Land cover classification

The importance of accurate and timely information describing the nature and extent of land resources and changes over time is increasing, especially in rapidly growing metropolitan areas. We have developed a methodology to map and monitor land cover change using multitemporal Landsat Thematic Mapper (TM) data in the seven-county Twin Cities Metropolitan Area of Minnesota for 1986Minnesota for , 1991Minnesota for , 1998Minnesota for , and 2002. The overall seven-class classification accuracies averaged 94% for the four years. The overall accuracy of land cover change maps, generated from post-classification change detection methods and evaluated using several approaches, ranged from 80% to 90%. The maps showed that between 1986 and 2002 the amount of urban or developed land increased from 23.7% to 32.8% of the total area, while rural cover types of agriculture, forest and wetland decreased from 69.6% to 60.5%. The results quantify the land cover change patterns in the metropolitan area and demonstrate the potential of multitemporal Landsat data to provide an accurate, economical means to map and analyze changes in land cover over time that can be used as inputs to land management and policy decisions. D

Before being used in scientific investigations and policy described. The objective of this article is to elucidate these fundamental structures. We describe the three basic com-decisions, thematic maps constructed from remotely sensed data should be subjected to a statistically rigorous accuponents of an accuracy assessment, the sampling design, the response design, and the estimation and analysis pro-racy assessment. The three basic components of an accuracy assessment are: 1) the sampling design used to select tocol, and we provide recommendations and general guidelines for a statistically rigorous assessment. the reference sample; 2) the response design used to obtain the reference land-cover classification for each sam-An accuracy assessment begins with the definition of the target population, which is the area or region repre-pling unit; and 3) the estimation and analysis procedures. We discuss options available for each of these composented by the land-cover map. The individual units or elements of this population are defined as pixels or poly-nents. A statistically rigorous assessment requires both a gons, depending on the map representation. A sample of probability sampling design and statistically consistent units is selected from this population for accuracy assessestimators of accuracy parameters, along with a response ment. Choosing the sampling unit and the sampling dedesign determined in accordance with features of the sign are two major decisions required when planning the mapping and classification process such as the land-cover sampling protocol. The reference or "true" classification classification scheme, minimum mapping unit, and spais obtained for each sampling unit based on interpreting tial scale of the mapping. ©Elsevier Science Inc., 1998 aerial photography or videography, a ground visit, or a combination of these sources. The methods used to determine this reference classification are called the "re-INTRODUCTION sponse design." The response design includes procedures Land-cover maps are used in numerous natural resource to collect information pertaining to the reference landapplications to describe the spatial distribution and pattern cover determination, and rules for assigning one or more of land-cover, to estimate areal extent of various cover reference classifications to each sampling unit. The landclasses, or as input into habitat suitability models, landcover classifications from the map are compared to the cover change analyses, hydrological models, and risk analyreference classifications, and the extent to which these ses. Accuracy assessment quantifies data quality so that two classifications agree is defined as map accuracy. map users may evaluate the utility of a thematic map for their intended applications. Despite the widespread ac-SELECTING THE REFERENCE SAMPLE ceptance of accuracy assessment and the numerous articles published on this topic, the basic structures of a sta-In this section, we discuss details of the sampling design tistically rigorous accuracy assessment have not been fully component of the assessment. The sampling design is the protocol by which the reference sample units are selected. A probability sampling design is a key element of a statisti-

A new algorithm for exploiting the nonlinear structure of hyperspectral imagery is developed and compared against the de facto standard of linear mixing. This new approach seeks a manifold coordinate system that preserves geodesic distances in the high-dimensional hyperspectral data space. Algorithms for deriving manifold coordinates, such as isometric mapping (ISOMAP), have been developed for other applications. ISOMAP guarantees a globally optimal solution, but is computationally practical only for small datasets because of computational and memory requirements. Here, we develop a hybrid technique to circumvent ISOMAP's computational cost. We divide the scene into a set of smaller tiles. The manifolds derived from the individual tiles are then aligned and stitched together to recomplete the scene. Several alignment methods are discussed. This hybrid approach exploits the fact that ISOMAP guarantees a globally optimal solution for each tile and the presumed similarity of the manifold structures derived from different tiles. Using land-cover classification of hyperspectral imagery in the Virginia Coast Reserve as a test case, we show that the new manifold representation provides better separation of spectrally similar classes than one of the standard linear mixing models. Additionally, we demonstrate that this technique provides a natural data compression scheme, which dramatically reduces the number of components needed to model hyperspectral data when compared with traditional methods such as the minimum noise fraction transform.

The Moderate Resolution Imaging Radiometer (MODIS) is the primary instrument in the NASA Earth Observing System for monitoring the seasonality of global terrestrial vegetation. Estimates of 8-day mean daily gross primary production (GPP) at the 1 km spatial resolution are now operationally produced by the MODIS Land Science Team for the global terrestrial surface using a production efficiency approach. In this study, the 2001 MODIS GPP product was compared with scaled GPP estimates (25 km 2 ) based on ground measurements at two forested sites. The ground-based GPP scaling approach relied on a carbon cycle process model run in a spatially distributed mode. Land cover classification and maximum annual leaf area index, as derived from Landsat ETM+ imagery, were used in model initiation. The model was driven by daily meteorological observations from an eddy covariance flux tower situated at the center of each site. Model simulated GPPs were corroborated with daily GPP estimates from the flux tower. At the hardwood forest site, the MODIS GPP phenology started earlier than was indicated by the scaled GPP, and the summertime GPP from MODIS was generally lower than the scaled GPP values. The fall-off in production at the end of the growing season was similar to the validation data. At the boreal forest site, the GPP phenologies generally agreed because both responded to the strong signal associated with minimum temperature. The midsummer MODIS GPP there was generally higher than the ground-based GPP. The differences between the MODIS GPP products and the ground-based GPPs were driven by differences in the timing of FPAR and the magnitude of light use efficiency as well as by differences in other inputs to the MODIS GPP algorithm-daily incident PAR, minimum temperature, and vapor pressure deficit. Ground-based scaling of GPP has the potential to improve the parameterization of light use efficiency in satellite-based GPP monitoring algorithms. D

