Classifiers vs. input variables—The drivers in image classification for land cover mapping

The study investigates the performance of image classifiers for landscape-scale land cover mapping and the relevance of ancillary data for the classification success in order to assess and to quantify the importance of these components in image classification. Specifically tested are the performance of maximum likelihood classification (MLC), artificial neural networks (ANN) and discriminant analysis (DA) based on Landsat7 ETM+ spectral data in combination with topographic measures and NDVI. ANN produced high accuracies of more than 75% also with limited input information, while MLC and DA produced comparable results only by incorporating ancillary data into the classification process. The superiority of ANN classification was less pronounced on the level of the single land cover classes. The use of ancillary data generally increased classification accuracy and showed a similar potential for increasing classification accuracy than the selection of the classifier. Therefore, a stronger focus on the development of appropriate and optimised sets of input variables is suggested. Also the definition and selection of land cover classes has shown to be crucial and not to be simply adaptable from existing land cover class schemes. A stronger research focus towards discriminating land cover classes by their typical spectral, topographic or seasonal properties is therefore suggested to advance image classification.     

遥感地表温度降尺度方法比较————性能对比及适应性评价

在归纳现有遥感地表温度降尺度方法的基础上, 选取3种代表性方法:Normalized Difference Vegetation Index (NDVI)、Pixel Block Intensity Modulation (PBIM)和Linear Spectral Mixture Model (LSMM)方法进行实验比较, 并建立了一种纹理相似性度量指标CO-RMSE (Co-Occurrence Root Mean Square Error)。结果表明:(1)NDVI方法受季节影响最严重, 不适于春、冬季, 其次为PBIM方法;(2)LSMM方法受分辨率限制最大, 低分辨率时丢失大量纹理信息, NDVI方法在较高分辨率时优于PBIM方法, 较低分辨率时则相反;(3)3种方法的适用区域分别为植被与裸土像元并存区域, 山区和反照率变化较大区域, 以及类别间温差较大区域;(4)NDVI方法操作最简单, LSMM方法最复杂。分析认为, 尺度因子是决定方法性能的关键, 应根据季节、分辨率、地表覆盖、应用目的和操作性等综合选择。     

Assessing land cover change resulting from large surface mining development

A remote sensing based land cover change assessment methodology is presented and applied to a case study of the Oil Sands Mining Development in Athabasca, Alta., Canada. The primary impact was assessed using an information extraction method applied to two LANDSAT scenes. The analysis based on derived land cover maps shows a decrease of natural vegetation in the study area (715,094 ha) for 2001 approximately −8.64% relative to 1992. Secondary assessment based on a key resources indicator (KRI), calculated using normalized difference vegetation index (NDVI measurements acquired by NOAA–AVHRR satellites), air temperature and global radiation was performed for a time period from 1990 to 2002. KRI trend analysis indicates a slightly decreasing trend in vegetation greenness in close proximity to the mining development. A good agreement between the time series of inter-annual variations in NDVI and air temperature is observed increasing the confidence of NDVI as an indicator for assessing vegetation productivity and its sensitivity to changes in local conditions.     

Vegetation discrimination using irs-p3 wifs temporal dataset - a case study from rampa forests, eastern ghats, a.p.

Multitemporal data sets from coarse resolution sensors of Indian Remote Sensing Satellites provides an opportunity to classify various forest types using their phenological attributes reflected in temporal NDVI profiles. The present study attempts to classify various vegetation classes using time integrated NDVI (T-NDVI) values derived from IRS-P3 WiFS data. The algorithm explores the differential characteristics in T-NDVI values of different features and the results suggest the possible use of the methodology for forest type classification.     

atmospheric corrections to thermal IR data for estimation of land surface temperature

Satellite data provides important inputs far estimating regional surface emisslviiy and surface temperature. The methodology for estimation of emissivity over heterogeneous areas is based on the calculation of fraction vegetation cover per pixel taking NDVI, reflectances of pure pixels as input. The surface temperature is calculated using a sptit-window equation, which depends on atmospheric water vapour, viewing angle and channel surface emissivities. In the present study model coefficients for atmospheric corrections to NOAA AVHRR thermal data Fqr tropical atmospheres have been derived with a view to operationally use the methodolpgy for generating land surface temperature information from satellite data. The results of the study show that the estimated temperature values are comparable with the ctimatological values over the region Suggesting the possible use of the methodology.     

