基于高光谱影像的三江源区不同退化程度高寒草甸分类研究

三江源地区是我国最重要的生态功能区之一。近年来,受全球气候变暖及日趋频繁的人类活动的影响,三江源地区高寒草甸生态系统退化现象明显。以三江源称多县清水河镇东北部地区为实验区,基于环境小卫星HJ\|1A HSI高光谱数据,结合不同退化程度高寒草甸地面光谱采集和样方调查,采用MLC和SAM方法对不同退化程度的高寒草甸开展了分类研究。结果表明:基于高光谱数据的不同退化程度高寒草甸采用SAM方法分类总体精度达到75%以上,证实了分类方法的可行性,基于高光谱数据分类能有效区分盖度相近、退化程度不同的草地类型,其中SAM分类结果更加精细准确,优于MLC方法,SAM方法对中度退化草甸区分能力最高,对其他退化程度草甸区分能力稍弱。     

基于HJ-1A高光谱数据的藏北高原草地分类方法对比

环境减灾星星座A星(HJ-1A)携带的超光谱仪填补了我国星载高光谱影像采集领域的空白,但目前国内关于该高光谱数据的应用较少.本文基于HJ-1A高光谱(HSI)数据预处理技术,以申扎县北部为研究区,采用SPCA-MLC和HSI-SAM分类方法,结合野外实测样本,将研究区分为沼泽草甸、高寒草甸、高寒草原、荒漠化草原和裸地5种类型,并结合分类精度和分类图对2种分类方法进行了对比分析,可得基于HJ-1A高光谱数据的藏北高原草地分类方法中SPCA-MLC法优于HSI-SAM法.2种方法的分类精度皆大于80％,证明了HJ-1A的HSI数据在实现藏北草地高精度分类方面的巨大潜力.     

CHRIS 高光谱图像森林类型分类方法比较研究

以长白山为试验区,选择CHRIS/PROBA高光谱零度角遥感数据,在对其进行预处理的基础上,通过应用最大似然法(MLC)、最小距离法、支持向量机法(SVM)和光谱角填图法(SAM)等几种常用的高光谱遥感分类方法对影像进行森林类型分类。利用混淆矩阵对分类结果进行验证,结果显示:在高光谱遥感森林类型分类中,SVM总体分类精度最高,为84.60％;其次是MLC,为 83.53％,最小距离法73.81％,SAM 56.49％。Kappa系数从高到底为:SVM 0.78,MLC 0.77,最小距离法0.68,SAM 0.52。经过比较分析,得出SVM分类方法精度最高,这表明该方法用于高光谱遥感森林分类中的实用性和优越性。     

基于HJ-1A高光谱遥感数据的湟水流域典型农作物分类研究

利用高光谱遥感技术识别农作物类型已经成为高光谱遥感研究的热点领域。以青海省湟水流域内油菜、小麦和青稞等典型农作物为分类对象,以HJ-1A HSI高光谱数据和GF-1 WFV高分辨率数据为数据源,探讨利用高光谱遥感影像进行农作物类型信息提取的方法。数据经预处理后,首先,利用WFV数据采用面向对象方法提取研究区农作物种植边界,并利用其对HSI高光谱影像进行种植区域提取;其次,将提取后的高光谱影像经数据形式变换获得包括:R、1/R、Log(R)、d(R)、d(Log(R))和CR共6种数据形式;最后,利用上述6种数据形式的全波段数据和经遗传算法GA-SVM进行光谱波段选取后的6种特征数据,采用支持向量机SVM方法进行农作物分类。结果表明:采用基于样本的面向对象分类方法提取耕地信息精度高且实现周期短;利用GA-SVM波段选取后的6种特征数据集进行农作物分类,其精度显著高于全波段数据集分类精度;6种数据变换形式中,d(Log(R))和CR是两种最优的高光谱分类数据形式,其全波段和特征波段数据进行农作物分类均能获得较好的分类精度,总体精度最高分别达88%和86%,而采用1/R、Log(R)和R数据形式需经GA-SVM光谱波段选取后才能获得较优分类精度。     

基于HJ星高光谱数据红边参数的冬小麦叶面积指数反演

针对我国HJ-1A星搭载的高光谱成像仪(HSI)数据,探索基于HJ星高光谱影像的LAI反演研究,本文利用inverted Gaussian模型提取红谷位置、红边位置、红边振幅以及红边斜率4个红边参数,结合2009年4月、5月两期同步地面观测LAI数据,经过回归分析构建了反演叶面积指数的最优红边参数模型.结果表明红边位置、红边斜率和红边振幅与叶面积指数都达到了极显著相关,R2分别为0.5592,0.7796和0.8107说明HJ星高光谱影像数据在叶面积指数反演方面有很大的应用潜力.     