A hybrid segmentation procedure to integrate contex-compared to traditional per-pixel maximum likelihood classification results. ©Elsevier Science Inc., 2000 tual information with per-pixel classification in a metropolitan area land cover classification project is described and evaluated. It is presented as a flexible tool within a

Fractional vegetation cover ( ) is needed in the modeling of the land-atmosphere exchanges of momentum, energy, water, and trace gases. From global 1-km, 10-day composite Advanced Very High Resolution Radiometer normalized difference vegetation index (NDVI) data from April 1992 to March 1993, global 1-km is derived based on the annual maximum NDVI value for each pixel in comparison with the NDVI value that corresponds to 100% vegetation cover for each International Geosphere-Biosphere Program land cover type. This dataset is pixel dependent but season independent, with the seasonal variation of vegetation greenness in a pixel accounted for by the leaf area index. The authors' algorithm is found to be insensitive to the use of a specific land cover classification. In comparison with an independent dataset derived by DeFries et al. by using a more sophisticated statistical approach, the current dataset has a similar spatial distribution but systematically smaller (particularly over shrublands and barren land cover). It also gives values that overall are consistent with those derived from higher-resolution aircraft and satellite data over Arizona and field-survey data over Germany.

In recent years, decision tree classifiers have been successfully used for land cover classification from remote sensing data. Their implementation as a per-pixel based classifier to produce hard or crisp classification has been reported in the literature. Remote sensing images, particularly at coarse spatial resolutions, are contaminated with mixed pixels that contain more than one class on the ground. The per-pixel approach may result in erroneous classification of images dominated by mixed pixels. Therefore, soft classification approaches that decompose the pixel into its class constituents in the form of class proportions have been advocated. In this paper, we employ a decision tree regression approach to determine class proportions within a pixel so as to produce soft classification from remote sensing data. Classification accuracy achieved by decision tree regression is compared with those achieved by the most widely used maximum likelihood classifier, implemented in the soft mode, and a supervised version of the fuzzy cmeans classifier. Root Mean Square Error (RMSE) and fuzzy error matrix based measures have been used for accuracy assessment of soft classification. D

The MODIS sensor to be launched on the EOS-AM ucts, there should be standardization of error analysis and metadata reporting. ©Elsevier Science Inc., 1999 platform will be the most important sensor for global vegetation mapping. Among the programmatic goals for the MODIS sensor are to assess and track changes in land use/landcover, leaf area index (LAI), and net pri-*Institute for Tropical Ecosystem Studies, University of Puerto Rico, San Juan et al., 1996), but the EOS program initiates a new level †Department of Forest Resources, University of Minnesota, St. of detail, both spatial and spectral, in the derivation of Paul global metrics currently obtained from AVHRR imagery ‡Forestry Sciences Laboratory, USDA Forest Service, Corvallis, Oregon (Barron et al., 1995). indicate the

The combination of multisource remote sensing and geographic data is believed to offer improved accuracies in land cover classification. For such classification, the conventional parametric statistical classifiers, which have been applied successfully in remote sensing for the last two decades, are not appropriate, since a convenient multivariate statistical model does not exist for the data. In this paper, several single and multiple classifiers, that are appropriate for the classification of multisource remote sensing and geographic data are considered. The focus is on multiple classifiers: bagging algorithms, boosting algorithms, and consensus-theoretic classifiers. These multiple classifiers have different characteristics. The performance of the algorithms in terms of accuracies is compared for two multisource remote sensing and geographic datasets. In the experiments, the multiple classifiers outperform the single classifiers in terms of overall accuracies.

As part of the activities of the Multi-Resolution Land Characteristics (MRLC) Interagency Consortium, an intermediate-scale land cover data set is being generated for the conterminous United States. This effort is being conducted on a region-by-region basis using U.S. Standard Federal Regions. To date, land cover data sets have been generated for Federal Regions 3 (Pennsylvania, West Virginia, Virginia, Maryland, and Delaware) and 2 (New York and New Jersey). Classification work is currently under way in Federal Region 4 (the southeastern United States), and land cover mapping activities have been started in Federal Regions 5 (the Great Lakes region) and 1 (New England). It is anticipated that a land cover data set for the conterminous United States will be completed by the end of 1999. A standard land cover classification legend is used, which is analogous to and compatible with other classification schemes. The primary MRLC regional classification scheme contains 23 land cover classes.

