Artificial neural network ensemble-based land-cover classifiers using MODIS data

Terra and Aqua, two satellites launched by the NASA-centered International Earth Observing System project, house MODIS (moderate resolution imaging spectroradiometer) sensors. Moderate-resolution remote sensing allows the quantifying of land-surface type and extent, which can be used to monitor changes in land cover and land use for extended periods of time. In this article, we propose land-surface classification by applying an ensemble technique based on fault masking among individual classifiers in N-version programming. An N-version programming ensemble of artificial neural networks is created, in which the majority vote result is used to predict land-surface cover from MODIS data. It is shown by experiment that an N-version programming ensemble of neural networks greatly improves the classification error rate of land-cover type.     

The use of temporal metrics for land cover change detection at coarse spatial scales

Successful land cover change analysis requires selection of an appropriate set of variables for measuring and characterizing change. Coarse spatial resolution satellite sensors offer the advantage of frequent coverage of large areas and this facilitates the monitoring of surface processes. Fine spatial resolution satellite sensors provide reliable land cover information on a local basis. This work examines the ability of several temporal change metrics to detect land cover change in sub-Saharan Africa using remote sensing data collected at a coarse spatial resolution over 16 test sites for which fine spatial resolution data are available. We model change in the fine-resolution data as a function of the coarse spatial resolution metrics without regard to the type of change. Results indicate that coarse spatial resolution temporal metrics (i) relate in a statistically significant way to aggregate changes in land cover, (ii) relate more strongly to fine spatial resolution change metrics when including a measure of surface temperature instead of a vegetation index alone, and (iii) are most effective as land cover change indicators when various metrics are combined in multivariate models.     

An automatic method for synthetically preprocessing multi‐source NOAA‐AVHRR series and FY‐1D 1A.5 data

Polar orbiting meteorological satellites have been used successfully in many fields in which applications, such as dynamic monitoring, need multi‐temporal remote sensing data imaged in the same region but at different times and with different sensors. The basis of these quantitative applications is automatic and synthetic preprocessing of different satellite data. In order to ensure the direct comparability of multi‐source satellite National Oceanic & Atmospheric Administration‐Advanced Very High Resolution Radiometer (NOAA‐AVHRR) series and FY‐1D 1A.5 data, based on its imaging mechanisms, the paper discusses radiant and geometric preprocessing methods used for visual and near‐infrared and thermal infrared channels. This is performed by analysing the technical parameters of the satellites and the sensors, theoretical numerical simulation and image data comparison. For synthetic radiance preprocessing, radiance calibration, spectral calibration for different sensors and calibration of the Sun elevation angles or the satellite zenith angles at the imaging time are included. Based on normalization of spatial resolution in scan‐line directions, rough geometric correction and precise registration are sequentially processed for geometric preprocessing with a higher precision. Combining the detailed steps introduced in the paper, automatic preprocessing of polar orbiting meteorological satellites can be realized.     

成像光谱技术在土地利用动态遥感监测中的应用研究

随着数字化调查技术的发展，国土资源管理对土地利用动态遥感监测提出了更高的要求，目前主要采用的多光谱数据由于受光谱分辨率限制以及“同谱异物，同物异谱”现象的影响，难以满足管理需要。成像光谱数据具有较高光谱分辨率。在类别细分方面具有一定的优势，在当前土地利用动态遥感监测中具有一定的应用潜质。该文针对成像光谱数据特点，探索了与成像光谱数据相适应的土地利用动态遥感监测方法，提出了异常光谱检测法，该方法在试验区应用中取得了良好效果。     

