基于ICESatGLAS的云南省森林地上生物量反演

结合机载、星载激光雷达对GLAS(地球科学激光测高系统)光斑范围内的森林地上生物量进行估测,并利用MODIS植被产品以及MERIS土地覆盖产品进行了云南省森林地上生物量的连续制图。机载LiDAR扫描的260个训练样本用于构建星载GLAS的森林地上生物量估测模型,模型的决定系数(R2)为0.52,均方根误差(RMSE)为31Mg/ha。研究结果显示,云南省总森林地上生物量为12.72亿t,平均森林地上生物量为94Mg/ha。估测的森林地上生物量空间分布情况与实际情况相符,森林地上生物量总量与基于森林资源清查数据的估测结果相符,表明了利用机载LiDAR与星载ICESatGLAS结合进行大区域森林地上生物量估测的可靠性。     

集成Landsat OLI和机载LiDAR条带数据的亚热带森林生物量制图

以亚热带天然次生林为研究对象,借助一个条带的少量LiDAR点云数据和覆盖整个研究区的免费Landsat OLI多光谱数据,并结合地面实测数据,探索森林生物量低成本高精度制图方法。首先,提取了OLI和LiDAR特征变量,并与地上和地下生物量进行相关分析以筛选变量;然后,借助LiDAR数据覆盖区的样地和条带LiDAR数据构建"LiDAR生物量模型";再从LiDAR反演生物量的结果中进行采样,结合OLI特征变量构建"LiDAR-OLI模型";最后,与单独使用OLI多光谱数据建立的"OLI估算模型"结果进行比较,分析精度并验证新方法的效果。结果表明,"LiDAR-OLI模型"对地上和地下生物量的模型拟合效果较好且均优于"OLI模型",且其交叉验证的精度也较高并优于"OLI模型",从而证明了新方法的可靠性及有效性。本研究为主、被动遥感技术在中小尺度上协同反演森林参数提供了实验基础,也为基于全覆盖免费OLI多光谱数据及条带LiDAR数据的低成本森林生物量制图探索了技术路线。     

WorldView-2纹理的森林地上生物量反演

使用高空间分辨率卫星WorldView-2的多光谱遥感影像,构建植被指数和纹理因子等遥感因子与森林地上生物量的关系方程,并计算模型估测精度和均方根误差,探索高分辨率数据的光谱与纹理信息在温带森林地上生物量估测应用中的潜力。以黑龙江省凉水自然保护区温带天然林及天然次生林为研究对象,通过灰度共生矩阵(GLCM)、灰度差分向量(GLDV)及和差直方图(SADH)对高分辨率遥感影像进行纹理信息提取,并利用外业调查的74个样地地上生物量与遥感因子建立参数估计模型。提取的遥感因子包括6种植被指数(比值植被指数RVI、差值植被指数DVI、规一化植被指数NDVI、增强植被指数EVI、土壤调节植被指数SAVI和修正的土壤调节植被指数MSAVI)以及3类纹理因子(GLCM、GLDV和SADH)。为避免特征变量个数较多对估测模型造成过拟合,利用随机森林算法对提取的遥感因子进行特征选择,将最优的特征变量输入模型参与建模估测。采用支持向量回归(SVR)进行生物量建模及验证,结果显示选入模型的和差直方图均值(sadh_mean)、灰度共生矩阵方差(glcm_var)和差值植被指数(DVI)等遥感因子对森林地上生物量有较好的解释效果;植被指数+纹理因子组合的模型获得较精确的AGB估算结果(R2=0.85,RMSE=42.30 t/ha),单独使用植被指数的模型精度则较低(R~2=0.69,RMSE=61.13 t/ha)。     

