Land cover mapping in support of LAI and FPAR retrievals from EOS-MODIS and MISR: Classification methods and sensitivities to errors

Land cover maps are used widely to parameterize the biophysical properties of plant canopies in models that describe terrestrial biogeochemical processes. In this paper, we describe the use of supervised classification algorithms to generate land cover maps that characterize the vegetation types required for Leaf Area Index (LAI) and Fraction of Photosynthetically Active Radiation (FPAR) retrievals from MODIS and MISR. As part of this analysis, we examine the sensitivity of remote sensing-based retrievals of LAI and FPAR to land cover information used to parameterize vegetation canopy radiative transfer models. Specifically, a decision tree classification algorithm is used to generate a land cover map of North America from Advanced Very High Resolution Radiometer (AVHRR) data with 1 km spatial resolution using a six-biome classification scheme. To do this, a time series of normalized difference vegetation index data from the AVHRR is used in association with extensive site-based training data compiled using Landsat Thematic Mapper (TM) and ancillary map sources. Accuracy assessment of the map produced via decision tree classification yields a cross-validated map accuracy of 73%. Results comparing LAI and FPAR retrievals using maps from different sources show that disagreement in land cover labels generally do not translate into strong disagreement in LAI and FPAR maps. Further, the main source of disagreement in LAI and FPAR maps can be attributed to specific biome classes that are characterized by a continuum of fractional cover and canopy structure.     

Assessing the extent of agriculture/pasture and secondary succession forest in the Brazilian Legal Amazon using SPOT VEGETATION data

There has been growing concern about land use/land cover change in tropical regions, as there is evidence of its influence on the observed increase in atmospheric carbon dioxide concentration and consequent climatic changes. Mapping of deforestation by the Brazil's National Space Research Institute (INPE) in areas of primary tropical forest using satellite data indicates a value of 587,727 km² up to the year 2000. Although most of the efforts have been concentrated in mapping primary tropical forest deforestation, there is also evidence of large-scale deforestation in the cerrado savanna, the second most important biome in the region.The main purpose of this work was to assess the extent of agriculture/pasture and secondary succession forest in the Brazilian Legal Amazon (BLA) in 2000, using a set of multitemporal images from the 1-km SPOT-4 VEGETATION (VGT) sensor. Additionally, we discriminated primary tropical forest, cerrado savanna, and natural/artificial waterbodies. Four classification algorithms were tested: quadratic discriminant analysis (QDA), simple classification trees (SCT), probability-bagging classification trees (PBCT), and k-nearest neighbors (K-NN). The agriculture/pasture class is a surrogate for those areas cleared of its original vegetation cover in the past, acting as a source of carbon. On the contrary, the secondary succession forest class behaves as a sink of carbon.We used a time series of 12 monthly composite images of the year 2000, derived from the SPOT-4 VGT sensor. A set of 19 Landsat scenes was used to select training and testing data. A 10-fold cross validation procedure rated PBCT as the best classification algorithm, with an overall sample accuracy of 0.92. High omission and commission errors occurred in the secondary succession forest class, due to confusion with agriculture/pasture and primary tropical forest classes. However, the PBCT algorithm generated the lower misclassification error in this class. Besides, this algorithm yields information about class membership probability, with ∼80% of the pixels with class membership probability greater or equal than 0.8. The estimated total area of agriculture/pasture and secondary succession forest in 2000 in the BLA was 966 × 10³ and 140 × 10³ km², respectively. Comparison with an existing land cover map indicates that agriculture/pasture occurred primarily in areas previously occupied by primary tropical forest (46%) and cerrado savanna (33%), and also in transition forest (19%), and other vegetation types (2%). This further confirms the existing evidence of extensive cerrado savanna conversion.This study also concludes that SPOT-4 VGT data are adequate for discriminating several major land cover types in tropical regions. Agriculture/pasture was mapped with errors of about 5%. Very high classification errors were associated with secondary succession forest, suggesting that a different methodology/sensor has to be used to address this difficult land cover class (namely with the inclusion of ancillary data). For the other classes, we consider that accurate maps can be derived from SPOT-4 VGT data with errors lower than 20% for the cerrado savanna, and errors lower than 10% for the other land cover classes. These estimates may be useful to evaluate impacts of land use/land cover change on the carbon and water cycles, biotic diversity, and soil degradation.     

