Finer resolution observation and monitoring of global land cover: first mapping results with Landsat TM and ETM+ data

We have produced the first 30 m resolution global land-cover maps using Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper Plus (ETM+) data. We have classified over 6600 scenes of Landsat TM data after 2006, and over 2300 scenes of Landsat TM and ETM+ data before 2006, all selected from the green season. These images cover most of the world's land surface except Antarctica and Greenland. Most of these images came from the United States Geological Survey in level L1T (orthorectified). Four classifiers that were freely available were employed, including the conventional maximum likelihood classifier (MLC), J4.8 decision tree classifier, Random Forest (RF) classifier and support vector machine (SVM) classifier. A total of 91,433 training samples were collected by traversing each scene and finding the most representative and homogeneous samples. A total of 38,664 test samples were collected at preset, fixed locations based on a globally systematic unaligned sampling strategy. Two software tools, Global Analyst and Global Mapper developed by extending the functionality of Google Earth, were used in developing the training and test sample databases by referencing the Moderate Resolution Imaging Spectroradiometer enhanced vegetation index (MODIS EVI) time series for 2010 and high resolution images from Google Earth. A unique land-cover classification system was developed that can be crosswalked to the existing United Nations Food and Agriculture Organization (FAO) land-cover classification system as well as the International Geosphere-Biosphere Programme (IGBP) system. Using the four classification algorithms, we obtained the initial set of global land-cover maps. The SVM produced the highest overall classification accuracy (OCA) of 64.9% assessed with our test samples, with RF (59.8%), J4.8 (57.9%), and MLC (53.9%) ranked from the second to the fourth. We also estimated the OCAs using a subset of our test samples (8629) each of which represented a homogeneous area greater than 500 m?×?500 m. Using this subset, we found the OCA for the SVM to be 71.5%. As a consistent source for estimating the coverage of global land-cover types in the world, estimation from the test samples shows that only 6.90% of the world is planted for agricultural production. The total area of cropland is 11.51% if unplanted croplands are included. The forests, grasslands, and shrublands cover 28.35%, 13.37%, and 11.49% of the world, respectively. The impervious surface covers only 0.66% of the world. Inland waterbodies, barren lands, and snow and ice cover 3.56%, 16.51%, and 12.81% of the world, respectively.     

Detection of fire impact and vegetation recovery over tropical peat swamp forest by satellite data and ground-based NDVI instrument

Multitemporal Principal Component Analysis (MPCA) was used for processing Landsat Thematic Mapper (TM)/Enhanced Thematic Mapper plus (ETM+) satellite images. MPCA was able to merge spectral data corresponding to TM-1996 (pre-fire in 1997), ETM-2000 (post-fire 1997 and pre-fire 2002) and ETM-2003 (post-fire in 2002), which was crucial for detecting the fire impact and vegetation recovery. Results indicate that the burnt areas of 1997 and 2002 were 89,086 ha (16.5%) and 31,859 ha (5.9%), respectively, within the study area of 540,000 ha. Satellite Pour 1’Observation de la Terre (SPOT)-VEGETATION 10-day Maximum Value Composite (MVC) data were also used and compared with Normalized Difference Vegetation Index (NDVI) from ground-based NDVI. Our research demonstrates the strong relationship between Landsat- TM/ETM+, SPOT-VEGETATION data and ground-based NDVI in identifying land-cover changes and vegetation recovery over the tropical peat swamp forest area in Central Kalimantan, Indonesia that is affected by forest fires that occurred in 1997 and 2002.     

An operational radiometric calibration procedure for the Landsat sensors based on pseudo-invariant target sites

