Variation and directional anisotropy of reflectance at the crown scale — Implications for tree species classification in digital aerial images

Tree species classification is still solved at insufficient reliability in airborne optical data. The variation caused by directional reflectance anisotropy hampers image-based solutions. In addition, trees show considerable within-species variation in reflectance properties. We examined these phenomena at the single-tree level, using the Leica ADS40 line sensor and XPro software, which constitute the first photogrammetric large-format multispectral system to provide target reflectance images. To analyze the influence of illumination conditions in the canopy, we developed a method in which the crown shape as well as between-tree occlusions and shading were modeled, using dense LiDAR data. The precision of the ADS40 reflectance images in well-defined surfaces was 5% as coefficient of variation when 1−4-km image data were fused. The range of reflectance anisotropy was ± 30% for trees near the solar principal plane, with differences between bands and species. Because of the anisotropy differences observed, the spectral separability of the tree species in different bands is dependent on the view-illumination geometry. The within-species variation was high; the coefficient of variation was 13−31%. The contribution of tree and stand variables to anisotropy-normalized reflectance variation was examined. The effects of the species composition of adjacent trees were substantial in NIR and this variation hampers spectral classification in mixed stands. We also studied species- and band-specific intracrown brightness patterns, and we suggest their use as high-order image features in species classification. A species classification accuracy of up to 80% was obtained using 4-km data, which showed the high potential of the ADS40.     

Tree species differentiation using intensity data derived from leaf-on and leaf-off airborne laser scanner data

Tree species identification is important for a variety of natural resource management and monitoring activities including riparian buffer characterization, wildfire risk assessment, biodiversity monitoring, and wildlife habitat assessment. Intensity data recorded for each laser point in a LIDAR system is related to the spectral reflectance of the target material and thus may be useful for differentiating materials and ultimately tree species. The aim of this study is to test if LIDAR intensity data can be used to differentiate tree species. Leaf-off and leaf-on LIDAR data were obtained in the Washington Park Arboretum, Seattle, Washington, USA. Field work was conducted to measure tree locations, tree species and heights, crown base heights, and crown diameters of individual trees for eight broadleaved species and seven coniferous species. LIDAR points from individual trees were identified using the field-measured tree location. Points from adjacent trees within a crown were excluded using a procedure to separate crown overlap. Mean intensity values of laser returns within individual tree crowns were compared between species. We found that the intensity values for different species were related not only to reflective properties of the vegetation, but also to a presence or absence of foliage and the arrangement of foliage and branches within individual tree crowns. The classification results for broadleaved and coniferous species using linear discriminant function with a cross validation suggests that the classification rate was higher using leaf-off data (83.4%) than using leaf-on data (73.1%), with highest (90.6%) when combining these two LIDAR data sets. The result also indicates that different ranges of intensity values between two LIDAR datasets didn't affect the result of discriminant functions. Overall results indicate that some species and species groups can be differentiated using LIDAR intensity data and implies the potential of combining two LIDAR datasets for one study.     

Intra- and inter-class spectral variability of tropical tree species at La Selva, Costa Rica: Implications for species identification using HYDICE imagery

Hyperspectral remote sensing provides great potential to monitor and study biodiversity of tropical forests through species identification and mapping. In this study, five species were selected to examine crown-level spectral variation within and between species using HYperspectral Digital Collection Experiment (HYDICE) data collected over La Selva, Costa Rica. Spectral angle was used to evaluate the spectral variation in reflectance, first derivative and wavelet-transformed spectral domains. Results indicated that intra-crown spectral variation does not always follow a normal distribution and can vary from crown to crown, therefore presenting challenges to statistically define the spectral variation within species using conventional classification approaches that assume normal distributions. Although derivative analysis has been used extensively in hyperspectral remote sensing of vegetation, our results suggest that it might not be optimal for species identification in tropical forestry using airborne hyperspectral data. The wavelet-transformed spectra, however, were useful for the identification of tree species. The wavelet coefficients at coarse spectral scales and the wavelet energy feature are more capable of reducing variation within crowns/species and capturing spectral differences between species. The implications of this examination of intra- and inter-specific variability at crown-level were: (1) the wavelet transform is a robust tool for the identification of tree species using hyperspectral data because it can provide a systematic view of the spectra at multiple scales; and (2) it may be impractical to identify every species using only hyperspectral data, given that spectral similarity may exist between species and that within-crown/species variability may be influenced by many factors.     

