PROSPECT5耦合4SAIL模型的亚热带典型森林冠层反射率时间序列模拟

冠层反射率在森林植被类型精确解译、森林碳同化关键参数如叶面积指数(LAI)、叶绿素等遥感反演等方面具有重要意义.本研究以亚热带毛竹林、雷竹林和常绿落叶阔叶混交林3种典型森林类型为研究对象,通过耦合PROSPECT5和4SAIL模型模拟其冠层反射率时间序列.首先,对PROSPECT5和4SAIL模型参数进行敏感性分析,探讨模型参数对冠层反射率的影响;其次,利用实测反射率对不敏感参数进行优化,并确定其参数值;最后,耦合PROSPECT5和4SAIL模型模拟3种亚热带森林冠层反射率,并与MODIS反射率进行对比.结果表明:LAI对第1、2、3、5、7波段最敏感,各波段的总敏感指数分别为0.80、0.83、0.94、0.66、0.47;叶绿素含量对第4波段最敏感,总敏感指数为0.59;叶片含水量对第6波段的敏感性最大,总敏感性指数为0.54;叶子结构参数、类胡萝卜素、热点参数、干物质含量和土壤干湿比等参数对各个波段都不敏感或敏感性较小.优化后的PROSPECT5和4SAIL模型模拟得到的冠层反射率能够真实反映3种典型森林的季节性变化规律,通过与MODIS反射率对比分析发现,模拟冠层反射率和MODIS反射率之间具有较高的决定系数,分别为0.86、0.90、0.93,均方根误差(RMSE)也较小,分别为0.09、0.07、0.05,且模拟反射率能在一定程度上解决MODIS反射率数据冬季易受雨雪、混合像元影响等问题.     

基于辐射传输模型的高光谱植被指数与叶绿素浓度及叶面积指数的线性关系改进

高光谱植被指数以其特有的精细光谱特征,能够获得非常细微的植被生理状况和环境胁迫差异,因而使遥感技术在精细农业中的应用.尤其是在叶绿素浓度和叶面积指数的反演上面有着广阔的应用前景.然而,现有的植被指数往往和这2个参数呈非线性关系,且只对某一区间的数值敏感,无法适用于其它植被覆盖程度的研究.为了寻找合适的波段位置以改善植被指数与叶绿素浓度和叶面积指数的线性关系,去除饱和区域,进而提高这2个参数的实际估算精度,该文选取了叶绿素浓度和叶面积指数,以辐射传输模型PROSPECT和SAIL为基础,模拟了这2个参数变化对3类高光谱植被指数(归一化植被指数(NDVI)、优化的简单比值指数(MSR)和优化的叶绿素吸收率指数(MCARI))的影响.叶绿素浓度变化敏感性分析结果表明,对这3类植被指数而言,750 nm和705 nm的叶片反射率更适合实际的叶绿素浓度反演.以750 nm和705 nm代替800 nm/700 nm和670 nm成功地提高了3类植被指数与叶绿素浓度的线性相关程度,其中MCARI705和叶绿素浓度基本呈线性关系.叶面积指数变化敏感性分析I口j样显示,以750 nm和705 nm组成的植被指数能够获取更可靠的叶面积指数信息,尤其对于高植被覆盖区域.其中MCARI705能较好地降低随叶面积指数变化的饱和程度,相比其它植被指数,当叶面积指数大于8时,MCARI705才出现明显的饱和.由于冠层的尺度效应,波段位置的选择对植被指数与叶面积指数线性关系的改善不及对叶绿素浓度明显.     

基于辐射传输模型的高光谱植被指数与叶绿素浓度及叶面积指数的线性关系改进

高光谱植被指数以其特有的精细光谱特征, 能够获得非常细微的植被生理状况和环境胁迫差异, 因而使遥感技术在精细农业中的应用, 尤其是在叶绿素浓度和叶面积指数的反演上面有着广阔的应用前景。然而, 现有的植被指数往往和这2个参数呈非线性关系, 且只对某一区间的数值敏感, 无法适用于其它植被覆盖程度的研究。为了寻找合适的波段位置以改善植被指数与叶绿素浓度和叶面积指数的线性关系, 去除饱和区域, 进而提高这2个参数的实际估算精度, 该文选取了叶绿素浓度和叶面积指数, 以辐射传输模型PROSPECT和SAIL为基础, 模拟了这2个参数变化对3类高光谱植被指数(归一化植被指数(NDVI)、 优化的简单比值指数(MSR)和优化的叶绿素吸收率指数(MCARI))的影响。叶绿素浓度变化敏感性分析结果表明, 对这3类植被指数而言, 750 nm 和705 nm 的叶片反射率更适合实际的叶绿素浓度反演。以750 nm 和705 nm代替 800 nm/700 nm 和670 nm成功地提高了3类植被指数与叶绿素浓度的线性相关程度, 其中MCARI705 和叶绿素浓度基本呈线性关系。叶面积指数变化敏感性分析同样显示, 以750 nm 和705 nm 组成的植被指数能够获取更可靠的叶面积指数信息, 尤其对于高植被覆盖区域。其中MCARI705 能较好地降低随叶面积指数变化的饱和程度, 相比其它植被指数, 当叶面积指数大于8时, MCARI705 才出现明显的饱和。由于冠层的尺度效应, 波段位置的选择对植被指数与叶面积指数线性关系的改善不及对叶绿素浓度明显。     

