Land cover classification

The characterization of landscape structure in space or time is fundamental to infer ecological processes (Ingegnoli, 2002). Landscape pattern arrangements strongly influence forest ecological functioning and biodiversity, as an example landscape fragmentation can induce habitat degradation reducing forest species populations or limiting their recolonization. Such arrangements are spatially correlated and scale-dependent, therefore they have distinctive operational-scales at which they can

Analysis of land-cover change in the seasonal tropical forests of the Southern Yucatán, Mexico presents a number of significant challenges for the fine-scale land-cover information required of land-change science. Subtle variation in mature forest types across the regional ecocline is compounded by vegetation transitions following agricultural landuses. Such complex mapping environments require innovation in multispectral classification methodologies. This research presents an application of a step-wise maximum likelihood/ in-Process Classification Assessment (IPCA) procedure. This hybrid supervised and unsupervised classification methodology allows for exploration of underlying characteristics of Landsat Thematic Mapper imagery in tropical environments. Once spectrally-separable classes have been identified, field data then determine the ecological definition of vegetation types with special attention paid to areas of unknown or mixed classes. A post-field assessment re-classification using the Dempster-Shafer method reduced the original 25 spectral classes to 14 ecologically-distinctive classes, providing the fine-tuned land-cover distinctions that are required for both environmental and socioeconomic research questions. The overall map accuracy was 87% with an average per-class accuracy of 86%. Per-class accuracy (omission errors) ranged from as low as 45% for pasture grass to a high of 100% for tall-stature evergreen upland forest (selva alta perennifolia), for semi-deciduous medium-stature forests (selva baja and mediana subcaducifolia) and deciduous forest (selva caducifolia).

Continuous land cover modification is an important part of spatial epidemiology because it can help identify environmental factors and Culex mosquitoes associated with arbovirus transmission and thus guide control intervention. The aim of this study was to determine whether remotely sensed data could be used to identify rice-related Culex quinquefasciatus breeding habitats in three rice-villages within the Mwea Rice Scheme, Kenya. We examined whether a land use land cover (LULC) classification based on two scenes, IKONOS at 4 m and Landsat Thematic Mapper at 30 m could be used to map different land uses and rice planted at different times (cohorts), and to infer which LULC change were correlated to high density Cx. quinquefasciatus aquatic habitats. We performed a maximum likelihood unsupervised classification in Erdas Imagine V8.7 and generated three land cover classifications, rice field, fallow and built environment. Differentially corrected global positioning systems (DGPS) grou...

The spatial variability of remotely sensed image values provides important information about the arrangement of objects and their spatial relationships within the image. The characterisation of spatial variability in such images, for example, to measure of texture, is of great utility for the discrimination of land cover classes. To this end, the variogram, a function commonly applied in geostatistics, has been used widely to extract image texture for remotely sensed data classification. The aim of this study was to assess the increase in accuracy that can be achieved by incorporating univariate and multivariate textural measures of Landsat TM imagery in classification models applied to large heterogeneous landscapes. Such landscapes which difficult to classify due to the large number of land cover categories and low inter-class separability. Madogram, rodogram and direct variogram for the univariate case, and cross-and pseudocross variograms for the multivariate one, together with multi-seasonal spectral information were used in a Random Forest classifier to map land cover types. The addition of spatial variability into multi-seasonal Random Forest models leads to an increase in the overall accuracy of 8%, and to an increase in the Kappa index of 9%, respectively. The increase in per categories Kappa for the textural Random Forest model reached 30% for certain categories. This study demonstrates that the use of information on spatial variability produces a fundamental increase in per class classification accuracy of complex land-cover categories.

Turkey is one of the few countries in the world with a favourable climate for hazelnut production. In addition, it has the leading position in world hazelnut production and export, supplying about 70% of world’s production. However, hazelnut production exceeds the demand and new some regulations have been enacted to create new land use policies in Turkey. By putting into practice regulations restricting hazelnut plantation areas, a more efficient and productive hazelnut harvest policy could be created. Samsun city is one of the most important hazelnut production centres in Central Black Sea region. The main objective of this study is to determine potential hazelnut areas in Samsun city located Central Black Sea Region according to current regulations using geographic information system technique regarding to support hazelnut policy developers and organizations. According to the criteria dictated by government regulations, potential hazelnut area in Samsun province was determined and...