The Statewide Landcover and Trees Study (SLATS) use both Landsat-7 ETM+ and Landsat-5 TM imagery to monitor short-term woody vegetation changes throughout Queensland, Australia. In order to analyse more subtle long-term vegetation change, time-based trends resulting from artefacts introduced by the sensor system must be removed. In this study, a reflectance-based vicarious calibration approach using high-reflectance, pseudo-invariant targets in western Queensland was developed. This calibration procedure was used to test the existing calibration models for ETM+ and TM, and develop a consistent operational calibration procedure which provides calibration information for the MSS sensors. Ground based data, sensor spectral response functions and atmospheric variables were used as input to MODTRAN radiative transfer code to estimate top-of-atmosphere radiance. The estimated gains for Landsat-7 ETM+ (1999-2003), -5 TM (1987-2004), -5 MSS (1984-1993) and -2 MSS (1979-1982) are presented. Results confirm the stability and accuracy of the ETM+ calibration, and the suitability of this data as a radiometric standard for cross-calibration with TM. Vicarious data support the use of the existing TM calibration model for the red and two shortwave-infrared bands. However, alternative models for blue, green and near-infrared bands are presented. The models proposed differ most noticeably at dates prior to 1995, with differences in estimated gains of up to 9.7%, 10.8% and 6.9% for the blue, green and near-infrared bands respectively. Vicarious gains for Landsat-2 MSS and Landsat-5 MSS are presented and are compared with those applied by the on-board calibration system. Updated calibration coefficients to scale MSS data to the SLATS vicarious measurements are given. The removal of time based calibration trends in the SLATS data archive will enable the measurement of vegetation changes over the 26 year period covered by Landsat -2, -5 and -7.     

Improving 30 m global land-cover map FROM-GLC with time series MODIS and auxiliary data sets: a segmentation-based approach

FROM-GLC (Fine Resolution Observation and Monitoring of Global Land Cover) is the first 30 m resolution global land-cover map produced using Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper Plus (ETM+) data. Due to the lack of temporal features as inputs in producing FROM-GLC, considerable confusion exists among land-cover types (e.g. agriculture lands, grasslands, shrublands, and bareland). The Moderate Resolution Imaging Spectrometer (MODIS) provides high-temporal frequency information on surface cover. Other auxiliary bioclimatic, digital elevation model (DEM), and world maps on soil-water conditions are possible sources for improving the accuracy of FROM-GLC. In this article, a segmentation-based approach was applied to Landsat imagery to down-scale coarser-resolution MODIS data (250 m) and other 1 km resolution auxiliary data to the segment scale based on TM data. Two classifiers (support vector machine (SVM) and random forest (RF)) and two different strategies for use of training samples (global and regional samples based on a spatial temporal selection criterion) were performed. Results show that RF based on the global use of training samples achieves an overall classification accuracy of 67.08% when assessed by test samples collected independently. This is better than the 64.89% achieved by FROM-GLC based on the same set of test samples. Accuracies for vegetation cover types are most substantially improved.     

Summary of current radiometric calibration coefficients for Landsat MSS, TM, ETM+, and EO-1 ALI sensors

This paper provides a summary of the current equations and rescaling factors for converting calibrated Digital Numbers (DNs) to absolute units of at-sensor spectral radiance, Top-Of-Atmosphere (TOA) reflectance, and at-sensor brightness temperature. It tabulates the necessary constants for the Multispectral Scanner (MSS), Thematic Mapper (TM), Enhanced Thematic Mapper Plus (ETM+), and Advanced Land Imager (ALI) sensors. These conversions provide a basis for standardized comparison of data in a single scene or between images acquired on different dates or by different sensors. This paper forms a needed guide for Landsat data users who now have access to the entire Landsat archive at no cost.     

The roles of textural images in improving land-cover classification in the Brazilian Amazon

Texture has long been recognized as valuable in improving land-cover classification, but how data from different sensors with varying spatial resolutions affect the selection of textural images is poorly understood. This research examines textural images from the Landsat Thematic Mapper (TM), ALOS (Advanced Land Observing Satellite) PALSAR (Phased Array type L-band Synthetic Aperture Radar), the SPOT (Satellite Pour l’Observation de la Terre) high-resolution geometric (HRG) instrument, and the QuickBird satellite, which have pixel sizes of 30, 12.5, 10/5, and 0.6 m, respectively, for land-cover classification in the Brazilian Amazon. GLCM (grey-level co-occurrence matrix)-based texture measures with various sizes of moving windows are used to extract textural images from the aforementioned sensor data. An index based on standard deviations and correlation coefficients is used to identify the best texture combination following separability analysis of land-cover types based on training sample plots. A maximum likelihood classifier is used to conduct the land-cover classification, and the results are evaluated using field survey data. This research shows the importance of textural images in improving land-cover classification, and the importance becomes more significant as the pixel size improved. It is also shown that texture is especially important in the case of the ALOS PALSAR and QuickBird data. Overall, textural images have less capability in distinguishing land-cover types than spectral signatures, especially for Landsat TM imagery, but incorporation of textures into radiometric data is valuable for improving land-cover classification. The classification accuracy can be improved by 5.2–13.4% as the pixel size changes from 30 to 0.6 m.     

