Leaf area index measurements at Fluxnet-Canada forest sites

Leaf area index (LAI) measurements made at 17 forest sites of the Fluxnet Canada Research Network are reported here. In addition to LAI, we also report other major structural parameters including the effective LAI, element clumping index, needle-to-shoot area ratio, and woody-to-total area ratio. Values of the fraction of photosynthetically active radiation (FPAR) absorbed by green leaves in these stands at noon of 15 August are also provided, and a procedure is suggested for using the effective LAI for estimating FPAR at various times of the day and year. Labour-intensive laboratory measurements of the needle-to-shoot area ratio were made for five conifer sites. For each site, 45 shoot samples were measured at three heights from three trees. LAI-2000, TRAC and digital hemispherical photography (DHP) were used in the field, and good agreements between these techniques were obtained. In particular, the low cost DHP technique agreed within 21% of LAI-2000 in terms of effective LAI measurements and 12% of TRAC in terms of element clumping index measurements, suggesting a possibility of using DHP alone for indirect LAI measurements. However, LAI-2000 and TRAC are still found to be more reliable than DHP because of some remaining technical issues with DHP. We confirm the correct method for determining the photographic exposure proposed in previous studies and suggest optimum zenith angle ranges in photograph processing to estimate the effective LAI and the clumping index.     

Using digital repeat photography and eddy covariance data to model grassland phenology and photosynthetic CO2 uptake

The continuous and automated monitoring of canopy phenology is of increasing scientific interest for the multiple implications of vegetation dynamics on ecosystem carbon and energy fluxes. For this purpose we evaluated the applicability of digital camera imagery for monitoring and modeling phenology and physiology of a subalpine grassland over the 2009 and 2010 growing seasons.We tested the relationships between color indices (i.e. the algebraic combinations of RGB brightness levels) tracking canopy greenness extracted from repeated digital images against field measurements of green and total biomass, leaf area index (LAI), greenness visual estimation, vegetation indices computed from continuous spectroradiometric measurements and CO₂ fluxes observed with the eddy covariance technique. A strong relationship was found between canopy greenness and (i) structural parameters (i.e., LAI) and (ii) canopy photosynthesis (i.e. Gross Primary Production; GPP). Color indices were also well correlated with vegetation indices typically used for monitoring landscape phenology from satellite, suggesting that digital repeat photography provides high-quality ground data for evaluation of satellite phenology products.We demonstrate that by using canopy greenness we can refine phenological models (Growing Season Index, GSI) by describing canopy development and considering the role of ecological factors (e.g., snow, temperature and photoperiod) controlling grassland phenology. Moreover, we show that canopy greenness combined with radiation use efficiency (RUE) obtained from spectral indices related to photochemistry (i.e., scaled Photochemical Reflectance Index) or meteorology (i.e., MOD17 RUE) can be used to predict daily GPP.Building on previous work that has demonstrated that seasonal variation in the structure and function of plant canopies can be quantified using digital camera imagery, we have highlighted the potential use of these data for the development and parameterization of phenological and RUE models, and thus point toward an extension of the proposed methodologies to the dataset collected within PhenoCam Network.     