This paper examines the role of soil fertility and land-use history on the rates of forest successional regrowth in ®ve regions of the Amazon Basin. Sites are located in the Bragantina Region, Tome  Ac Ëu  Region, Altamira Region and Marajo  Region of the State of Para Â, Brazil and in a region of the Colombian Vaupe Âs. Methods used included vegetation inventories of successional and mature forests, soil sampling, ethnographic assessment of land-use histories, and land cover classi®cation based on multi-temporal Landsat Thematic Mapper digital satellite data. The paper examines inter-regional differences, intraregional differences, and Basin-wide differences in rates of forest regrowth. Inter-regional differences are best explained by the differences between areas in soil fertility, whereas intra-regional differences are best explained by the differential impact of land-use history on forest recovery. Basin-wide differences in rates of succession can best be captured by differences in tree height, and secondarily by differences in basal area. In inter-regional comparisons we found that ultisols, oxisols, and spodosols present similar rates of regrowth, but considerably slower rates when compared to al®sols. During the ®rst 5±10 years of regrowth, al®sol areas have average stand height 1 m higher, and this difference doubles after 15 years of regrowth. In intra-regional comparisons, using one region as an example, we found that land-use differences are most able to explain differences. Areas that had been in swidden agriculture grew back at a rate of 1.5 m per year, as compared with 0.45 m per year for areas that had experienced mechanized land preparation, and 0.62 m per year for areas that had been in pasture. In Basin-wide comparisons we have been able to discriminate three distinct stages of secondary succession using structural criteria, across both soil and land-use types with height as the most predictive criteria of overall structural development. By comparing our ®ve sites' data with that of other investigators, the paper proposes Basin-wide patterns of regrowth for each of three structural stages of secondary succession that facilitate linking the ®eld vegetation data to remotely-sensed data of land cover. #

An understanding of land use/land cover change at local, regional, and global scales is important in an increasingly human-dominated biosphere. Here, we report on an under-appreciated complexity in the analysis of land cover change important in arid and semi-arid environments. In these environments, some land cover types show a high degree of inter-annual variability in productivity. In this study, we show that ecosystems dominated by non-native cheatgrass (Bromus tectorum) show an inter-annual amplified response to rainfall distinct from native shrub/bunch grass in the Great Basin, US. This response is apparent in time series of Landsat and Advanced Very High Resolution Radiometer (AVHRR) that encompass enough time to include years with high and low rainfall. Based on areas showing a similar amplified response elsewhere in the Great Basin, 20,000 km 2 , or 7% of land cover, are currently dominated by cheatgrass. Inter-annual patterns, like the high variability seen in cheatgrass-dominated areas, should be considered for more accurate land cover classification. Land cover change science should be aware that high inter-annual variability is inherent in annual dominated ecosystems and does not necessarily correspond to active land cover change.

Accurate and timely information about land cover pattern and change in urban areas is crucial for urban land management decision-making, ecosystem monitoring and urban planning. This paper presents the methods and results of an object-based classification and post-classification change detection of multitemporal high-spatial resolution Emerge aerial imagery in the Gwynns Falls watershed from 1999 to 2004. The Gwynns Falls watershed includes portions of Baltimore City and Baltimore County, Maryland, USA. An object-based approach was first applied to implement the land cover classification separately for each of the two years. The overall accuracies of the classification maps of 1999 and 2004 were 92.3% and 93.7%, respectively. Following the classification, we conducted a comparison of two different land cover change detection methods: traditional (i.e., pixel-based) post-classification comparison and object-based post-classification comparison. The results from our analyses indicated that an objectbased approach provides a better means for change detection than a pixel based method because it provides an effective way to incorporate spatial information and expert knowledge into the change detection process. The overall accuracy of the change map produced by the object-based method was 90.0%, with Kappa statistic of 0.854, whereas the overall accuracy and Kappa statistic of that by the pixel-based method were 81.3% and 0.712, respectively. Sensors 2008, 8 1614

The MODIS land science team produces a number of standard products, including land cover and leaf area index (LAI). Critical to the success of MODIS and other sensor products is an independent evaluation of product quality. In that context, we describe a study using field data and Landsat ETM+ to map land cover and LAI at four 49-km 2 sites in North America containing agricultural cropland (AGRO), prairie grassland (KONZ), boreal needleleaf forest, and temperate mixed forest. The purpose was to: (1) develop accurate maps of land cover, based on the MODIS IGBP (International Geosphere -Biosphere Programme) land cover classification scheme; (2) derive continuous surfaces of LAI that capture the mean and variability of the LAI field measurements; and (3) conduct initial MODIS validation exercises to assess the quality of early (i.e., provisional) MODIS products. ETM+ land cover maps varied in overall accuracy from 81% to 95%. The boreal forest was the most spatially complex, had the greatest number of classes, and the lowest accuracy. The intensive agricultural cropland had the simplest spatial structure, the least number of classes, and the highest overall accuracy. At each site, mapped LAI patterns generally followed patterns of land cover across the site. Predicted versus observed LAI indicated a high degree of correspondence between field-based measures and ETM+ predictions of LAI. Direct comparisons of ETM+ land cover maps with Collection 3 MODIS cover maps revealed several important distinctions and similarities. One obvious difference was associated with image/map resolution. ETM+ captured much of the spatial complexity of land cover at the sites. In contrast, the relatively coarse resolution of MODIS did not allow for that level of spatial detail. Over the extent of all sites, the greatest difference was an overprediction by MODIS of evergreen needleleaf forest cover at the boreal forest site, which consisted largely of open shrubland, woody savanna, and savanna. At the agricultural, temperate mixed forest, and prairie grassland sites, ETM+ and MODIS cover estimates were similar. Collection 3 MODIS-based LAI estimates were considerably higher (up to 4 m 2 m À 2 ) than those based on ETM+ LAI at each site. There are numerous probable reasons for this, the most important being the algorithms' sensitivity to MODIS reflectance calibration, its use of a prelaunch AVHRR-based land cover map, and its apparent reliance on mainly red and near-IR reflectance. Samples of Collection 4 LAI products were examined and found to consist of significantly improved LAI predictions for KONZ, and to some extent for AGRO, but not for the other two sites. In this study, we demonstrate that MODIS reflectance data are highly correlated with LAI across three study sites, with relationships increasing in strength from 500 to 1000 m spatial resolution, when shortwave-infrared bands are included. D