Object-based land cover change detection for cross-sensor images

Accurate and timely land cover change detection at regional and global scales is necessary for both natural resource management and global environmental change studies. Satellite remote sensing has been widely used in land cover change detection over the past three decades. The variety of satellites which have been launched for Earth Observation (EO) and the large volume of remotely sensed data archives acquired by different sensors provide a unique opportunity for land cover change detection. This article introduces an object-based land cover change detection approach for cross-sensor images. First, two images acquired by different sensors were stacked together and principal component analysis (PCA) was applied to the stacked data. Second, based on the Eigen values of the PCA transformation, six principal bands were selected for further image segmentation. Finally, a land cover change detection classification scheme was designed based on the land cover change patterns in the study area. An image–object classification was implemented to generate a land cover change map. The experiment was carried out using images acquired by Landsat 5 TM and IRS-P6 LISS3 over Daqing, China. The overall accuracy and kappa coefficient of the change map were 83.42% and 0.82, respectively. The results indicate that this is a promising approach to produce land cover change maps using cross-sensor images.     

改进SeMask主干网络的高分辨率遥感影像变化检测模型

随着城市化进程的加速和人口不断增加, 土地资源的利用和管理变得愈发重要. 高分辨率遥感影像技术的发展为土地覆盖类别变化检测提供了新的途径. 目前, 多数遥感影像变化检测任务主要针对显著建筑物的变化检测, 缺少对土地覆盖类别变化检测任务的研究, 本研究基于公开数据集, 对更多土地覆盖类别变化情况进行标注. 在原语义分割主干网络的基础上结合孪生网络结构, 提出适用于土地覆盖类别变化检测任务的检测模型, 该模型在网络的特征提取阶段加入变化引导模块, 以辅助网络关注两时相影像中的变化信息, 并在网络不同阶段加入通道信息交互模块, 以增强不同特征图的信息融合. 同时, 在特征提取阶段最后一层加入特征对齐模块, 以缓解下采样过程导致的特征偏移. 在土地覆盖类别变化检测数据集上的实验结果表明, 本文提出的方法可以有效提取影像中的变化信息, 并提高分割精度.     

Dryland observation at local and regional scale — Comparison of Landsat TM/ETM+ and NOAA AVHRR time series

During the last few decades, many regions have experienced major land use transformations, often driven by human activities. Assessing and evaluating these changes requires consistent data over time at appropriate scales as provided by remote sensing imagery. Given the availability of small and large-scale observation systems that provide the required long-term records, it is important to understand the specific characteristics associated with both observation scales. The aim of this study was to evaluate the potentials and limits of remote sensing time series for change analysis of drylands. We focussed on the assessment and monitoring of land change processes using two scales of remote sensing data. Special interest was given to the influence of the spatial and temporal resolution of different sensors on the derivation of enhanced vegetation related variables, such as trends in time and the shift of phenological cycles. Time series of Landsat TM/ETM+ and NOAA AVHRR covering the overlapping time period from 1990 to 2000 were compared for a study area in the Mediterranean. The test site is located in Central Macedonia (Greece) and represents a typical heterogeneous Mediterranean landscape. It is undergoing extensification and intensification processes such as long-term, gradual processes driven by changing rangeland management and the extension of irrigated arable land. Time series analysis of NOAA AVHRRR and Landsat TM/ETM+ data showed that both sensors are able to detect this kind of land cover change in complementary ways. Thereby, the high temporal resolution of NOAA AVHRR data can partially compensate for the coarse spatial resolution because it allows enhanced time series methods like frequency analysis that provide complementary information. In contrast, the analysis of Landsat data was able to reveal changes at a fine spatial scale, which are associated with shifts in land management practice.     