DEVELOPMENT OF A VEGETATION INDEX FOR ESTIMATION OF LEAF AREA INDEX BASED ON SIMULATION MODELING

Leaf area index (LAI) is an important structural variable for quantitative analysis of the energy and mass exchange characteristics of a terrestrial ecosystem. The objective of the research was to use the Scattering by Arbitrarily Inclined Leaves (SAIL) model to develop a new vegetation index for estimating LAI based on the Ratio Vegetation Index (RVI) and Perpendicular Vegetation Index (PVI). In the study, RVIs and PVIs were derived from the SAIL-simulated reflectance, and several potential limitations of RVI and PVI in LAI estimation were identified. First, for a given LAI level, a dark soil background resulted in higher RVI values and overestimated LAI values. The reverse was true for light colored soils. On the contrary, the PVI tended to underestimate LAI for dark soil background and overestimate LAI for light soil background. The RVI behaves oppositely to PVI in LAI estimation for same soil background. Based on these results, a new vegetation index (RMPVI: RVI Multiplied by PVI Vegetation Index) was constructed, and the sensitivity of this index to LAI was then evaluated and the performance of RMPVI in LAI estimation was compared with those of other vegetation indices. The results show that the RMPVI can greatly minimize the soil background influences, and is more sensitive to LAI than other indices, especially when LAI is greater than 2. As for LAI estimation, RMPVI can yield highest R² than other vegetation indices used in the study, with a root mean square error (RMSE) of 0.16, which shows RMVPI is an efficient index for LAI estimation.     

基于多源无人机影像特征融合的冬小麦LAI估算

为探讨无人机多源影像特征融合估测作物叶面积指数的能力,该研究以冬小麦为研究对象,利用多旋翼无人机搭载高清数码相机和UHD185成像光谱仪获取研究区冬小麦关键生育期（扬花期、灌浆期）的可见光和高光谱影像。综合考虑可见光、高光谱影像特征与冬小麦叶面积指数的相关性及影像特征重要性进行特征筛选,然后,以可见光植被指数、纹理特征、可见光植被指数+纹理特征、高光谱波段、高光谱植被指数及高光谱波段+植被指数分别作为输入变量构建多元线性回归、支持向量回归和随机森林回归的叶面积指数估测模型（单传感器数据源）;以优选的两种影像特征结合支持向量回归、随机森林回归构建叶面积指数估测模型（两种传感器数据源）,比较分析单源与多源影像特征监测冬小麦叶面积指数的性能。进一步地,考虑到小区土壤空间异质性会影响冬小麦叶面积指数估测结果,该研究探讨了不同影像采样面积下基于单源遥感数据构建的小麦叶面积指数估测模型精度。研究结果表明：在扬花期和灌浆期,使用两种影像优选特征构建的随机森林回归估测模型精度最佳,验证集决定系数分别为0.733和0.929,均方根误差为0.193和0.118。可见光影像采样面积分别为30%和50%,高光谱影像采样面积为65%时,基于单源影像特征构建的随机森林回归估测模型在扬花期和灌浆期效果最好。综上,该研究结果可为无人机遥感监测作物生理参数提供有价值的依据和参考。     

规则林地覆盖度照相测量法的精度分析

为定量研究照相法测量规则林地覆盖度中采样策略、相机非垂直拍摄等因素对覆盖度计算结果的影响,该文对河北怀来的一片梨树林进行了实地测量,通过对测量数据的统计计算,利用植被结构生成软件Onyx tree professional重建梨树林场景,采用基于物理的光线追踪算法模拟相机拍摄的照片,提出了规则林地覆盖度照相法测量的采样方案,并通过计算机模拟检验了采样方案。通过精确控制模拟相机的拍摄条件,进一步分析了非垂直拍摄对测量精度的影响。结果表明,相机的非竖直拍摄对覆盖度计算结果的精度有较大的影响,误差最大可达27.9%。该文提出的测量方案为规则林地的覆盖度实地测量提供了参考。     

通量贡献区叶面积指数空间分布的测定

使用循环采样设计方案,对海北沼泽化湿地生态系统涡度相关通量观测塔的通量贡献区内叶面积指数进行了实地调查。根据样方叶面积指数的实地测量值和样方的GPS空间定位信息,利用空间插值方法绘制了通量贡献区内的叶面积指数空间分布图。并基于数字摄影与地理信息系统技术,提出了一种测量速度快、计算精度高、适合多种植物叶片叶面积测量的新方法。该方法在ArcGIS的GRID模块下对叶片图像进行格式转化和重新采样处理,使用自行编写的色阶诊断程序提取图像中叶片的叶面积。研究结果表明:此种方法叶面积测量结果与LI-3000A叶面积仪所测的结果具有很好的吻合性,两者的线性回归方程决定系数R2为0.98,叶面积的测量精度完全可以达到实际应用中的测量要求。     

