基于混合像元分解的 MODIS 绿潮覆盖面积精细化提取方法研究

在绿潮遥感业务化监测中,250 m 分辨率的 MODIS 卫星数据是主要数据源,归一化差值植被指数（ND-VI）是绿潮卫星遥感信息提取的主要方法。研究发现,由于 MODIS 空间分辨率较低,存在大量的混合像元,导致提取的绿潮覆盖面积明显偏大。针对该问题,本文在 MODIS 绿潮 NDVI 计算的基础上,首先对大于 NDVI 阈值的像元进行混合像元分解,得到 MODIS NDVI 混合像元分解后的绿潮面积,然后以准同步的30 m 分辨率 HJ-1 CCD 影像提取的绿潮覆盖面积为真值,建立了 MODIS NDVI 混合像元分解得到的绿潮面积与 HJ-1提取的绿潮面积之间的关系模型,以实现绿潮面积的精细化提取。与传统的 NDVI 阈值法和混合像元分解法相比,该方法提取的绿潮覆盖面积更接近于“真值”,面积约为“真值”的96％,而传统的 NDVI 阈值法和混合像元分解方法提取的面积分别为“真值”的2．96倍和45％。另外,与传统的 NDVI 阈值法相比,新方法对 NDVI 阈值变化不敏感,在相同的 NDVI 阈值变化区间内,前者提取的绿潮覆盖面积变化了41％,而新方法的变化仅为11％。本文的工作在很大程度上解决了 MODIS 空间分辨率低导致的绿潮监测结果不准确的问题,为精细化的绿潮卫星遥感业务监测提供了参考。     

基于MODIS数据黄海绿潮覆盖面积精细化提取

针对MODIS数据绿潮提取存在大量混合像元导致绿潮覆盖面积偏大这一问题,基于3 m分辨率的机载SAR高分影像,结合归一化植被指数(NDVI),对MODIS研究区中大于NDVI阈值的像元进行混合像元分解得到绿潮的"覆盖面积";获取的准同步3 m机载SAR提取的绿潮面积为"真实值",然后建立二者之间的关系模型,并选取不同的样本区域对该模型进行了验证。实验结果表明:NDVI等传统算法所提取的绿潮覆盖面积约为"真实值"的2.68倍;基于混合像元分解的方法所提取的绿潮面积较"真实值"偏小,约为"真值"的0.56倍;与传统的NDVI等多波段比值法相比,该精细化模型方法提取的绿潮覆盖面积更接近于"真实值",与"真实值"误差仅为6.7%。     

浒苔遥感监测研究进展

浒苔大规模集聚形成的绿潮灾害是海洋生态系统主要生态环境问题之一,基于卫星遥感影像监测浒苔及其扩展动态已成为一种及时有效的手段。对国内外浒苔遥感监测方面文献进行归纳整理,认为光学遥感数据、多波段比值法是最常用的遥感数据和监测方法。对遥感监测浒苔机理进行了阐述,并对分类方法进行评价认为监督分类法解译精度不高。目前单波段阈值法和多波段比值法应用广泛,但在监测漂浮浒苔和混合象元解译存在不足。辐射传输模型法能有效提高信息解译的精度,但还处于起步阶段。遥感监测浒苔灾害的未来发展需要提高影像空间分辨率,深入研究监测方法,进行多种平台和多源遥感数据相结合,并由定性走向定量,从而建立健全遥感监测预警系统。     

测量的区域土地覆盖格局研基于多尺度遥感究

 利用1km、4km和8km 3种空间分辨率的NOAA/AVHRR数字影像,对中国NECT样带西部地区进行了土地覆盖分类及其景观特征的比较研究。重点比较了几种空间分辨率遥感数据分类结果边界的一致性和空间差异,以及影像所记录的景观格局的差异。为进一步在不同尺度上研究景观变化过程以及尺度转换研究奠定了基础。研究表明：3种空间分辨率的遥感影像所反映的区域土地覆盖的宏观空间格局是一致的,但类型的边界、每一类型斑块的形状和数量均产生较大的差异;经过对反映景观空间结构的4种指标（分维数、破碎度、多样性、优势度）的比较显示出随着遥感影像空间分辨率的变化,影像所反映的景观结构发生了较大的变化。其中,各覆盖类型的分维数表现出最大差异,表征着空间分辨率的变化对斑块复杂程度的影响最大。     

