Land Cover Classification System Research Articles

Semantically based reclassification of Danish land-use and land-cover information

In recent years, the availability of georeferenced data has increased substantially, as have the number of producers and users of this information. As a consequence, there is a growing need for harmonization of data, not least in its classification descriptions. Unfortunately, inadequate metadata hampers understanding of how data sets are produced and what data classes represent. This study describes how five different categorical geodata sets for Denmark, ranging from habitat registrations through maps of agricultural land use to national topographic data, are integrated and how the integrated data set is reclassified to land-use and land-cover classes. All five data sets differ with respect to data acquisition, and description and classification methodologies, and none of the data distinguish between land use and land cover. The purpose of the reclassification was to produce maps of land use and land cover, with classes being compatible with the land cover classification system (LCCS) from the Food and Agriculture Organization of the United Nations. We identified land-cover and land-use classes from the LCCS that matched Danish conditions and cross-tabulated those classes with classes from the integrated Danish data set. Based on the semantic meaning of the class names from the integrated data set, we used heuristic associational knowledge to estimate their membership in the land-use and land-cover classes. The results are three land-use maps and five land-cover maps, indicating qualitative estimates of the presence of land-cover classes measured on an ordinal scale.

Use of various remote sensing land cover products for plant functional type mapping over Siberia

Abstract. High-latitude ecosystems play an important role in the global carbon cycle and in regulating the climate system and are presently undergoing rapid environmental change. Accurate land cover data sets are required to both document these changes as well as to provide land-surface information for benchmarking and initializing Earth system models. Earth system models also require specific land cover classification systems based on plant functional types (PFTs), rather than species or ecosystems, and so post-processing of existing land cover data is often required. This study compares over Siberia, multiple land cover data sets against one another and with auxiliary data to identify key uncertainties that contribute to variability in PFT classifications that would introduce errors in Earth system modeling. Land cover classification systems from GLC 2000, GlobCover 2005 and 2009, and MODIS collections 5 and 5.1 are first aggregated to a common legend, and then compared to high-resolution land cover classification systems, vegetation continuous fields (MODIS VCFs) and satellite-derived tree heights (to discriminate against sparse, shrub, and forest vegetation). The GlobCover data set, with a lower threshold for tree cover and taller tree heights and a better spatial resolution, tends to have better distributions of tree cover compared to high-resolution data. It has therefore been chosen to build new PFT maps for the ORCHIDEE land surface model at 1 km scale. Compared to the original PFT data set, the new PFT maps based on GlobCover 2005 and an updated cross-walking approach mainly differ in the characterization of forests and degree of tree cover. The partition of grasslands and bare soils now appears more realistic compared with ground truth data. This new vegetation map provides a framework for further development of new PFTs in the ORCHIDEE model like shrubs, lichens and mosses, to represent the water and carbon cycles in northern latitudes better. Updated land cover data sets are critical for improving and maintaining the relevance of Earth system models for assessing climate and human impacts on biogeochemistry and biophysics. The new PFT map at 5 km scale is available for download from the PANGAEA website at 10.1594/PANGAEA.810709.

Harmonization of the Land Cover Classification System (LCCS) with the General Habitat Categories (GHC) classification system

Monitoring land cover and habitat change is a key issue for conservation managers because of its potential negative impact on biodiversity. The Land Cover Classification System (LCCS) and the General Habitat Categories (GHC) System have been proposed by the remote sensing and ecological research community, respectively, for the classification of land covers and habitats across various scales. Linking the two systems can be a major step forward towards biodiversity monitoring using remote sensing. The translation between the two systems has proved to be challenging, largely because of differences in definitions and related difficulties in creating one-to-one relationships between the two systems. This paper proposes a system of rules for linking the two systems and additionally identifies requirements for site-specific contextual and environmental information to enable the translation. As an illustration, the LCCS classification of the Le Cesine protected area in Italy is used to show rules for translating the LCCS classes to GHCs. This study demonstrates the benefits of a translation system for biodiversity monitoring using remote sensing data but also shows that a successful translation is often depending on the degree of ecological knowledge of the habitats and its relationship with land cover and contextual information.