基于HJ-CCD数据和随机森林算法的小麦叶面积指数反演

为给小麦长势的遥感监测提供技术支持,该文运用随机森林回归(RF,random forest)算法建立小麦叶面积指数(LAI)遥感反演模型。首先基于2010-2013年江苏地区小麦环境减灾卫星HJ-CCD的影像数据,提取拔节、孕穗和开花3个生育期的卫星植被指数,进而根据各生育期植被指数和相应实测LAI数据,利用RF算法构建各期小麦LAI反演模型,并以人工神经网络(ANN,artificial neural network)模型为参比模型进行预测精度的比较。结果表明:RF算法模型在3个生育期的预测结果均好于同期的ANN模型。拔节、孕穗和开花3个生育期RF模型预测值与地面实测值的R2分别为0.79,0.67和0.59,对应的RMSE分别为0.57,0.90和0.78;ANN模型的R2分别为0.67,0.31和0.30,对应的RMSE分别为0.82,1.94和1.43。该研究结果为提高大田尺度下的小麦LAI遥感预测精度提供了技术和方法。     

基于高光谱维数约简与植被指数估算冬小麦叶面积指数的比较

高光谱遥感反演LAI时,由于实际样本数远小于光谱维数,易导致基于全谱段建立的模型不稳定。针对该问题,该文提出将基于原始光谱反射率与LAI相关性和基于光谱曲线特征的2种波段选择方式分别与主成分回归(PCR)或偏最小二乘回归(PLSR)结合的高光谱维数约简方法,估算冬小麦LAI。并选择归一化植被指数(NDVI)、增强型植被指数(EVI)、重归一化植被指数(RDVI)、修正土壤调节植被指数(MSAVI)和三角形植被指数(TVI)5种代表性植被指数,利用2009、2010年实测大田冬小麦冠层高光谱和LAI数据,将提出的基于维数约简的方法与基于植被指数的LAI估算方法进行了比较,独立样本集验证结果和交叉验证结果均表明,提出的基于维数约简的方法比基于植被指数方法的估算精度高,在交叉验证结果中,基于维数约简的方法R2最高达到0.818,相应RMSE为0.685。该研究可为后续基于高光谱的LAI估算提供参考。     

Documenting improvement in leaf area index estimates from MODIS using hemispherical photos for Australian savannas

This paper compares estimates of Leaf Area Index (LAI) obtained from the MODIS (Moderate Resolution Imaging Spectroradiometer) collections 4.8 (MC4) and 5.0 (MC5) with ground-based measurements taken along a 900 km north-south transect through savanna in the Northern Territory, Australia. There was excellent agreement for both the magnitude and timing in the annual variation in LAI from MC5 and biometric estimates at Howard Springs, near Darwin, whereas MC4 overestimated LAI by 1-2 m² m⁻² for the first 200 days of the year. Estimates of LAI from MC5 were also compared with those obtained from the analysis of digital hemispherical photographs taken during the dry season (September 2008) based on algorithms that included random and clumped distribution of leaves. Linear regression of LAI from MC5 versus that using the clumping algorithm yielded a slope close to 1 (m = 0.98). The regression based on a random distribution of leaves yielded a slope significantly different from 1 (m = 1.37), with higher Mean Absolute Error (MAE) and bias compared to the clumped analysis. The intercept for either analysis was not significantly different from zero but inclusion of five additional sites that were visually bare or without green vegetation produced a statistically significant offset of +0.16 m² m⁻² by MC5. Overall, our results show considerable improvement of MC5 over MC4 LAI and good agreement between MC5 and ground-based LAI estimates from hemispherical photos incorporating clumping of leaves.     

