基于多季相Sentinel-2影像的白洋淀湿地信息提取

白洋淀湿地是华北平原上重要的浅水湖泊湿地,对雄安新区绿色发展具有重要的生态价值。对白洋淀高度异质化的景观格局进行分类,能够为白洋淀湿地资源的遥感监测提供指导意义。针对湿地季节变化的特点,对白洋淀每个季节选取一期具有代表性的Sentinel-2影像,采用分类与回归树（CART）、支持向量机（SVM）、随机森林（RF）3种常用的机器学习分类器对15种季相组合实验方案进行分类,分析不同季相遥感影像及其组合对白洋淀湿地信息提取的优劣。结果表明：相较于使用单一季相影像分类,多季相影像的组合能够显著提高分类精度,春&夏季相组合能够得到最优的分类效果,相对单季影像总体分类精度提高了10.9%~25.5%,Kappa系数提高了0.09~0.29;SVM分类器的分类表现较为稳定,能够得到最高的平均分类精度,CART分类器在处理高维特征的能力不如随机森林和SVM;不同特征类型对湿地信息提取的贡献度从高到底依次是红边光谱特征、传统光谱特征、缨帽变换特征、主成分分析特征、纹理特征。实验成果能为湿地信息的遥感识别提供依据。     

植物污染胁迫遥感监测研究综述

我国工农业在过去几十年内迅速发展,随之而来的是大批环境问题的涌现,其中包括工农业对植物的污染问题。由于需要监测的范围越来越广、监测的对象越来越多,利用遥感技术监测受污植物受到越来越多的重视。首先提出了影响植被光谱的主要因素有叶绿素含量、细胞结构以及水含量,分析了植物在受到污染胁迫时光谱特征的变化,进而介绍了当前遥感监测植物污染胁迫的主要指标：植被指数和红边参数,并对主要植物污染物--重金属和烃类物质污染应用进行了详细阐述。指出了找到合适的植被指数和利用合适方法精确提取植被红边是监测的关键。最后对用遥感技术监测植物污染提出展望。     

基于KNN-FIFS的内蒙古根河森林郁闭度遥感估测研究

为探索国产高分一号宽幅（GF-1 Wide Field of View，GF-1 WFV）数据以及具有宽覆盖、红边波段（Red-Edge band，RE）的高分六号（GF-6）卫星数据在森林郁闭度（Forest Canopy Closure，FCC）定量反演中的潜力，本研究以GF-1 WFV多光谱数据为基础，添加哨兵2号（Sentinel-2A）红边波段，模拟GF-6红边波段特性，并提取相关纹理信息（Texture Information，TI）、植被指数（Vegetation Index，VI）和红边指数（Red- edge Index，RI），同时添加太阳入射角的余弦值cosi和1/cosi进一步探究了地形因素（Topographic Factors，TF）对FCC估测的影响，利用快速迭代特征选择的k-NN（k-Nearest Neighbor with Fast Iterative Features Selection，KNN-FIFS）模型，实现了内蒙古大兴安岭根河研究区FCC的定量反演，并对比逐步多元线性回归（Stepwise Multiple Linear Regressions，SMLR）和支持向量机（Support Vector Machine，SVM）估测结果。通过44块调查样地实测数据验证发现：基于GF-1 WFV估测的FCC与实测数据具有很好的一致性，R²=0.52，RMSE=0.08；GF-1 WFV+VI+TI估测结果为R²=0.56，RMSE=0.08；GF-1 WFV+RE+RI+TI的精度明显提高，R²=0.63，RMSE=0.07；GF-1 WFV+RE+RI+TI+TF的精度最高，R²=0.68，RMSE=0.07，并高于SMLR（R²=0.39，RMSE=0.10）和SVM（R²=0.49，RMSE=0.10）方法。KNN-FIFS方法比SMLR和SVM方法更适用于FCC遥感估测，且添加红边信息经地形校正后，能有效提高FCC的估测精度。     