Recent production of land surface anisotropy, diffuse bihemispherical (white-sky) albedo and direct beam directional hemispherical (blacksky) albedo from observations acquired by the MODIS instruments aboard NASA's Terra and Aqua satellite platforms have provided researchers with unprecedented spatial, spectral, and temporal information on the land surface's radiative characteristics. Cloud cover, which curtails retrievals, and the presence of ephemeral and seasonal snow limit the snow-free data to approximately half the global land surfaces on an annual equal-angle basis. This precludes the MOD43B3 albedo products from being used in some remote sensing and ground-based applications, climate models, and global change research projects.

A significant proportion of high spatial resolution imagery in urban areas can be affected by shadows. Considerable research has been conducted to investigate shadow detection and removal in remotely sensed imagery. Few studies, however, have evaluated how applications of these shadow detection and restoration methods can help eliminate the shadow problem in land cover classification of high spatial resolution images in urban settings. This paper presents a comparison study of three methods for land cover classification of shaded areas from high spatial resolution imagery in an urban environment. Method 1 combines spectral information in shaded areas with spatial information for shadow classification. Method 2 applies a shadow restoration technique, the linear-correlation correction method to create a "shadow-free" image before the classification. Method 3 uses multisource data fusion to aid in classification of shadows. The results indicated that Method 3 achieved the best accuracy, with overall accuracy of 88%. It provides a significantly better means for shadow classification than the other two methods. The overall accuracy for Method 1 was 81.5%, slightly but not significantly higher than the 80.5% from Method 2. All of the three methods applied an object-based classification procedure, which was critical as it provides an effective way to address the problems of radiometric difference and spatial misregistration associated with multisource data fusion (Method 3), and to incorporate thematic spatial information (Method 1).

Remotely sensed images and processing techniques are a primary tool for mapping changes in tropical forest types important to biodiversity and environmental assessment. Detailed land cover data are lacking for most wet tropical areas that present special challenges for data collection. For this study, we utilize decision tree (DT) classifiers to map 32 land cover types of varying ecological and economic importance over an 8000 km 2 study area and biological corridor in Costa Rica. We assess multivariate QUEST DTs with unbiased classification rules and linear discriminant node models for integrated vegetation mapping and change detection. Predictor variables essential to accurate land cover classification were selected using importance indices statistically derived with classification trees. A set of 35 variables from SRTM-DEM terrain variables, WorldClim grids, and Landsat TM bands were assessed.

The objective of this study was to explore the use of multi-temporal Landsat TM data from the same growing season for the classification of land cover types in the south-western portion of the Argentine Pampas. Investigations were made on how many dates are necessary to obtain an accurate classification and, given a fixed number of dates, which is the particular combination of dates that yield the best results. Additionally, the effect of using the NDVI instead of all the bands available on the classification accuracy and the use of a moving window filter over the classified image were tested. Scenes acquired in spring, early summer, late summer and early fall of the 1996-1997 growing season were used. Land cover information for the same period was collected from farms and ranches and this information was included in a GIS. Supervised classifications were performed using all the 15 possible ways to combine the four dates. At least two scenes are needed for a satisfactory classification. These scenes must embrace the shift between winter and summer crops (i.e. one spring and one summer image). Using the NDVI instead of Landsat TM bands 3, 4 and 5 increased the biological interpretability of the results but caused a decrease in accuracy.

Current studies of land cover change and landscape fragmentation rely predominantly on land cover classifications derived from remotely sensed images. However, limitations of traditional land cover classifications are numerous and well known. This research compares classification-based techniques (discrete data) to the use of vegetation indices (continuous data) for land cover modeling and analyses of landscape fragmentation for a study area in western Honduras. The study area typifies many regions of tropical developing countries, where a complex interaction of social and environmental factors have given rise to a dynamic mosaic of patches of reforestation and deforestation. Understanding the complex human dimensions of land use and land cover change in these parts of the world continues to present a challenge for researchers.

Urban and peri-urban environments are composed of a wide variety of materials, making land cover classification challenging. The objective of this research is to determine how effectively multi-season Landsat Enhanced Thematic Mapper Plus (ETM+) and single-season Advanced Land Observing Satellite (ALOS) Phased Array type L-band Synthetic Aperture Radar (PALSAR) data can be combined to map 17 land cover categories in the Greater Boston area of, Massachusetts, USA. The key goal of this work is to test the integration of radar and optical data. The contribution of different dimensions of input data to a random forest classifier was evaluated with map accuracy statistics. PALSAR data produced a 30.99% overall classification accuracy. Higher classification accuracy (72.24%) was achieved by adding texture variables derived from the PALSAR data. A September Landsat image produced a map accuracy of 77.96%. The inclusion of Landsat images from other three seasons increased map accuracy to 86.86% and Landsat derived texture variables further increased the map accuracy to 92.69%. The highest map accuracy (93.82%) was achieved by combining Landsat and PALSAR. Though combining PALSAR and Landsat only increased the overall accuracy by 1.1%, it was a statistically significant increase, whose magnitude was limited by the high accuracy already achieved with Landsat data. Moreover, confusion matrices and land cover maps indicated that most of this increase was from three urban land cover types (low density residential, high density residential, and commercial/industrial). The results demonstrate the value of combining multitemporal Landsat imagery, ALOS PALSAR data, and texture variables for land cover classification in urban and peri-urban environments.