基于可见光图像和卷积神经网络的冬小麦苗期长势参数估算

针对目前基于计算机视觉估算冬小麦苗期长势参数存在易受噪声干扰且对人工特征依赖性较强的问题,该文综合运用图像处理和深度学习技术,提出一种基于卷积神经网络(convolutional neural network, CNN)的冬小麦苗期长势参数估算方法。以冬小麦苗期冠层可见光图像作为输入,构建了适用于冬小麦苗期长势参数估算卷积神经网络模型,通过学习的方式建立冬小麦冠层可见光图像与长势参数的关系,实现了农田尺度冬小麦苗期冠层叶面积指数(leaf area index,LAI)和地上生物量(above ground biomass, AGB)的准确估算。为验证方法的有效性,该研究采用以冠层覆盖率(canopy cover, CC)作为自变量的线性回归模型和以图像特征为输入的随机森林(random forest, RF)、支持向量机回归(support vectormachinesregression,SVM)进行对比分析,采用决定系数(coefficientofdetermination,R2)和归一化均方根误差(normalized root mean square error, NRMSE)定量评价估算方法的准确率。结果表明:该方法估算准确率均优于对比方法,其中AGB估算结果的R2为0.7917,NRMSE为24.37%,LAI估算结果的R2为0.8256,NRMSE为23.33%。研究可为冬小麦苗期长势监测与田间精细管理提供参考。     

Slope correction for LAI estimation from gap fraction measurements

Digital hemispherical photography poses specific problems when deriving leaf area index (LAI) over sloping terrain. This study proposes a method to correct from the slope effect. It is based on simple geometrical considerations to account for the path length variation within the canopy for cameras pointing vertically. Simulations over sloping terrain show that gap fraction increases up-slope while decreasing down-slope. As a consequence of this balance between up- and down-slope effects, effective LAI estimates derived from inversion of the Poisson model are marginally affected for low to medium slopes (<25°) and LAI (LAI < 2). However, for larger slopes and LAI values, estimated LAI values may be strongly underestimated. The proposed correction was evaluated over four forested sites located over sloping terrain. Results indicate that in these conditions (LAI between 0.6 up to 3.0, clumped canopies with relatively erectophile leaf distribution), the effect of the slope (between 25° and 36°) was moderate as compared to other potential sources of problems when deriving LAI from gap fraction measurements, including clumping, leaf angle inclination and spatial sampling.     

基于低空无人机成像光谱仪影像估算棉花叶面积指数

农作物叶面积指数(leaf area index,LAI)遥感监测具有快速、无损的优势。该文以低空无人机作为遥感平台,使用新型成像光谱仪获取的农田高光谱影像数据对棉花LAI进行反演。利用影像高光谱分辨率的特点,针对传统固定波段植被指数(fixed-bandvegetation index,F_VI)进行改进,通过动态搜索相应植被指数定义所使用波段范围内的反射率极值的方法,计算与各类植被指数对应的极值植被指数(extremum vegetation index,E_VI)。分别以原始全波段光谱反射率、连续投影算法(successive projections algorithm,SPA)提取的有效波段反射率以及各类F_VI和E_VI作为自变量,使用最小二乘和偏最小二乘(partial least squares,PLS)回归等方法构建LAI遥感估算模型。结果显示:1)以植被指数为自变量的模型估算效果(验证R2最高为0.85)优于以光谱反射率作为自变量的模型(验证R2最高为0.59);2)使用E_VI作为自变量能够显著提高LAI的估测精度(验证R2最大提高了0.11);3)使用PLS回归算法结合多个E_VI建立的LAI-E_VIs-PLS模型精度最高。使用LAI-E_VIs-PLS模型对棉花地块高光谱影像进行反演,制作棉花LAI空间分布图,取得良好的估算结果(验证R2=0.88,RMSE=0.29),为农作物LAI遥感监测提供了新的技术手段。     