COSMO雷达数据在多云多雾地区土地变化监测研究

近年来新型成像雷达遥感(极化、干涉)及数据处理技术的发展,SAR遥感影像上获得的地表信息越来越多,如何利用雷达信息探测土地变化成为研究的新课题。但是雷达影像不同于光学影像,目前雷达数据解译仍存在着一些困难。本文主要针对多云多雾地区雷达数据土地变化监测,以四川成都地区COSMO数据为例,利用雷达相干影像,后向散射强度,强度比值影像,提出一种新的雷达处理手段,减少了雷达数据土地变化监测的工作量,提高工作效率。     

Passive optical remote sensing of cyanobacteria and other intense phytoplankton blooms in coastal and inland waters

Increased frequency and extent of potentially harmful blooms in coastal and inland waters world-wide require the development of methods for operative and reliable monitoring of the blooms over vast coastal areas and a large number of lakes. Remote sensing could provide the tool. An overview of the literature in this field suggests that operative monitoring of the extent of some types of blooms (i.e. cyanobacteria) is relatively straightforward. Operative monitoring of inland waters is currently limited to larger lakes or using airborne and hand-held remote sensing instruments as there are no satellite sensors with sufficient spatial resolution to provide daily coverage. Extremely high spatial and vertical variability in biomass during blooms of some phytoplankton species and the strong effects of this on the remote sensing signal suggest that water sampling techniques and strategies have to be redesigned for highly stratified bloom conditions, especially if the samples are collected for algorithm development and validation of remote sensing data. Comparing spectral signatures of different bloom-forming species with the spectral resolution available in most satellites and taking into account variability in optical properties of different water bodies suggests that developing global algorithms for recognizing and quantitative mapping of (harmful) algal blooms is questionable. On the other hand some authors cited in the present paper have found particular cases where satellites with coarse spectral and spatial resolution can be used to recognize phytoplankton blooms even at species level. Thus, the algorithms and methods to be used depend on the optical complexity of the water to which they will be applied. The aim of this paper is to summarize different methods and algorithms available in an attempt to assist in selecting the most appropriate method for a particular site and problem under investigation.     

Haze removal for new generation optical sensors

Haze and cloud contamination is a common problem in optical remote-sensing imagery, as it can lead to the inaccurate estimation of physical properties of the surface derived from remote sensing and reduced accuracy of land cover classification and change detection. Haze optimized transform (HOT) is a methodology applicable to radiometric compensation of additive haze effects in visible bands that exhibits a spatially complex distribution over an image. The generic approach of HOT allows for the use of older satellite imaging sensors that include at least two visible bands (e.g. Landsat Thematic Mapper (TM) and Landsat Multispectral Scanner (MSS) sensors). This study proposes modifications to extend HOT applicability to new sensors. The improvements and extended functionality adapt the method to the higher radiometric resolution specifications of newer generation sensors and use percentile-based minimum in the correction procedure to avoid causing fake minimum. Alternative filters are also evaluated to smooth raw HOT output and the cloud mask is generated as an additional output. A Landsat 8 scene of Los Angeles is used to demonstrate the improved methodology. The methodology is applicable to sensors such as QuickBird, Worldview 2/3. More than 20 additional scenes were used to evaluate the effectiveness of the methodology.     

Spatial and temporal probabilities of obtaining cloud‐free Landsat images over the Brazilian tropical savanna

Remotely sensed data are the best and perhaps the only possible way for monitoring large‐scale, human‐induced land occupation and biosphere‐atmosphere processes in regions such as the Brazilian tropical savanna (Cerrado). Landsat imagery has been intensively employed for these studies because of their long‐term data coverage (>30 years), suitable spatial and temporal resolutions, and ability to discriminate different land‐use and land‐cover classes. However, cloud cover is the most obvious constraint for obtaining optical remote sensing data in tropical regions, and cloud cover analysis of remotely sensed data is a requisite step needed for any optical remote sensing studies. This study addresses the extent to which cloudiness can restrict the monitoring of the Brazilian Cerrado from Landsat‐like sensors. Percent cloud cover from more than 35 500 Landsat quick‐looks were estimated by the K‐means unsupervised classification technique. The data were examined by month, season, and El Niño Southern Oscillation event. Monthly observations of any part of the biome are highly unlikely during the wet season (October–March), but very possible during the dry season, especially in July and August. Research involving seasonality is feasible in some parts of the Cerrado at the temporal satellite sampling frequency of Landsat sensors. There are several limitations at the northern limit of the Cerrado, especially in the transitional area with the Amazon. During the 1997 El Niño event, the cloudiness over the Cerrado decreased to a measurable but small degree (5% less, on average). These results set the framework and limitations of future studies of land use/land cover and ecological dynamics using Landsat‐like satellite sensors.     

