面向对象的旱区植被遥感精细分类研究

针对旱区植被分类尺度过大、种群无法准确提取的问题,该文提出了面向对象的CFS-RF分类模型,即利用CFS算法对先验样本数据集进行特征优选,结合随机森林构建分类规则,完成分类过程。以新疆阿勒泰为研究区,利用GF-2数据,通过CFS、ReliefF两种不同特征选择方法和J48、SVM、RF 3种分类算法构造出6种面向对象分类方案来实现小尺度植被种群提取。结果表明,经过特征选择,上述分类方案的精度和效率均得到了提升。其中,CFS-RF算法最优,总体精度达到92.41%,Kappa系数为0.90,更适用于旱区植被遥感精细分类。     

Examining deforestation and agropasture dynamics along the Brazilian TransAmazon Highway using multitemporal Landsat imagery

This research aims to understand the difference of major land-cover change results caused in various time periods and to examine the impacts of human-induced factors on land-cover changes along the TransAmazon Highway region. The Landsat Thematic Mapper and Operational Land Imager data from 2011, 2014, and 2017 and our previous land-cover classification results in 1991, 2000, and 2008 were used to examine land-cover dynamics. A classification system consisting of five land-cover classes – primary forest (PF), secondary forest (SF), agropasture (AP), urban area, and water – were chosen. The hierarchical-based classification method was used to generate land-cover classification results, and the post-classification comparison approach was used to produce detailed “from-to” conversions for each detection period. The emphasis was on deforestation of PF, dynamic change of SF and AP, and urbanization over time. The impacts of human-induced factors such as population and economic conditions on urban expansion, AP expansion, and deforestation were examined. This research indicated that selection of a suitable time period was critical for effectively detecting land-cover changes; that is, too long time period (i.e., 9 years) cannot accurately capture some land-cover changes such as the AP and SF in this research. Although deforestation – the conversion from PF to SF and AP – accounted for a large proportion of land-cover changes, the changes between SF and AP became more important than PF conversion, and required a short time period (i.e., 3 years here) for effectively reflecting their dynamics. Human-induced factors play important roles in deforestation, dynamic changes between AP and SF, and urbanization.     

结合随机森林面向对象的森林资源分类

针对森林资源分类研究较少且缺少相对简单有效的方法的情况,提出一种结合面向对象和随机森林的森林资源分类方法。面向对象分割技术可减少“椒盐效应”,随机森林分类算法具有高准确度、抗噪声能力强、性能稳定等优势。鉴于此,通过调整面向对象的分割参数,构造最优特征空间及估算随机森林中决策树的数量等,构建了最优的面向对象随机森林分类模型。另外,选择了SVM算法作对比试验。试验结果显示,本文分类算法的总体精度达到83.34%,Kappa系数为0.7892,明显高于SVM,能够有效提高森林资源分类的精度。     

Evaluation of Goddard’s LiDAR,hyperspectral, and thermal data products for mapping urban land-cover types

Goddard’s LiDAR (Light Detection And Ranging), hyperspectral and thermal (G-LiHT) airborne imager is a new system to advance concepts of data fusion for worldwide applications. A recent G-LiHT mission conducted in June 2016 over an urban area opens a new opportunity to assess the G-LiHT products for urban land-cover mapping. In this study, the G-LiHT hyperspectral and LiDAR-canopy height model (LiDAR-CHM) products were evaluated to map five broad land-cover types. A feature/decision-level fusion strategy was developed to integrate two products. Contemporary data processing techniques were applied, including object-based image analysis, machine-learning algorithms, and ensemble analysis. Evaluation focused on the capability of G-LiHT hyperspectral products compared with multispectral data with similar spatial resolution, the contribution of LiDAR-CHM, and the potential of ensemble analysis in land-cover mapping. The results showed that there was no significant difference between the application of the G-LiHT hyperspectral product and simulated Quickbird data in the classification. A synthesis of G-LiHT hyperspectral and LiDAR-CHM products achieved the best result with an overall accuracy of 96.3% and a Kappa value of 0.95 when ensemble analysis was applied. Ensemble analysis of the three classifiers not only increased the classification accuracy but also generated an uncertainty map to show regions with a robust classification as well as areas where classification errors were most likely to occur. Ensemble analysis is a promising tool for land-cover classification.     