LAI assessment of wheat and potato crops by VENμS and Sentinel-2 bands

Leaf Area Index (LAI) is an important variable that governs canopy processes and can be monitored by satellites. The current study aims at exploring the potential and limitations of using the red-edge spectral bands of the forthcoming superspectral satellites, namely—Vegetation and Environmental New micro Spacecraft (VENμS) and Sentinel-2, for assessing LAI in field crops. The research was conducted in experimental plots of wheat and potato in the northwestern Negev, Israel. Continuous spectral data were collected by a field spectrometer and LAI data were obtained by a ceptometer. The spectral data were resampled to the superspectral VENμS and Sentinel-2 resolutions. The data were divided into seven datasets (four seasons, two crops, and one including all data). The LAI prediction abilities by Partial Least Squares (PLS) models for continuous spectra and the resampled spectra were compared and evaluated. For wheat and potato of the continuous, VENμS, and Sentinel-2 data formations, the PLS correlation coefficients (r) values were 0.93, 0.93, and 0.92, respectively. In most cases, the red-edge region was found to be the most important spectral region for the three data formations, according to the Variable Importance in Projection (VIP) analysis. Additionally, Normalized Difference Vegetation Index (NDVI) and the Red-Edge Inflection Point (REIP) were computed for the three data formations in order to observe relation to as well as prediction accuracy in retrieving LAI values. The prediction abilities of the calculated indices by the data formations were compared, peaking for wheat, with r values of 0.91 for the REIP for the three data formations. Therefore, it is concluded that VENμS and Sentinel-2 can spectrally assess LAI as good as a hyperspectral sensor. The REIP was found to be a significantly better predictor than NDVI for wheat data and therefore can potentially be implemented for future LAI monitoring applications by superspectral sensors that contain four red-edge bands.     

Use of hyperspectral derivative ratios in the red-edge region to identify plant stress responses to gas leaks

Hyperspectral features in the red-edge region were tested as an index of plant stress responses to soil-oxygen depletion. The aim was to provide the basis for a warning system to identify natural gas leakage by the spectral responses of plants growing in the affected soil.Elevated concentrations of natural gas in the soil atmosphere were used to deplete oxygen concentrations around the roots of grass, wheat (Hordeum vulgare cv Claire) and bean (Vicia faba cv Clipper) growing in a field facility. Visible symptoms due to the natural gas included reduced growth of the plants and chlorosis of the leaves.Spectral responses included increased reflectance in the visible wavelengths and decreased reflectance in the near infra-red. Derivative analysis identified features within the red-edge at 720-730 and 702 nm. Ratios of the magnitude of the derivative at 725 to that at 702 nm were less in areas where gas was present. This ratio enabled identification of stress due to gas leakage up to 7 days before visible symptoms were observed and also at the edges of gassed plots where visible symptoms were not expressed. The technique was able to identify stress responses to long-term leaks in all the crops tested but to short-term leaks only in grass. This study therefore suggests that under appropriate conditions remote sensing could be used to detect pipeline gas leaks from decreases in the ratio of peaks within the red-edge.     

基于KNN-FIFS的内蒙古根河森林郁闭度遥感估测研究

为探索国产高分一号宽幅（GF-1 Wide Field of View,GF-1 WFV）数据以及具有宽覆盖、红边波段（Red-Edge band,RE）的高分六号（GF-6）卫星数据在森林郁闭度（Forest Canopy Closure,FCC）定量反演中的潜力,本研究以GF-1 WFV多光谱数据为基础,添加哨兵2号（Sentinel-2A）红边波段,模拟GF-6红边波段特性,并提取相关纹理信息（Texture Information,TI）、植被指数（Vegetation Index,VI）和红边指数（Red- edge Index,RI）,同时添加太阳入射角的余弦值cosi和1/cosi进一步探究了地形因素（Topographic Factors,TF）对FCC估测的影响,利用快速迭代特征选择的k-NN（k-Nearest Neighbor with Fast Iterative Features Selection,KNN-FIFS）模型,实现了内蒙古大兴安岭根河研究区FCC的定量反演,并对比逐步多元线性回归（Stepwise Multiple Linear Regressions,SMLR）和支持向量机（Support Vector Machine,SVM）估测结果。通过44块调查样地实测数据验证发现：基于GF-1 WFV估测的FCC与实测数据具有很好的一致性,R²=0.52,RMSE=0.08;GF-1 WFV+VI+TI估测结果为R²=0.56,RMSE=0.08;GF-1 WFV+RE+RI+TI的精度明显提高,R²=0.63,RMSE=0.07;GF-1 WFV+RE+RI+TI+TF的精度最高,R²=0.68,RMSE=0.07,并高于SMLR（R²=0.39,RMSE=0.10）和SVM（R²=0.49,RMSE=0.10）方法。KNN-FIFS方法比SMLR和SVM方法更适用于FCC遥感估测,且添加红边信息经地形校正后,能有效提高FCC的估测精度。     