Improved calibration of Landsat-5 TM applicable for high-latitude and dark areas

The Landsat-5 Thematic Mapper (TM) has provided data for more than 17 years, making it one of the most successful missions so far. TM sensor degradation is well documented and although efforts are made to account for this degradation when calibrating the sensor, such calibrations are often done over high reflective and bright surfaces in combination with a high solar irradiance. These conditions are not found in high-latitude and dark areas, making the calibration coefficients inappropriate to use. In this study reflectances obtained from TM bands 1–4 over a high-arctic area from 1987–1998 are compared to reflectances obtained from the Landsat-7 Enhanced Thematic Mapper Plus (ETM + ) over the same area in 1999–2000. From the reflectance comparison it was found that a correction of the calibration gain could be described by a power function using days since launch as the controlling variable. From the power function, updated and lifetime calibration coefficients for TM bands 1–4 applicable for high-latitude and dark areas were determined. Furthermore, the study showed a continuous decrease of the TM sensor response with band 1 being the most affected and band 4 the most stable. The study also showed the possibility of using ETM+ to determine updated calibration coefficients for TM by a cross-calibration even though the ETM+ and TM scenes are not coincident.     

Land-cover classification in a moist tropical region of Brazil with Landsat Thematic Mapper imagery

This research aims to improve land-cover classification accuracy in a moist tropical region in Brazil by examining the use of different remote-sensing-derived variables and classification algorithms. Different scenarios based on Landsat Thematic Mapper (TM) spectral data and derived vegetation indices and textural images and different classification algorithms, maximum likelihood classification (MLC), artificial neural network (ANN), classification tree analysis (CTA) and object-based classification (OBC), were explored. The results indicate that a combination of vegetation indices as extra bands into Landsat TM multi-spectral bands did not improve the overall classification performance, but the combination of textural images was valuable for improving vegetation classification accuracy. In particular, the combination of both vegetation indices and textural images into TM multi-spectral bands improved the overall classification accuracy (OCA) by 5.6% and the overall kappa coefficient (OKC) by 6.25%. Comparison of the different classification algorithms indicated that CTA and ANN have poor classification performance in this research, but OBC improved primary forest and pasture classification accuracies. This research indicates that use of textural images or use of OBC are especially valuable for improving the vegetation classes such as upland and liana forest classes that have complex stand structures and large patch sizes.     

Mapping impervious surface expansion using medium-resolution satellite image time series: a case study in the Yangtze River Delta,China

Cities have been expanding rapidly worldwide, especially over the past few decades. Mapping the dynamic expansion of impervious surface in both space and time is essential for an improved understanding of the urbanization process, land-cover and land-use change, and their impacts on the environment. Landsat and other medium-resolution satellites provide the necessary spatial details and temporal frequency for mapping impervious surface expansion over the past four decades. Since the US Geological Survey opened the historical record of the Landsat image archive for free access in 2008, the decades-old bottleneck of data limitation has gone. Remote-sensing scientists are now rich with data, and the challenge is how to make best use of this precious resource. In this article, we develop an efficient algorithm to map the continuous expansion of impervious surface using a time series of four decades of medium-resolution satellite images. The algorithm is based on a supervised classification of the time-series image stack using a decision tree. Each imerpervious class represents urbanization starting in a different image. The algorithm also allows us to remove inconsistent training samples because impervious expansion is not reversible during the study period. The objective is to extract a time series of complete and consistent impervious surface maps from a corresponding times series of images collected from multiple sensors, and with a minimal amount of image preprocessing effort. The approach was tested in the lower Yangtze River Delta region, one of the fastest urban growth areas in China. Results from nearly four decades of medium-resolution satellite data from the Landsat Multispectral Scanner (MSS), Thematic Mapper (TM), Enhanced Thematic Mapper plus (ETM+) and China–Brazil Earth Resources Satellite (CBERS) show a consistent urbanization process that is consistent with economic development plans and policies. The time-series impervious spatial extent maps derived from this study agree well with an existing urban extent polygon data set that was previously developed independently. The overall mapping accuracy was estimated at about 92.5% with 3% commission error and 12% omission error for the impervious type from all images regardless of image quality and initial spatial resolution.     