利用静止轨道卫星开展绿潮业务化监测的可行性研究

针对光学遥感手段无法探测云雾覆盖下绿潮分布区域这一现状问题,本文着力于开展绿潮时空分布的监测研究,获取了近三年五月中旬到八月中旬的GOCI(geostationary ocean color imager)数据,基于该数据进行绿潮范围提取、绿潮漂移路径分析、精细化云区域提取和统计云覆盖情况,分析云量覆盖对利用静止轨道卫星监测绿潮的影响程度,进而从探测能力和动态能力两方面论证利用静止轨道卫星开展绿潮业务化监测的可行性。     

测量的区域土地覆盖格局研基于多尺度遥感究

利用1km、4km和8km 3种空间分辨率的NOAA/AVHRR数字影像,对中国NECT样带西部地区进行了土地覆盖分类及其景观特征的比较研究。重点比较了几种空间分辨率遥感数据分类结果边界的一致性和空间差异,以及影像所记录的景观格局的差异。为进一步在不同尺度上研究景观变化过程以及尺度转换研究奠定了基础。研究表明：3种空间分辨率的遥感影像所反映的区域土地覆盖的宏观空间格局是一致的,但类型的边界、每一类型斑块的形状和数量均产生较大的差异;经过对反映景观空间结构的4种指标（分维数、破碎度、多样性、优势度）的比较显示出随着遥感影像空间分辨率的变化,影像所反映的景观结构发生了较大的变化。其中,各覆盖类型的分维数表现出最大差异,表征着空间分辨率的变化对斑块复杂程度的影响最大。     

基于线性混合模型和NDVI阈值法的MODIS影像黄海浒苔监测

水体富营养化等原因,导致浒苔(Enteromorpha prolifera)灾害自2007年在中国黄海海域频发,成为黄海最严重的生态灾害。卫星遥感具有大范围监测、瞬时优势,成为浒苔灾害最重要的监测手段之一。中分辨率成像光谱仪(moderate-resolution imaging spectroradiometer,MODIS)影像因其幅宽大、时间分辨率高、免费分发,成为浒苔业务化监测的主要数据源。由于空间分辨率(250 m)较低,混合像元的存在导致传统阈值法获取的浒苔面积误差较大。本文结合线性混合模型(linear mixing model,LMM)和归一化差值植被指数(normalized different vegetation index,NDVI)阈值法对250 m空间分辨率的MODIS影像进行浒苔面积提取。选择1个大区及其内部3个小区,以准同步5.8 m空间分辨率资源三号(ZY-3)卫星影像提取浒苔结果为准进行精度评价。发现NDVI阈值为0.04提取的浒苔像元对线性混合模型分解结果掩膜所得浒苔面积误差最小,大区及其3个小区的误差分别是7.86%、14.59%、-7.65%、-0.15%。应用本文提供方法可有效排除浒苔混合像元和非浒苔像元对浒苔面积信息提取的干扰,与阈值法相比大幅提高了反演精度,且在不同区域精度较稳定,可为浒苔生态灾害的处置决策和评估提供支撑。     

基于中分辨率TM数据的湿地水生植被提取

利用湿地水生植被生长旺盛、光谱反射较强、光谱信息比较丰富的8月份中分辨率Landsat TM和ETM+多光谱遥感影像,采用面向对象的分类方法,进行野鸭湖湿地水生植被的提取。研究表明:在提取过程中,通过对原始影像进行主成分变换和穗帽变换,将主要信息与噪声分离,不仅减小了数据冗余和波段间的相关性,而且增大了影像上湿地水生植被与其他地物类型光谱和空间信息的差异性,并结合野外水生植被光谱特征分析,选择归一化植被指数NDVI与归一化水体指数NDWI辅助分类,构建特征波段或波段组合,然后,确定适当的隶属度函数和阈值范围,构建分类决策树,完成湿地水生植被的自动分类,提高了影像分割与面向对象分类的精度,取得了较为理想的湿地水生植被提取结果。2002年和2008年两景影像的总体分类精度分别达到86.5%和85.44%,表明中分辨率TM影像可以满足湿地水生植被提取的需要,又因为其具有较高的波谱分辨率、极为丰富的信息量、相对较低的价格、长时间序列,可以作为近20a湿地水生植被提取和动态变化监测的主要数据源。     