Enhanced Vegetation Index (EVI) and Leaf Area Index (LAI) data. Results of the simulations reveal four main groups of time series similarity measures with different sensitivities: (i) D M an , D E , D P CA , D F k quantify the difference in time series values, (ii) D M ah accounts for temporal correlation and non-stationarity of variance, (iii) D CC measures the temporal correlation, and (iv) the Fourier based D F F T and D ξ show their specific sensitivity based on the selected Fourier components. The difference measures show relatively the highest sensitivity to amplitude effects, whereas the correlation based measures are highly sensitive to variations in timing and noise. The Fourier based measures, finally, depend highly on the signal to noise ratio and the balance between amplitude and phase dominance. The heterogeneity in sensitivity of each D stresses the importance of (i) understanding the time series characteristics before applying any classification of change detection approach and (ii) defining the variability one wants to identify/account for. This requires an understanding of the ecosystem dynamics and time series characteristics related to the baseline, amplitude, timing, noise and variability of the ecosystem time series. This is also illustrated in the quantitative comparison, where the different sensitivities of D for the NDVI, EVI, and LAI data relate specifically to the temporal characteristics of each data set. Additionally, the effect of noise and intra-and interclass variability is demonstrated in a case study based on land cover classification.

Global change issues are high on the current international political agenda. A variety of global protocols and conventions have been established aimed at mitigating global environmental risks. A system for monitoring, evaluation and compliance of these international agreements is needed, with each component requiring comprehensive analytical work based on consistent datasets. Consequently, scientists and policymakers have put faith in earth observation data for improved global analysis. Land cover provides in many aspects the foundation for environmental monitoring . Proceedings of the FAO/UNEP Expert Consultation on Strategies for Global Land Cover Mapping and Monitoring. FAO, Rome, Italy, 38 pp.]. Despite the significance of land cover as an environmental variable, our knowledge of land cover and its dynamics is poor [Foody, G.M., 2002. Status of land cover classification accuracy assessment. Rem. Sens. Environ. 80,[185][186][187][188][189][190][191][192][193][194][195][196][197][198][199][200][201]. This study compares four satellite derived 1 km land cover datasets freely available from the internet and in wide use among the scientific community. Our analysis shows that while these datasets have in many cases reasonable agreement at a global level in terms of total area and general spatial pattern, there is limited agreement on the spatial distribution of the individual land classes. If global datasets are used at a continental or regional level, agreement in many cases decreases significantly. Reasons for these differences are many-ranging from the classes and thresholds applied, time of data collection, sensor type, classification techniques, use of in situ data, etc., and make comparison difficult. Results of studies based on global land cover datasets are likely influenced by the dataset chosen. Scientists and policymakers should be made aware of the inherent limitations in using current global land cover datasets, and would be wise to utilise multiple datasets for comparison. #

Light detection and ranging (lidar) provides high-resolution vertical and horizontal spatial data and has become an important technology for generating digital elevation models (DEMs) and digital surface models (DSMs). The latest terrestrial lidar sensors record intensity and echo information for each pulse in addition to range. In this study, lidar height and intensity data were used to classify land cover using an object-oriented approach. The study area was selected based on the variety of land cover types present and consists of urban, mixed forest, and wetland-estuary coastal environments. Surfaces constructed from the lidar points included DSM, DEM, intensity, multiple echos, and normalized height. These surfaces were segmented and classified using object rule based classification. Ten classes were extracted from the lidar data, including saturated and non-saturated intertidal sediments, saturated or stressed and lush ground cover vegetation, low and tall deciduous and coniferous trees, roads and bare soil, bright-roofed structures, dark-roofed structures, and water. The accuracy of the classification was assessed using independent ground reference polygons interpreted from colour orthophotographs and intensity images. The average accuracy of the 10 classes was 94%, but improved to 98% when the classification results were aggregated into seven classes. The results indicate that accurate land cover maps can be generated from a single lidar survey using the derived surfaces, and that image object segmentation and rule-based classification techniques allow the exploitation of spectral and spatial attributes of the lidar data. Résumé. La haute définition des données spatiales altimétriques et horizontales fournies par lidar (détection et télémétrie par ondes lumineuses) fait de lui une technologie clé dans l'élaboration de modèles altimétriques (MAN) et surfaciques numériques (MSN). Les capteurs lidar terrestres de dernière heure enregistrent des données non seulement sur la distance, mais aussi sur l'intensité et l'écho radar de chaque impulsion radar. Au cours de cette étude les données lidar altimétriques et d'intensité ont été utilisées dans une approche centré-objet pour catégoriser les couvertures terrestres. En effet, la zone de l'étude a eté sélectionnée en raison de l'éventail des couvertures y présentes : milieu urbain, forêt mixte, et estuaire marécageux côtier. Les surfaces créées à partir des retours d'impulsions lidar comprenaient des MAN, des MSN et des surfaces d'intensité, d'échos multiples, et de hauteur normalisée. Ces surfaces ont été segmentées et catégorisées selon un schéma d'objets fondé sur des règles. Dix classes de couverture ont été tirées des données lidar, comprenant les sédiments intertidaux saturés et non-saturés; les tapis végataux saturés, agressés et luxuriants; les arbres conifères et décidus, bas et hauts; les chausées et le sol nu; les bâtiments aux toitures lumineuses et sombres; et l'eau. L'exactitude de ce classement a été évaluée en utilisant des polygones de référence au sol élaborés à base d'orthographies en couleur et d'images d'intensité. L'exactitude moyenne des dix classes étaient de 94 %, chiffre qui passait à 98 % si on consolidait les résultats en sept classes. Les résultats donnent à croire : qu'il est possible de cartographier la couverture terrestre à partir d'une seule levée lidar en employant les surfaces dérivées ; et que la segmentation de l'image en objets et les techniques de classification fondées sur des règles permettent d'exploiter les attributs spatial et spectral des données lidar. [Traduit par la Rédaction] Brennan and Webster 172 162