Classifying individual tree genera using stepwise cluster analysis based on height and intensity metrics derived from airborne laser scanner data

This paper evaluates the ability of small footprint, multiple return and pulsed airborne scanner data to classify tree genera hierarchically using stepwise cluster analysis. Leaf-on and leaf-off airborne scanner datasets obtained in the Washington Park Arboretum, Seattle, Washington, USA were used for tree genera classification. Parameters derived from structure and intensity data from the leaf-on and leaf-off laser scanning datasets were compared to ground truth data. Relative height percentiles and simple crown shapes using the ratio of a crown length to width were computed for the structure variables. Selected structure variables from the leaf-on dataset had higher classification rate (74.9%) than those from the leaf-off dataset (50.2%) for distinguishing deciduous from coniferous genera using linear discriminant functions.Unsupervised stepwise cluster analysis was conducted to find groupings of similar genera at consecutive steps using k-medoid algorithm. The three stepwise cluster analyses using different seasonal laser scanning datasets resulted in different outcomes, which imply that genera might be grouped differently depending on the timing of the data collection. When combining leaf-on and leaf-off LIDAR datasets, the cluster analysis could separate the deciduous genera from evergreen coniferous genera and could make further separations between evergreen coniferous genera. When using the leaf-on LIDAR dataset only, the cluster analysis did not separate deciduous from evergreen genera. The overall results indicate the importance of the timing of laser scanner data acquisition for tree genera separation and suggest that the potential of combining two LIDAR datasets for improved classification.     

Identifying species of individual trees using airborne laser scanner

Individual trees can be detected using high-density airborne laser scanner data. Also, variables characterizing the detected trees such as tree height, crown area, and crown base height can be measured. The Scandinavian boreal forest mainly consists of Norway spruce (Picea abies L. Karst.), Scots pine (Pinus sylvestris L.), and deciduous trees. It is possible to separate coniferous from deciduous trees using near-infrared images, but pine and spruce give similar spectral signals. Airborne laser scanning, measuring structure and shape of tree crowns could be used for discriminating between spruce and pine. The aim of this study was to test classification of Scots pine versus Norway spruce on an individual tree level using features extracted from airborne laser scanning data. Field measurements were used for training and validation of the classification. The position of all trees on 12 rectangular plots (50×20 m²) were measured in field and tree species was recorded. The dominating species (>80%) was Norway spruce for six of the plots and Scots pine for six plots. The field-measured trees were automatically linked to the laser-measured trees. The laser-detected trees on each plot were classified into species classes using all laser-detected trees on the other plots as training data. The portion correctly classified trees on all plots was 95%. Crown base height estimations of individual trees were also evaluated (r=0.84). The classification results in this study demonstrate the ability to discriminate between pine and spruce using laser data. This method could be applied in an operational context. In the first step, a segmentation of individual tree crowns is performed using laser data. In the second step, tree species classification is performed based on the segments. Methods could be developed in the future that combine laser data with digital near-infrared photographs for classification with the three classes: Norway spruce, Scots pine, and deciduous trees.     

基于时序Landsat数据的三江平原植被地表类型变化遥感探测研究

Landsat 卫星遥感数据具有分辨率较高,数据积累时间长的特点,在探测地表覆盖变化和地物分类中得到广泛应用。首先,对获取的Landsat TM/ETM+时间序列数据进行了定量化处理,获取了三江平原七台河市1989~2012年时间序列Landsat地表反射率图像。其次,设计了林地指数和湿地指数,提取了三江平原七台河区域地物光谱和时序特征,同时设计构建了地表覆盖分类和植被地表类型变化探测的决策树算法,实现了1989~2012年七台河区域的植被地表覆盖变化的动态监测,提取了森林覆盖变化的空间分布与变化时间。最后,对七台河区域地表覆盖与植被地表类型变化进行了精度检验,分类总体精度达到90.04%,Kappa系数达0.88。研究结果表明:基于定量化的Landsat时间序列数据的分类算法能克服单时相影像分类的缺陷,实现区域地物自动分类和地表覆盖变化的动态监测。
     