A comparison of the IGBP DISCover and University of Maryland 1 km global land cover products

Two global 1 km land cover data sets derived from 1992-1993 Advanced Very High Resolution Radiometer (AVHRR) data are currently available, the International Geosphere-Biosphere Programme Data and Information System (IGBP-DIS) DISCover and the University of Maryland (UMd) 1 km land cover maps. This paper makes a preliminary comparison of the methodologies and results of the two products. The DISCover methodology employed an unsupervised clustering classification scheme on a per-continent basis using 12 monthly maximum NDVI composites as inputs. The UMd approach employed a supervised classification tree method in which temporal metrics derived from all AVHRR bands and the NDVI were used to predict class membership across the entire globe. The DISCover map uses the IGBP classification scheme, while the UMd map employs a modified IGBP scheme minus the classes of permanent wetlands, cropland/natural vegetation mosaic and ice and snow. Global area totals of aggregated vegetation types are very similar and have a per-pixel agreement of 74%. For tall versus short/no vegetation, the per-pixel agreement is 84%. For broad vegetation types, core areas map similarly, while transition zones around core areas differ significantly. This results in high regional variability between the maps. Individual class agreement between the two 1 km maps is 49%. Comparison of the maps at a nominal 0.5 resolution with two global ground-based maps shows an improvement of thematic concurrency of 46% when viewing average class agreement. The absence of the cropland mosaic class creates a difficulty in comparing the maps, due to its significant extent in the DISCover map. The DISCover map, in general, has more forest, while the UMd map has considerably more area in the intermediate tree cover classes of woody savanna/ woodland and savanna/wooded grassland.     

Using remote sensing to estimate the change of carbon storage: A case study in the estuary of Yellow River delta

Land cover change can exert a crucial effect on the terrestrial carbon cycle. To estimate changes in the carbon pools and carbon fluxes to the atmosphere, Landsat Thematic Mapper (TM) data of 1992 and 1996 were used to calculate the extent of different land cover types and their changes in the estuary area of the Yellow River delta. Image processing and the unsupervised classification allowed accurate land cover maps for 1992 and 1996 to be generated, by which the changes in the carbon pools were detected. Estimation of the carbon pools and the carbon fluxes to the atmosphere was carried out employing the results of Landsat image analysis and the published data on carbon stocks in vegetation and soil. By calculating the area changed between different types of vegetation and their different carbon stocks, the quantity of the terrestrial carbon cycle in the estuary area of the Yellow River delta was acquired. The results shows that the vegetation carbon storage was 11.43 2 10 11 g and soil carbon pool 7.24 2 10 12 g in 1992, and the vegetation carbon pool increased by 3.77 2 10 11 g during the 4 years from 1992 to 1996.     

Mapping land cover types in the Amazon Basin using 1 km JERS-1 mosaic

During the Global Rain Forest Mapping (GRFM) project, the JERS-1 SAR (Synthetic Aperture Radar) satellite acquired wall-to-wall image coverage of the humid tropical forests of the world. The rationale for the project was to demonstrate the application of spaceborne L-band radar in tropical forest studies. In particular, the use of orbital radar data for mapping land cover types, estimating the area of floodplains, and monitoring deforestation and forest regeneration were of primary importance. In this paper we examine the information content of the JERS-1 SAR data for mapping land cover types in the Amazon basin. More than 1500 high-resolution (12.5 m pixel spacing) images acquired during the low flood period of the Amazon river were resampled to 100 m resolution and mosaicked into a seamless image of about 8 million km², including the entire Amazon basin. This image was used in a classifier to generate a 1 km resolution land cover map. The inputs to the classifier were 1 km resolution mean backscatter and seven first-order texture measures derived from the 100 m data by using a 10 x 10 independent sampling window. The classification approach included two interdependent stages. First, a supervised maximum a posteriori Baysian approach classified the mean backscatter image into five general cover categories: terra firme forest (including secondary forest), savanna, inundated vegetation, open deforested areas and open water. A hierarchical decision rule based on texture measures was then applied to attempt further discrimination of known subcategories of vegetation types based on taxonomic information and woody biomass levels. True distributions of the general categories were identified from the RADAMBRASIL project vegetation maps and several field studies. Training and validation test sites were chosen from the JERS-1 image by consulting the RADAM vegetation maps. After several iterations and combining land cover types, 14 vegetation classes were successfully separated at the 1 km scale. The accuracy of the classification methodology was estimated to be 78% when using the validation sites. The results were also verified by comparison with the RADAM- and AVHRR-based 1 km resolution land cover maps.     