基于多时相中巴资源卫星影像的冬小麦分类精度

中巴资源卫星2号星（CBERS-02）具有较高的空间分辨率和较丰富的光谱信息,对植被有较强的探测能力.利用2006—2007年北京地区冬小麦生育期早期的5景CBERS-02卫星影像,计算了各时相和不同时相组合的主要地物类型及冬小麦的光谱可分性距离,进行了监督分类,同时,结合高分辨率航空和卫星遥感影像,构建了训练样本和验证样本,对利用CBERS-02卫星提取的生育早期的冬小麦进行了时相分析和精度评价,并与同期TM影像提取结果进行对比.结果表明：时相是影响冬小麦分类的主要因素,不同光学传感器的遥感影像也会影响分类精度;多时相组合有利于提高冬小麦的提取精度,与单时相冬小麦提取的最高精度相比,最佳时相组合的制图精度提高了20.0%、用户精度提高了7.83%;与TM数据相比, CBERS-02卫星影像的冬小麦分类精度略低.     

基于星载通道光谱指数与小麦冠层叶片氮素营养指标的定量关系

利用空间遥感信息大面积监测小麦冠层氮素营养状况和生产力指标具有重要意义和应用前景.本研究基于不同施氮水平下小麦冠层反射光谱信息,利用响应函数模拟基于不同卫星通道构建的光谱指数(包括单波段、比值光谱指数和归一化光谱指数),分析基于星载通道的光谱指数与小麦冠层叶片氮素营养指标的定量关系,确定监测小麦冠层叶片氮素营养的较好卫星传感器和光谱波段,建立小麦冠层氮素营养指标监测方程.结果表明:利用NDVI(MSS7,MSS5)、NDVI(RBV3,RBV2)、TM4 、CH2、MODIS1和MODIS2遥感数据可以预估小麦叶片氮含量(LNC),其决定系数(R2)在0.60以上;应用NDVI(PB4,PB2)、NDVI(CH2,CH1)、NDVI(MSS7,MSS5)、RVI(MSS7,MSS5)、MODIS1和MODIS2可以预测小麦叶片氮积累量(LNA),其R2大于0.86.比较而言,NDVI(MSS7,MSS5)和NDVI(PB4,PB2)分别为预测小麦LNC和LNA的适宜星载通道光谱参数.     

Predicting mountain plant richness and rarity from space using satellite-derived vegetation indices

Can species richness and rarity be predicted from space? If satellite‐derived vegetation indices can provide us with accurate predictions of richness and rarity in an area, they can serve as an excellent tool in diversity and conservation research, especially in inaccessible areas. The increasing availability of high‐resolution satellite images is enabling us to study this question more carefully. We sampled plant richness and rarity in 34 quadrats (1000 m²) along an elevation gradient between 300 and 2200 m focusing on Mount Hermon as a case study. We then used 10 Landsat, Aster, and QuickBird satellite images ranging over several seasons, going up to very high resolutions, to examine the relationship between plant richness, rarity, and vegetation indices calculated from the images. We used the normalized difference vegetation index (NDVI), one of the most commonly used vegetation indexes, which is strongly correlated to primary production both globally and locally (in more seasonal and in drier and/or colder environments that have wide ranges of NDVI values). All images showed a positive significant correlation between NDVI and both plant species richness and percentage tree cover (with R² as high as 0.87 between NDVI and total plant richness and 0.89 for annual plant richness). The high resolution images enabled us to examine spatial heterogeneity in NDVI within our quadrats. Plant richness was significantly correlated with the standard deviation of NDVI values (but not with their coefficient of variation) within quadrats and between images. Contrary to richness, relative range size rarity was negatively correlated with NDVI in all images, this result being significant in most cases. Thus, given that they are validated by fieldwork, satellite‐derived indices can shed light on richness and even rarity patterns in mountains, many of which are important biodiversity centres.     