机载多光谱LiDAR的随机森林地物分类

机载多光谱LiDAR技术利用激光进行探测和测距，不仅可以快速获取地面物体的三维坐标，还可以获得多个波段的地物光谱信息，可广泛用于地形测绘、土地覆盖分类、环境建模、森林资源调查等。本文提出了多光谱LiDAR的随机森林地物分类方法。该方法通过对LiDAR强度数据和高程数据提取分类特征，完成多光谱LiDAR的随机森林地物分类；并分析随机森林的特征贡献度特性，采用后向特征选择方法实现分类特征选择。通过对加拿大Optech Titan多光谱LiDAR数据的试验表明：随机森林方法可以获得较好的地物分类精度，而且可以适当地去除部分冗余和相关的特征，从而有效提高分类精度。     

The Forest Types and Ages Cleared for Land Development in Puerto Rico

On the Caribbean island of Puerto Rico, forest, urban/built-up, and pasture lands have replaced most formerly cultivated lands. The extent and age distribution of each forest type that undergoes land development, however, is unknown. This study assembles a time series of four land cover maps for Puerto Rico. The time series includes two digitized paper maps of land cover in 1951 and 1978 that are based on photo interpretation. The other two maps are of forest type and land cover and are based on decision tree classification of Landsat image mosaics dated 1991 and 2000. With the map time series we quantify land-cover changes from 1951 to 2000; map forest age classes in 1991 and 2000; and quantify the forest that undergoes land development (urban development or surface mining) from 1991 to 2000 by forest type and age. This step relies on intersecting a map of land development from 1991 to 2000 (from the same satellite imagery) with the forest age and type maps. Land cover changes from 1991 to 2000 that continue prior trends include urban expansion and transition of sugar cane, pineapple, and other lowland agriculture to pasture. Forest recovery continues, but it has slowed. Emergent and forested wetland area increased between 1977 and 2000. Sun coffee cultivation appears to have increased slightly. Most of the forests cleared for land development, 55%, were young (1-13 yr). Only 13% of the developed forest was older (41-55+ yr). However, older forest on rugged karst lands that long ago reforested is vulnerable to land development if it is close to an urban center and unprotected.     

基于GF-1号影像的南方水稻种植信息提取

针对南方复杂地区水稻遥感信息提取研究中机器自动学习分类研究较少、分类精度不高的问题,以福建省三明市建宁县溪口镇为研究区,基于GF-1号卫星影像,采用面向对象的随机森林遥感分类算法对研究区内水稻田信息进行提取。首先通过优化面向对象分割参数和随机森林分类模型参数,提取并调用了影像中的多种特征;再对光谱特征、植被指数特征、纹理特征、几何特征进行特征空间优选;最后通过设置4种特征优选试验进行对比,得到最优分类模型。实验结果显示,基于特征空间优选的面向对象随机森林分类算法的水稻提取精度高达90%,分类总体精度可达87%,Kappa系数为0.85;与其他试验结果相比,漏分和误分现象较少,实现了南方地区水稻信息高精度自动识别。该方法计算特征少、实现简便,对于国产高分卫星影像在南方复杂地区作物自动提取中的应用具有参考性。     

Coupling high-resolution satellite imagery with ALS-based canopy height model and digital elevation model in object-based boreal forest habitat type classification

We developed a classification workflow for boreal forest habitat type mapping. In object-based image analysis framework, Fractal Net Evolution Approach segmentation was combined with random forest classification. High-resolution WorldView-2 imagery was coupled with ALS based canopy height model and digital terrain model. We calculated several features (e.g. spectral, textural and topographic) per image object from the used datasets. We tested different feature set alternatives; a classification accuracy of 78.0% was obtained when all features were used. The highest classification accuracy (79.1%) was obtained when the amount of features was reduced from the initial 328 to the 100 most important using Boruta feature selection algorithm and when ancillary soil and land-use GIS-datasets were used. Although Boruta could rank the importance of features, it could not separate unimportant features from the important ones. Classification accuracy was bit lower (78.7%) when the classification was performed separately on two areas: the areas above and below 1 m vertical distance from the nearest stream. The data split, however, improved the classification accuracy of mire habitat types and streamside habitats, probably because their proportion in the below 1 m data was higher than in the other datasets. It was found that several types of data are needed to get the highest classification accuracy whereas omitting some feature groups reduced the classification accuracy. A major habitat type in the study area was mesic forests in different successional stages. It was found that the inner heterogeneity of different mesic forest age groups was large and other habitat types were often inside this heterogeneity.     