多源数据支持下的森林地上生物量估算方法

以Landsat8 OLI（operational land imager）为遥感数据源，森林资源二类调查和地理国情数据为主要辅助数据，对森林地上生物量（aboveground biomass，AGB）进行了反演和估算。以安徽省金寨县的天然林为研究对象，通过计算覆盖研究区Landsat8 OLI的光谱、纹理和地形特征，利用森林资源二类调查、地理国情普查与监测和外业调查数据建立AGB定量反演模型，以此为基础分析了不同特征对于AGB估算的影响。结果表明，基于所采用的方法得到的金寨县的森林地上生物量，最优反演模型的实测值与估算值相对误差为0.708 718，均方根误差为1.318 983，精度较高。依据该模型计算得到金寨县的生物总量为4 723 728 530 t，结果与实际情况符合。该研究对AGB定量反演和研究所采用的方法对于大范围监测森林资源具有可用性。     

基于遥感的某市城市圈森林碳储量估算与空间分布特征

森林生态系统碳储量占有整个陆地生态系统碳储量约50%,利用遥感数据进行森林碳储量估算对加快实现“碳达峰”和“碳中和”具有重要意义。本研究利用Landsat 8 OLI遥感影像和DEM数据提取植被指数和地形因子,转换净初级生产力数据为生物量数据,并利用多元逐步回归分析法建立武汉城市圈森林植被碳储量遥感估算模型。根据统计数据和估算模型得出,武汉城市圈碳储量空间分布表现为东北部和南部山脉区域的碳储量和碳密度较高,而中东部武汉市和黄石市中心区域相对较低,且植被碳密度主要集中在中海拔地区。     

基于Google Earth Engine与机器学习的省级尺度零散分布草地生物量估算

大区域草地地上生物量估算对草地资源利用管理及全球碳循环研究具有重要意义。为高效快速地估算大区域零散分布草地地上生物量，本文选取安徽省为研究区，在谷歌地球云引擎（Google Earth Engine）平台的支撑下，通过机器学习方法建立Landsat 8 OLI及其他辅助数据与地面实测草地地上生物量之间的联系，开展了草地零散分布地区省级尺度草地地上生物量高分辨率估算，并与传统的基于归一化植被指数（NDVI）回归模型进行了比较。研究结果表明，综合利用光谱与地形因子的机器学习方法，估算零散化分布草地地上生物量的精度可以达到65%以上，其中分类回归树（CART）模型R²=0.57，预测精度为68.60%，支持向量机（SVM）模型R²=0.59，预测精度为75.74%，而使用NDVI的回归分析产生的误差较大，R²=0.37，预测精度为57.51%，因此机器学习方法相对于传统基于NDVI的回归分析具有明显优势。另外，谷歌地球云引擎平台数据来源广泛、获取方便，可以高效地实现海量影像数据的预处理及计算分析，大大提升了工作效率，与地面调查数据的结合可实现更大区域乃至全国尺度上的零散分布草地地上生物量高分辨率遥感估算。     

基于GLAS激光雷达反演森林生物量

森林生物量是森林生态系统的重要指标。GLAS大光斑回波信息与森林结构参数存在较强的相关性,适用于森林生物量的反演。本文简要介绍了GLAS激光雷达系统及其特点,利用GLAS的9波形参数对小兴安岭部分地区进行针叶林与阔叶林的生物量估算,结果显示,引入纠正参数后生物量估测模型的决定系数R²由0.657提高到0806,均方根误差（RMSE）减小为35 Mg/ha,表明利用GLAS进行森林地上生物量估测时,需要考虑地形因素对反演精度的影响。     