基于多光谱影像的森林树种识别及其空间尺度响应

当前,不同空间分辨率卫星影像对森林类型识别结果中普遍存在的尺度效应,而且纹理参量对不同尺度下树种识别精度的影响仍缺乏广泛认知.本研究以中国东北旺业甸林场为研究区,采用观测时相同步、地理坐标匹配的GF-1 PMS、GF-2 PMS、GF-1 WFV,以及Landsat-8 OLI卫星传感器数据组成空间尺度观测序列(1、2、4、8、16、30 m),并结合支持向量机(SVM)模型,探讨了区域内5种优势树种遥感识别结果的尺度变化规律及其纹理特征参数的影响,同时检验了基于尺度上推转换影像的树种识别结果差异.结果表明: 影像空间分辨率对区域树种识别结果具有显著影响,其中,研究区森林树种识别的最佳影像分辨率为4 m,当分辨率降低至30 m时,树种识别结果最差.在1～8 m影像分辨率范围内,增加纹理信息能够显著提高不同优势树种的识别精度,使总分类精度提升了2.0%～3.6%,但纹理信息对16～30 m影像的识别结果没有显著影响.与真实尺度卫星影像相比,基于升尺度转换影像的树种识别结果及其尺度响应特征存在显著差异,表明在面向多个空间尺度的遥感观测和应用研究中,需要采用真实分辨率影像以确保结果的准确性.     

Pheno-climatic profiles of vegetation based on multitemporal analysis of satellite data

Satellite Remote Sensing offers numerous advantages: study of large areas in a short time, study of areas with not easy accessibility, synoptic observation of territory, multitemporal observations of the same area, monitoring land modifications and change detection studies. The effectiveness of using satellite images for studying and mapping vegetation and land use has been stressed since the early 1980s. The photosynthetically active vegetation presents a very characteristic spectral response. In fact, leaves absorb red radiation (RED) in order to do photosynthetic process and reflect almost completely near infrared (NIR) wavelengths. The most diffused index for quantifying photosynthetically active biomass is the NDVI (Normalized Difference Vegetation Index): NDVI = (NIR-RED)/(NIR+RED). The NDVI is calculated, for each pixel of the images analysed, through an appropriate software. Low values of NDVI correspond to scarcely vegetated areas, while high values indicate densely vegetated ones. In order to distinguish among vegetation typologies we need some images of the same territory, well distributed during the year, showing seasonal variations of vegetation photosynthetic activity. Then it will be e.g. very easy distinguish between evergreen species (with NDVI almost steady during the year) and deciduous ones. Several types of sensors aboard some satellites allow different investigations to be done. AVHRR sensor on NOAA and TM sensor on Landsat are among the best known sensors available. They have different characteristics as for spectral resolution (number of spectral bands), spatial resolution (size of each elementary cell) and temporal resolution (the period of the satellite passes on the same territory). Vegetation phenology (including biomass and photosynthetic activity) heavily depends on climatic factors. The most important are: solar radiance, with an annual cycle and maximum at summer solstice; air temperature, (depending on solar radiance) with an annual cycle and maximum more than one month later; water availability, which is strongly dependent on rainfalls; in the Mediterranean area they can have an annual cycle (maximum during winter) or a six-monthly one (maxima near the equinoxes). Having a set of multitemporal satellite data (e.g. 12 monthly NOAA-AVHRR images) we can use a mathematical model able to discriminate annual and six-monthly cycles. Through Fourier analysis, the mathematical model calculate, for each pixel of the image, the parameters of the annual NDVI profile and create a synthetic image (pheno-climatic map), in which the values of the three RGB components (Red, Green, Blue ) are proportional to the integral of the NDVI profile for the following three periods: B=Nov-Feb G=Mar-Jun R=Jul-Oct. A similarly analysis is possible with Landsat satellite data, which have a higher spatial resolution, given that some shrewdness are taken. In fact, it is necessary to select satellite images according to the presence of cloud cover, which is--over the Italian peninsula--quite common during the March-April and October-November intervals. The purpose of carrying out pheno-climatic maps can be accomplished using 6 Landsat-TM images well-distributed during a year, every two months, even if the images have been taken during different years.     