Extracting Urban Subzonal Land Uses through Morphological and Spatial Arrangement Analyses Using Geographic Data and Remotely Sensed Imagery

Land use (LU) information is of significant value for various urban studies and is needed for a wide variety of decision-making initiatives in the range of global, regional, and urban areas. However, it has been difficult to collect LU information in an efficient and cost-effective way in the past, which forces planners and engineers to use other information (e.g., population and employment) as proxy in their modeling practice. This paper intends to establish a hybrid geographic information and remote sensing (GI/RS) expert system to extract LU at a very detailed subzonal level. Three LU classification algorithms including fuzzy-decision tree (FDT), logistic-decision tree (LDT), and artificial neural network (ANN) are designed to classify urban subzones, dissemination blocks (DBs), the smallest census zone, into single LUs using very high resolution (VHR) aerial imagery and geographic vector data. A novel, hybrid pixel- and object-based land-cover classification system is developed to extract the information of parking lot, bare soil, and vegetation from aerial imagery. Morphological properties at the zonal level are derived from the geographic data and the land-cover classification results. Statistical analyses, such as scatter graph and nonparametric Kruskal-Wallis test, are used to examine the separability of each pair LUs with respect to the derived DB properties. Selected morphological properties are then used as either independent or input variables of the designed FDT, LDT, and ANN classification algorithms. FDT and LDT are used in a five-level decision tree system and ANN is used directly for LU recognition. The performances of the three designed classification systems are then compared through accuracy assessments; the logistic-decision tree has the best performance with an overall accuracy of 97.05%. In addition, spatial arrangement analysis is used to study the interrelationships of buildings within zones (DB) based on nearest neighbor and Gabriel graph analysis, which show a significant potential of extracting different LUs from mixed-LU zones.

Translating land cover/land use classifications to habitat taxonomies for landscape monitoring: a Mediterranean assessment

Periodic monitoring of biodiversity changes at a landscape scale constitutes a key issue for conservation managers. Earth observation (EO) data offer a potential solution, through direct or indirect mapping of species or habitats. Most national and international programs rely on the use of land cover (LC) and/or land use (LU) classification systems. Yet, these are not as clearly relatable to biodiversity in comparison to habitat classifications, and provide less scope for monitoring. While a conversion from LC/LU classification to habitat classification can be of great utility, differences in definitions and criteria have so far limited the establishment of a unified approach for such translation between these two classification systems. Focusing on five Mediterranean NATURA 2000 sites, this paper considers the scope for three of the most commonly used global LC/LU taxonomies—CORINE Land Cover, the Food and Agricultural Organisation (FAO) land cover classification system (LCCS) and the International Geosphere-Biosphere Programme to be translated to habitat taxonomies. Through both quantitative and expert knowledge based qualitative analysis of selected taxonomies, FAO-LCCS turns out to be the best candidate to cope with the complexity of habitat description and provides a framework for EO and in situ data integration for habitat mapping, reducing uncertainties and class overlaps and bridging the gap between LC/LU and habitats domains for landscape monitoring—a major issue for conservation. This study also highlights the need to modify the FAO-LCCS hierarchical class description process to permit the addition of attributes based on class-specific expert knowledge to select multi-temporal (seasonal) EO data and improve classification. An application of LC/LU to habitat mapping is provided for a coastal Natura 2000 site with high classification accuracy as a result.

Land Cover Map Delineation, for Agriculture Development, Case Study in North Sinai, Egypt Using SPOT4 Data and Geographic Information System

Land cover map for a part of North Sinai was produced using the FAO—Land Cover Classification System (LCCS) of 2004. The standard FAO classification scheme provides a standardized system of classification that can be used to analyze spatial and temporal land cover variability in the study area. This approach also has the advantage of facilitating the integration of Sinai land cover mapping products to be included with the regional and global land cover datasets. The total study area is 7450 km2 (1,773,842) feddans. The landscape classification was performed on SPOT4 data acquired in 2011 using combined multi-spectral bands of 20 meter spatial resolution. Geographic Information System (GIS) was used to edit the classification result in order to reach the maximum possible accuracy. GIS was also used to include all necessary information. The identified vegetative land cover classes of the study area are irrigated herbaceous crops, irrigated tree crops and rain fed tree crops. The non-vegetated land covers in the study area include: bare rock, bare soil, bare soil stony, bare soil very stony, bare soil salt crusts, loose and shifting sands and sand dunes. The water bodies were classified as artificial perennial water bodies (fish ponds and irrigated canals) and natural perennial water bodies as lakes (standing) and rivers (flowing). Artificial surfaces in the study area include linear and non-linear. The produced maps and the statistics of the different land covers are included in the following sub-sections.