Mapping forest successional stages following deforestation in Brazilian Amazonia using multi‐temporal Landsat images

Tropical forest successional stages have been mapped previously with multi‐temporal satellite sensor imagery. The precise identification and classification of such stages, however, has proved difficult. This Letter presents a new method for the classification of forest successional stages following deforestation in Brazilian Amazonia. Multi‐temporal Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper Plus (ETM+) and derived fraction images and field data were used in a semi‐automatic classification approach. The results were encouraging and signal the application of the method for the entire Brazilian Amazonia.     

Logistic regression modelling of multitemporal Thematic Mapper data for burned area mapping

This study focused on the development of a logistic regression model for burned area mapping using two Landsat-5 Thematic Mapper (TM) images. Logistic regression models were structured using the spectral channels of the two images as explanatory variables. The overall accuracy of the results and other statistical indications denote that logisticregression modelling can be usedsuccessfully for burned area mapping. The model that consisted of the spectral channels TM4, TM7 and TM1 and had an overall accuracy of 97.62%, proved to be the most suitable. Moreover, the study concluded that the spectral channel TM4 was the most sensitive to alterations of the spectral response of the burned category pixels, followed by TM7.     

The application of ensemble techniques for land-cover classification in arid lands

Robust classification approaches are required for accurate classification of complex land-use/land-cover categories of desert landscapes using remotely sensed data. Machine-learning ensemble classifiers have proved to be powerful for the classification of remotely sensed data. However, they have not been evaluated for classifying land-cover categories in desert regions. In this study, the performance of two machine-learning ensemble classifiers – random forests (RF) and boosted artificial neural networks – is explored in the context of classification of land use/land cover of desert landscapes. The evaluation is based on the accuracy of classification of remotely sensed data, with and without integration of ancillary data. Landsat-5 Thematic Mapper data captured for a desert landscape in the north-western coastal desert of Egypt are used with ancillary variables derived from a digital terrain model to classify 13 different land-use/land-cover categories. Results show that the two ensemble methods produce accurate land-cover classifications, with and without integrating spectral data with ancillary data. In general, the overall accuracy exceeded 85% and the kappa coefficient (κ) attained values over 0.83. The integration of ancillary data improved the performance of the boosted artificial neural networks by approximately 5% and the random forests by 9%. The latter showed overall higher accuracy; however, boosted artificial neural networks showed better generalization ability and lower overfitting tendencies. The results reveal the merit of applying ensemble methods to integrated spectral and ancillary data of similar desert landscapes for achieving high classification accuracies.     

Land-cover classification in a moist tropical region of Brazil with Landsat TM imagery

This research aims to improve land-cover classification accuracy in a moist tropical region in Brazil by examining the use of different remote sensing-derived variables and classification algorithms. Different scenarios based on Landsat Thematic Mapper (TM) spectral data and derived vegetation indices and textural images, and different classification algorithms - maximum likelihood classification (MLC), artificial neural network (ANN), classification tree analysis (CTA), and object-based classification (OBC), were explored. The results indicated that a combination of vegetation indices as extra bands into Landsat TM multispectral bands did not improve the overall classification performance, but the combination of textural images was valuable for improving vegetation classification accuracy. In particular, the combination of both vegetation indices and textural images into TM multispectral bands improved overall classification accuracy by 5.6% and kappa coefficient by 6.25%. Comparison of the different classification algorithms indicated that CTA and ANN have poor classification performance in this research, but OBC improved primary forest and pasture classification accuracies. This research indicates that use of textural images or use of OBC are especially valuable for improving the vegetation classes such as upland and liana forest classes having complex stand structures and having relatively large patch sizes.     