在水分胁迫下棉花冠层叶片全氮含量的高光谱遥感估算模型研究

研究利用美国产ASD地物光谱仪,获取新疆北部地区棉花冠层关键生育时期的高光谱数据,采用红边积分面积变量估测棉花冠层叶片的全氮含量,对反射光谱进行一阶微分,应用一阶微分光谱数据,衍生出基于光谱位置变量的分析方法,以红边积分面积(SDr)为自变量,冠层全氮(TN)含量为因变量,做相关分析与处理,构建新陆早6号红边积分面积与冠层叶片TN含量的相关数学模型。研究在不同水处理条件下,对棉花冠层单叶叶绿素含量和单叶全氮含量做相关分析,结果表明:叶绿素含量与TN含量呈显著的正相关(R=0.8723,n=39),叶绿素含量能有效的估计棉花单叶TN含量;红边积分面积变量与冠层TN含量呈显著的相关性,相关系数是0.7394(n=40),利用构建的相关模型可以较为精确地估测棉花两个品种新陆早6号与8号冠层叶片的全氮含量,RMSE分别为0.3859和0.4272。研究认为红边积分面积变量具有预测棉花冠层全氮含量的应用潜力,研究得出利用3边面积变量构造的数学模型对反演作物冠层TN含量有较高应用价值。研究认为,红边位移现象结合红边幅度的变化的研究,用于诊断棉花水分胁迫也是可行的,关键是建立相应合理的诊断指标体系。研究结果证明:①随着棉花的生长发育,叶片的生理生化参数发生变化,冠层的生理生化参数随之发生变化;②.棉花叶片叶绿素含量与叶片的全氮含量相关性显著(R=0.8723,n=38),通过建立数学模型,可以估测叶片中全氮的含量;③由一阶微分光谱衍生出基于光谱“红边”位置变量的分析方法,使我们认识到“红边”的变幅、形状和面积包含了各个波段的信息,这些波段综合产生的变量所构造的模型,为棉花氮素营养参数的估计提供了预测能力;④如果棉花叶绿素含量高,说明水分充足、氮代谢旺盛,植株处于生长旺盛时期,红边向蓝光方向发生了位移。利用红边位移现象结合红边幅度的变化的研究,用于诊断棉花水分胁迫也是可行的,关键是建立相应合理的诊断指标体系。     

在水分胁迫下棉花冠层叶片全氮含量的高光谱遥感估算模型研究

研究利用美国产ASD地物光谱仪,获取新疆北部地区棉花冠层关键生育时期的高光谱数据,采用红边积分面积变量估测棉花冠层叶片的全氮含量,对反射光谱进行一阶微分,应用一阶微分光谱数据,衍生出基于光谱位置变量的分析方法,以红边积分面积(SDr)为自变量,冠层全氮(TN)含量为因变量,做相关分析与处理,构建新陆早6号红边积分面积与冠层叶片TN含量的相关数学模型。研究在不同水处理条件下,对棉花冠层单叶叶绿素含量和单叶全氮含量做相关分析,结果表明:叶绿素含量与TN含量呈显著的正相关(R=0.8723,n=39),叶绿素含量能有效的估计棉花单叶TN含量;红边积分面积变量与冠层TN含量呈显著的相关性,相关系数是0.7394(n=40),利用构建的相关模型可以较为精确地估测棉花两个品种新陆早6号与8号冠层叶片的全氮含量,RMSE分别为0.3859和0.4272。研究认为红边积分面积变量具有预测棉花冠层全氮含量的应用潜力,研究得出利用3边面积变量构造的数学模型对反演作物冠层TN含量有较高应用价值。研究认为,红边位移现象结合红边幅度的变化的研究,用于诊断棉花水分胁迫也是可行的,关键是建立相应合理的诊断指标体系。研究结果证明:①随着棉花的生长发育,叶片的生理生化参数发生变化,冠层的生理生化参数随之发生变化;②.棉花叶片叶绿素含量与叶片的全氮含量相关性显著(R=0.8723,n=38),通过建立数学模型,可以估测叶片中全氮的含量;③由一阶微分光谱衍生出基于光谱“红边”位置变量的分析方法,使我们认识到“红边”的变幅、形状和面积包含了各个波段的信息,这些波段综合产生的变量所构造的模型,为棉花氮素营养参数的估计提供了预测能力;④如果棉花叶绿素含量高,说明水分充足、氮代谢旺盛,植株处于生长旺盛时期,红边向蓝光方向发生了位移。利用红边位移现象结合红边幅度的变化的研究,用于诊断棉花水分胁迫也是可行的,关键是建立相应合理的诊断指标体系。