A hybrid approach to urban land use/cover mapping using Landsat 7 Enhanced Thematic Mapper Plus (ETM+) images

A hybrid method that incorporates the advantages of supervised and unsupervised approaches as well as hard and soft classifications was proposed for mapping the land use/cover of the Atlanta metropolitan area using Landsat 7 Enhanced Thematic Mapper Plus (ETM+) data. The unsupervised ISODATA clustering method was initially used to segment the image into a large number of clusters of pixels. With reference to ground data based on 1?:?40?000 colour infrared aerial photographs in the form of Digital Orthophoto Quarter Quadrangle (DOQQ), homogeneous clusters were labelled. Clusters that could not be labelled because of mixed pixels were clipped out and subjected to a supervised fuzzy classification. A final land use/cover map was obtained by a union overlay of the two partial land use/cover maps. This map was evaluated by comparing with maps produced using unsupervised ISODATA clustering, supervised fuzzy and supervised maximum likelihood classification methods. It was found that the hybrid approach was slightly better than the unsupervised ISODATA clustering in land use/cover classification accuracy, most probably because of the supervised fuzzy classification, which effectively dealt with the mixed pixel problem in the low-density urban use category of land use/cover. It was suggested that this hybrid approach can be economically implemented in a standard image processing software package to produce land use/cover maps with higher accuracy from satellite images of moderate spatial resolution in a complex urban environment, where both discrete and continuous land cover elements occur side by side.     

Hierarchical mapping of Northern Eurasian land cover using MODIS data

The Northern Eurasian land mass encompasses a diverse array of land cover types including tundra, boreal forest, wetlands, semi-arid steppe, and agricultural land use. Despite the well-established importance of Northern Eurasia in the global carbon and climate system, the distribution and properties of land cover in this region are not well characterized. To address this knowledge and data gap, a hierarchical mapping approach was developed that encompasses the study area for the Northern Eurasia Earth System Partnership Initiative (NEESPI). The Northern Eurasia Land Cover (NELC) database developed in this study follows the FAO-Land Cover Classification System and provides nested groupings of land cover characteristics, with separate layers for land use, wetlands, and tundra. The database implementation is substantially different from other large-scale land cover datasets that provide maps based on a single set of discrete classes. By providing a database consisting of nested maps and complementary layers, the NELC database provides a flexible framework that allows users to tailor maps to suit their needs. The methods used to create the database combine empirically derived climate–vegetation relationships with results from supervised classifications based on Moderate Resolution Imaging Spectroradiometer (MODIS) data. The hierarchical approach provides an effective framework for integrating climate–vegetation relationships with remote sensing-based classifications, and also allows sources of error to be characterized and attributed to specific levels in the hierarchy. The cross-validated accuracy was 73% for the land cover map and 73% and 91% for the agriculture and wetland classifications, respectively. These results support the use of hierarchical classification and climate–vegetation relationships for mapping land cover at continental scales.     

Land cover characterization of Temperate East Asia using multi-temporal VEGETATION sensor data

Temperate East Asia (TEA) is characterized by diverse land cover types, including forest and agricultural lands, one of the world's largest temperate grasslands, and extensive desert and barren landscapes. In this paper, we explored the potential of SPOT-4 VEGETATION (VGT) data for the classification of land cover types in TEA. An unsupervised classification was performed using multi-temporal (March–November 2000) VGT-derived spectral indices (Land Surface Water Index [LSWI] and Enhanced Vegetation Index [EVI]) to generate a land cover map of TEA (called VGT-TEA). Land cover classes from VGT-TEA were aggregated to broad, general class types, and then compared and validated with classifications derived from fine-resolution (Landsat) data. VGT-TEA produced reasonable results when compared to the Landsat products. Analysis of the seasonal dynamics of LSWI and EVI allows for the identification of distinct growth patterns between different vegetation types. We suggest that LSWI seasonal curves can be used to define the growing season for temperate deciduous vegetation, including grassland types. Seasonal curves of EVI tend to have a slightly greater dynamic range than LSWI during the peak growing season and can be useful in discriminating between vegetation types. By using these two complementary spectral indices, VGT data can be used to produce timely and detailed land cover and phenology maps with limited ancillary data needed.     