Burned area mapping using logistic regression modeling of a single post-fire Landsat-5 Thematic Mapper image

Logistic regression modeling was applied, as an alternative classification procedure, to a single post-fire Landsat-5 Thematic Mapper image for burned land mapping. The nature of the classification problem in this case allowed the structure and application of logistic regression models, since the dependent variable could be expressed in a dichotomous way. The two logistic regression models consisted of the TM 4, TM 7, TM 1 and TM 4, TM 7, TM 2 presented an overall accuracy of 97.37% and 97.30%, respectively and proved to be the most well performing three-channel color composites. The discriminator ability in respect to burned area mapping of each one of the six spectral channels of Thematic Mapper, which was achieved by applying six logistic regression models, agreed with the results taken from the separability indices Jeffries-Matusita and Transformed Divergence.     

Circa 2014 African land-cover maps compatible with FROM-GLC and GLC2000 classification schemes based on multi-seasonal Landsat data

A new African land-cover data set has been developed using multi-seasonal Landsat Operational Land Imager (OLI) imagery mainly acquired around 2014, supplemented by Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper Plus (ETM+). Each path/row location was covered by five images, including one in the growing season of vegetation and the others in four meteorological seasons (i.e. spring, summer, autumn, and winter), choosing the image with the least cloud coverage. The data set has two classification schemes, i.e. Finer Resolution Observation and Monitoring – Global Land Cover (FROM-GLC) and Global Land Cover 2000 (GLC2000), providing greater flexibility in product comparisons and applications. Random forest was used as the classifier in this project. Overall accuracies were 71% and 67% for the maps in the FROM-GLC classification scheme at level 1 and level 2, respectively, and 70% for the map in the GLC2000 classification scheme. The newly developed African land-cover map achieved a greater improvement in accuracy compared to previous products in the FROM-GLC project. Multi-seasonal imagery helped increase the mapping accuracy by better differentiating vegetation types with similar spectral features in one specific season and identifying vegetation with a shorter growing season. Night light data with 1 km resolution was used to identify the potential area of impervious surfaces to avoid overestimating the distribution of impervious surfaces without decreasing the spatial resolution. Stacking multi-seasonal mapping results could adequately minimize the disturbance of cloud and shade.     

Changes in snow and glacier cover in an arid watershed of the western Kunlun Mountains using multisource remote-sensing data

Snow and glaciers in the mountain watersheds of the Tarim River basin in western China provide the primary water resources to cover the needs of downstream oases. Remote sensing provides a practical approach to monitoring the change in snow and glacier cover in those mountain watersheds. This study investigated the change in snow and glacier cover in one such mountain watershed using multisource remote-sensing data, including the Moderate Resolution Imaging Spectroradiometer (MODIS), Landsat (Multispectral Scanner (MSS), Thematic Mapper (TM), and Enhanced Thematic Mapper Plus (ETM+)), Corona, and Google Earth^TM imagery. With 10 years’ daily MODIS snow-cover data from 2002 to 2012, we used two de-cloud methods before calculating daily snow-cover percentage (SCP), annual snow-cover frequency (SCF), and annual minimum snow-cover percentage (AMSCP) for the watershed. Mann–Kendall analysis showed no significant trend in any of those snow-cover characterizations. With a total of 22 Landsat images from 1967 to 2011, we used band ratio and supervised classification methods for snow classification for Landsat TM/ETM+ images and MSS images, respectively. The Landsat snow-cover data were divided into two periods (1976–2002 and 2004–2011). Statistical tests indicated no significant difference in either the variance or mean of SCPs between the two periods. Three glaciers were identified from Landsat images of 1998 and 2011, and their total area increased by 12.6%. In addition, three rock glaciers were also identified on both the Corona image of 1968 and the Google high-resolution image of 2007, and their area increased by 2.5%. Overall, based on multisource remote-sensing data sets, our study found no evidence of significant changes in snow and glacier cover in the watershed.     