Performance analysis of image processing algorithms for classification of natural vegetation in the mountains of Southern California

Abstract

The Earth's forests fix carbon from the atmosphere during photosynthesis. Scientists are concerned that massive forest removals may promote an increase in atmospheric carbon dioxide, with possible global warming and related environmental effects. Space-based remote sensing may enable the production of accurate world forest maps needed to examine this concern objectively. To test the limits of remote sensing for large-area forest mapping, we use LANDSAT data acquired over a site in the forested mountains of southern California to examine the relative capacities of a variety of popular image processing algorithms to discriminate different forest types. Results indicate that certain algorithms are best suited to forest classification. Differences in performance between the algorithms tested appear related to variations in their sensitivities to spectral variations caused by background reflectance, differential illumination, and spatial pattern by species. Results emphasize the complexity between the land-cover regime, remotely sensed data and the algorithms used to process these data.     

Classification of Australian forest communities using aerial photography, CASI and HyMap data

Within Australia, the discrimination and mapping of forest communities has traditionally been undertaken at the stand scale using stereo aerial photography. Focusing on mixed species forests in central south-east Queensland, this paper outlines an approach for the generation of tree species maps at the tree crown/cluster level using 1 m spatial resolution Compact Airborne Spectrographic Imager (CASI; 445.8 nm–837.7 nm wavelength) and the use of these to generate stand-level assessments of community composition. Following automated delineation of tree crowns/crown clusters, spectral reflectance from pixels representing maxima or mean-lit averages of channel reflectance or band ratios were extracted for a range of species including Acacia, Angophora, Callitris and Eucalyptus. Based on stepwise discriminant analysis, classification accuracies of dominant species were greatest (87% and 76% for training and testing datasets; n = 398) when the mean-lit spectra associated with a ratio of the reflectance (ρ) at 742 nm (ρ₇₄₂) and 714 nm (ρ₇₁₄) were used. The integration of 2.6 m HyMap (446.1 nm–2477.8 nm) spectra increased the accuracy of classification for some species, largely because of the inclusion of shortwave infrared wavebands. Similar increases in accuracy were achieved when classifications of field spectra resampled to CASI and HyMap wavebands were compared. The discriminant functions were applied subsequently to classify crowns within each image and produce maps of tree species distributions which were equivalent or better than those generated through aerial photograph interpretation. The research provides a new approach to tree species mapping, although some a priori knowledge of the occurrence of broad species groups is required. The tree maps have application to biodiversity assessment in Australian forests.     

Spectral reflectance of coral reef benthos and substrate assemblages on Heron Reef,Australia

Studies investigating the spectral reflectance of coral reef benthos and substrates have focused on the measurement of pure endmembers, where the entire field of view (FOV) of a spectrometer is focused on a single benthos or substrate type. At the spatial scales of the current satellite sensors, the heterogeneity of coral reefs even at a sub-metre scale means that many individual image pixels will be made up of a mixture of benthos and substrate types. If pure endmember spectra are used as training data for image classification, there is a spatial discrepancy, because many pixels will have a mixed endmember spectral reflectance signature. This study investigated the spectral reflectance of coral reef benthos and substrates at a spatial scale directly linked to the pixel size of high spatial resolution imaging systems, by incorporating multiple benthos and substrate types into the spectrometer FOV in situ. A total of 334 spectral reflectance signatures were measured of 19 assemblages of the coral reef benthos and substrate types. The spectra were analysed for separability using first derivative values, and a discrimination decision tree was designed to identify the assemblages. Using the decision tree, it was possible to identify 15 assemblages with a mean overall classification accuracy of 62.6%.     