双台河口国际重要湿地芦苇地上生物量遥感估算

基于多时相的Landsat8 OLI卫星遥感数据,采用面向对象的分类方法,提取双台河口国际重要湿地芦苇分布信息。通过对归一化植被指数(normalized difference vegetation index,NDVI)等6个植被指数与野外实测芦苇地上生物量数据间的统计分析,比较不同植被指数对芦苇地上生物量的敏感性,构建双台河口国际重要湿地芦苇地上生物量遥感反演模型;应用该模型对芦苇地上生物量进行遥感反演以及空间格局分析。结果表明:双台河口国际重要湿地芦苇分布面积为4.39×104hm2,约占该研究区总面积32.96%;选取的6个植被指数均与芦苇地上生物量显著相关(p0.05),其中,以NDVI为变量的幂指数形式的估算模型为芦苇地上生物量遥感估算最优模型,模拟精度为79%,决策系数为0.76;双台河口国际重要湿地芦苇地上生物量呈东高西低和北高南低的分布格局,其平均地上生物量为4 785.5 g/m2,总地上生物量为2.06×106t;本研究结果可为双台河口国际重要湿地生态系统管理和生物多样性保护提供数据支持与科学指导。     

基于WOFOST模型与UAV数据的玉米生长后期地上生物量估算

地上生物量能够有效反映作物的生长状态，其信息的实时估算对产量预测和农田生产管理都有重要意义。作物生长模型因其详尽的生理生化基础和对生长过程数字化描述能力，成为生物量估算的理想模型。近年来，研究人员利用数据同化算法将时间序列遥感数据同化到作物生长模型中，实现了作物模型由基于气象站的点模拟到区域尺度面模拟的外推，使生物量模拟结果同时具备大范围和机理性两个方面的特点。这一模式下，时间序列的遥感数据质量将对生物量模拟精度产生直接影响，作物生长后期受到光谱饱和的影响，遥感数据的作物冠层信息获取能力会出现明显下降，因此有必要对该阶段遥感数据和作物模型的结合方式进行优化，提升生物量模拟精度。本文针对东北地区春玉米生物量遥感估算存在的问题，提出了利用WOFOST作物模型结合无人机（UAV）遥感数据实现作物生长后期生物量准确估算的新思路。新思路首先利用多光谱遥感数据获取WOFOST模型具备较高空间异质性的土壤速效养分参数以提升模型的空间信息模拟能力，使其能在一定程度上摆脱点尺度模拟的限制。同时，结合集合卡尔曼滤波算法将生长前期无人机（UAV）遥感数据同化到模型中，以缩短模型单独运行时间，减少模型运行过程中的参数误差累积，实现无遥感数据参与下的短期作物生长模拟，并输出生长后期相应的生物量模拟结果。最后，本文利用地面实测数据对新方法的生物量模拟精度进行了评价。结果表明，与全生育期数据同化相比，新方法的生物量估算精度有了明显的提升（全生育期同化：R² = 0.45，RMSE = 4254.30 kg/ha；新方法：R²= 0.86，RMSE = 2216.79 kg/ha）。     

高光谱和LiDAR联合反演森林生物量研究

估算森林地上生物量(AGB)对于全球实现碳中和目标至关重要。本文以美国缅因州Howland森林为研究区域,借助地面实测样地数据,对比分析协同不同数据源(高光谱和LiDAR)和机器学习算法(随机森林、支持向量机、梯度提升决策树和K最邻近回归)的研究,以改善Howland森林的生物量估计精度。结果表明,采用LiDAR和高光谱植被指数变量模型的最佳精度分别为0.874和0.868,协同高光谱和LiDAR变量并采用梯度提升决策树回归模型的精度为0.927,即多源遥感数据要优于单一数据源。高光谱和LiDAR数据的协同使用对于提高类似于Howland地区或更广泛区域的生物量估计的准确性,具有普遍的适用性与一定的应用前景。     

Geostatistical modeling using LiDAR-derived prior knowledge with SPOT-6 data to estimate temperate forest canopy cover and above-ground biomass via stratified random sampling