Land-cover classification of China: Integrated analysis of AVHRR imagery and geophysical data

Over last two decades, numerous studies have used remotely sensed data from the Advanced Very High Resolution Radiometer (AVHRR) sensors to map land use and land cover at large spatial scales, but achieved only limited success. In this paper, we employed an approach that combines both AVHRR images and geophysical datasets (e.g. climate, elevation). Three geophysical datasets are used in this study: annual mean temperature, annual precipitation, and elevation. We first divide China into nine bio-climatic regions, using the long-term mean climate data. For each of nine regions, the three geophysical data layers are stacked together with AVHRR data and AVHRR-derived vegetation index (Normalized Difference Vegetation Index) data, and the resultant multi-source datasets were then analysed to generate land-cover maps for individual regions, using supervised classification algorithms. The nine land-cover maps for individual regions were assembled together for China. The existing land-cover dataset derived from Landsat Thematic Mapper (TM) images was used to assess the accuracy of the classification that is based on AVHRR and geophysical data. Accuracy of individual regions varies from 73% to 89%, with an overall accuracy of 81% for China. The results showed that the methodology used in this study is, in general, feasible for large-scale land-cover mapping in China.     

Towards a common validation sample set for global land-cover mapping

Validating land-cover maps at the global scale is a significant challenge. We built a global validation data-set based on interpreting Landsat Thematic Mapper (TM) and Enhanced TM Plus (ETM+) images for a total of 38,664 sample units pre-determined with an equal-area stratified sampling scheme. This was supplemented by MODIS enhanced vegetation index (EVI) time series data and other high-resolution imagery on Google Earth. Initially designed for validating 30 m-resolution global land-cover maps in the Finer Resolution Observation and Monitoring of Global Land Cover (FROM-GLC) project, the data-set has been carefully improved through several rounds of interpretation and verification by different image interpreters, and checked by one quality controller. Independent test interpretation indicated that the quality control correctness level reached 90% at level 1 classes using selected interpretation keys from various parts of the USA. Fifty-nine per cent of the samples have been verified with high-resolution images on Google Earth. Uncertainty in interpretation was measured by the interpreter’s perceived confidence. Only less than 7% of the sample was perceived as low confidence at level 1 by interpreters. Nearly 42% of the sample units located within a homogeneous area could be applied to validating global land-cover maps whose resolution is 500 m or finer. Forty-six per cent of the sample whose EVI values are high or with little seasonal variation throughout the year can be applied to validate land-cover products produced from data acquired in different phenological stages, while approximately 76% of the remaining sample whose EVI values have obvious seasonal variation was interpreted from images acquired within the growing season. While the improvement is under way, some of the homogeneous sample units in the data-set have already been used in assessing other classification results or as training data for land-cover mapping with coarser-resolution data.     

Annual land-cover mapping based on multi-temporal cloud-contaminated landsat images

Landsat images, which have fine spatial resolution, are an important data source for land-cover mapping. Multi-temporal Landsat classification has become popular because of the abundance of free-access Landsat images that are available. However, cloud cover is inevitable due to the relatively low temporal frequency of the data. In this paper, a novel approach for multi-temporal Landsat land-cover classification is proposed. The land cover for each Landsat acquisition date was first classified using a Support Vector Machine (SVM) and then the classification results were combined using different strategies, with missing observations allowed. Three strategies, including the majority vote (MultiSVM-MV), Expectation Maximisation (MultiSVM-EM) and joint SVM probability (JSVM), were used to merge the multi-temporal classification maps. The three algorithms were then applied to a region of the path/row 143/31 scene using 2010 Landsat-5 Thematic Mapper (TM) images. The results demonstrated that, for these three algorithms, the average overall accuracy (OA) improved with the increase in temporal depth; also, for a given temporal depth, the performance of JSVM was clearly better than that of MultiSVM-MV and MultiSVM-EM, and the performance of MultiSVM-EM was slightly better than that of MultiSVM-MV. The OA values for the three classification results, which use all epochs, were 70.28%, 72.40% and 74.80% for MultiSVM-MV, MultiSVM-EM and JSVM, respectively. In comparison, two other annual composite image-based classification methods, annual maximum Normalised Difference Vegetation Index (NDVI) composite image-based classification and annual best-available-pixel (BAP) composite image-based classification, gave OA values of 68.08% and 69.92%, respectively, meaning that our method produced a better performance. Therefore, the novel multi-temporal Landsat classification method presented in this paper can deal with the cloud-contamination problem and produce accurate annual land-cover mapping using multi-temporal cloud-contaminated images, which is of importance for regional and global land-cover mapping.