Assessing tundra-taiga boundary with multi-sensor satellite data

Monitoring the dynamics of the circumpolar boreal forest (taiga) and Arctic tundra boundary is important for understanding the causes and consequences of changes observed in these areas. This ecotone, the world's largest, stretches for over 13,400 km and marks the transition between the northern limits of forests and the southern margin of the tundra. Because of the inaccessibility and large extent of this zone, remote sensing data can play an important role for mapping the characteristics and monitoring the dynamics. Basic understanding of the capabilities of existing space borne instruments for these purposes is required. In this study we examined the use of several remote sensing techniques for characterizing the existing tundra-taiga ecotone. These include Landsat-7, MISR, MODIS and RADARSAT data. Historical cover maps, recent forest stand measurements and high-resolution IKONOS images were used for local ground truth.It was found that a tundra-taiga transitional area can be characterized using multi-spectral Landsat ETM+ summer images, multi-angle MISR red band reflectance images, RADARSAT images with larger incidence angle, or multi-temporal and multi-spectral MODIS data. Because of different resolutions and spectral regions covered, the transition zone maps derived from different data types were not identical, but the general patterns were consistent.     

基于MODIS温度和植被指数产品的山东省土地覆盖变化研究

地表温度(LST)与归一化植被指数(NDVI)构成的NDVI-Ts特征空间具有丰富的地学和生态学内涵。MODIS数据因其优越的时间分辨率、波谱分辨率,已被广泛地运用于各个领域。在本研究中,运用遥感技术和GIS技术相结合的手段,利用NASA提供的MODIS温度产品和NDVI产品,以山东省土地利用图、山东省TM遥感影像图和基于3S技术的山东省森林资源调查项目的外业调查数据为参考和评价标准,以NDVI-Ts时间序列为指标,在进行土地覆盖分类的基础上,分析比较了山东省土地覆盖从2000年到2006年的变化情况。研究结果表明,利用MODIS产品将NDVI-Ts时间序列作为分类特征,在较大尺度范围的土地覆盖分类中具有较高的分类精度,有利于对土地覆盖变化进行动态监测。     

高分辨率影像配准误差对土地覆盖分类和变化检测的影响

近年来高分辨率影像技术发展迅速,土地专题信息的提取对高分辨率数据处理的质量提出了更高的要求,针对配准误差对影像处理和应用的影响研究,有助于专题信息提取过程中遥感数据处理质量控制指标的确定。选取北京市通州区不同时相的IKONOS影像作为实验数据进行模拟研究,在实验研究中通过产生具有不同配准误差的图像,从影像融合、土地覆盖分类和变化检测等角度,分析不同的配准误差对遥感应用的影响。结果表明:随着配准误差的增大,融合图像的可分辨性降低,配准误差增加到3个像元时,土地覆盖分类精度降低2~3%,土地覆盖变化检测中增加了5%的伪变化信息,虚检率增大。     

全球民用陆地成像卫星概览

陆地成像卫星发展迅速, 目前已有17 个国家的中/ 高分辨率的卫星在轨道上运行, 2010 年将达到20 个国家。美国摄影测量与遥感协会( The American Society for Photog rammet ry and RemoteSensing , ASPRS) 以网上公开资源为信息源, 收集整理了全球运行中和计划发射的中/ 高分辨率民用陆地成像卫星资料, 制成一目了然的图表, 为ASPRS 成员和广大遥感使用者提供言简意赅的《陆地成像卫星指南》。主要对该指南做了编译, 并选取有代表性的5 幅图表, 为国内相关领域的研究者提供参考。     