一种改进的遥感图像融合方法:LFF

通过遥感图像融合,可获得更丰富的信息,常用遥感图像融合方法如HIS彩色变换可将不同平台、不同光谱响应范围的高空间分辨率的遥感数据与多光谱遥感数据进行融合,但要求这两组数据的光谱响应范围一致,否则便会产生光谱扭曲的现象,从而影响了地物的识别。针对不同平台、不同光谱响应范围的遥感数据,该文在HSI变换的基础上,提出了一种改进的方法,即LFF融合法,首先对高几何分辨率的全色波段进行LoG滤波,而后将LoG滤波后的全色波段与多光谱经HSI正变换后的强度分量进行灰度直方图匹配,并替换之,经HSI逆变换便得到融合图像。论文从灰度变化指数和分类精度两方面分析了光谱保持性能和融合图像的分类精度,分析结果表明:LFF融合法的光谱保持性能优于HSI变换法,LFF融合后图像的分类精度高于HSI融合后的图像,LFF融合法是一种能较好地保持光谱特性的融合方法。     

基于MODIS NDVI时间序列数据的藏北草地类型识别

选取西藏自治区申扎县北部部分区域作为藏北草原的典型代表区进行研究。对区域内2009年全年23期MODIS NDVI数据收集处理,得到23个波段的时间序列数据。通过Savitzky-Golay滤波算法优化时间序列数据并对不同类型的时间序列曲线做对比分析,提取分类的特征波段。最后采用非监督分类法把研究区分为沼泽草甸、高寒草甸、中盖度草原、低盖度草原、裸地和水体6种类型,并结合地形数据分析了其中4种草地类型的空间分布特征。一系列的处理结果及分析内容证明了MODIS NDVI时间序列数据在藏北草地类型识别方面的可行性和适用性。     

采用分段主成分和PPI的高光谱影像分类

高光谱遥感影像波段多且存在混合像元,特征提取以及端元提取都是高光谱影像分类必不可少的工作,分类方法的选择也是因地适宜。以福建省泉州市德化县下属某一地区的CASI影像为实验数据,基于分段主成分(segmental principal component analysis,SPCA)和纯净像元指数法(pure pixel index,PPI),提出了最小距离(minimum distance classification,MDC)和二进制编码(binary encoding,BE)的高光谱影像分类方法。实验结果表明,MDC的总体精度为69.71%,BE的总体精度为70.88%。对单一地物精度而言2种方法各有其长,MDC对道路的分类精度更高,为98.08%;而植被、耕地和水体采用BE方法的分类精度更高,分别为94.12%、98.08%、98.11%。本文提出的方法应用于CASI高光谱影像,对该研究区的地物分类研究有一定的实用性和参考价值。     

基于面向对象的多源卫星遥感影像玉米倒伏面积提取

风灾引起的玉米倒伏可能导致玉米大量减产,利用遥感技术准确监测玉米倒伏面积与空间分布信息对灾情的评估非常重要。利用Planet和Sentinel-2影像分别结合面向对象与基于像元方法提取研究区玉米倒伏,同时评估了不同影像特征（光谱特征、植被指数和纹理特征）与不同分类方法（支持向量机法SVM、随机森林法RF和最大似然法MLC）对玉米倒伏提取精度的影响。结果表明：①使用高空间分辨率的Planet影像进行玉米倒伏提取的精度普遍高于Sentinel-2影像;②从分类精度和面积精度来看,Planet影像的光谱特征+植被指数+均值特征结合面向对象RF分类,总体精度和Kappa系数分别为93.77%和0.87,面积的平均误差最低为4.76%;③采用Planet和Sentinel-2影像结合面向对象分类提取玉米倒伏精度高于基于像元分类。研究不仅分析了面向对象方法的优势,还评估了使用不用影像数据结合面向对象方法的适用性,可以为遥感提取作物倒伏相关研究提供一定的借鉴。     

藏北中部地区草地退化遥感监测

为了探讨藏北中部地区近20年来的草地退化状况及主要驱动力,基于1990、2000及2010年的Landsat影像,采用混合分类方法对研究区主要景观类型进行分类,并利用景观指数分析景观格局及草地退化特征。另外,基于气象数据,利用灰色关联度方法分析草地退化的主要驱动力。结果表明:草地退化面积在1990~2000年呈增加趋势,在2000~2010年呈减少趋势,2000年之后研究区草地退化状况趋于缓和。造成草地退化的主要气候因素为降水量、气温、日照时数、平均相对湿度和蒸发量等,其中,高寒草甸草地主要受平均相对湿度及降水量的影响,高寒草原草地受蒸发量、日照时数、平均最低气温、降水量、平均相对湿度和平均最高气温的影响均较大,高寒荒漠草地与平均最高气温、日照时数和蒸发量的关系最为密切。     