基于辐射传输模型的叶绿素含量定量反演

利用基于叶片内部辐射传输机制的PROSPECT模型模拟大量不同生化含量和叶肉结构的叶片光谱,研究利用高光谱植被指数定量反演叶绿素含量的可行性和精度,并比较各指数的稳定性和抗干扰能力。结果显示,各指数在对叶绿素的敏感性方面相差不大,除三角植被指数(TVI)外,其它指数均随叶绿素含量的增加而减小。叶片水分含量的差异对各指数的影响很小,干物质次之,叶肉结构影响最大。在抵抗干物质影响和叶肉结构影响方面,结构无关色素指数(SIPI)明显优于其它四种指数,吸收中心波深归一化后的面积指数(ABNC)次之。通过使用叶片光学模型的模拟光谱来研究叶绿素含量变化的光谱响应及其影响因素和反演策略,具有较强的理论性和普适性。研究结果与实际观测相吻合,方法简单易行。     

秋茄类胡萝卜素含量高光谱反演

类胡萝卜素(Car)作为植物主要色素,对诊断植被生理状态有重要作用。于2013年4月和7月采集闽江口秋茄(Kandelia candel)叶片,室内测定其叶片正面和背面反射光谱,同时测定其Car含量\[单位面积(μg·cm-2)和单位质量(mg·g-1)\]。选取常见Car含量估算的光谱参数,同时分析确定最佳比值植被指数(SR),基于回归分析,建立秋茄叶片Car含量估算与验证模型。结果表明,叶片光谱反射率表现为叶片背面大于正面(350~2350 nm);基于叶面背面光谱计算的SR与叶片Car含量(μg·cm-2)的相关系数优于其他组合,相关系数较高的区域分布在520~540 nm与1000~1100 nm波段组合,700~720 nm与800~1100 nm波段组合;基于背面光谱计算的大部分光谱参数与Car含量(μg·cm-2)的相关系数要高于基于正面光谱计算的。因此,以叶片背面光谱作为Car含量估算的光谱数据,以单位面积Car含量为估算量纲建立反演模型。本研究表明,光谱指数LCI、DD、NDVI(770,713)、NDVI(773,562)、SR(723,770)和SR(1000,700)均可实现Car含量的反演,估算与检验模型的R2均0.65,RMSE均1.52;并且新构建的SR(1000,700)估算精度最好,模型和检验R2分别为0.77和0.87,模型和检验RMSE分别为1.08和1.11。这些预味着基于高光谱遥感对闽江河口湿地秋茄Car含量进行估算是可行的。     

植物叶绿素含量的遥感模型及用陆地卫星数据所作的二维分布图

为了确定全球和区域尺度的植物叶绿素含量的多少和分布,遥感方法是一种很好的途径。有些遥感科学家直接利用归一化植被指数(NDVI)等植被指数推算叶绿素含量。实践表明,这种做法的缺点有:1.植被指数NDVI中的近红外波段反射率与叶绿素含量基本无关,这样NDVI中的主信息并没有表达叶绿素含量,其信噪比显著降低;2.存在明显的信息不确定性,同一个植被指数NDVI可能有两种状态:较高叶面积指数伴有较低叶绿素含量或者是较低叶面积指数伴有较高叶绿素含量。通过实践,作者推导出可用LANDSAT和NOAA影像图灰度值推算植被叶绿素含量遥感模型。并在中国科学院禹城遥感实验场实际应用中得到验证:用其模型推算植被叶绿素含量的精度比NDVI有较明显提高,不确定性有明显的降低。     