基于遥感和GIS的中国土地利用/土地覆盖的现状研究

通过利用土地利用程度指数和植被指数等指标。采用ＧＩＳ工具和数学模型方法,对中国土地利用／土地覆盖的现状进行分析,结论为：中国现有土 况代表着中国总的土地利用程度指数为２０２的一种开发利用水平。从土地利用程度指数极限为４００看,中国土地资源总的开发程度仅达开发极限的一半;东部湿润区以耕地型覆盖为主,开发利用程度高;     

Temporal optimisation of image acquisition for land cover classification with Random Forest and MODIS time-series

The analysis and classification of land cover is one of the principal applications in terrestrial remote sensing. Due to the seasonal variability of different vegetation types and land surface characteristics, the ability to discriminate land cover types changes over time. Multi-temporal classification can help to improve the classification accuracies, but different constraints, such as financial restrictions or atmospheric conditions, may impede their application. The optimisation of image acquisition timing and frequencies can help to increase the effectiveness of the classification process. For this purpose, the Feature Importance (FI) measure of the state-of-the art machine learning method Random Forest was used to determine the optimal image acquisition periods for a general (Grassland, Forest, Water, Settlement, Peatland) and Grassland specific (Improved Grassland, Semi-Improved Grassland) land cover classification in central Ireland based on a 9-year time-series of MODIS Terra 16 day composite data (MOD13Q1). Feature Importances for each acquisition period of the Enhanced Vegetation Index (EVI) and Normalised Difference Vegetation Index (NDVI) were calculated for both classification scenarios. In the general land cover classification, the months December and January showed the highest, and July and August the lowest separability for both VIs over the entire nine-year period. This temporal separability was reflected in the classification accuracies, where the optimal choice of image dates outperformed the worst image date by 13% using NDVI and 5% using EVI on a mono-temporal analysis. With the addition of the next best image periods to the data input the classification accuracies converged quickly to their limit at around 8–10 images. The binary classification schemes, using two classes only, showed a stronger seasonal dependency with a higher intra-annual, but lower inter-annual variation. Nonetheless anomalous weather conditions, such as the cold winter of 2009/2010 can alter the temporal separability pattern significantly. Due to the extensive use of the NDVI for land cover discrimination, the findings of this study should be transferrable to data from other optical sensors with a higher spatial resolution. However, the high impact of outliers from the general climatic pattern highlights the limitation of spatial transferability to locations with different climatic and land cover conditions. The use of high-temporal, moderate resolution data such as MODIS in conjunction with machine-learning techniques proved to be a good base for the prediction of image acquisition timing for optimal land cover classification results.     

Mapping land cover gradients through analysis of hyper-temporal NDVI imagery

The green cover of the earth exhibits various spatial gradients that represent gradual changes in space of vegetation density and/or in species composition. To date, land cover mapping methods differentiate at best, mapping units with different cover densities and/or species compositions, but typically fail to express such differences as gradients. Present interpretation techniques still make insufficient use of freely available spatial-temporal Earth Observation (EO) data that allow detection of existing land cover gradients. This study explores the use of hyper-temporal NDVI imagery to detect and delineate land cover gradients analyzing the temporal behavior of NDVI values. MODIS-Terra MVC-images (250 m, 16-day) of Crete, Greece, from February 2000 to July 2009 are used. The analysis approach uses an ISODATA unsupervised classification in combination with a Hierarchical Clustering Analysis (HCA). Clustering of class-specific temporal NDVI profiles through HCA resulted in the identification of gradients in landcover vegetation growth patterns. The detected gradients were arranged in a relational diagram, and mapped. Three groups of NDVI-classes were evaluated by correlating their class-specific annual average NDVI values with the field data (tree, shrub, grass, bare soil, stone, litter fraction covers). Multiple regression analysis showed that within each NDVI group, the fraction cover data were linearly related with the NDVI data, while NDVI groups were significantly different with respect to tree cover (adj. R² = 0.96), shrub cover (adj. R² = 0.83), grass cover (adj. R² = 0.71), bare soil (adj. R² = 0.88), stone cover (adj. R² = 0.83) and litter cover (adj. R² = 0.69) fractions. Similarly, the mean Sorenson dissimilarity values were found high and significant at confidence interval of 95% in all pairs of three NDVI groups. The study demonstrates that hyper-temporal NDVI imagery can successfully detect and map land cover gradients. The results may improve land cover assessment and aid in agricultural and ecological studies.     