The leaf area index (LAI) product from the Moderate Resolution Imaging Spectroradiometer (MODIS) is important for monitoring and modelling global change and terrestrial dynamics at many scales. The algorithm relies on spectral reflectances and a six biome land cover classification. Evaluation of the specific behaviour and performance of the product for regions of the globe such as Australia are needed to assist with product refinement and validation. We made an assessment of Collection 4 of the MODIS LAI product using four approaches: (a) assessment against a continental scale Structural Classification of Australian Vegetation (SCAV); (b) assessment against a continental scale land use classification (LUC); (c) assessment against historical field-based measurement of LAI collected prior to the Terra Mission; and (d) direct comparison of MODIS LAI with coincident field measurements of LAI, mostly from hemispherical photography. The MODIS LAI product produced a wide variety of geographically and structurally specific temporal response profiles between different classes and even for sub-groups within classes of the SCAV. Historical and concurrent field measurements indicated that MODIS LAI was giving reasonable estimates for LAI for most cover types and land use types, but that major overestimation of LAI occurs in some eastern Australian open forests and woodlands. The six biome structural land cover classification showed some significant deviations in class allocation compared to the SCAV particularly where grasslands are allocated to shrubland, savanna woodlands are allocated to shrubland, savanna and broadleaf forest, and open forests are allocated to savanna and broadleaf forest. The land cover and LAI products could benefit from some additional examination of Australian data addressing the structural representation of Eucalypt canopies in the "space of canopy realisation" for savanna and broadleaf forest classes.

The leaf area index (LAI) product from the Moderate Resolution Imaging Spectroradiometer (MODIS) is important for monitoring and modelling global change and terrestrial dynamics at many scales. The algorithm relies on spectral reflectances and a six biome land cover classification. Evaluation of the specific behaviour and performance of the product for regions of the globe such as Australia are needed to assist with product refinement and validation. We made an assessment of Collection 4 of the MODIS LAI product using four approaches: (a) assessment against a continental scale Structural Classification of Australian Vegetation (SCAV); (b) assessment against a continental scale land use classification (LUC); (c) assessment against historical field-based measurement of LAI collected prior to the Terra Mission; and (d) direct comparison of MODIS LAI with coincident field measurements of LAI, mostly from hemispherical photography. The MODIS LAI product produced a wide variety of geographically and structurally specific temporal response profiles between different classes and even for sub-groups within classes of the SCAV. Historical and concurrent field measurements indicated that MODIS LAI was giving reasonable estimates for LAI for most cover types and land use types, but that major overestimation of LAI occurs in some eastern Australian open forests and woodlands. The six biome structural land cover classification showed some significant deviations in class allocation compared to the SCAV particularly where grasslands are allocated to shrubland, savanna woodlands are allocated to shrubland, savanna and broadleaf forest, and open forests are allocated to savanna and broadleaf forest. The land cover and LAI products could benefit from some additional examination of Australian data addressing the structural representation of Eucalypt canopies in the "space of canopy realisation" for savanna and broadleaf forest classes.

Classifying land use/land cover (LULC) with sufficient accuracy in heterogeneous landscapes is challenging using only satellite imagery. To improve classification accuracy inclusion of features from auxiliary geospatial datasets in classification models is applied since 1980s. However, the method is mostly limited to pixel-based classifications, and the coverage, accuracy and resolution of free and open-access auxiliary datasets have been poor until recent years. We evaluated how recent global coverage open-access geospatial datasets improve object-based LULC classification accuracy compared to using only spectral and texture features from satellite images. We applied feature sets topography, population, soil, canopy cover, distance to watercourses and spectral-temporal metrics from Landsat-8 time series on the southern foothills and savanna of Mt. Kilimanjaro, Tanzania, where the landscape is characterized by heterogeneous and fragmented mosaic of disturbed savanna vegetation, croplands, and settlements. The classification was based on image objects (groups of spectrally similar pixels) derived from segmentation of four Formosat-2 scenes with 8 m spatial resolution using 1370 ground reference points for training, validation, and for defining 17 LULC classes. We built six Random Forest classification models with different sets of object features in each. The baseline model having only spectral and texture features was compared with five other models supplemented with auxiliary features. Inclusion of auxiliary features significantly improved classification overall accuracy (OA). The baseline model gave a median OA of 60.7%, but auxiliary features in other models increased median OA between 6.1 and 16.5 percentage points. The best OA was achieved with a model including all features of which elevation was the most important auxiliary feature followed by Enhanced Vegetation Index temporal range and slope degree. Applying object-based classification to geospatial information on topography, soil, settlement patterns and vegetation phenology, the discriminatory potential of challenging LULC classes can be significantly improved. We demonstrated this for the first time, and the technique shows good potential for improving LULC mapping across a multitude of fragmented landscapes worldwide.