Harmonizing and Combining Existing Land Cover/Land Use Datasets for Cropland Area Monitoring at the African Continental Scale

Mapping cropland areas is of great interest in diverse fields, from crop monitoring to climate change and food security. Recognizing the value of a reliable and harmonized crop mask that entirely covers the African continent, the objectives of this study were to (i) consolidate the best existing land cover/land use datasets, (ii) adapt the Land Cover Classification System (LCCS) for harmonization, (iii) assess the final product, and (iv) compare the final product with two existing datasets. Ten datasets were compared and combined through an expert-based approach in order to create the derived map of cropland areas at 250 m covering the whole of Africa. The resulting cropland mask was compared with two recent cropland extent maps at 1 km: one derived from MODIS and one derived from five existing products. The accuracy of the three products was assessed against a validation sample of 3,591 pixels of 1km regularly distributed over Africa and interpreted using high resolution images, which were collected using the Geo-Wiki tool. The comparison of the resulting crop mask with existing products shows that it has a greater agreement with the expert validation dataset, in particular for places where the cropland represents more than 30% of the area of the validation pixel.

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.

Mapping Congo Basin vegetation types from 300 m and 1 km multi-sensor time series for carbon stocks and forest areas estimation

Abstract. This study aims to contribute to the understanding of the Congo Basin forests by delivering a detailed map of vegetation types with an improved spatial discrimination and coherence for the whole Congo Basin region. A total of 20 land cover classes were described with the standardized Land Cover Classification System (LCCS) developed by the FAO. Based on a semi-automatic processing chain, the Congo Basin vegetation types map was produced by combining 19 months of observations from the Envisat MERIS full resolution products (300 m) and 8 yr of daily SPOT VEGETATION (VGT) reflectances (1 km). Four zones (north, south and two central) were delineated and processed separately according to their seasonal and cloud cover specificities. The discrimination between different vegetation types (e.g. forest and savannas) was significantly improved thanks to the MERIS sharp spatial resolution. A better discrimination was achieved in cloudy areas by taking advantage of the temporal consistency of the SPOT VGT observations. This resulted in a precise delineation of the spatial extent of the rural complex in the countries situated along the Atlantic coast. Based on this new map, more accurate estimates of the surface areas of forest types were produced for each country of the Congo Basin. Carbon stocks of the Basin were evaluated to a total of 49 360 million metric tons. The regional scale of the map was an opportunity to investigate what could be an appropriate tree cover threshold for a forest class definition in the Congo Basin countries. A 30% tree cover threshold was suggested. Furthermore, the phenology of the different vegetation types was illustrated systematically with EVI temporal profiles. This Congo Basin forest types map reached a satisfactory overall accuracy of 71.5% and even 78.9% when some classes are aggregated. The values of the Cohen's kappa coefficient, respectively 0.64 and 0.76 indicates a result significantly better than random.

Land Cover Classification Using IRS-1D Data and a Decision Tree Classifier

Land cover is one of basic data layers in geographic information system for physical planning and environmentalmonitoring. Digital image classification is generally performed to produce land cover maps from remote sensing data,particularly for large areas. In the present study the multispectral image from IRS LISS-III image along with ancillary datasuch as vegetation indices, principal component analysis and digital elevation layers, have been used to perform imageclassification using maximum likelihood classifier and decision tree method. The selected study area that is located innorth-east Iran represents a wide range of physiographical and environmental phenomena. In this study, based on LandCover Classification System (LCCS), seven land cover classes were defined. Comparison of the results using statisticaltechniques showed that while supervised classification (i.e. MLC) produces an overall accuracy of about 72%; thedecision tree method, which improves the classification accuracy, can increase the results by about 7 percent to 79%. Theresults illustrated that ancillary data, especially vegetation indices and DEM, are able to improve significantlyclassification accuracy in arid and semi arid regions, and also the mountainous or hilly areas.

Accuracy assessments of the GLOBCOVER dataset using global statistical inventories and FLUXNET site data