光谱指数趋势面的城市地表温度降尺度转换

针对城市及周边区域建造区和自然地表交织分布的特点,探讨了利用归一化植被指数(NDVI)和归一化建造指数(NDBI)构造趋势面的地表温度(LST)降尺度方法,以北京市市区及周边较平坦区域为例实现了LST自960 m向120 m的降尺度转换。分析了LST空间分布特征及NDVI、NDBI对地物的指示性特征;以北京市四至六环为界分析NDVI、NDBI趋势面对地表温度的拟合程度及各自的适用区域;在120 m、240 m、480 m和960 m 4个尺度上评价了NDVI、NDBI和NDVI+NDBI趋势面对LST的拟合程度和趋势面转换函数的尺度效应;对NDVI、NDBI和NDVI NDBI等3种方法的降尺度结果分覆盖类型、分区域对比评价。实验结果表明结合两种光谱指数的NDVI NDBI方法降尺度转换精度有所改善,改善程度取决于地表覆盖类型组合。     

Bi-temporal characterization of land surface temperature in relation to impervious surface area,NDVI and NDBI,using a sub-pixel image analysis

As more than 50% of the human population are situated in cities of the world, urbanization has become an important contributor to global warming due to remarkable urban heat island (UHI) effect. UHI effect has been linked to the regional climate, environment, and socio-economic development. In this study, Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper Plus (ETM+) imagery, respectively acquired in 1989 and 2001, were utilized to assess urban area thermal characteristics in Fuzhou, the capital city of Fujian province in south-eastern China. As a key indicator for the assessment of urban environments, sub-pixel impervious surface area (ISA) was mapped to quantitatively determine urban land-use extents and urban surface thermal patterns. In order to accurately estimate urban surface types, high-resolution imagery was utilized to generate the proportion of impervious surface areas. Urban thermal characteristics was further analysed by investigating the relationships between the land surface temperature (LST), percent impervious surface area, and two indices, the Normalized Difference Vegetation Index (NDVI) and Normalized Difference Built-up Index (NDBI). The results show that correlations between NDVI and LST are rather weak, but there is a strong positive correlation between percent ISA, NDBI and LST. This suggests that percent ISA, combined with LST, and NDBI, can quantitatively describe the spatial distribution and temporal variation of urban thermal patterns and associated land-use/land-cover (LULC) conditions.     

Development of a 1-km landcover dataset of China using AVHRR data

This paper describes the development of a 1-km landcover dataset of China by using monthly NDVI data spanning April 1992 through March 1993. The method used combined unsupervised and supervised classification of NDVI data from AVHRR. It is composed of five steps: (a) unsupervised clustering of monthly AVHRR NDVI maximum value composites is performed using the ISOCLASS algorithm; (b) preliminary identification is carried out with the addition of digital elevation models, eco-region data and a collection of other landcover/vegetation reference data to identify the clusters with single landcover classes; (c) re-clustering is performed of clusters with size greater than a given threshold value and containing two or more disparate landcover classes; (d) cluster combining is performed to combine all clusters with a single landcover class in one cluster, and all other clusters into one mixed cluster; and (e) supervised classification is used to carry out post-classification of the mixed cluster generated in the previous step by using the maximum likelihood algorithm and the identified single landcover classes of the previous step as training data. The classification is based on extensive use of computer-assisted image processing and tools, as well as the skills of the human interpreter to take the final decisions regarding the relationship between spectral classes defined using unsupervised methods and landscape characteristics that are used to define landcover classes.     

Independent two-step thresholding of binary images in inter-annual land cover change/no-change identification

Binary images from one or more spectral bands have been used in many studies for land-cover change/no-change identification in diverse climatic conditions. Determination of appropriate threshold levels for change/no-change identification is a critical factor that influences change detection result accuracy. The most used method to determine the threshold values is based on the standard deviation (SD) from the mean, assuming the amount of change (due to increase or decrease in brightness values) to be symmetrically distributed on a standard normal curve, which is not always true. Considering the asymmetrical nature of distribution histogram for the two sides, this study proposes a relatively simple and easy ‘Independent Two-Step’ thresholding approach for optimal threshold value determination for spectrally increased and decreased part using Normalized Difference Vegetation Index (NDVI) difference image. Six NDVI differencing images from 2007 to 2009 of different seasons were tested for inter-annual or seasonal land cover change/no-change identification. The relative performances of the proposed and two other methods towards the sensitivity of distributions were tested and an improvement of ∼3% in overall accuracy and of ∼0.04 in Kappa was attained with the Proposed Method. This study demonstrated the importance of consideration of normality of data distributions in land-cover change/no-change analysis.     