Leaf area index uncertainty estimates for model-data fusion applications

Estimates of data uncertainties are required to integrate different observational data streams as model constraints using model-data fusion. We describe an approach with which random and systematic uncertainties in optical measurements of leaf area index [LAI] can be quantified. We use data from a measurement campaign at the spruce-dominated Howland Forest AmeriFlux site for illustrative purposes. We made measurements along two transects (one in a mature stand, one in a recently harvested shelterwood) before sunset on successive days using both the Li-Cor LAI-2000 plant canopy analyzer and digital hemispherical photography (DHP). The random measurement uncertainty (1σ) at a given point for a single measurement is about 5% for LAI-2000 and 10% for DHP. These uncertainties are small compared to potential systematic biases due to instrument calibration errors and data processing decisions, which are estimated to be 10-20% for each instrument. Sampling uncertainty (due to the spatial variability along each transect where we conducted our measurements) is an additional, but again relatively small, uncertainty. Assumptions about clumping parameters, for which standard literature values are typically used, remain large sources of uncertainty. This analysis can also be used to develop strategies to reduce measurement uncertainties.     

基于多光谱影像和机器学习算法的红树林树种LAI估算

针对红树林叶面积指数（Leaf Area Index,LAI）实地测量难度大、无法快速大范围LAI估算的问题。该研究以广西北部湾红树林为研究对象,以无人机（Unmanned Aerial Vehicle,UAV）和哨兵二号（Sentinel-2A,S2）多光谱影像为数据源,整合原始光谱波段、植被指数和组合植被指数构建高维数据集,并进行数据降维和特征优选。定量评估6种机器学习算法（XGBoost、前馈反向传播神经网络（Back Propagation,BP）、支持向量机（SVM）、岭回归（Ridge）、Lasso和弹性网络（ElasticNet））对不同红树林树种LAI的估算能力;探究UAV和Sentinel-2A影像对红树林树种LAI估算的精度差异。研究结果表明：1）基于XGBoost算法构建的模型实现了红树林LAI高精度估算,R2均高于0.70,RMSE均低于0.349;2）在UAV和Sentinel-2A影像下,XGBoost模型对不同红树林树种LAI的估算精度（R2）比其他5种模型分别提高了0.105～0.365和0.283～0.540,RMSE降低了0.100～0.392和0.102～0.518;3）UAV影像数据与XGBoost算法构建的模型对海榄雌LAI的估算精度优于其他组合（R2=0.821、RMSE=0.288）,Sentinel-2A影像数据与XGBoost算法构建的模型对秋茄和桐花树LAI的估算精度优于其他组合（R2=0.940～0.979、RMSE=0.142～0.104）,不同红树林树种LAI的估算精度依次为桐花树>秋茄>海榄雌;4）SNAP-SL2P算法整体性低估红树林LAI值,UAV影像红树林树种LAI的平均估算精度（R2=0.677～0.713）均优于Sentinel-2A影像,实现了不同红树林树种LAI的高精度估算。     

基于混合像元分解模型的森林叶面积指数反演

在叶面积指数（LAI）遥感估算中,常用的基于统计的遥感反演方法难以处理"同物异谱,同谱异物"的难题,该文从研究地物组分物理结构着手,采用像元分解的方法建立LAI的遥感反演模型,不仅能很好的解决这个难题,而且反演方法简便可行、具有普适性。该研究先对TM数据做最小噪声分离（MNF）并基于影像本身选择端元,经混合像元分解得到研究区植被覆盖度,再根据植被覆盖度与LAI的关系,用多次散射过程冠层模型求解迭代的方法逐步逼近准确的LAI值。最终选择植被、土壤、水体和水泥建筑4个端元,采用非限制性线性混合像元分解模型来分解影像,4个端元分解影像的平均误差为0.0028,端元质量好,分解效果较好。结果证明：混合像元分解模型和多次散射过程冠层模型相结合来反演森林叶面积指数的方法,能很好的预测森林LAI;研究区样点实测值与预测值的拟合度R2为0.8219,均方根误差RMSE为0.368,两者存在很强的相关性。该研究可为森林资源遥感定量估算提供技术支撑,为森林资源空间配置的优化调整与辅助解决方案的提出提供参考。     