Forest canopy cover (CC) and above-ground biomass (AGB) are important ecological indicators for forest monitoring and geoscience applications. This study aimed to estimate temperate forest CC and AGB by integrating airborne LiDAR data with wall-to-wall space-borne SPOT-6 data through geostatistical modeling. Our study involved the following approach: (1) reference maps of CC and AGB were derived from wall-to-wall LiDAR data and calibrated by field measurements; (2) twelve discrete LiDAR flights were simulated by assuming that LiDAR data were only available beneath these flights; (3) training/testing samples of CC and AGB were extracted from the reference maps inside and outside the simulated flights using stratified random sampling; (4) The simple linear regression, ordinary kriging and regression kriging model were used to extend the sparsely sampled CC/AGB data to the entire study area by incorporating a selection of SPOT-6 variables, including vegetation indices and texture variables. The regression kriging model was superior at estimating and mapping the spatial distribution of CC and AGB, as it featured the lowest mean absolute error (MAE; 11.295% and 18.929 t/ha for CC and AGB, respectively) and root mean squared error (RMSE; 17.361% and 21.351 t/ha for CC and AGB, respectively). The predicted and reference values of both CC and AGB were highly correlated for the entire study area based on the estimation histograms and error maps. Finally, we concluded that the regression kriging model was superior and more effective at estimating LiDAR-derived CC and AGB values using the spatially-reduced samples and the SPOT-6 variables. The presented modeling workflow will greatly facilitate future forest growth monitoring and carbon stock assessments for large areas of temperate forest in northeast China. It also provides guidance on how to take full advantage of future sparsely collected LiDAR data in cases where wall-to-wall LiDAR coverage is not available from the perspective of geostatistics.     

Assessing Performance of Tomo-SAR and Backscattering Coefficient for Hemi-Boreal Forest Aboveground Biomass Estimation

Tomo-SAR technique has been used for hemi-boreal forest height and further forest biomass estimation through allometric equation. Backscattering coefficient especially in longer wavelength (L- or P-band) is thought as a useful parameter for hemi-boreal forest biomass retrieval. The aim of this paper is to assess the performance of vertical backscattering power and backscattering coefficient for hemi-boreal forest aboveground biomass (AGB) estimation with airborne P-band data. The test site locates in southern Sweden called Remningstorp test site, and the in-situ forest AGB ranges from 14 t/ha to 245 t/ha at stand level. Multi-baseline P-band Pol-InSAR data in repeat-path mode collected during March and May in 2007 at Remningstorp test site was used. We found that the correlation coefficient (R) between backscattering coefficient of P-band HH polarization and the in-situ forest biomass reached 0.87. The R for P-band VV backscattering power at 5 m is 0.71 and 10 m is 0.72. Backscattering coefficient in HH polarization and vertical backscattering power at 5 m and 10 m were applied to construct a model for hemi-boreal forest AGB estimation by backward step-wise regression and cross-validation approach. The results showed that the estimated forest AGB ranges from 19 to 240 t/ha, and the constructed model obtained a higher R and smaller RMSE, the value of R is 0.91, RMSE is 30.43 t/ha at Remningstorp test site.     

Stratified aboveground forest biomass estimation by remote sensing data

Remote sensing-assisted estimates of aboveground forest biomass are essential for modeling carbon budgets. It has been suggested that estimates can be improved by building species- or strata-specific biomass models. However, few studies have attempted a systematic analysis of the benefits of such stratification, especially in combination with other factors such as sensor type, statistical prediction method and sampling design of the reference inventory data. We addressed this topic by analyzing the impact of stratifying forest data into three classes (broadleaved, coniferous and mixed forest). We compare predictive accuracy (a) between the strata (b) to a case without stratification for a set of pre-selected predictors from airborne LiDAR and hyperspectral data obtained in a managed mixed forest site in southwestern Germany. We used 5 commonly applied algorithms for biomass predictions on bootstrapped subsamples of the data to obtain cross validated RMSE and r² diagnostics. Those values were analyzed in a factorial design by an analysis of variance (ANOVA) to rank the relative importance of each factor. Selected models were used for wall-to-wall mapping of biomass estimates and their associated uncertainty. The results revealed marginal advantages for the strata-specific prediction models over the unstratified ones, which were more obvious on the wall-to-wall mapped area-based predictions. Yet further tests are necessary to establish the generality of these results. Input data type and statistical prediction method are concluded to remain the two most crucial factors for the quality of remote sensing-assisted biomass models.     