Data fusion and classifier ensemble techniques for vegetation mapping in the coastal Everglades

This study examined the applicability of data fusion and classifier ensemble techniques for vegetation mapping in the coastal Everglades. A framework was designed to combine these two techniques. In the framework, 20-m hyperspectral imagery collected from Airborne Visible/Infrared Imaging Spectrometer was first merged with 1-m Digital Orthophoto Quarter Quads using a proposed pixel/feature-level fusion strategy. The fused data set was then classified with an ensemble approach based on two contemporary machine learning algorithms: Random Forest and Support Vector Machine. The framework was applied to classify nine vegetation types in a portion of the coastal Everglades. An object-based vegetation map was produced with an overall accuracy of 90% and Kappa value of 0.86. Per-class classification accuracy varied from 61% for identifying buttonwood forest to 100% for identifying red mangrove scrub. The result shows that the framework is promising for automated vegetation mapping in the Everglades.     

利用遥感进行退耕还林成活率及长势监测方法的研究

本文以张家口退耕还林工程的新造经济林为例,提出了一种利用高分辨率遥感技术监测新造林成活率及长势的方法。主要采用面向对象的信息提取技术提取退耕地新造林的树冠信息。开发了基于树冠分布图的树冠因子提取程序,计算树冠因子,统计造林成活率,从而掌握新造林地的现状。最后,根据实际测量的数据进行误差检验,由遥感数据自动提取的树冠冠幅平均误差为:东西冠幅为0.337m,南北冠幅为0.433m;计算新造林成活率的精度达到了89.837%。为退耕还林工程科学,高效的管理及决策支持提供了依据。     

Detecting nutrient deficiency in spruce forests using multispectral satellite imagery

Nutrient deficiency in forest stands has a negative impact on timber production. Although there are numerous studies investigating nutrient deficiency in forests using remote sensing, research has usually focused on extracting nutrient/pigment concentrations using hyperspectral imagery. Results of studies using this method of assessment are uncertain at the canopy level. This study proposes using freely available multispectral imagery to identify nutrient deficiency in commercially managed forest plantations. A classification map of nutrient deficient, healthy, and a third class, other, for State spruce forests in the Republic of Ireland was constructed using multispectral Sentinel 2 images from Spring and a Random Forest model. The forest area of interest (AOI) was Sitka spruce or Norway spruce plantations greater than 12 years old. Results showed that the overall accuracy was 89% and the associated Kappa Index of agreement was 79%. An unbiased area estimator was vital for an accurate estimate of the scale of nutrient deficiency, which concluded that 23% of the AOI was nutrient deficient. Early detection of nutrient deficiency is crucial to mitigate negative impacts on productivity so a time series analysis of the spectral response of healthy and nutrient deficient classes using Google Earth Engine's Landsat 5, 7, and 8 archive was carried out. A control of known nutrient deficient sites, as identified through foliar analysis, was used for comparison with the nutrient deficient and healthy training data. The spectral response showed a decrease through time for all of the foliar analysis and training data using the green (520–600 nm), red (630–690 nm), and SWIR spectra (1550–1700 nm) during Spring. This decreasing trend is due to the growth of foliage, with the difference in spectral response between nutrient deficient and healthy stands being attributed to the presence of chlorosis in stands suffering from nutrient deficiency. Spectral thresholds using digital numbers for nutrient deficient stands were identified for an operational optimum age cohort of between 10–12 years old which will be used for early detection.     

Unmanned aerial vehicles (UAV) for assessment of qualitative classification of Norway spruce in temperate forest stands

Abstract

The study investigates the potential of UAV-based remote sensing technique for monitoring of Norway spruce health condition in the affected forest areas. The objectives are: (1) to test the applicability of UAV visible an near-infrared (VNIR) and geometrical data based on Z values of point dense cloud (PDC) raster to separate forest species and dead trees in the study area; (2) to explore the relationship between UAV VNIR data and individual spruce health indicators from field sampling; and (3) to explore the possibility of the qualitative classification of spruce health indicators. Analysis based on NDVI and PDC raster was successfully applied for separation of spruce and silver fir, and for identification of dead tree category. Separation between common beech and fir was distinguished by the object-oriented image analysis. NDVI was able to identify the presence of key indicators of spruce health, such as mechanical damage on stems and stem resin exudation linked to honey fungus infestation, while stem damage by peeling was identified at the significance margin. The results contributed to improving separation of coniferous (spruce and fir) tree species based on VNIR and PDC raster UAV data, and newly demonstrated the potential of NDVI for qualitative classification of spruce trees. The proposed methodology can be applicable for monitoring of spruce health condition in the local forest sites.     