Temporal and spatial variations of grassland desertification monitoring in Tibet of China

Tibet, the largest region of the Qinghai–Tibet Plateau, is undergoing extensive grassland deterioration and desertification due to both human and natural factors. Alpine meadow and grassland restoration is difficult after degradation; consequently, the desertification of the Tibetan grassland has attracted substantial social attention. This article considered Amdo, Baingoin, Coqên, and Zhongba counties in Tibet as the study areas, employed remote-sensing data, and developed Tibetan grassland desertification classification indices based on field surveys. Moreover, this study used spectral mixture analysis (SMA) methods to interpret remote-sensing image data from the study areas during three periods (1990, 2000, and 2009) and considered the bare sand (gravel) area proportion as the main basis for the evaluation of grassland desertification. The results of this study demonstrate that the slightly, moderately, and severely desertified grasslands of the monitoring zone covered a total area of 114,113.16 km² in 1990, accounting for 82.12% of the study area. The area exhibited no change in 2000 and decreased by 4472.31 km² in 2009. The severely desertified grassland area declined from 1990 to 2009. The degree of grassland desertification in these four Tibetan counties diminished from 1990 to 2009, and the grassland desertification area exhibited a gradual reduction during the same period. Regarding other soil coverage types, the ice and snow area markedly changed and declined to approximately one-third of its original extent during these 20 years, and most of the ice and snow area was converted to bare land and various types of desertified grassland.     

Mapping of native plant species and noxious weeds to investigate grassland degradation in the Three-River Headwaters region using HJ-1A/HSI imagery

ABSTRACT

Characterized by an alpine meadow, the ecological environment system in the ‘Three-River Headwaters’ region (TRHR) is considered to be a typical fragile ecological system. Numerous observations and research results have indicated that grassland degradation has occurred in the TRHR in recent years. However, research related to utilize the species information of grass communities to monitor grassland degradation remains rare. Therefore, the aim of this study is to produce the distribution maps of native plant species and noxious weeds to investigate grassland degradation for livestock farming perspective. In this study, the fused HJ-1A/HSI data was combined with field investigation samples to define the coverage of native plant species and noxious weeds at different coverage levels. Then, coverage distribution maps of native plant species and noxious weeds were produced by using support vector machine (SVM) classification and random forests (RF) regression methods. Meanwhile, the overall accuracy (OA) and root-mean-square error (RMSE) of each coverage map were assessed. Finally, a grassland degradation map was derived according to the native plant species and noxious weeds cover information. The experimental results show that (1) the spectral feature of native plant species and noxious weeds can be distinguished based on field measurement spectra in the TRHR; (2) the fractional coverage of native plant species and noxious weeds can be relatively accurately estimated when coverage is divided into nine levels; (3) the grass coverage estimation accuracies of SVM classification are similar with these of RF regression method. The OAs of SVM classification are 69.7% at nine grassland coverage levels for native plant species and noxious weeds, and corresponding RMSEs are 8.2% and 8.0%, respectively; and (4) the coverage of native plant species is generally higher than that of the noxious weeds in the study area.     

Mapping an inland wetland complex using hyperspectral imagery

The goal is to determine the extent to which heterogeneous inland wetland vegetation communities and their dominant species, as well as adjacent upland vegetation types, can be mapped using 4‐m hyperspectral Compact Airborne Spectrographic Imager (CASI) data. Two classification algorithms, the maximum‐likelihood classifier (MLC) and the spectral angle mapper (SAM), are applied to CASI data acquired over an inland wetland complex located in southern Ontario, Canada. Application of the MLC algorithm to all bands of the CASI data produced better classification results than use of the SAM. Using the MLC, 10 classes were identified with an overall accuracy of 92%. This approach permitted differentiation between areas of shrub‐dominated vegetation communities, floating aquatic communities, emergent aquatics and shallow open water. In the SAM classification, 11 image‐derived spectral endmembers were generated. Wetland classes identified were shrub‐dominated wetlands, floating aquatic vegetation communities, shallow open water and moderately turbid shallow open water. Upland vegetation types were accurately mapped with both algorithms. Reasons why the SAM did not perform as well as the MLC in this complex environment are suggested. It is concluded that high‐resolution hyperspectral data can provide information needed by wetland managers about inland wetland plant communities and their dominant species.     