Assessment of deforestation in the Lower Amazon floodplain using historical Landsat MSS/TM imagery

The floodplain forests bordering the Amazon River have outstanding ecological, economic, and social importance for the region. However, the original distribution of these forests is not well known, since they have suffered severe degradation since the 16th century. The previously published vegetation map of the Amazon River floodplain (Hess et al., 2003), based on data acquired in 1996, shows enormous difference in vegetation cover classes between the regions upstream and downstream of the city of Manaus. The upper floodplain is mostly covered by forests, while the lower floodplain is predominantly occupied by grasses and shrubs.This study assesses deforestation in the Lower Amazon floodplain over a ~ 30 year period by producing and comparing a historical vegetation map based on MSS/Landsat images acquired in the late 1970s with a recent vegetation map produced from TM/Landsat images obtained in 2008. The maps were generated through the following steps: 1) normalization and mosaicking of images for each decade; 2) application of a linear mixing model transformation to produce vegetation, soil and shade fraction-images; and 3) object-oriented image analysis and classification. For both maps, the following classes were mapped: floodplain forest, non-forest floodplain vegetation, bare soil and open water. The two maps were combined using object-level Boolean operations to identify time transitions among the mapped classes, resulting in a map of the land cover change occurred over ~ 30 years. Ground information collected at 168 ground points was used to build confusion matrices and calculate Kappa indices of agreement. A survey strategy combining field observations and interviews allowed the collection of information about both recent and historical land cover for validation purposes. Kappa values (0.77, 0.75 and 0.75) indicated the good quality of the maps, and the error estimates were used to adjust the estimated deforested area to a value of 3457 km² ± 1062 km² (95% CI) of floodplain deforestation over the ~ 30 years.     

Incorporating domain knowledge and spatial relationships into land cover classifications: a rule‐based approach

For some tropical regions, remote sensing of land cover yields unacceptable results, particularly as the number of land cover classes increases. This research explores the utility of incorporating domain knowledge and multiple algorithms into land cover classifications via a rule‐based algorithm for a series of satellite images. The proposed technique integrates the fundamental, knowledge‐based interpretation elements of remote sensing without sacrificing the ease and consistency of automated, algorithm‐based processing. Compared with results from a traditional maximum likelihood algorithm, classification accuracy was improved substantially for each of the six land cover classes and all three years in the image series. Use of domain knowledge proved effective in accurately classifying problematic tropical land covers, such as tropical deciduous forest and seasonal wetlands. Results also suggest that ancillary data may be most useful in the classification of historic images, where the greatest improvement was observed relative to results from maximum likelihood. The cost of incorporating contextual knowledge and extensive spatial data sets may be justified, since results from the proposed technique suggest a considerable improvement in accuracy may be achieved.     

Land-use survey of Idukki District

The growing population pressure and limited availability of land necessitates proper utilization of the available land through scientific land-use planning. Realizing this fact, a project using multiband aerial photography was taken up with following objectives: identification and mapping of areas under various land uses, study of geomorphology, identification of potential groundwater areas and identification of areas requiring soil conservation and reclamation measures. The major land-use classes mapped are agriculture, forest, wasteland, water bodies and public use. These are further subclassified. Some of the important observations are the reduction in the area under forest and the existence of a large area under wasteland. Geomorphological maps prepared depict slope, relief, drainage and vegetation. With the help of structural and geomorphological maps, the technique of identifying potential groundwater areas has been demonstrated. The land-use maps have a classification accuracy of 86 per cent whereas the accuracy of slopes in geomorphological maps is 72 per cent. The locational accuracy of the land features is 50-150 m on the ground. Visual interpretation of LANDSAT images led to the identification of nine land-cover classes while digital analysis enabled identification of twelve land-cover classes. Significant changes in land cover over a decade (1973-1982) have been noticed. This information has been used for preparation of a detailed land-use plan for the Idukki District.     

Integrating LIDAR data and multispectral imagery for enhanced classification of rangeland vegetation: A meta analysis