Using AVIRIS data and multiple-masking techniques to map urban forest tree species

Tree type and species information are critical parameters for urban forest management, benefit cost analysis and urban planning. However, traditionally, these parameters have been derived based on limited field samples in urban forest management practice. In this study we used high-resolution Airborne Visible Infrared Imaging Spectrometer (AVIRIS) data and multiple-spectral masking techniques to identify and map urban forest trees. Trees were identified based on their spectral character difference in AVIRIS data. The use of multiple-masking techniques shift the focus to the target land cover types only, thus reducing confounding noise during spectral analysis. The results were checked against ground reference data and by comparison to tree information in an existing geographical information system (GIS) database. At the tree type level, mapping was accomplished with 94% accuracy. At the tree species level, the average accuracy was 70% but this varied with both tree type and species. Of the four evergreen tree species, the average accuracy was 69%. For the 12 deciduous tree species, the average accuracy was 70%. The relatively low accuracy for several deciduous species was due to small tree size and overlapping among tree crowns at the 3.5 m spatial resolution of AVIRIS data. This urban forest tree species mapping method has the potential to increase data update intervals and accuracy while reducing costs compared to field sampling or other traditional methods.     

Multi-angular reflectance properties of a hemiboreal forest: An analysis using CHRIS PROBA data

Forest types differ in their hyperspectral anisotropy patterns mainly due to species-specific geometrical structure, spatial arrangement of canopies and subsequent shadow patterns. This paper examines the multi-angular, hyperspectral reflectance properties of typical hemiboreal forests during summer time using three simultaneous CHRIS PROBA (mode 3) scenes and stand inventory data from the Järvselja Training and Experimental Forestry District in southeastern Estonia. We investigated the magnitude and reasons for the differences in the anisotropy patterns of deciduous and coniferous stands at three backward viewing angles. A forest reflectance model (FRT) was used as a tool to provide a theoretical basis to the discussion, and to estimate the directional contribution of scattering from crowns and ground to total stand reflectance for the two forest types. The FRT model simulated successfully the HDRF (hemispherical–directional reflectance factor) curves of the study stands to match those obtained from the CHRIS image, yet it produced a smaller and less wavelength-dependent angular reflectance effect than was observed in the satellite image. The main results of this study provide new information for separating the spectral contribution of the forest floor (or understory layer) from the tree canopy layer: (1) the red edge domain was identified to have the largest contribution from forest understory, and (2) the more oblique the viewing angle, the smaller the contribution from the understory. In addition, coniferous stands were observed to have a specific angular effect at the red and red edge domain, possibly as a result of the hierarchical structure and arrangement of coniferous canopies.     

A comparison of supervised maximum likelihood and decision tree classification for crop cover estimation from multitemporal LANDSAT MSS data

Abstract

Supervised maximum likelihood classification was compared with a supervised binary decision tree for crop classification from multitemporal LANDSAT MSS data. Similar levels of classification accuracy were obtained using both algorithms, but the ease of training and computational simplicity of the binary decision tree suggest that this algorithm may be a viable alternative to the maximum likelihood for the analysis of data sets with high dimensionality such as multitemporal LANDSAT MSS data.     

Classification of remotely-sensed imagery using an indicator kriging approach: application to the problem of calcite-dolomite mineral mapping

A new image classification technique for analysis of remotely-sensed data based on geostatistical indicator kriging is introduced. Conventional classification techniques require ground truth information, use only the spectral characteristics of an unknown pixel in comparison, rely on a Gaussian probability distribution for the spectral signature of the training data, and work on a pixel support or spatial resolution without allowing classification on larger or smaller volumes. The indicator kriging classifier overcomes such problems because: (1) it relies on spectral information from laboratory studies rather than on ground truth data, (2) through the kriging estimation variances an estimate of uncertainly is derived, (3) it incorporates spatial aspects because it uses local estimation techniques, (4) it is distribution-free, (5) and may be applied on different supports if the data are corrected for support changes. Comparison of classification results applied to the problem of mapping calcite and dolomite from GER imaging spectrometry data shows that indicator kriging performs better than the conventional classification algorithms and gives insight in the accuracy of the results without prior field knowledge     

Combining high spatial resolution multi-temporal satellite data with leaf-on LiDAR to enhance tree species discrimination at the crown level