Remote sensing needs to clarify the strengths of different methods so they can be consistently applied in forest management and ecology. Both the use of phenological information in satellite imagery and the use of vegetation indices have independently improved classifications of north temperate forests. Combining these sources of information in change detection has been effective for land cover classifications at the continental scale based on Advanced Very High Resolution Radiometer (AVHRR) imagery. Our objective is to test if using vegetation indices and change analysis of multiseasonal imagery can also improve the classification accuracy of deciduous forests at the landscape scale. We used Landsat Thematic Mapper (TM) scenes that corresponded to Populus spp. leaf-on and Quercus spp. leaf-off (May), peak summer (August), Acer spp. peak color (September), Acer spp. and Populus spp. leaf-off (October). Input data files derived from the imagery were:

One of key inputs to hydrological modeling is the potential evapotranspiration, either from the interception (PET 0 ) or from the soil water of root zone (PET). The Shuttleworth-Wallace (S-W) model is developed for their estimation. In this parameterization, neither experimental measurement nor calibration is introduced. Based on IGBP land cover classification, the typical thresholds of vegetation parameters are drawn from the literature. The spatial and temporal variation of vegetation LAI is derived from the composite NOAA-AVHRR NDVI using the SiB2 method. The CRU database supplies with the required meteorological data. They are all publicly available. The developed S-W model is applicable at the global scale, particularly to the data-poor or ungauged large basins.

Geosciences Laser Altimeter System (GLAS) space LiDAR data are used to attribute a MODerate resolution Imaging Spectrometer (MODIS) 500 m land cover classification of a 10°latitude by 12°longitude study area in south-central Siberia. Timber volume estimates are generated for 16 forest classes, i.e., four forest cover types × four canopy density classes, across this 811,414 km 2 area and compared with a ground-based regional volume estimate. Two regional GLAS/MODIS timber volume products, one considering only those pulses falling on slopes ≤10°and one utilizing all GLAS pulses regardless of slope, are generated. Using a two-phase (GLAS-ground plot) sampling design, GLAS/MODIS volumes average 163.4 ± 11.8 m 3 /ha across all 16 forest classes based on GLAS pulses on slopes ≤ 10°and 171.9 ± 12.4 m 3 /ha considering GLAS shots on all slopes. The increase in regional GLAS volume per-hectare estimates as a function of increasing slope most likely illustrate the effects of vertical waveform expansion due to the convolution of topography with the forest canopy response. A comparable, independent, ground-based estimate is 146 m 3 /ha . Phytomass (live biomass) and carbon of Siberian forests. Biomass and Bioenergy, 14, 21-31], a difference of 11.9% and 17.7% for GLAS shots on slopes ≤10°and all GLAS shots regardless of slope, respectively. A ground-based estimate of total volume for the entire study area, 7.46 × 10 9 m 3 , is derived using Shepashenko et al.'s per-hectare volume estimate in conjunction with forest area derived from a 1990 forest map [Grasia, M.G. (ed.). (1990). Forest Map of USSR. Soyuzgiproleskhoz, Moscow, RU. Scale: 1:2,500,000]. The comparable GLAS/MODIS estimate is 7.38 × 10 9 m 3 , a difference of less than 1.1%. Results indicate that GLAS data can be used to attribute digital land cover maps to estimate forest resources over subcontinental areas encompassing hundreds of thousands of square kilometers.

The problem of classification of hyperspectral images containing mixed pixels is addressed. Hyperspectral imaging is a continuously growing area of remote sensing applications. The wide spectral range of such imagery, providing a very high spectral resolution, allows to detect and classify surfaces and chemical elements of the observed image. The main problem of hyperspectral data is the (relatively) low spatial resolution, which can vary from a few to tens of meters. Many factors make the spatial resolution one of the most expensive and hardest to improve in imaging systems. For classification, the major problem caused by low spatial resolution are the mixed pixels, i.e., parts of the image where more than one land cover map lie in the same pixel. In this work we propose a method to address the problem of mixed pixels and to obtain a finer spatial resolution of the land cover classification maps. The method exploits the advantages of both soft classification techniques and spectral unmixing algorithms, in order to determine the fractional abundances of the classes at a sub-pixel scale. Spatial regularization by Simulated Annealing is finally performed to spatially locate the obtained classes. Experiments carried out on synthetic real data sets show excellent results both from a qualitative and quantitative point of view.

We define "operational" here as producing information on a regular basis, and "spatial water resources monitoring systems" (SWRMS) as software that integrates observations into models to produce spatial estimates of current (and past) water resources distribution.