Broad-scale spatial pattern of forest landscape types in the Guiana Shield

Detecting broad scale spatial patterns across the South American rainforest biome is still a major challenge. Although several countries do possess their own, more or less detailed land-cover map, these are based on classifications that appear largely discordant from a country to another. Up to now, continental scale remote sensing studies failed to fill this gap. They mostly result in crude representations of the rainforest biome as a single, uniform vegetation class, in contrast with open vegetations. A few studies identified broad scale spatial patterns, but only when they managed to map a particular forest characteristic such as biomass. The main objective of this study is to identify, characterize and map distinct forest landscape types within the evergreen lowland rainforest at the sub-continental scale of the Guiana Shield (north-east tropical South-America 10° North-2° South; 66° West-50° West). This study is based on the analysis of a 1-year daily data set (from January 1st to December 31st, 2000) from the VEGETATION sensor onboard the SPOT-4 satellite (1-km spatial resolution). We interpreted remotely sensed landscape classes (RSLC) from field and high resolution remote sensing data of 21 sites in French Guiana. We cross-analyzed remote sensing data, field observations and environmental data using multivariate analysis. We obtained 33 remotely sensed landscape classes (RSLC) among which five forest-RSLC representing 78% of the forested area. The latter were classified as different broad forest landscape types according to a gradient of canopy openness. Their mapping revealed a new and meaningful broad-scale spatial pattern of forest landscape types. At the scale of the Guiana Shield, we observed a spatial patterns similarity between climatic and forest landscape types. The two most open forest-RSLCs were observed mainly within the north-west to south-east dry belt. The three other forest-RSLCs were observed in wetter and less anthropized areas, particularly in the newly recognized “Guianan dense forest arch”. Better management and conservation policies, as well as improvement of biological and ecological knowledge, require accurate and stable representations of the geographical components of ecosystems. Our results represent a decisive step in this way for the Guiana Shield area and contribute to fill one of the major shortfall in the knowledge of tropical forests.     

Change detection and classification of land cover at Hustai National Park in Mongolia

Land cover types of Hustai National Park (HNP) in Mongolia, a hotspot area with rare species, were classified and their temporal changes were evaluated using Landsat MSS TM/ETM data between 1994 and 2000. Maximum-likelihood classification analysis showed an overall accuracy of 88.0% and 85.0% for the 1994 and 2000 images, respectively. Kappa coefficients associated with the classification were resulted to 0.85 for 1994 and 0.82 for 2000 image. Land cover types revealed significant temporal changes in the classification maps between 1994 and 2000. The area has increased considerably by 166.5 km² for mountain steppe and by 12 km² for a sand dune. By contrast, agricultural areas and degraded areas affected by human being activity were decreased by 46.1 km² and 194.8 km² over the 6-year span, respectively. These areas were replaced by mountain steppe area. Specifically, forest area was noticeably fragmented, accompanied by the decrease of ∼400 ha. The forest area revealed a pattern with systematic gain and loss associated with the specific phenomenon called as ‘forest free-south slope’. We discussed the potential environmental conditions responsible for the systematic pattern and addressed other biological impacts by outbreaks of forest pests and ungulates.     

Mapping the distributions of C3 and C4 grasses in the mixed-grass prairies of southwest Oklahoma using the Random Forest classification algorithm

The objective of this paper is to demonstrate a new method to map the distributions of C₃ and C₄ grasses at 30 m resolution and over a 25-year period of time (1988–2013) by combining the Random Forest (RF) classification algorithm and patch stable areas identified using the spatial pattern analysis software FRAGSTATS. Predictor variables for RF classifications consisted of ten spectral variables, four soil edaphic variables and three topographic variables. We provided a confidence score in terms of obtaining pure land cover at each pixel location by retrieving the classification tree votes. Classification accuracy assessments and predictor variable importance evaluations were conducted based on a repeated stratified sampling approach. Results show that patch stable areas obtained from larger patches are more appropriate to be used as sample data pools to train and validate RF classifiers for historical land cover mapping purposes and it is more reasonable to use patch stable areas as sample pools to map land cover in a year closer to the present rather than years further back in time. The percentage of obtained high confidence prediction pixels across the study area ranges from 71.18% in 1988 to 73.48% in 2013. The repeated stratified sampling approach is necessary in terms of reducing the positive bias in the estimated classification accuracy caused by the possible selections of training and validation pixels from the same patch stable areas. The RF classification algorithm was able to identify the important environmental factors affecting the distributions of C₃ and C₄ grasses in our study area such as elevation, soil pH, soil organic matter and soil texture.     