ABSTRACT

The long-standing goal of discriminating tree species at the crown-level from high spatial resolution imagery remains challenging. The aim of this study is to evaluate whether combining (a) high spatial resolution multi-temporal images from different phenological periods (spring, summer and autumn), and (b) leaf-on LiDAR height and intensity data can enhance the ability to discriminate the species of individual tree crowns of red oak (Quercus rubra), sugar maple (Acer saccharum), tulip poplar (Liriodendron tulipifera), and black cherry (Prunus serotina) in the Fernow Experimental Forest, West Virginia, USA. We used RandomForest models to measure a loss of classification accuracy caused by iteratively removing from the classification one or more groups from six groups of variables: spectral reflectance from all multispectral bands in the (1) spring, (2) summer, and (3) autumn images, (4) vegetation indices derived from the three multispectral datasets, (5) canopy height and intensity from the LiDAR imagery, and (6) texture related variables from the panchromatic and LiDAR datasets. We also used ANOVA and decision tree analyses to elucidate how the multispectral and LiDAR datasets combine to help discriminate tree species based on their unique phenological, spectral, textural, and crown architectural traits. From these results, we conclude that combing high spatial resolution multi-temporal satellite data with LiDAR datasets can enhance the ability to discriminate tree species at the crown level.     

全球典型植被叶片光谱特征及其对LAI反演的影响分析

在全球范围长时间序列LAI遥感产品反演算法中,植被冠层反射率模型仅使用少量叶片光谱特征代表全球植被全年的典型植被光谱特征,叶片光谱的不确定性导致LAI遥感产品存在一定的误差。目前全球已经构建了多个典型植被叶片波谱数据集,这些数据集包含多个植被物种、不同空间地域及多时相叶片光谱数据,为定量分析叶片光谱特征提供了数据支持。主要利用LOPEX’93、ANGERS’03、中国典型地物波谱数据库和野外实测的叶片光谱数据,以黄边参数、红边参数和叶片光谱指数作为分析指标,探讨不同植被物种、不同气候区和不同物候期的叶片光谱特征差异,及其对植被冠层反射率、LAI反演的影响,为发展考虑现实叶片光谱差异的LAI反演算法提供研究基础。结果表明:植被叶片光谱存在多样性,叶片光谱特征差异主要影响MODIS传感器近红外波段和绿波段反射率值,其中,绿波段反射率值对叶片光谱变化最为敏感;在LAI反演算法中,如果只考虑植被类型而不考虑物种叶片光谱差异,可能会给LAI反演带来大于3的误差。     

基于ENVISat ASAR影像的湿地植被冠层下淹水范围提取以扎龙保护区为例

以扎龙自然保护区湿地为例,结合ENVISat ASAR多极化(HH/HV)雷达影像与传统的光学影像Landsat TM (band1~5,7),分析雷达影像后向散射系数与Landsat TM影像不同波段反射率在淹水植被、非淹水植被、明水面和裸土不同地表覆被类型的差异。选择训练样本,采用分类回归树(Classification and Regression Tree,CART)模型,分别对两种影像进行分类,可视化表达湿地植被淹水范围空间分布情况。基于实测的植被冠层下淹水范围与非淹水范围样本点对两种数据源的分类结果进行精度验证。结果表明:HH/HV极化影像中,植被覆盖下水体的后向散射系数与其他地表覆被类型有明显区别,分类结果总精度为79.49%,Kappa系数为0.70,湿地植被淹水范围提取精度较高。而TM影像分类结果中,由于部分地区植被覆盖水体,淹水植被分类误差较高。将雷达影像引入沼泽湿地研究,提高了植被淹水范围提取效果,为有效分析湿地生态水文过程提供基础,对湿地水资源合理利用及生物多样性保护具有重要意义。     

Object-based fusion for urban tree species classification from hyperspectral,panchromatic and nDSM data