The spatio-temporal distribution of land cover provides fundamental data for global climate and environmental change research. In recent decades, five global land cover maps have been produced based on remote sensing data sources and methodologies. Related research have shown that the availability and quality of the first four global land cover datasets are poor at the regional or the continental scale for a variety of reasons. There is still no consensus on the accuracy of the latest global land cover map. Based on comparison of the land cover dataset with the statistical cropland data from FAO and the FLUXNET site data, this paper discusses the accuracy of the fifth global land cover map, namely, the GLOBCOVER dataset, at different spatial scales. At the global scale, the cropland area obtained from the GLOBCOVER dataset is greater than that of the FAO statistical data by 47.06–84.49%, and the land cover types of the GLOBCOVER dataset have a 65.02% consistency with that of the FLUXNET site data. At the continental scale, the difference between cropland areas obtained from the GLOBCOVER dataset and the statistical cropland area vary from −43.42% to 502.36%; continents that have a more accurate cropland area compared to the FAO statistical data tend to be less consistent with the FLUXNET site data. In general, North America has a higher accuracy and Oceania has a lower accuracy. At the country scale, the accuracy estimates vary sharply over a wide range: between −100.00% and 190670.37%. It is recommended that future studies should pay careful attention to the data validation step before using the GLOBCOVER dataset for any particular problem. Future studies are also required for the development of a universal land cover classification system and advanced algorithms for remote sensing classification of global land cover maps.

Conventional and fuzzy comparisons of large scale land cover products: Application to CORINE, GLC2000, MODIS and GlobCover in Europe

One of the major drawbacks of land cover products is the lack of interoperability among them. Since their development was driven by different national or international initiatives, they were developed for different purposes and hold diverse technical characteristics. Thus, comparison among products and quality monitoring is necessary in assessing their usefulness. This paper provides a methodology to compare global land cover maps that allows for differences in legend definitions among products.Two different approaches were considered for map comparison, a Boolean approach and a new methodology based on fuzzy set theory in which the Land Cover Classification System (LCCS) acted as a general bridging system. The fuzzy approach let us reconcile legends in terms of a set of nine selected attributes. The methodology is illustrated over Europe using four different land cover products: CORINE, GLC2000, MODIS land cover (MODISLC) and GlobCover.Overall accuracies between datasets based on a generalized nomenclature ranged from 35% (GlobCover–MODISLC) to 57% (CORINE–GLC2000). A further assessment based on a flexible Boolean comparison allowed us to minimize part of the uncertainty introduced by ambiguity of legends, resulting in an overall increase of agreement of around 10% in absolute terms. Reduction of positional errors produced an agreement increase of around 5%. The fuzzy approach allowed us to examine certain criteria used in classification systems to better understand the causes of discrepancies among datasets and express them in terms of relevant land attributes, irrespective of class names. This method served to highlight the consistency of the land cover maps under consideration, especially for datasets that showed higher divergences. In particular, it allows the identification of areas that are very close in land attributes, even when a Boolean comparison shows them as different.

A methodology to generate a synergetic land-cover map by fusion of different land-cover products

The main goal of this study is to develop a general framework for building a hybrid land-cover map by the synergistic combination of a number of land-cover classifications with different legends and spatial resolutions. The proposed approach assesses class-specific accuracies of datasets and establishes affinity between thematic legends using a common land-cover language such as the UN Land-Cover Classification System (LCCS). The approach is illustrated over a large region in Europe using four land-cover datasets (CORINE, GLC2000, MODIS and GlobCover), but it can be applied to any set of existing products. The multi-classification map is expected to improve the performance of individual classifications by reconciling their best characteristics while avoiding their main weaknesses. The intermap comparison reveals improved agreement of the hybrid map with all other land-cover products and therefore indicates the successful exploration of synergies between the different products. The approach offers also estimates for the classification confidence associated with the pixel label and flexibility to shift the balance between commission and omission errors, which are critical in order to obtain a desired reliable map.

A comparative analysis of ALOS PALSAR L-band and RADARSAT-2 C-band data for land-cover classification in a tropical moist region

This paper explores the use of ALOS (Advanced Land Observing Satellite) PALSARL-band (Phased Array type L-band Synthetic Aperture Radar) and RADARSAT-2 C-band data for land-cover classification in a tropical moist region. Transformed divergence was used to identify potential textural images which were calculated with the gray-level co-occurrence matrix method. The standard deviation of selected textural images and correlation coefficients between them were then used to determine the best combination of texture images for land-cover classification. Classification results based on different scenarios with maximum likelihood classifier were compared. Based on the identified best scenarios, different classification algorithms – maximum likelihood classifier, classification tree analysis, Fuzzy ARTMAP (a neural-network method), k-nearest neighbor, object-based classification, and support vector machine were compared for examining which algorithm was suitable for land-cover classification in the tropical moist region. This research indicates that the combination of radiometric images and their textures provided considerably better classification accuracies than individual datasets. The L-band data provided much better land-cover classification than C-band data but neither L-band nor C-band was suitable for fine land-cover classification system, no matter which classification algorithm was used. L-band data provided reasonably good classification accuracies for coarse land-cover classification system such as forest, succession, agropasture, water, wetland, and urban with an overall classification accuracy of 72.2%, but C-band data provided only 54.7%. Compared to the maximum likelihood classifier, both classification tree analysis and Fuzzy ARTMAP provided better performances, object-based classification and support vector machine had similar performances, and k-nearest neighbor performed poorly. More research should address the use of multitemporal radar data and the integration of radar and optical sensor data for improving land-cover classification.