应用卫星与气象数据及其关系研究黄河流域的荒漠化现状

本文应用20年（1981—2000年）的卫星数据反演归一化差值植被指数（NDVI）,同时获取地面格网的温度与降雨数据,并分析这些数据之间的关系。基于地面的温度和降雨格网数据将研究区划分为8个气候区域,再利用NDVI数据把降雨量最少的3个气候区——区1,2,3各划分为10个等级。此外,分析这3个气候区在1983—1998年15年间的NDVI变化状况,结果显示出研究区荒漠化状况的加剧。     

Characterization of biophysical land units using remote sensing and gis

In the present study, an attempt has been made to characterize the biophysical land units in Kanholi bara river basin of sub-humid tropical ecosystem of central India using remotely sensed data, field surveys and GIS based multi-criteria overlay analysis. The geo-spatial database on elevation, slope, landforms, soil depth, soil erosion, land use/land cover and hydrogeomorphological parameters has been generated using IRS-ID LISS-III satellite data coupled with soil survey data in GIS. The methodology followed in characterization of biophysical land units in GIS includes assigning scores for different classes of the layers and weighatges for different layers based on their characteristics and degree of influence on desired output. GIS based ‘multi criteria overlay’ analysis reveals seventeen distinct biophysical land units in the river basin. Severe (50.5-59.5) to very severe (59.5) biophysical stress units are found in plateau spurs, isolated mounds, linear ridges, dissected plateau and escarpments. These zones are associated with severe to very severe erosion, steep to very steep, extremely shallow soils, poor to very poor groundwater prospects, wastelands and scrublands. The characterization of biophysical land units helps in analysis of their potentials, problems and stress environment to plan and execute site-specific landscape management practices and maximize the productivity from each biophysical land unit. The present study demonstrates that generation of geo-spatial database based on remotely sensed data and field surveys in GIS and their analysis helps great extent in characterization of biophysical land units and analysis of their stress environment for management.     

Vegetational response of rainfall in Rajasthan using AVHRR imagery

The purpose of this paper is to demonstrate the suitability of NOAA-AVHRR imagery for monitoring vegetal cover. Of course, climate is recognized as the most important control on natural vegetation. However, spatial and seasonal changes in land cover are dependent on several factors including the crop species, soil hydrology, grazing conditions, harvest timing and above all on the frequency and amount of rainfall. The analysis is based on six maximum value composite (MVC) NDVI images with a spatial resolution of approximately 5.5 km. An attempt has also been made to examine the serial relationship between the NDVI and the rainfall occurrences at different stations. ‘Thar Desert’ is clearly visible in the images of summer months. A transitional zone is situated between the arid and semi-arid areas along the western fringe of the Aravalli Range. The elimination of time lag in vegetational response of rainfall indicates expected positive relationship between vegetal cover and rainfall.     

Analysis of land degradation processes on a tiger bush plateau in South West Niger using MODIS and LANDSAT TM/ETM+ data

Remote sensing digital image analysis has been applied to monitor land clearing and degradation processes on a plateau covered by tiger bush near Niamey in South West Niger, where signs of severe landscape degradation due to fuelwood supply have been observed in the last decades. A MODIS NDVI dataset (2000–2015) and five LANDSAT images (1986–2012) were used to identify spatial and temporal dynamics and to emphasize areas of greater degradation. The study indicates that the land clearing found by previous investigations in the second part of the 20^th century is still ongoing, with a decreasing trend of MODIS NDVI values recorded in the period 2000–2015. This trend appeared to be linked to an increase in bare soil areas that was demonstrated by analysis of LANDSAT SAVI images. The investigation also indicated that rates of degradation are stronger in more deteriorated areas like those located nearer Niamey; degradation patterns also tend to increase from the inner areas to the edges of the plateau. These results attest to the urgency to develop effective environmental preservation policies and find alternative solutions for domestic energy supply.     

Evaluation of potential usefulness of irs-1c wifs simulated data for dryland rabi sorghum crop discrimination

Two band simulad WiFS data for five dates correspfonding to rabi sorghun growing season of 1993-94 has been generated for Aurangabad district of Maharashtra. Ground truth data has been used for supervised classificatioa of one date raw image and five date NDVI of simulated WiFS data and the results were compared with those derived from single date IRS LISS I data. Analysis of classification accuracies indicate that single date WIFS data gives slightly lower accuracy of 79 per cent against 81 per cent obtained for single date LISS I data. Overall accuracy for 5-date WiFS data is 96 per cent which shows that classification performance of five date WiFS NDVI data is far superior to the single date data of the IRS-IC WiFS as well as the IRS LISS I. The study thus shows the importance of temporal domain of data acquisition in sorghum crop discrimination, Growth profile for sorghum and other crop classes were generated from multidate WiFS derived NDVI data. Differences in growth profiles of sorghum vigour classes as well as amongst different crop types and forests corroborate the premise of better discrimination of crop types and their vigour on multidate remotely sensed data.     