单极化合成孔径雷达影像在土地利用分类中的潜力分析

从我国土地利用调查应用出发,为了解决我国多云多雨地区土地利用分类及遥感动态监测问题,以面向对象影像分割、分类软件--Definiens Developer作为处理平台,对中分辨率星载合成孔径雷达（SAR）（以ENVISAT ASAR和Radarsat-1为例）、高分辨率星载SAR（以TerraSAR-X为例）进行分类处理,分析了单极化星载中、高分辨率星载SAR在土地利用分类中的能力,并对该模式星载SAR在土地利用分类中的影像特征和可解析程度进行了小结。     

深度学习对不同分辨率影像冬小麦识别的适用性研究

定量分析遥感影像尺度与分类精度之间的关系是进行土地覆盖分类的基础。深度学习具有从底层到高层特征非监督学习的能力,解决了传统分类模型中需要人工选择特征的问题。这种新型的分类方法的分类精度是否受到不同分辨率尺度影响,有待研究。利用深度卷积神经网络（Deep Convolutional Neural Network, DCNN）——金字塔场景分析网络（Pyramid Scene Parsing Network, PSPNet）进行4种分辨率（8、3.2、2和0.8 m）的米级、亚米级影像冬小麦分类。实验结果表明: PSPNet能够有效地进行大样本的学习训练,非监督提取出空间特征信息,实现“端—端”的冬小麦自动化分类。不同于传统分类器分类精度与分类尺度之间的关系,随着影像分辨率的逐步增高,地物表达特征越来越清晰,PSPNet识别的冬小麦精度会逐步增高,识别地块结果也越来越规整,不受分辨率尺度的影响。这对于选择甚高亚米级影像提高作物分类精度提供了实验基础。     

深度学习对不同分辨率影像冬小麦识别的适用性研究

定量分析遥感影像尺度与分类精度之间的关系是进行土地覆盖分类的基础。深度学习具有从底层到高层特征非监督学习的能力，解决了传统分类模型中需要人工选择特征的问题。这种新型的分类方法的分类精度是否受到不同分辨率尺度影响，有待研究。利用深度卷积神经网络（Deep Convolutional Neural Network, DCNN）——金字塔场景分析网络（Pyramid Scene Parsing Network, PSPNet）进行4种分辨率（8、3.2、2和0.8 m）的米级、亚米级影像冬小麦分类。实验结果表明: PSPNet能够有效地进行大样本的学习训练，非监督提取出空间特征信息，实现“端—端”的冬小麦自动化分类。不同于传统分类器分类精度与分类尺度之间的关系，随着影像分辨率的逐步增高，地物表达特征越来越清晰，PSPNet识别的冬小麦精度会逐步增高，识别地块结果也越来越规整，不受分辨率尺度的影响。这对于选择甚高亚米级影像提高作物分类精度提供了实验基础。     

Land cover changes around a major east African wildlife reserve: The Mara Ecosystem (Kenya)

The dynamics of savannah vegetation are still poorly understood. This study aims at analysing land cover changes over the past 20 years in the rangelands area of Narok District, Kenya. To analyse the impact of inter-annual climate variability and human activities on land cover modifications in the area, change detection techniques based on remote sensing data at different spatial and temporal resolutions were used. Coarse spatial, high temporal resolution NOAA (National Oceanic and Atmospheric Administration) data were analysed to investigate the role of inter-annual climate variations on the ecosystem. A combination of time contextual and spatial contextual change detection approaches was used on a set of three high spatial resolution Landsat images to map land cover modifications over the past 20 years. Both datasets are highly complementary in the detection of land cover dynamics. On the one hand, the coarse spatial resolution data detected areas that are sensitive to inter-annual climate fluctuations, but are not subjected to land cover conversion. On the other hand, the high spatial resolution data allowed the detection of land cover conversions or modifications between two consecutive dates that have a more permanent character and are independent of climate-induced fluctuations in surface attributes.