Evaluation of the variation in semantic contents of class sets on modelling dynamics of land-use changes

Understanding the scale of interaction and the scale of different environmental and social processes is of paramount importance to define and explain the interaction of human–environment systems. There are three dimensions of scale: space, time and the organisational hierarchy as constructed by the observer. The latter is synonymous with the variation in semantic contents of data expressed as differences in categorisation. This dimension of scale has received little attention. In this article the relationship between the semantic contents of data and modelling dynamics is explored using two land-cover data sets for Romania, one based on the Land-Cover Classification System and the other as used in the EURURALIS study. Three levels of semantic contents of the LCCS data and the single semantic level present in the EURURALIS data are used to establish empirical relations between the land-cover class and its explaining factors. The analysis results show that the variations in semantic contents of data within one data set and between two data sets lead to different sets of spatial determinants for land cover. We did not recognise patterns when establishing the organisational hierarchy. Future policy and decision-making depend to a great extent on which organisational hierarchy is present in the data set used to formulate a policy or to make an informed decision. This would mean that if the same results would be found in other data sets using different models not only multi-scale but also multi-semantic analysis are needed in order to make meaningful predictions of spatially explicit land change.

Large-scale land cover mapping with the integration of multi-source information based on the Dempster–Shafer theory

Land cover type is a crucial parameter that is required for various land surface models that simulate water and carbon cycles, ecosystem dynamics, and climate change. Many land use/land cover maps used in recent years have been derived from field investigations and remote-sensing observations. However, no land cover map that is derived from a single source (such as satellite observation) properly meets the needs of land surface simulation in China. This article presents a decision-fuse method to produce a higher-accuracy land cover map by combining multi-source local data based on the Dempster–Shafer (D–S) evidence theory. A practical evidence generation scheme was used to integrate multi-source land cover classification information. The basic probability values of the input data were obtained from literature reviews and expert knowledge. A Multi-source Integrated Chinese Land Cover (MICLCover) map was generated by combining multi-source land cover/land use classification maps including a 1:1,000,000 vegetation map, a 1:100,000 land use map for the year 2000, a 1:1,000,000 swamp-wetland map, a glacier map, and a Moderate-Resolution Imaging Spectroradiometer land cover map for China in 2001 (MODIS2001). The merit of this new map is that it uses a common classification system (the International Geosphere-Biosphere Programme (IGBP) land cover classification system), and it has a unified 1 km resolution. The accuracy of the new map was validated by a hybrid procedure. The validation results show great improvement in accuracy for the MICLCover map. The local-scale visual comparison validations for three regions show that the MICLCover map provides more spatial details on land cover at the local scale compared with other popular land cover products. The improvement in accuracy is true for all classes but particularly for cropland, urban, glacier, wetland, and water body classes. Validation by comparison with the China Forestry Scientific Data Center (CFSDC)–Forest Inventory Data (FID) data shows that overall forest accuracies in five provinces increased to between 42.19% and 88.65% for our MICLCover map, while those of the MODIS2001 map increased between 27.77% and 77.89%. The validation all over China shows that the overall accuracy of the MICLCover map is 71%, which is higher than the accuracies of other land cover maps. This map therefore can be used as an important input for land surface models of China. It has the potential to improve the modeling accuracy of land surface processes as well as to support other aspects of scientific land surface investigations in China.

Development of a modified neural network-based land cover classification system using automated data selector and multiresolution remotely sensed data

Integrating multiple images with artificial neural networks (ANN) improves classification accuracy. ANN performance is sensitive to training datasets. Complexity and errors compound when merging multiple data, pointing to needs for new techniques. Kohonen's self-organizing mapping (KSOM) neural network was adapted as an automated data selector (ADS) to replace manual training data processes. The multilayer perceptron (MLP) network was then trained using automatically extracted datasets and used for classification. Two hypotheses were tested: ADS adapted from the KSOM network provides adequate and reliable training datasets, improving MLP classification performance; and fusion of Landsat thematic mapper (TM) and SPOT images using the modified ANN approach increases accuracy. ADS adapted from the KSOM network improved training data quality and increased classification accuracy and efficiency. Fusion of compatible multiple data can improve performance if appropriate training datasets are collected. This proved to be a viable classification scheme particularly where acquiring sufficient and reliable training datasets is difficult.