基于Sentinel-2多光谱数据的棉花叶面积指数估算

棉花叶面积指数(leaf are index, LAI)的快速、准确获取对棉花长势监测、发育期诊断、面积提取以及产量估算等遥感监测具有重要意义。该研究利用2017年和2018年的Sentinel-2多光谱卫星数据及大面积田间试验观测获取的棉花不同发育期LAI实测数据,构建了基于单波段反射率及各类植被指数的棉花不同发育期及全发育期LAI估算模型,并采用留一验证(LOOCV, leave-one-out cross validation)和交叉验证对模型精度进行了检验。结果表明:1)对于单波段反射率,基于中心波长为842 nm波宽为145 nm的B8近红外波段对不同发育期LAI估算精度最优均方根误差(RMSE, root mean square error, RMSE=0.378);2)对于各类植被指数,花蕾期(20170616)和花铃期(20170802)时增强植被指数(EVI, enhanced vegetation index,)表现最佳(RMSE分别为0.352和0.367),开花期(20180623)时校正土壤调节植被指数(MSAVI2, modified soil adjusted vegetation index 2,)估算精度最高(RMSE=0.323);3)单波段反射率和各类植被指数对全发育期LAI的估算均要优于对单个发育期LAI的估算,其中基于IRECI指数的(invertedred-edge chlorophyllindex)全发育期LAI估算模型精度最佳,LOOCV检验RMSE=0.425,交叉检验RMSE=0.368;将基于IRECI的全发育期LAI估算模型应用到单个发育期LAI估算并与各单个发育期LAI估算模型精度对比,发现交叉验证RMSE平均值仅比LOOCV验证RMSE平均值高0.07,反映了全发育期LAI估算模型良好的普适性。该研究为农作物LAI估算提供了新的数据选择,完善了Sentinel-2卫星数据在LAI估算中的应用领域。     

基于赤池信息量准则的冬小麦叶面积指数高光谱估测

冬小麦叶面积指数(leaf area index,LAI)是描述冠层结构的重要参数之一,对评价其长势和预测产量具有重要意义。该文利用灰色关联分析(grey relational analysis,GRA)对植被指数进行排序,用偏最小二乘法(partial least squares regression,PLS)选择不同的植被指数个数作为自变量进行回归建模,通过赤池信息量准则(Akaike’s information criterion,AIC)选择AIC值最小的模型作为冬小麦LAI最优估算模型,即GRA、PLS和AIC 3种方法整合建立冬小麦LAI最优估算模型。使用2008-2009年在中国北京通州区和顺义区获取的整个生育期冬小麦LAI和配套的光谱数据进行建模,利用2009-2010相关数据进行验证。研究表明:采用GRA评价标准与冬小麦LAI关联度最大的植被指数是VOG1,关联度最小的植被指数是SR;通过AIC建立的以8个植被指数作为自变量的冬小麦LAI模型效果最优,建模集的决定系数R2和标准误SE分别为0.76和0.009,验证集的R2和相对均方根误差RRMSE分别为0.63和0.004,预测模型和验证模型均具有较高的精度和可靠性。结果表明采用GRA-PLS-AIC方法进行冬小麦LAI反演是可行的,为提高冬小麦LAI遥感预测精度提供了一种有效的方法。     