资料误差对大尺度NDVI-气温关系的影响

大量研究利用PathfinderAVHRR NDVI资料分析植被状况与气温、降水等气候要素之间的关系。许多分析指出Pathfinder资料包含误差 ,并分析这些资料误差对大尺度NDVI 气温耦合关系检测结果的影响。利用奇异值分解方法 (SVD) ,通过比较不同NDVI资料误差情况下北半球春季NDVI对气温变化响应的时空特征的差异 ,对资料误差造成的分析结果的可靠性进行判断。考虑了 4种误差形式 ,分别是不同强度的连续误差、不连续误差、强火山喷发造成的误差及趋势误差。分析结果表明 ,利用SVD分析大尺度的NDVI 气温耦合特征时 ,允许的NDVI资料误差的最大上限阈值大致在 0 5σ左右。PathfinderAVHRR NDVI原始资料包含的误差很可能低于此阈值 ,得到的分析结果有较高的可信度。此外 ,在不知道NDVI原始资料误差的情况下进行植被对气候变化响应的检测时 ,可以借鉴此方法对结果的可靠性进行检查和验证。     

Retrieval of land surface temperature (LST) from landsat TM6 and TIRS data by single channel radiative transfer algorithm using satellite and ground-based inputs

The present study proposes land surface temperature (LST) retrieval from satellite-based thermal IR data by single channel radiative transfer algorithm using atmospheric correction parameters derived from satellite-based and in-situ data and land surface emissivity (LSE) derived by a hybrid LSE model. For example, atmospheric transmittance (τ) was derived from Terra MODIS spectral radiance in atmospheric window and absorption bands, whereas the atmospheric path radiance and sky radiance were estimated using satellite- and ground-based in-situ solar radiation, geographic location and observation conditions. The hybrid LSE model which is coupled with ground-based emissivity measurements is more versatile than the previous LSE models and yields improved emissivity values by knowledge-based approach. It uses NDVI-based and NDVI Threshold method (NDVI^THM) based algorithms and field-measured emissivity values. The model is applicable for dense vegetation cover, mixed vegetation cover, bare earth including coal mining related land surface classes. The study was conducted in a coalfield of India badly affected by coal fire for decades. In a coal fire affected coalfield, LST would provide precise temperature difference between thermally anomalous coal fire pixels and background pixels to facilitate coal fire detection and monitoring. The derived LST products of the present study were compared with radiant temperature images across some of the prominent coal fire locations in the study area by graphical means and by some standard mathematical dispersion coefficients such as coefficient of variation, coefficient of quartile deviation, coefficient of quartile deviation for 3rd quartile vs. maximum temperature, coefficient of mean deviation (about median) indicating significant increase in the temperature difference among the pixels. The average temperature slope between adjacent pixels, which increases the potential of coal fire pixel detection from background pixels, is significantly larger in the derived LST products than the corresponding radiant temperature images.     

THE DEPARTMENT OF PHOTOGRAMMETRY AND REMOTE SENSING AT WUHAN TECHNICAL UNIVERSITY,PEOPLE'S REPUBLIC OF CHINA

Many remote sensing applications are predicated on the fact that there is a known relationship between climate and vegetation dynamics as monitored from space. However, few studies investigate vegetation index variation on individual homogeneous land cover units as they relate to specific climate and environmental influences at the local scale. This study focuses on the relationship between the Palmer Drought Severity Index (PDSI) and different vegetation types through the derivation of vegetation indices from Landsat 7 ETM+ data (NDVI, Tasseled Cap, and SAVI). A series of closely spaced through time images from 1999 to 2002 were selected, classified, and analyzed for an area in northeastern Ohio. Supervised classification of the images allowed us to monitor the response in individual land cover classes to changing climate conditions, and compare these individual changes to those over the entire larger areas. Specifically, the images were compared using linear regression techniques at various time lags to PDSI values for these areas collected by NOAA. Although NDVI is a robust indicator of vegetation greenness and vigor, it may not be the best index to use, depending on the type of vegetation studied and the scale of analysis used. A combination of NDVI and other prominent vegetation indices can be used to detect subtle drought conditions by specifically identifying various time lags between climate condition and vegetation response.