Multiple subpixel shifted images (MSIs) from the same area can be incorporated to improve the accuracy of softthen-hard subpixel mapping (STHSPM). In this paper, a novel method that derives higher resolution MSIs with more spatialspectral information (MSI-SS) is proposed. First, coarse MSIs produce two high-resolution MSIs for each class respectively by two paths at the same time. The spatial path produces the highresolution MSIs by soft classification followed by interpolation. And, the other high-resolution MSIs are derived from the spectral path by interpolation followed by soft classification. Then the higher resolution MSIs with more spatial-spectral information for each class are derived by integrating the aforementioned two kinds of high-resolution MSIs by the appropriate weight. Finally, the integrated higher resolution MSIs for each class are used to allocate hard class labels to subpixels. The proposed method is fast and takes more spatial-spectral information of the original MSIs into account. Experiments on three real hyperspectral remote sensing images show that the proposed method produce higher SPM accuracy result.

Semantic segmentation requires methods capable of learning high-level features while dealing with large volume of data. Towards such goal, Convolutional Networks can learn specific and adaptable features based on the data. However, these networks are not capable of processing a whole remote sensing image, given its huge size. To overcome such limitation, the image is processed using fixed size patches. The definition of the input patch size is usually performed empirically (evaluating several sizes) or imposed (by network constraint). Both strategies suffer from drawbacks and could not lead to the best patch size. To alleviate this problem, several works exploited multi-context information by combining networks or layers. This process increases the number of parameters resulting in a more difficult model to train. In this work, we propose a novel technique to perform semantic segmentation of remote sensing images that exploits a multi-context paradigm without increasing the number of parameters while defining, in training time, the best patch size. The main idea is to train a dilated network with distinct patch sizes, allowing it to capture multi-context characteristics from heterogeneous contexts. While processing these varying patches, the network provides a score for each patch size, helping in the definition of the best size for the current scenario. A systematic evaluation of the proposed algorithm is conducted using four high-resolution remote sensing datasets with very distinct properties. Our results show that the proposed algorithm provides improvements in pixelwise classification accuracy when compared to state-of-the-art methods.

Multiple subpixel shifted images (MSIs) from the same area can be incorporated to improve the accuracy of softthen-hard subpixel mapping (STHSPM). In this paper, a novel method that derives higher resolution MSIs with more spatialspectral information (MSI-SS) is proposed. First, coarse MSIs produce two high-resolution MSIs for each class respectively by two paths at the same time. The spatial path produces the highresolution MSIs by soft classification followed by interpolation. And, the other high-resolution MSIs are derived from the spectral path by interpolation followed by soft classification. Then the higher resolution MSIs with more spatial-spectral information for each class are derived by integrating the aforementioned two kinds of high-resolution MSIs by the appropriate weight. Finally, the integrated higher resolution MSIs for each class are used to allocate hard class labels to subpixels. The proposed method is fast and takes more spatial-spectral information of the original MSIs into account. Experiments on three real hyperspectral remote sensing images show that the proposed method produce higher SPM accuracy result.

In recent years, sparse representations have been widely studied in the context of remote sensing image analysis. In this paper, we propose to exploit sparse representations of morphological attribute profiles for remotely sensed image classification. Specifically, we use extended multiattribute profiles (EMAPs) to integrate the spatial and spectral information contained in the data. EMAPs provide a multilevel characterization of an image created by the sequential application of morphological attribute filters that can be used to model different kinds of structural information. Although the EMAPs' feature vectors may have high dimensionality, they lie in class-dependent low-dimensional subpaces or submanifolds. In this paper, we use the sparse representation classification framework to exploit this characteristic of the EMAPs. In short, by gathering representative samples of the low-dimensional class-dependent structures, any given sample may by sparsely represented, and thus classified, with respect to the gathered samples. Our experiments reveal that the proposed approach exploits the inherent low-dimensional structure of the EMAPs to provide state-of-the-art classification results for different multi/hyperspectral data sets.