This study aims at identifying the best object-based fusion strategy that takes advantage of the complementarity of several heterogeneous airborne data sources for improving the classification of 15 tree species in an urban area (Toulouse, France). The airborne data sources are: hyperspectral Visible Near-Infrared (160 spectral bands, spatial resolution of 0.4 m) and Short-Wavelength Infrared (256 spectral bands, 1.6 m), panchromatic (14 cm), and a normalized Digital Surface Model (12.5 cm). Object-based feature and decision level fusion strategies are proposed and compared when applied to a reference site where the species are previously identified during ground truth collection. This allows the best fusion strategy to be selected with a view to introducing the method in an automatic process (tree crown delineation and species classification) on a test site, independent of the reference site used for learning. In particular, a decision level fusion is selected: based on the Support Vector Machine algorithm, Visible Near-Infrared and Short-Wavelength Infrared classifications use Minimum Noise Fraction components at the original spatial resolution, whereas panchromatic and normalized Digital Surface Model classifications use, respectively, Haralick’s and structural features computed at the object scale. After the computation of a decision profile for each source at the object level based on the classification algorithms’ membership probabilities, these decision profiles are combined and a decision rule is applied to predict the species. Focusing on the reference site, the Visible Near-Infrared exhibits the best performances with F-score values higher than 60% for 13 species out of 15. The Short-Wavelength Infrared is the most powerful for three species with F-score greater than 60% for seven common species with the Visible Near-Infrared. The panchromatic and normalized Digital Surface Model contribute marginally. The best fusion strategy (decision fusion) does not improve significantly the overall accuracy with 77% (kappa = 74%) against 75% (kappa = 72%) for the Visible Near-Infrared but in general, it improves the results for cases where complementarities have been observed. When applied to the test site and assessed for the two majority species (Tilia tomentosa and Platanus x hispanica), the selected approach gives consistent results with an overall accuracy of 63% against 55% for the Visible Near-Infrared.     

基于面向对象方法的沙化土地遥感信息提取技术研究

针对基于像元光谱特征提取沙化土地信息分类精度偏低的问题,以Landsat\|5 TM为数据源,基于面向对象的方法对沙化土地遥感信息提取技术进行研究。首先采用多尺度分割法对影像进行分割以获得同质区域,然后结合野外调查数据制成不同地物类型的多种特征图,从而确定提取目标地物的特征并建立沙化和非沙化地物提取决策树,最后对影像进行模糊分类,并对分类结果进行精度评价。结果表明,基于面向对象提取沙化土地信息的总精度达84.89%,Kappa系数为0.8077。研究结果为后续深入研究奠定了基础。     

The effect of spectral and spatial degradation of hyperspectral imagery for the Sclerophyll tree species classification

Hyperspectral imaging can be a useful remote-sensing technology for classifying tree species. Prior to the image classification stage, effective mapping endeavours must first identify the optimal spectral and spatial resolutions for discriminating the species of interest. Such a procedure may contribute to improving the classification accuracy, as well as the image acquisition planning. In this work, we address the effect of degrading the original bandwidth and pixel size of a hyperspectral and hyperspatial image for the classification of Sclerophyll forest tree species. A HySpex-VNIR 1600 airborne-based hyperspectral image with submetric spatial resolution was acquired in December 2009 for a native forest located in the foothills of the Andes of central Chile. The main tree species of this forest were then sampled in the field between January and February 2010. The original image spectral and spatial resolutions (160 bands with a width of 3.7 nm and pixel sizes of 0.3 m) were systematically degraded by resampling using a Gaussian model and a nearest neighbour method, respectively (until reaching 39 bands with a width of 14.8 nm and pixel sizes of 2.4 m). As a result, 12 images with different spectral and spatial resolution combinations were created. Subsequently, these images were noise-reduced using the minimum noise fraction procedure and 12 additional images were created. Statistical class separabilities from the spectral divergence measure and an assessment of classification accuracy of two supervised hyperspectral classifiers (spectral angle mapper (SAM) and spectral information divergence (SID)) were applied for each of the 24 images. The best overall and per-class classification accuracies (>80%) were observed when the SAM classifier was applied on the noise-reduced reflectance image at its original spectral and spatial resolutions. This result indicates that pixels somewhat smaller than the tree canopy diameters were the most appropriate to represent the spatial variability of the tree species of interest. On the other hand, it suggests that noise-reduced bands derived from the full image spectral resolution rendered the best discrimination of the spectral properties of the tree species of interest. Meanwhile, the better performance of SAM over SID may result from the ability of the former to classify tree species regardless of the illumination differences in the image. This technical approach can be particularly useful in native forest environments, where the irregular surface of the uppermost canopy is subject to a differentiated illumination.