Assessing effects of temporal compositing and varying observation periods for large-area land-cover mapping in semi-arid ecosystems: Implications for global monitoring

Land-cover is an important parameter in analyzing the state and dynamics of natural and anthropogenic terrestrial ecosystems. Land-cover classes related to semi-arid savannas currently exhibit among the greatest uncertainties in available global land cover datasets. This study focuses on the Kalahari in northeastern Namibia and compares the effects of different composite lengths and observation periods with class-wise mapping accuracies derived from multi-temporal MODIS time series classifications to better understand and overcome quality gaps in mapping semi-arid land-cover types. We further assess the effects of precipitation patterns on mapping accuracy using Tropical Rainfall Measuring Mission (TRMM) observation data. Botanical field samples, translated into the UN Land Cover Classification System (LCCS), were used for training and validation. Different sets of composites (16-day to three-monthly) were generated from MODIS (MOD13Q1) data covering the sample period from 2004 to 2007. Land-cover classifications were performed cumulatively based on annual and inter-annual feature sets with the use of random forests. Woody vegetation proved to be more stable in terms of omission and commission errors compared to herbaceous vegetation types. Generally, mapping accuracy increases with increasing length of the observation period. Analyses of variance (ANOVA) verified that inter-annual classifications significantly improved class-wise mapping accuracies, and confirmed that monthly composites achieved the best accuracy scores for both annual and inter-annual classifications. Correlation analyses using piecewise linear models affirmed positive correlations between cumulative mapping accuracy and rainfall and indicated an influence of seasonality and environmental cues on the mapping accuracies. The consideration of the inter-seasonal variability of vegetation activity and phenology cycling in the classification process further increases the overall classification performance of savanna classes in large-area land-cover datasets. Implications for global monitoring frameworks are discussed based on a conceptual model of the relationship between observation period and accuracy.

Production of global land cover data – GLCNMO

Global land cover is one of the fundamental contents of Digital Earth. The Global Mapping project coordinated by the International Steering Committee for Global Mapping has produced a 1-km global land cover dataset – Global Land Cover by National Mapping Organizations. It has 20 land cover classes defined using the Land Cover Classification System. Of them, 14 classes were derived using supervised classification. The remaining six were classified independently: urban, tree open, mangrove, wetland, snow/ice, and water. Primary source data of this land cover mapping were eight periods of 16-day composite 7-band 1-km MODIS data of 2003. Training data for supervised classification were collected using Landsat images, MODIS NDVI seasonal change patterns, Google Earth, Virtual Earth, existing regional maps, and expert's comments. The overall accuracy is 76.5% and the overall accuracy with the weight of the mapped area coverage is 81.2%. The data are available from the Global Mapping project website (http://www.iscgm.org/). The MODIS data used, land cover training data, and a list of existing regional maps are also available from the CEReS website. This mapping attempt demonstrates that training/validation data accumulation from different mapping projects must be promoted to support future global land cover mapping.

Ecosystem mapping at the African continent scale using a hybrid clustering approach based on 1-km resolution multi-annual data from SPOT/VEGETATION

The goal of this study is to propose a new classification of African ecosystems based on an 8-year analysis of Normalized Difference Vegetation Index (NDVI) data sets from SPOT/VEGETATION. We develop two methods of classification. The first method is obtained from a k-nearest neighbour (k-NN) classifier, which represents a simple machine learning algorithm in pattern recognition. The second method is hybrid in that it combines k-NN clustering, hierarchical principles and the Fast Fourier Transform (FFT). The nomenclature of the two classifications relies on three levels of vegetation structural categories based on the Land Cover Classification System (LCCS). The two main outcomes are: (i) The delineation of the spatial distribution of ecosystems into five bioclimatic ecoregions at the African continental scale; (ii) Two ecosystem maps were made sequentially: an initial map with 92 ecosystems from the k-NN, plus a deduced hybrid classification with 73 classes, which better reflects the bio-geographical patterns. The inclusion of bioclimatic information and successive k-NN clustering elements helps to enhance the discrimination of ecosystems. Adopting this hybrid approach makes the ecosystem identification and labelling more flexible and more accurate in comparison to straightforward methods of classification. The validation of the hybrid classification, conducted by crossing-comparisons with validated continental maps, displayed a mapping accuracy of 54% to 61%.