高光谱与拟合多光谱植被指数反演武夷山亚高山草甸LAI的对比研究

植被叶面积指数(Leaf Area Index, LAI)是重要的生态学参数, 被广泛用于指示植被密度、生物量、碳、氮物质循环以及气候变化对生态系统的影响, 也作为生态过程模型的重要输入参数。地面实测高光谱遥感数据能以更高的空间分辨率及更高的光谱分辨率监测植物的光谱特征, 为精准反演LAI提供了基础。本项研究以武夷山国家公园黄岗山顶的亚高山草甸为研究对象, 通过建立多种高光谱植被指数和拟合多光谱植被指数反演叶面积指数的统计模型, 并比较高光谱与多光谱对叶面积指数反演的效果, 阐明用于反演高覆盖率亚高山草甸的最适高光谱和拟合多光谱植被指数。结果表明: 高光谱新植被指数(NVI)对于反演LAI有最好的效果, R2 = 0.85, P < 0.01; 依据高光谱NVI拟合而成的多光谱NVI反演结果次之, R2 = 0.82, P < 0.01。几种常用比值植被指数NDVI、MSR、RVI和GNDVI在高光谱和拟合多光谱反演结果中相差不大, 表现较好, R2都在0.65以上。通过对比高光谱和拟合Sentinel-2A和Landsat-8两种多光谱卫星波段的反演结果发现, 光谱响应函数中具有更窄波段范围的近红外、红、绿、蓝波段构成的植被指数可以得到更好的反演结果, 而固定波段的高光谱植被指数未必在每种植被指数中都具有最好的反演效果。同时, 发现当某种植被指数反演LAI的线性回归方程的斜率越大, 说明这种植被指数越有可能随LAI的增大而出现饱和现象, 相反的, 斜率越小则说明该种植被指数没有出现饱和现象。此外, 在研究区内使用高光谱和拟合多光谱波段植被指数法反演LAI, NDVI都获得了较好的效果, 存在很好的线性关系, 之前的很多研究和判断都认为NDVI不适用于反演高覆盖植被的LAI, 这个发现是具有意义的, 表明高覆盖植被的叶面积指数在一定范围内是能够被NDVI(应用最广泛的植被指数)较好的反演, 进一步扩展了NDVI反演LAI的适用性和可能性。     

辐射传输模型多尺度反演植被理化参数研究进展

植被是生态系统最重要的组成成分之一,许多与植被有关的物质能量交换过程都与植被的理化参数密切相关,因此定量估算植被的理化参数含量对监测植被生长状况、森林火灾预警以及研究全球碳氮循环过程等都具有重要意义.在众多定量反演植被理化参数的方法中,基于数学、物理学以及生物学的基本理论建立起来的辐射传输模型受到越来越多的关注.辐射传输模型描述了植被与入射辐射之间的相互作用过程和特征,相对于传统的经验/半经验方法,辐射传输模型物理意义明确,具有稳定性和可移植性强的特点.在分析国内外最新相关研究的基础上,首先从植被叶片、冠层和像元3个不同的尺度阐述反演植被理化参数的辐射传输模型.叶片尺度上主要介绍PROSPECT模型和LIBERTY模型;冠层尺度上主要介绍SAIL冠层辐射传输模型以及PROSPECT与SAIL耦合的PROSAIL叶片-冠层辐射传输模型;像元尺度的植被理化参数反演目前主要采用冠层尺度的辐射传输模型.其次,分析尺度变化下植被理化参数遥感反演所面临的主要问题,如不同尺度下模型参数敏感性的变化、辐射传输模型的选取以及混合像元的影响等.最后,总结展望植被理化参数反演多模型与多种数据源相互结合的研究趋势,以及将来具有高空间分辨率的高光谱遥感卫星升空后所带来的发展前景.     