This study compared the suitability of LIDAR (LIght Detection And Ranging) data, three-band multispectral data, and LIDAR data integrated with multispectral information, for classifying spatially complex vegetation in the Aspen Parkland of western Canada. Classifications were performed for both a) general vegetation classes limited to three major formations of deciduous forest, shrubland and grassland, and b) eight detailed vegetation classes including upland mixed prairie and fescue grasslands, closed and semi-open aspen forests, western snowberry and silverberry shrublands, and fresh and saline riparian (lowland) meadows. A Digital Elevation Model (DEM) and Surface Elevation Model (SEM) developed from LIDAR data incorporated both topographic and biological biases in community positioning across the landscape. Using multispectral data, the original digital image mosaic, its hybrid color composite, and an intensity-hue-saturation (IHS) image were each tested. Final vegetation classification was done through integration of information from both digital images and LIDAR data to evaluate the improvement in classification accuracy. Among the land cover schedules with three and eight classes of vegetation, classification from the multispectral imagery, specifically the hybrid color composite image, had the highest accuracy, peaking at 74.6% and 59.4%, respectively. In contrast, the LIDAR classification schedules led to an average classification accuracy of 64.8% and 52.3%, respectively, for the general and detailed vegetation data. Subsequent integration of the LIDAR and digital image classification schedules resulted in accuracy improvements of 16 to 20%, resulting in a superior final accuracy of 91% and 80.3%, respectively, for the three and eight classes of vegetation. A final land cover map including 8 classes of vegetation, fresh and saline water, as well as bare ground, was created for the study area with an overall accuracy of 83.9%, highlighting the benefit of integrating LIDAR and multispectral imagery for enhanced vegetation classification in heterogenous rangeland environments.     

Land cover discrimination from multitemporal ERS images and multispectral Landsat images: A study case in an agricultural area in France

More and more remote sensing data corresponding to various wavelength domains is becoming available. Visible/infrared data were first used for land cover classification. However, radar data are becoming more widely used for hydrological and agricultural applications. This paper discusses the performance, for land cover type discrimination, of an optical image acquisition and a multitemporal radar series. For the majority of land cover types existing within the test site (representative of northern European agricultural areas), both ERS multitemporal SAR and Landsat multispectral visible/infrared classifications lead to good results, with the latter being more robust. For better identification of cultures that are less represented, the complementarity of the two datasets may be exploited using an efficient data fusion algorithm based on the Dempster-Shafer evidence theory. The performance of this combination was verified on two successive vegetation cycles.     

Optimization at multi-spectral land cover classification with fuzzy clustering and the Kohonen feature map

The number and structure of land cover classes separatable in a region on the basis of multi-spectral satellite images are usually known. The method presented here is capable of combining information from the spectral and nonspectral channels of multi-spectral images in a way that offers a possibility to cover all spectrally separable classes. The method also enables the determination of the representativeness of the available training areas with respect to these classes. In the course of the procedure, 5-35 classes are iteratively determined in succession for different channel combinations with the Kohonen feature map and the fuzzy c-means clustering algorithm. The maximum Jeffries-Matusita (JM) distance between these classes indicates the optimum class number of unambiguously separable classes for the entire multi-spectral image. In a simple step, the available training areas are grouped by the fuzzy c-means clustering algorithm. In this case, the maximum JM distance indicates the class number that ensures optimum separation of training areas. If this class number is smaller than the optimum class number determined for the entire scene, then further training areas must be defined in order to improve the results of a subsequent supervised land cover classification. The method's efficiency is demonstrated by the example of land cover classification for the region around Perpignan.     

Coastal wetland vegetation classification with a Landsat Thematic Mapper image

Coastal wetland vegetation classification with remotely sensed data has attracted increased attention but remains a challenge. This paper explored a hybrid approach on a Landsat Thematic Mapper (TM) image for classifying coastal wetland vegetation classes. Linear spectral mixture analysis was used to unmix the TM image into four fraction images, which were used for classifying major land covers with a thresholding technique. The spectral signatures of each land cover were extracted separately and then classified into clusters with the unsupervised classification method. Expert rules were finally used to modify the classified image. This research indicates that the hybrid approach employing sub-pixel information, an analyst's knowledge and characteristics of coastal wetland vegetation distribution shows promise in successfully distinguishing coastal vegetation classes, which are difficult to separate with a maximum likelihood classifier (MLC). The hybrid method provides significantly better classification results than MLC.     