Numerous classifiers have been developed and different classifiers have their own characteristics. Controversial results often occurred depending on the landscape complexity of the study area and the data used. Therefore, this paper aims to find a suitable classifier for the tropical land cover classification. Five classifiers -minimum distance classifier (MDC), maximum likelihood classifier (MLC), fisher linear discriminant (FLD), extraction and classification of homogeneous objects (ECHO), and linear spectral mixture analysis (LSMA) -were tested using Landsat Thematic Mapper (TM) data in the Amazon basin using the same training sample data sets. Seven land cover classes -mature forest, advanced succession forest, initial secondary succession forest, pasture, agricultural lands, bare lands, and water -were classified. Overall classification accuracy and kappa analysis were calculated. The results indicate that LSMA and ECHO classifiers provided better classification accuracies than the MDC, MLC, and FLD in the moist tropical region. The overall accuracy of LSMA approach reaches 86% associated with 0.82 kappa coefficient.

In this study we evaluated changes in land cover and rainfall in the Upper Gilgel Abbay catchment in the Upper Blue Nile basin and how changes affected stream flow in terms of annual flow, high flows and low flows. Land cover change assessment was through classification analysis of remote sensing based land cover data while assessments on rainfall and stream flow data are by statistical analysis. Results of the supervised land cover classification analysis indicated that 50.9 % and 16.7 % of the catchment area was covered by forest in 1973 and 2001, respectively. This significant decrease in forest cover is mainly due to expansion of agricultural land. By use of a change detection procedure, three periods were identified for which changes in rainfall and stream flow were analyzed. Rainfall was analyzed at monthly base by use of the Mann-Kendall test statistic and results indicated a statistically significant, decreasing trend for most months of the year. However, for the wet season months of June, July and August rainfall has increased. In the period 1973-2005, the annual flow of the catchment decreased by 12.1 %. Low flow and high flow at daily base were analyzed by a low flow and a high flow index that is based on a 95 % and 5 % exceedance probability. Results of the low flow index indicated decreases of 18.1 % and 66.6 % for the periods 1982-2000 and 2001-2005 respectively. Results of high flows indicated an increase of 7.6 % and 46.6 % for the same periods. In this study it is concluded that over the period 1973-2005 stream flow has changed in the Gilgel Abbay catchment by changes in land cover and changes in rainfall.

Improvement in remote sensing techniques in spatial/spectral resolution strengthens their applicability for urban environmental study. Unfortunately, high spatial resolution imagery also increases internal variability in land cover units and can cause a 'salt-and-pepper' effect, resulting in decreased accuracy using pixel-based classification results. Region-based classification techniques, using an image object (IO) rather than a pixel as a classification unit, appear to hold promise as a method for overcoming this problem. Using IKONOS high spatial resolution imagery, we examined whether the IO technique could significantly improve classification accuracy compared to the pixel-based method when applied to urban land cover mapping in Tampa Bay, FL, USA. We further compared the performance of an artificial neural network (ANN) and a minimum distance classifier (MDC) in urban detailed land cover classification and evaluated whether the classification accuracy was affected by the number of extracted IO features. Our analysis methods included IKONOS image data calibration, data fusion with the pansharpening (PS) process, Hue-Intensity-Saturation (HIS) transferred indices and textural feature extraction, and feature selection using a stepwise discriminant analysis (SDA). The classification results were evaluated with visually interpreted data from high-resolution (0.3 m) digital aerial photographs. Our results indicate a statistically significant difference in classification accuracy between pixel-and object-based techniques; ANN outperforms MDC as an objectbased classifier; and the use of more features (27 vs. 9 features) increases the IO classification accuracy, although the increase is statistically significant for the MDC but not for the ANN.

The Boreal Ecosystem Atmosphere Study (BOREAS) was a large, multiyear internationally supported study designed to improve our understanding of the boreal forest biome and its interactions with the atmosphere, biosphere, and the carbon cycle in the face of global climate change. In the initial phase of this study (early 1990s), remote sensing played a key role by providing products needed for planning and modeling. During and after the main BOREAS field campaigns (1994 and 1996), innovative remote sensing approaches and analyses expanded our understanding of the boreal forest in four key areas: (1) definition of vegetation structure, (2) land-cover classification, (3) assessment of the carbon balance, and (4) links between surface properties, weather, and climate. In addition to six BOREAS special issues and over 500 journal papers, a principal legacy of BOREAS is its well-documented and publicly available database, which provides a lasting scientific resource and opportunity to further advance our understanding of this critical northern biome. D

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 Á 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.

Floods are a common feature in rapidly urbanizing Dhaka and its adjoining areas. Though Greater Dhaka experiences flood almost in every year, flood management policies are mostly based on structural options including flood walls, dykes, embankments etc. Many shortcomings of the existing flood management systems are reported in numerous literatures. The objective of this paper is to assess flood hazard in Greater Dhaka for the historical flood event of 1998 using Synthetic Aperture Radar (SAR) data with GIS data. Flood-affected frequency and flood depth calculated from the multi-date SAR imageries were used as hydrologic parameters. Elevation heights, land cover classification, geomorphic division and drainage network data generated from optical remote sensing and analogue maps were used through GIS approach. Using a ranking matrix in three dimensional multiplication mode, flood hazard was assessed. All possible combination of flood hazard maps was prepared using land-cover, geomorphology and elevation heights for flood-affected frequency and floodwater depth. Using two hazard maps which produced the highest congruence for flood frequency and flood depth, a new flood hazard map was developed by considering the interactive effect of flood-affected frequency and floodwater depth, simultaneously. This new hazard map can provide more safety for flood countermeasures because pixels belonging to higher hazard degrees were increased due to the