Accurate mapping of forest types using dense seasonal Landsat time-series

An accurate map of forest types is important for proper usage and management of forestry resources. Medium resolution satellite images (e.g., Landsat) have been widely used for forest type mapping because they are able to cover large areas more efficiently than the traditional forest inventory. However, the results of a detailed forest type classification based on these images are still not satisfactory. To improve forest mapping accuracy, this study proposed an operational method to get detailed forest types from dense Landsat time-series incorporating with or without topographic information provided by DEM. This method integrated a feature selection and a training-sample-adding procedure into a hierarchical classification framework. The proposed method has been tested in Vinton County of southeastern Ohio. The detailed forest types include pine forest, oak forest, and mixed-mesophytic forest. The proposed method was trained and validated using ground samples from field plots. The three forest types were classified with an overall accuracy of 90.52% using dense Landsat time-series, while topographic information can only slightly improve the accuracy to 92.63%. Moreover, the comparison between results of using Landsat time-series and a single image reveals that time-series data can largely improve the accuracy of forest type mapping, indicating the importance of phenological information contained in multi-seasonal images for discriminating different forest types. Thanks to zero cost of all input remotely sensed datasets and ease of implementation, this approach has the potential to be applied to map forest types at regional or global scales.     

Mapping post-fire forest regeneration and vegetation recovery using a combination of very high spatial resolution and hyperspectral satellite imagery

Careful evaluation of forest regeneration and vegetation recovery after a fire event provides vital information useful in land management. The use of remotely sensed data is considered to be especially suitable for monitoring ecosystem dynamics after fire. The aim of this work was to map post-fire forest regeneration and vegetation recovery on the Mediterranean island of Thasos by using a combination of very high spatial (VHS) resolution (QuickBird) and hyperspectral (EO-1 Hyperion) imagery and by employing object-based image analysis. More specifically, the work focused on (1) the separation and mapping of three major post-fire classes (forest regeneration, other vegetation recovery, unburned vegetation) existing within the fire perimeter, and (2) the differentiation and mapping of the two main forest regeneration classes, namely, Pinus brutia regeneration, and Pinus nigra regeneration. The data used in this study consisted of satellite images and field observations of homogeneous regenerated and revegetated areas. The methodology followed two main steps: a three-level image segmentation, and, a classification of the segmented images. The process resulted in the separation of classes related to the aforementioned objectives. The overall accuracy assessment revealed very promising results (approximately 83.7% overall accuracy, with a Kappa Index of Agreement of 0.79). The achieved accuracy was 8% higher when compared to the results reported in a previous work in which only the EO-1 Hyperion image was employed in order to map the same classes. Some classification confusions involving the classes of P. brutia regeneration and P. nigra regeneration were observed. This could be attributed to the absence of large and dense homogeneous areas of regenerated pine trees in the study area.     

面向对象的高空间分辨率影像分类研究

采用面向对象遥感影像分类方法,进行了高空间分辨率遥感影像信息提取试验,分析了其与基于像元方法的信息提取结果的差异,试验研究表明,在目视效果上,传统方法的分类结果图中椒盐现象非常明显,而面向对象方法可以有效地避免椒盐现象;在分类精度上,面向对象方法分类结果的总体精度、Kappa系数、生产者精度、用户精度、Hellden精度和Short精度均明显高于传统方法,各类地物提取效果显著提高。面向对象方法在高空间分辨率遥感影像信息提取中具有明显的优势。     

Object-oriented crop mapping and monitoring using multi-temporal polarimetric RADARSAT-2 data

The aim of this paper is to assess the accuracy of an object-oriented classification of polarimetric Synthetic Aperture Radar (PolSAR) data to map and monitor crops using 19 RADARSAT-2 fine beam polarimetric (FQ) images of an agricultural area in North-eastern Ontario, Canada. Polarimetric images and field data were acquired during the 2011 and 2012 growing seasons. The classification and field data collection focused on the main crop types grown in the region, which include: wheat, oat, soybean, canola and forage. The polarimetric parameters were extracted with PolSAR analysis using both the Cloude–Pottier and Freeman–Durden decompositions. The object-oriented classification, with a single date of PolSAR data, was able to classify all five crop types with an accuracy of 95% and Kappa of 0.93; a 6% improvement in comparison with linear-polarization only classification. However, the time of acquisition is crucial. The larger biomass crops of canola and soybean were most accurately mapped, whereas the identification of oat and wheat were more variable. The multi-temporal data using the Cloude–Pottier decomposition parameters provided the best classification accuracy compared to the linear polarizations and the Freeman–Durden decomposition parameters. In general, the object-oriented classifications were able to accurately map crop types by reducing the noise inherent in the SAR data. Furthermore, using the crop classification maps we were able to monitor crop growth stage based on a trend analysis of the radar response. Based on field data from canola crops, there was a strong relationship between the phenological growth stage based on the BBCH scale, and the HV backscatter and entropy.