要求 ,判读模型可应用于森林连续清查间隔期内广东省马尾松林分蓄积量的估算。     

Estimating foliage biomass in Scots pine (Pinus sylvestris) and Norway spruce (Picea abies) plots

Dynamic decomposition models are needed to estimate changes in the carbon stock of boreal soil because these changes are difficult to measure directly. An important aboveground carbon flux to the soil is foliage litterfall. To estimate this flux, both the amount and the turnover rate of the foliage biomass component must be known. Several methods for estimating foliage biomass of Scots pine (Pinus sylvestris L.) and Norway spruce (Picea abies (L.) Karst.), including biomass equations and biomass expansion factors (BEFs), were compared with predicted foliage biomass based on forest inventory plot-level measurements. Measured foliage biomass was up-scaled from the branch-level to the plot-level by combining forest inventory variables (diameter, height, height at the crown base and crown base diameter) based on the assumptions of pipe model theory. Combining the foliage biomass: cross-sectional area ratio with the forest inventory variables provided accurate estimates of foliage biomass at the plot-level for plots in southern Finland. The results emphasize the need to test biomass equations with independent data, especially when the equations applied are based on neighboring regions.     

Predicting mortality for five California conifers following wildfire

Fire injury was characterized and survival monitored for 5677 trees >25 cm DBH from five wildfires in California that occurred between 2000 and 2004. Logistic regression models for predicting the probability of mortality 5-years after fire were developed for incense cedar (Calocedrus decurrens (Torr.) Florin), white fir (Abies concolor (Gord. & Glend.) Lindl. ex Hildebr.), sugar pine (Pinus lambertiana Douglas), Jeffrey pine (P. jeffreyi Balf.), and ponderosa pine (P. ponderosa C. Lawson). Differences in crown injury variables were also compared for Jeffrey and ponderosa pine. Most mortality (70–88% depending on species) occurred within 2 years post-wildfire and had stabilized by year 3. Crown length and crown volume injury variables predicted tree mortality equally well; however, the variables were not interchangeable. Crown injury and cambium kill rating was significant in predicting mortality in all models. DBH was only a significant predictor of mortality for white fir and the combined ponderosa and Jeffrey pine models developed from the McNally Fire; these models all predicted increasing mortality with increasing tree size. Red turpentine beetle (Dendroctonus valens) was a significant predictor variable for sugar pine, ponderosa pine, and Jeffrey pine; ambrosia beetle (Trypodendron and Gnathotrichus spp.) was a significant predictor variable for white fir. The mortality models and post-fire tree survival characteristics provide improved prediction of 5-year post-wildfire tree mortality for several California conifers. The models confirm the overall importance of crown injury in predicting post-fire mortality compared to other injury variables for all species. Additional variables such as cambium kill, bark beetles, and tree size improved model accuracies, but likely not enough to justify the added expense of data collection.     

天然林区小班森林资源数据的更新模型

以吉林省汪清林业局为例,根据１９９７年森林经理调查的８４８块固定样地数据,与全林整体模型方法相结合,建立了适合于天然林区林业局（场）无人为干预小班森林资源数据更新的林分级生长模型组。该组模型包括林分密度指数,平均高,断面积,形高,郁闭度等林分测算因子的生长或变化模型。     

GROWTH AND YIELD MODELS FOR DAHURIAN LARCH PLANTATIONS

Growth and yield models were developed for individual tress and stands based on336 temporary plots with 405 stem analysis trees of dahurian larch(Larix gmelinii(Rupr.)Rupr.)plantations throughout Daxing’anling mountains.Several equations were selected using nonlinearregression analysis.Results showed that the Richards equation was the best model for estimatingtree height,stand mean helght and stand dominant height from age; The Power equation was thebest model for prediction tree volume from DBH and tree height; The logarithmic stand volumeequation was good for predicting stand volume from age,mean height,basal area and other standvariables.These models can be used to construct the volume table, the site index table and other for-estry tables for dahurian larch plantations.     

ROTATION

The computer model ROTATION was developed to calculate and compare optimal rotation ages for even-aged forest stands according to mean annual increment, money yield table, forest rent, land expectation value, present net worth, internal rate of return, and financial maturity criteria. The program was written in Microsoft QuickBasic and h e input variables consist of volume yield data, stumpage price, land cost, stand establishment cost, stand management cost, and the rate of interest. Results are displayed in tabular format and values indicating the optimal rotation age based on each of the seven criteria are presented. A representative example is included which incorporates normal yield data for ponderosa pine (Pinus oonderosa Doual. ex Laws.) and commercial timber management revenues and expenditures approximating those currently countered in the Sierra Nevada and southern Cascades, USA. Potential users of ROTATION include forest managers and natural resource educators.