Retrieval of land-surface temperature (LST) using data from the METEOSAT Second Generation-1 (MSG) Spinning Enhanced Visible and Infrared Imager (SEVIRI) requires adequate estimates of land-surface emissivity (LSE). In this context, LSE maps for SEVIRI channels IR3.9, IR8.7, IR10.8, and IR12.0 were developed based on the vegetation cover method. A broadband LSE map (3-14 m) was also developed for estimating longwave surface fluxes that may prove to be useful in both energy balance and climate modeling studies. LSE is estimated from conventional static land-cover classifications, LSE spectral data for each land cover, and fractional vegetation cover (FVC) information. Both International Geosphere-Biosphere Program (IGBP) Data and Information System (DIS) and Moderate Resolution Imaging Spectrometer (MODIS) MOD12Q1 land-cover products were used to build the LSE maps. Data on LSE were obtained from the Johns Hopkins University and Jet Propulsion Laboratory spectral libraries included in the Advanced Spaceborne Thermal Emission and Reflection Radiometer spectral library, as well as from the MODIS University of California-Santa Barbara spectral library. FVC data for each pixel were derived based on the normalized differential vegetation index. Depending on land cover, the LSE errors for channels IR3.9 and IR8.7 spatially vary from 0 6% to 24% and 0 1% to 33%, respectively, whereas the broadband spectrum errors lie between 0 3% and 7%. In the case of channels IR10.8 and IR12.0, 73% of the land surfaces within the MSG disk present relative errors less than 1 5%, and almost all (26%) of the remaining areas have relative errors of 2 0%. Developed LSE maps provide a first estimate of the ranges of LSE in SEVIRI channels for each surface type, and obtained results may be used to assess the sensitivity of algorithms where an a priori knowledge of LSE is required.

The biodiversity of the Andean Chocó in western Ecuador and Colombia is threatened by anthropogenic changes in land cover. The main goal of this study was to contribute to conservation of 12 threatened species of amphibians at a cloud forest site in northwestern Ecuador, by identifying and proposing protection of critical areas. We used Geographic Information Systems (GIS) and remote sensing techniques to quantify land cover changes over 35 years and outline important areas for amphibian conservation. We performed a supervised classification of an IKONOS satellite image from 2011 and two aerial photographs from 1977 and 2000. The 2011 IKONOS satellite image classification was used to delineate areas important for conservation of threatened amphibians within a 200 m buffer around rivers and streams. The overall classification accuracy of the three images was ≥ 80%. Forest cover was reduced by 17% during the last 34 years. However, only 50% of the study area retained the initial (1977) forest cover, as land was cleared for farming and eventually reforested. Finally, using the 2011 IKONOS satellite image, we delineated areas of potential conservation interest that would benefit the long term survival of threatened amphibian species at the Ecuadorian cloud forest site studied.

A refined land cover classification for the arid Phoenix (Arizona, USA) metropolitan area and some simple modifications to the surface energetics were introduced in the fifth-generation PSU/NCAR mesoscale meteorological model MM5. The single urban category in the existing 24-category United States Geological Survey (USGS) land cover classification used in MM5 was divided into three classes to account for heterogeneity of urban land cover. Updated land cover data were derived from 1998 LANDSAT Thematic Mapper satellite images. The composition of the urban land use classes in terms of typical fractions of vegetation and man-made surfaces was determined from ground truth information allowing to vary moisture availability for evaporation by land cover class. Bulk approaches for characteristics of the urban surface energy budget such as heat storage, the production of anthropogenic heat and radiation trapping were introduced in MM5's Medium Range Forecast boundary layer scheme and slab land surface model. A 72-hour simulation was performed with MM5 on a 2 km x 2 km grid during the early summer of 1998. The new land cover classification had a significant impact on the turbulent heat fluxes and the evolution of the boundary layer and improved the capability of MM5 to simulate the daytime part of the diurnal temperature cycle in the urban area, while the nighttime near surface air temperatures were improved significantly by adding radiation trapping, heat storage and anthropogenic heating to the model.

The methodologies used by the Satellite Application Facility on Land Surface Analysis (Land SAF) for retrieving emissivity are presented here. In the first approach, i.e., the vegetation cover method (VCM), the land surface emissivity (EM) is computed for Spinning Enhanced Visible and Infrared Imager (SEVIRI) infrared channels and for the 3-to 14-µm range using information on the pixel fraction of vegetation cover (FVC). The VCM uses a lookup table, which takes into account the channel's spectral response function, and laboratory reflectance spectra for different materials. The accuracy of the VCM depends on the reliability of FVC and the land cover classification. The EM for SEVIRI split-window channels is primarily used as an internal product by Land SAF for land surface temperature (LST) estimations. However, sensitivity studies show that LST often fails to meet the required accuracy of 2 K over desert and semiarid regions, where the VCM is unable to model the EM spatial variability, which is mostly associated with soil composition. Moreover, it is also over such areas where the atmosphere is generally dry that the impact of EM uncertainties on LST is largest. A second approach to determine the EM for SEVIRI split-window channels is currently being tested. This methodology allows the simultaneous retrieval of LST and channel EMs with the assumption that the latter remain constant. The channel EMs are then averaged over a 22-day period to filter out the noise in the retrievals. A first analysis of the maps obtained for an area within Northern Africa shows spatial patterns with features also present in the surface albedo.