Biomass mapping using forest type and structure derived from Landsat TM imagery

A method for mapping forest biomass was developed and tested on a study area in western Newfoundland, Canada. The method, BIOmass from Cluster Labeling Using Structure and Type (BioCLUST), involves: (i) hyperclustering a Landsat TM image, (ii) automatically labeling the clusters with information about forest type and structure, and (iii) applying stand-level equations that estimate biomass as a function of height and crown closure within forest species-type classes. BioCLUST was validated with biomass values measured at geo-referenced field plots and mapped across the study area using an existing forest management photo-inventory. Root mean square error (RMSE) values ranged from 43 to 79 tonnes/ha, and were lowest for intermediate height classes when validated with field plots. Overall bias was negative at 10 tonnes/ha compared with a negative bias of 3 tonnes/ha estimated for the photo-inventory. Validation of the biomass map gave RMSE values of 37–47 tonnes/ha and overall landscape biomass estimates within 0.4% of biomass mapped by the photo-inventory. BioCLUST offers an alternative to other biomass mapping methods when scene-specific plot data are limited and a photo-inventory is available for a representative portion of a Landsat scene.     

GLAS星载激光雷达和Landsat/ETM+数据的森林生物量估算

基于大脚印激光雷达数据和野外观测数据,该文提出一种获取脚印点内森林生物量的新思路,并结合陆地卫星数据应用于长白山地区森林地上生物量估算。首先,基于3种森林类型(针叶林、阔叶林和针阔混交林),采用多元逐步回归方法建立激光雷达波形指数与脚印点内实测平均树高的回归模型,估算全部脚印点内的平均树高;然后根据脚印点内样方的野外观测数据(平均树高和平均胸径)以及它们与样方生物量的拟合方程估算没有野外调查数据对应的脚印点的生物量;最后对3种森林类型的脚印点森林生物量在各森林覆盖度条件下进行分层分区统计得到生物量等级图。验证比较遥感估算的生物量与野外调查数据推算的生物量,总体误差在0~30(t·hm~(-2))之间,均方根误差为14.66(t·hm~(-2))。     

Estimating forest biomass using small footprint LiDAR data: An individual tree-based approach that incorporates training data

A new individual tree-based algorithm for determining forest biomass using small footprint LiDAR data was developed and tested. This algorithm combines computer vision and optimization techniques to become the first training data-based algorithm specifically designed for processing forest LiDAR data. The computer vision portion of the algorithm uses generic properties of trees in small footprint LiDAR canopy height models (CHMs) to locate trees and find their crown boundaries and heights. The ways in which these generic properties are used for a specific scene and image type is dependent on 11 parameters, nine of which are set using training data and the Nelder–Mead simplex optimization procedure. Training data consist of small sections of the LiDAR data and corresponding ground data. After training, the biomass present in areas without ground measurements is determined by developing a regression equation between properties derived from the LiDAR data of the training stands and biomass, and then applying the equation to the new areas. A first test of this technique was performed using 25 plots (radius = 15 m) in a loblolly pine plantation in central Virginia, USA (37.42N, 78.68W) that was not intensively managed, together with corresponding data from a LiDAR canopy height model (resolution = 0.5 m). Results show correlations (r) between actual and predicted aboveground biomass ranging between 0.59 and 0.82, and RMSEs between 13.6 and 140.4 t/ha depending on the selection of training and testing plots, and the minimum diameter at breast height (7 or 10 cm) of trees included in the biomass estimate. Correlations between LiDAR-derived plot density estimates were low (0.22 ≤ r ≤ 0.56) but generally significant (at a 95% confidence level in most cases, based on a one tailed test), suggesting that the program is able to properly identify trees. Based on the results it is concluded that the validation of the first training data-based algorithm for determining forest biomass using small footprint LiDAR data was a success, and future refinement and testing are merited.