植被生化组分光谱模型抗土壤背景能力分析

应用LOPEX'93(Leaf Optical Properties Experiment)数据,分析了统计回归模型在进行植被叶绿素和水分反演中抗土壤背景影响的能力,模型参数分别使用了:反射率及其变化形式、光谱位置变量、植被指数。在LOPEX'93数据库的植被波谱中分别加入10%-90%的实测土壤光谱信息,得到植被与土壤的混合光谱,并分析混合光谱对植被生化组分的响应。结果表明:应用反射率及其变化形式进行植被叶绿素反演时,以730nm和400nm组合的反射率和反射率倒数的对数为参数的模型具有最高的抗土壤背景能力,在土壤背景所占比例从低到高的变化过程中,以二者反射率组合为参数的模型与叶绿素的相关系数,始终保持在0.645附近,以二者反射率倒数的对数为参数的模型与植被叶绿素的相关系数保持在0.650附近;应用反射率及其变化形式进行植被含水量反演时,以1100,1170,1000,1040,1080nm组合的反射率为参数的模型以及以1170,960,1210,1090,1080,950,1220,1210nm反射率倒数的对数组合为参数的模型具有较高的稳定性,在土壤组分变化的过程中,以上模型与植被含水量的相关系数均稳定的高于0.99;对于光谱位置变量的分析中,以红边-绿峰-红谷组合的模型与植被叶绿素含量具有较高、而且稳定的相关系数,在土壤背景所占比例变化的情况下,相关系数稳定在0.53附近;在应用植被指数进行叶绿素的反演过程中,植被指数与叶绿素的相关系数在土壤背景所占比例变化的情况下变化较大,抗土壤背景的能力均较差;在应用植被指数进行植被水分含量的反演时,以水分指数Ratio975和Ratio1200相关系数最高,且在不同比例土壤背景变化下稳定,相关系数分别分布在0.980附近和0.960附近。该结果可用于指导不同植被覆盖条件下植被冠层参数的反演,提高反演的稳定性和准确性。     

基于Sentinel-2A影像干旱区棉花叶片SPAD数字制图

植被叶片叶绿素是农业遥感反演的重要参数,叶绿素含量的变化与植被生长环境的胁迫程度、生理变化密切相关,故将植被叶绿素进行实时、动态监测对农业生产极为重要。然而,传统经验模型及叶绿素精准测量存在困难。基于高分辨率的Sentinel-2A数据,在机器学习框架下,利用光谱信息、最适光谱指数和基于PROSAIL辐射传输模型的生物协变量构建3种建模方案（方案1：光谱信息和最适光谱指数联合,方案2：光谱信息和物理模型生物协变量联合,方案3：光谱信息、最适光谱指数和物理模型生物协变量联合）。最终基于优选出的建模方案进行棉花叶片叶绿素相对含量的空间数字制图。结果表明：（1）红边波段参与的最适光谱指数比值植被指数（RVI）与棉花叶片SPAD值相关性最高r=0.767,P^**=0.195;（2）将构建的17个变量进行重要性分析可知,构建的最适光谱指数比值植被指数（RVI）与物理模型生物协变量LAI-Cab对估算模型的精度贡献率较大;（3）建模方案构建植被指数时红边波段被确定为最优波段,在增加精度方面起到决定性作用;通过模型评价标准来分析3种方案可知,预测精度大小顺序为模型方案3 > 模型方案1 > 模型方案2,其中方案3的决定系数R²最高为0.826,即估算模型方案3对棉花叶片SPAD值具有最好的预测能力,可以为干旱区农作物的生理参数反演提供新的思路,为农业安全监测,合理水肥配置提供科学数据支持。     

Quantitative estimation of vegetation changes by comparing two vegetation maps

The main goal of our work was to estimate how large the errors are associated with repeated vegetation mapping. We compared two vegetation maps (both 1:10 000) of the Pieniny National Park—ca 2300 ha (southern Poland), the first made in 1966 and the second in 2001. We superimposed—using the ARC-INFO software—a dense grid of points (50 × 50 m) upon each map, and we determined the identity of vegetation unit in each point for the year 1966 and for the year 2001. That procedure was repeated 100 times, each time changing the position of the grid by a random vector. To estimate the size of mapping errors, we compared the patches of communities which should not change their location during 35 years: vegetation of rocky outcrops and local wet depressions, and fertile beech forest, considered a climax community for the Pieniny Mountains. Overlapping small vegetation patches (average patch size below 0.5 ha) yielded highly erroneous results, while the reliability of overlapping the communities with large patches is much higher, exceeding 80% for average patch size of 5 ha. Taking into consideration the communities of average patch size of 1 ha, we can estimate that the vegetation has undergone profound changes: some communities expanded, while others shrunk. The area of meadows remained about the same, but majority of meadows in 2001 was located in former arable fields and previous meadow areas become forested. Among beech forests, we recorded an increase of area covered by floristically rich variants at the expense of floristically poor variants. We conclude that some information about vegetation changes may be obtained only by comparing sequential vegetation maps, but the reliability of the results strongly depends on the size of vegetation patches.     

Ecological vegetation maps

On ecological vegetation maps, the distribution of vegetation is related to one or more features of the environment. Tolerance, competition, map scales and the environment are discussed with regard to their bearing on the geographical distribution of phytocenoses and their portrayal on maps. There are two types of ecological vegetation maps: those relating the vegetation to one environmental quality, and those with two or more such qualities. The interpretation of ecological vegetation maps is relatively simple when plant communities are related to a single quality of the biotope and difficult but usually more useful when related to several qualities. Perfection is not possible but can be approached asymptotically.     