Efficient parallel algorithm for pixel classification in remote sensing imagery

An important approach for image classification is the clustering of pixels in the spectral domain. Fast detection of different land cover regions or clusters of arbitrarily varying shapes and sizes in satellite images presents a challenging task. In this article, an efficient scalable parallel clustering technique of multi-spectral remote sensing imagery using a recently developed point symmetry-based distance norm is proposed. The proposed distributed computing time efficient point symmetry based K-Means technique is able to correctly identify presence of overlapping clusters of any arbitrary shape and size, whether they are intra-symmetrical or inter-symmetrical in nature. A Kd-tree based approximate nearest neighbor searching technique is used as a speedup strategy for computing the point symmetry based distance. Superiority of this new parallel implementation with the novel two-phase speedup strategy over existing parallel K-Means clustering algorithm, is demonstrated both quantitatively and in computing time, on two SPOT and Indian Remote Sensing satellite images, as even K-Means algorithm fails to detect the symmetry in clusters. Different land cover regions, classified by the algorithms for both images, are also compared with the available ground truth information. The statistical analysis is also performed to establish its significance to classify both satellite images and numeric remote sensing data sets, described in terms of feature vectors.     

Land cover classification with coarse spatial resolution data to derive continuous and discrete maps for complex regions

Regularly updated land cover information at continental or national scales is a requirement for various land management applications as well as biogeochemical and climate modeling exercises. However, monitoring or updating of map products with sufficient spatial detail is currently not widely practiced due to inadequate time-series coverage for most regions of the Earth. Classifications of coarser spatial resolution data can be automatically generated on an annual or finer time scale. However, discrete land cover classifications of such data cannot sufficiently quantify land surface heterogeneity or change. This study presents a methodology for continuous and discrete land cover mapping using moderate spatial resolution time series data sets. The method automatically selects sample data from higher spatial resolution maps and generates multiple decision trees. The leaves of decision trees are interpreted considering the sample distribution of all classes yielding class membership maps, which can be used as estimates for the diversity of classes in a coarse resolution cell. Results are demonstrated for the heterogeneous, small-patch landscape of Germany and the bio-climatically varying landscape of South Africa. Results have overall classification accuracies of 80%. A sensitivity analysis of individual modules of the classification process indicates the importance of appropriately chosen features, sample data balanced among classes, and an appropriate method to combine individual classifications. The comparison of classification results over several years not only indicates the method's consistency, but also its potential to detect land cover changes.     

Land-use land-cover classification analysis of Giba catchment using hyper temporal MODIS NDVI satellite images

Landsat-based land-use land-cover (LULC) mapping studies were previously conducted in Giba catchment, comprising an area of 4019 km². No attempt has been done to map LULC of this catchment through the analysis of Moderate Resolution Imaging Spectroradiometer (MODIS) normalized difference vegetation index (NDVI) time-series data. This article is aimed to see whether time-series MODIS NDVI data set is applicable for LULC mapping of Giba catchment or not. MODIS NDVI data sets of the year 2010 were used for classification analysis. The original data were subjected to MODIS Reproduction Tool and stacking. The re-projected and stacked images were filtered using Harmonic Analysis of Time-Series filtering algorism to remove the effects of cloud and other noises. The MODIS NDVI data sets (16-day maximum value composite) were classified using the ISODATA clustering algorithm available under ERDAS IMAGINE software. A series of unsupervised classification runs were carried out with a pre-defined number of classes (5–24). From this classification, the optimal numbers of classes were determined to be eight after checking for average divergence analysis. The classification result became eight LULC classes namely: bare land, grass land, irrigated land, cultivated land, area closure, shrub land, bush land, and forest land with an overall accuracy of 87.7%. It was therefore concluded that MODIS NDVI time-series image is applicable for mapping large watersheds.     

Land cover update by supervised classification of segmented ASTER images

The revision of the 1995 land cover dataset for the Vale do Sousa region, in the northwest of Portugal, was carried out by supervised classification of a multi‐spectral image from the Advanced Spaceborne Thermal Emission and Reflectance Radiometer (ASTER) sensor. The nine reflective bands of ASTER were used, covering the spectral range from 0.52–2.43?µm. The image was initially ortho‐rectified and segmented into 51?186 objects, with an average object size of 135?pixels (about 3?ha). A total of 582 of these objects were identified for training nine land cover classes. The image was classified using an algorithm based on a fuzzy classifier, Support Vector Machines (SVM), K Nearest Neighbours (K‐NN) and a Logistic Discrimination (LD) classifier. The results from the classification were evaluated using a set of 277 validation sites, independently gathered. The overall accuracy was 44.6% for the fuzzy classifier, 70.5% for the SVM, 60.9% for the K‐NN and 72.2% for the LD classifier. The difficulty in discriminating between some of the forest land cover classes was examined by separability analysis and unsupervised classification with hierarchical clustering. The forest classes were found to overlap in the multi‐spectral space defined by the nine ASTER bands used.