利用无人机影像构建作物表面模型估测甘蔗LAI

为探讨从作物表面模型(crop surface models,CSMs)中提取株高来估算糖料蔗叶面积指数(leaf area index,LAI)的可行性,该文采用无人机-RGB高清数码相机构成的低空遥感平台,以广西糖料蔗为研究对象,采集了糖料蔗全生育期的高清数码影像,分别在有无地面控制点条件下建立各生育期CSMs并提取株高。此外,该文利用高清数码影像计算了6种可见光植被指数并建立LAI估算模型,用以对比从CSMs提取的株高对LAI的估算效果。结果表明:全生育期CSMs提取的株高与实测株高显著相关(P0.01),株高预测值与实测值高度拟合(R2=0.961 2,RMSE=0.215 2)。选取的6种可见光植被指数中,绿红植被指数对糖料蔗伸长末期以前的LAI的估测效果最好(R2=0.779 0,RMSE=0.556 1,MRE=0.168 0)。相同条件下,株高对LAI有更高的估测精度,其中CSMs提取的株高估测效果优于地面实测株高,预测模型R2=0.904 4,RMSE=0.366 2,MRE=0.124 3。研究表明,使用无人机拍摄RGB影像来提取株高并运用于糖料蔗重要生育期LAI的估算是可行的,CSMs提取的株高拥有较高的精度。该研究可为大区域进行精准快速的农情监测提供参考。     

不同生育期水稻叶面积指数的高光谱遥感估算模型

2011年和2012年通过大田试验,利用便携式野外光谱仪实测水稻冠层不同生育时期的高光谱数据,同时使用SUNSCAN冠层分析系统采集水稻冠层叶面积指数（LAI）;采用光谱微分技术和统计分析技术,分别分析高光谱反射率及其植被指数与LAI之间的关系,建立LAI估算模型并进行模拟结果对比。结果表明：水稻抽穗-成熟期,利用光谱值的对数形式对LAI值的模拟效果较好,分蘖-抽穗期利用光谱反射率模拟LAI变化过程的效果不理想。 在利用各种植被指数估算LAI方法中,水稻分蘖-抽穗期以修改型土壤调整植被指数MSAVI[758,805]对LAI的估算效果最好,模拟值与实测值的相关系数通过了0.01水平的显著性检验（R=0.7754）,估算精度较高。在抽穗-成熟期,也以修改型土壤调整植被指数MSAVI[758,817]对LAI的模拟效果最好,模拟值与实测值的相关系数通过了0.01水平的显著性检验（R=0.6488）,估算精度较高。说明修改型土壤调整植被指数（MSAVI）能更好地模拟水稻不同生育期的叶面积指数,按照分蘖-抽穗期、抽穗-成熟期两个生育阶段分别建立水稻冠层LAI的高光谱估算模型能够提高LAI估算的准确度,研究结果也证实了分生育阶段建模的必要性。     

漓江上游山区水源林叶面积指数变化的遥感监测

为了定量评价漓江上游山区复杂地形水源林叶面积指数(LAI)的变化,对阔叶林、针叶林、竹林样地以TRAC仪器测定LAI,利用遥感数据计算归一化植被指数(NDVI)、比值植被指数(SR)、减化比值植被指数(RSR)、土壤调整植被指数(SAVI)、增强植被指数(EVI),并从DEM数据获取高程、坡度、坡向,提出并建立复杂地形最优多植被指数组合估算山区林地LAI的神经网络模型,利用模型对1989–2009年6景TM/ETM遥感图像估算LAI空间分布。结果表明,神经网络解决了LAI与多植被指数的非线性回归方程无法引入地形因素、且方程系数较多较难确定的问题,提高了LAI的估算精度。研究区成熟阔叶林减少代之以大片种植经济幼林,是导致林地LAI变化的原因。1989-2000年,LAI≥6.0的林地面积比例从78.8%逐年急剧下降到44.1%,LAI在1.0~6.0的林地面积比例从20.8%大幅上升到55.4%;2000-2009年,随着幼林的生长、竹林的速生,LAI≥6.0的林地面积比例逐渐上升恢复到74.5%,但仍未恢复到1989年的面积比例,相应LAI在1.0~6.0的林地面积比例逐渐下降到25.1%。研究成果为漓江上游水源林生态评估提供参考。