Using a sub-pixel mapping model to improve the accuracy of landscape pattern indices

The assessment of landscape spatial patterns is a key issue in landscape management. Landscape pattern indices (LPIs) are tools appropriate for analyzing landscape spatial patterns. LPIs are often derived from raster land cover maps that are extracted from remotely sensed data through hard classification. However, pixel-based hard classification methods suffer from the mixed pixel problem (in which pixels contain more than one land cover class), making for inaccurate classification maps and LPIs. In addition, LPIs generated by hard classification methods are characterized by grain sizes (the sampling unit sizes) that limit the derived landscape pattern to a certain scale. Sub-pixel mapping (SPM) models can enable fine-scale estimation of the spatial patterns of land cover classes without requiring additional data; hence, this is an appropriate downscaling method for land cover mapping. The fraction images generated by soft classification estimate the area proportion of each land cover class within each pixel, and using these images as input enables SPM models to alleviate the mixed pixel problem. At the same time, by transforming fraction images into a finer-scaled hard classification map, SPM models can minimize the influence of grain size on LPIs calculation. In this research, simulated landscape thematic patterns that can provide different landscape spatial patterns, eight commonly used LPIs and a SPM model that maximizes the spatial dependence between neighbouring sub-pixels were applied to assess the efficiency of deriving LPIs from sub-pixel model maps. Results showed that the SPM model can more precisely characterize landscape patterns than hard classification methods can. Landscape fragmentation, class abundance, the uncertainty in SPM, and the spatial resolution of the remotely sensed data influenced LPIs derived from sub-pixel maps. The largest patch index, landscape division, and patch cohesion derived from remotely sensed data with different spatial resolutions through the SPM model were suitable for inter-comparison, whereas the patch density, mean patch area, edge density, landscape shape index, and area-weighted mean shape index derived from the sub-pixel maps were sensitive to the spatial resolution of the remotely sensed data.     

Modeling gross primary productivity of alpine meadow in the northern Tibet Plateau by using MODIS images and climate data

The Vegetation Photosynthesis Model (VPM) was used to simulate the gross primary productivities (GPP) of the alpine meadow ecosystem in the northern Tibet Plateau at three different spatial resolutions of 0.5 km, 1.5 km and 2.5 km, respectively. The linear relationships between enhanced vegetation indices (EVI) and GPP, with higher correlative coefficients, were better than those between normalized difference vegetation indices (NDVI) and GPP at the three resolutions. VPM could well simulate the seasonal changes and inter-annual variations of GPP, with similar trends at the three resolutions. There were significant differences (P < 0.0001) among the three modeled GPP with the three resolutions. Therefore, the modeled GPP at high resolution could not be directly extrapolated to low resolution, and vice versa. The contribution levels of different model parameters, including photosynthetically active radiation (PAR), air temperature (Ta), NDVI, EVI and land surface water indices (LSWI), to modeled GPP could vary with spatial resolution based on multiple stepwise linear regression analysis. This indicated that it was important to choose parameters properly and consider their effects on modeled GPP.     

Modeling gross primary productivity of alpine meadow in the northern Tibet Plateau by using MODIS images and climate data

The Vegetation Photosynthesis Model (VPM) was used to simulate the gross primary productivities (GPP) of the alpine meadow ecosystem in the northern Tibet Plateau at three different spatial resolutions of 0.5 km, 1.5 km and 2.5 km, respectively. The linear relationships between enhanced vegetation indices (EVI) and GPP, with higher correlative coefficients, were better than those between normalized difference vegetation indices (NDVI) and GPP at the three resolutions. VPM could well simulate the seasonal changes and inter-annual variations of GPP, with similar trends at the three resolutions. There were significant differences (P < 0.0001) among the three modeled GPP with the three resolutions. Therefore, the modeled GPP at high resolution could not be directly extrapolated to low resolution, and vice versa. The contribution levels of different model parameters, including photosynthetically active radiation (PAR), air temperature (Ta), NDVI, EVI and land surface water indices (LSWI), to modeled GPP could vary with spatial resolution based on multiple stepwise linear regression analysis. This indicated that it was important to choose parameters properly and consider their effects on modeled GPP.     

The remote sensing approach in broad‐scale phenological studies

Abstract. Satellite imagery provides a unique tool for monitoring seasonal dynamics of the Earth's vegetation on a global scale. The combination of the normalized difference vegetation index (NDVI) data derived from the Advanced Very High Resolution Radiometer (AVHRR) with a daily repeat cycle and 1 km spatial resolution makes weather satellites operated by the National Oceanic and Atmospheric Administration very well suited for deriving broad‐scale phenological metrics from satellite images. In this paper, similarities and differences between remotely sensed phenological studies and traditional symphenological studies conducted by ground‐based observations are summarized. Finally, major shortcomings in deriving phenological metrics from NDVI time series are discussed.