基于NDVI及DEM的青海湖北岸景观格局空间自相关分析

本文以青海湖北岸为研究区,利用2006年Landsat-5 TM数据,结合植被类型图、SRTM高程数据、1∶5万地形图等数据,获取了该区12类景观类型,分类总体精度为82.91%,Kappa系数为0.81。以NDVI均值、坡向均值、高程均值为区域观测值,计算研究区景观类型的空间自相关性的Moran指数,再根据局部Moran指数的空间集聚分布特征,分析这些景观类型之间的结构稳定性,表现为如下特征:第一,较稳定景观,格局类型包括河谷灌丛、山地灌丛、高寒沼泽、高寒草甸、河流、裸岩;第二,欠稳定景观,格局类型包括湖泊、温性草原、湖滨沼泽;第三,极不稳定景观,格局类型有沙地、石砾地、裸土地,如鸟岛、金沙湾,尕海地区的沙地。     

HJ-1A高光谱数据高效大气校正及应用潜力初探

环境与灾害监测预报小卫星于2009年3月30日开始正式交付使用,A星上搭载了我国自主研制的空间调制型干涉高光谱成像仪（HSI）,作为一种新型传感器,HSI数据的应用在我国还处于探索阶段。要充分发挥超光谱数据优势、进行有效的遥感应用,首先需要消除遥感成像过程中的大气影响,获得不同波段的地物真实反射辐射信息。通过使用FLAASH大气辐射传输模型对HSI数据进行大气校正,并与表观反射率进行对比分析,证明了校正后获得的地表光谱反射率的有效性。同时基于校正后得到的光谱反射率图像,进行改良型土壤调整植被指数（MSAVI）与叶面积指数（LAI）的反演,初步展现了HSI数据的实际应用效果。     

Effects of topography and human activity on the net primary productivity (NPP) of alpine grassland in northern Tibet from 1981 to 2004

Using the Carnegie–Ames–Stanford Approach (CASA) model on remote-sensing (RS), climatic, and other related data from 1981 to 2004, the researchers estimated the net primary productivity (NPP) of alpine grassland in northern Tibet. Geographical information system (GIS) techniques were used to analyse the spatial pattern of change in the NPP of alpine grassland and its response to the intensity of human activity. The researchers found that the mean values of NPP on flat (slope gradients <1°) and sunny slopes were relatively lower. Between 1981 and 2004, the NPP of alpine grassland in northern Tibet tended to decrease, but with relatively large annual fluctuations. In northern Tibet, the alpine grassland NPP for high-elevation regions has a greater proportion of area (over 26%) showing a decreasing trend. The change is more significant in areas where the slope is 15–30° and aspect has little influence on the extent of the change. The negative effects of local residential areas on the rate of change of alpine grassland NPP are smaller than those of roads.     

Effective feature extraction through segmentation-based folded-PCA for hyperspectral image classification

The remote sensing hyperspectral images (HSIs) usually comprise many important information of the land covers capturing through a set of hundreds of narrow and contiguous spectral wavelength bands. Appropriate classification performance can only offer the required knowledge from these immense bands of HSI since the classification result is not reasonable using all the original features (bands) of the HSI. Although it is not easy to calculate the intrinsic features from the bands, band (dimensionality) reduction techniques through feature extraction and feature selection are usually applied to increase the classification result and to fix the curse of dimensionality problem. Though the Principal Component Analysis (PCA) has been commonly adopted for the feature reduction of HSI, it can often fail to extract the local useful characteristics of the HSI for effective classification as it considers the global statistics of the HSI. Consequently, Segmented-PCA (SPCA), Spectrally-Segmented-PCA (SSPCA), Folded-PCA (FPCA) and Superpixelwise PCA (SuperPCA) have been introduced for better feature extraction of HSI in diverse ways. In this paper, feature extraction through SPCA & FPCA and SSPCA & FPCA, termed as Segmented-FPCA (SFPCA) and Spectrally-Segmented-FPCA (SSFPCA) respectively, has further been improved through applying FPCA on the highly correlated or spectrally separated bands’ segments of the HSI rather than not applying the FPCA on the entire dataset directly. The proposed methods are compared and analysed for a real mixed agricultural and an urban HSI classification using per-pixel SVM classifier. The experimental result shows that the classification performance using SSFPCA and SFPCA outperforms that of using conventional PCA, SPCA, SSPCA, FPCA, SuperPCA and using the entire original dataset without employing any feature reduction. Moreover, the proposed feature extraction methods provide the least memory and computation cost complexity.