Potential replacement vegetation: an approach to vegetation mapping of cultural landscapes

Abstract. The concept of mapping potential replacement vegetation (PRV) is proposed as a parallel to potential natural vegetation (PNV). Potential replacement vegetation (PRV) is an abstract and hypothetical vegetation which is in balance with climatic and soil factors currently affecting a given habitat, with environmental factors influencing the habitat from outside such as air pollution, and with an abstract anthropogenic influence (management) of given type, frequency and intensity. For every habitat, there is a series of possible PRV-types corresponding to the different anthropogenic influences, e.g. grazing, mowing, trampling or growing cereals. The PRV-concept is especially useful in large-scale mapping (scales > 1 : 25 000) of small areas where replacement vegetation is the focus of attention for managers and land-use planners, for example in nature reserves where the aim is conservation of replacement vegetation managed in a traditional way, or in restoration ecology where the concept may be used for defining restoration goals and evaluating the success of restoration efforts. At smaller scales, PRV-mapping may be useful for revealing the biogeographical patterns of larger areas which may be different from the corresponding PNV patterns, because replacement vegetation and natural vegetation may respond to environmental gradients at different scales. An example of medium-scale PRV-mapping through the coincidence of diagnostic species of vegetation types, based on species distribution grid data, is presented. In cultural landscapes, the advantage of using the PRV-concept instead of PNV is its direct relationship to the replacement vegetation. In the habitat mapping with respect to the replacement vegetation, the PRV concept yields more valuable results than the mapping of actual vegetation, as the latter is strongly affected by spatially variable anthropogenic influences which may be largely independent from climatic and soil factors.     

Climate and vegetation in China III water balance and distribution of vegetation

Distributions of 29 vegetation types in China as a function of climatic humidity or aridity were analysed using Thornthwaite's system, by employing meteorological records from 671 stations in China. The annual potential evapotranspiration and the humidity/aridity indices were calculated for every station, and distribution maps of water deficiency, water surplus and moisture index (Im) were constructed. The Im map showed that arid areas (Im<0) occupied about 56% of the country. The effect of the difference in soil water storage capacity on Thornthwaite's indices was examined, and Im values were found to differ little, although some differences were observed in actual annual evapotranspiration, water deficiency and water surplus values. Correlations between Im values and distributions of 29 vegetation types, identified from a vegetation map with a scale of 1/4000000, were investigated. The distributions of desert, steppe, woodland, deciduous forest and evergreen forest corresponded to Im values of below −40, −40–−20, −20-0, 0–60 and over 60, respectively. In addition, climatic factors delimiting the northern distribution of evergreen broadleaf forest were investigated, and it was clarified that the northern limit was restricted by combined hydrothermal conditions, and not by the low temperature in winter.     

基于可见光植被指数的面向对象湿地水生植被提取方法

利用ESP分割工具确定最佳分割尺度,通过多尺度分割算法创建最优分割影像,基于微型无人机影像数据生成可见光植被指数,从一系列可见光植被指数中选取一组最优植被指数,建立决策树规则,利用隶属度函数对研究区自动分类,生成水生植被分布图.结果表明: 监督分类法的总体精度为53.7%,面向对象分类法总体精度为91.7%,与基于像元的监督分类法相比,面向对象分类法显著改善了影像分类结果,并大大提高了水生植被提取精度,监督分类法的Kappa系数为0.4,而面向对象分类法的Kappa系数为0.9.这表明利用微型无人机数据生成的可见光植被指数结合面向对象分类方法提取水生植被在该研究区是可行的,并能够应用到其他类似区域.     

Invasion models of vegetation dynamics

Since communities change as a result of their successful invasion by new species it seems logical to attempt to predict future vegetation change by focussing on the invasion process. Several such invasion models are reviewed, and one particular model, based on dynamic game theory is developed further. This model can be used as an alternative to linear (e.g. Markov chain) models for the prediction of vegetation dynamics, and also to compare invasive abilities of species and resistance to invasion of communities. The main advantage of the model lies in the fact that it operates at a sufficiently high level of integration to allow for model calibration (in spite of the large number of underlying processes), and yet has an obvious population biological interpretation (in terms of the success of invading populations). The model can be calibrated using either time course data or experimental data, and it may be helpful in understanding what determines the fate of an invading population. It is used here to analyze two published vegetation dynamics data sets.This work was financially supported by operating grant A8115 from the Natural Sciences and Engineering Research Council of Canada.