The leaf area index (LAI) product from the Moderate Resolution Imaging Spectroradiometer (MODIS) is important for monitoring and modelling global change and terrestrial dynamics at many scales. The algorithm relies on spectral reflectances and a six biome land cover classification. Evaluation of the specific behaviour and performance of the product for regions of the globe such as Australia are needed to assist with product refinement and validation. We made an assessment of Collection 4 of the MODIS LAI product using four approaches: (a) assessment against a continental scale Structural Classification of Australian Vegetation (SCAV); (b) assessment against a continental scale land use classification (LUC); (c) assessment against historical field-based measurement of LAI collected prior to the Terra Mission; and (d) direct comparison of MODIS LAI with coincident field measurements of LAI, mostly from hemispherical photography. The MODIS LAI product produced a wide variety of geographically and structurally specific temporal response profiles between different classes and even for sub-groups within classes of the SCAV. Historical and concurrent field measurements indicated that MODIS LAI was giving reasonable estimates for LAI for most cover types and land use types, but that major overestimation of LAI occurs in some eastern Australian open forests and woodlands. The six biome structural land cover classification showed some significant deviations in class allocation compared to the SCAV particularly where grasslands are allocated to shrubland, savanna woodlands are allocated to shrubland, savanna and broadleaf forest, and open forests are allocated to savanna and broadleaf forest. The land cover and LAI products could benefit from some additional examination of Australian data addressing the structural representation of Eucalypt canopies in the "space of canopy realisation" for savanna and broadleaf forest classes.

Michigan has significant forest resources, and there is much debate about the use and protection of the forests. Managers, analysts, and the public have multiple sources for data regarding Michigan's forests, but these sources vary in how accurately they characterize forests. A close examination of three data sources shows considerable differences in accuracies of public land data sets. Forest Inventory and Analysis (FIA) plots were treated as the reference classification system, given their consistent application procedures on the ground. National forest and state forest tactical data sets and the Michigan Department of Natural Resources' remotely sensed data sets were compared with 753 FIA plots. The accuracies of the on-the-ground tactical data sets exceeded 82%, but remotely sensed data were accurate only 65% of the time. The latter may serve as a proxy for private lands that do not have tactical inventory data sets. Aggregating remotely sensed data and combining tactical and remotely sensed data would improve overall accuracy across landscapes with both public and private lands.

Forest cover monitoring using satellite imagery is important to U.S. military terrain analysis. Mobility models, cover and concealment, and precise surface feature delineation all rely on an accurate forest/nonforest cover layer. However, the challenge remains in distinguishing trees from other vegetative land covers when relying on single date imagery. A Forest Cover Index (FCI) algorithm was previously developed on Worldview-2 imagery and was designed to separate forest cover from nonforest covered areas. In this research, the FCI algorithm was applied to Landsat-7 imagery using the analogous red (636-673 nanometers (nm)) and near-infrared bands (851-878 nm) from a peak summer image on 16 August 2012. The results obtained with Landsat-7 imagery proved satisfactory with an overall accuracy (tree cover versus non-tree cover) of > 83 percent, according to testing with two different accuracy assessment tests. The application of the FCI to Landsat-7 imagery broadens the applicability of the FCI to freely available imagery. The contents of this report are not to be used for advertising, publication, or promotional purposes. Citation of trade names does not constitute an official endorsement or approval of the use of such commercial products. All product names and trademarks cited are the property of their respective owners. The findings of this report are not to be construed as an official Department of the Army position unless so designated by other authorized documents. DESTROY THIS REPORT WHEN NO LONGER NEEDED. DO NOT RETURN IT TO THE ORIGINATOR.

Reliable estimation of hydrological soil moisture state is of critical importance in operational hydrology to improve the flood prediction and hydrological cycle description. Although there have been a number of soil moisture products, they cannot be directly used in hydrological modelling. This paper attempts for the first time to build a soil moisture product directly applicable to hydrology using multiple data sources retrieved from SAC-SMA (soil moisture), MODIS (land surface temperature), and SMOS (multi-angle brightness temperatures in H-V polarisations). The simple yet effective local linear regression model is applied for the data fusion purpose in the Pontiac catchment. Four schemes according to temporal availabilities of the data sources are developed, which are pre-assessed and best selected by using the well-proven feature selection algorithm gamma test. The hydrological accuracy of the produced soil moisture data is evaluated against the Xinanjiang hydrological model's soil moisture deficit simulation. The result shows that a superior performance is obtained from the scheme with the data inputs from all sources (NSE = 0.912, r = 0.960, RMSE = 0.007 m). Additionally, the final daily-available hydrological soil moisture product significantly increases the Nash-Sutcliffe efficiency by almost 50 % in comparison with the two most popular soil moisture products. The proposed method could be easily applied to other catchments and fields with high confidence. The misconception between the hydrological soil moisture state variable and the real-world soil moisture content, and the potential to build a global routine hydrological soil moisture product are discussed.

and Conclusions 1. Is there urban enhancement of N deposition?: Yes -deposition of direct reaction products such as NO 2 are increased by up to one order of magnitude close to the urban core areas where they are produced. Our model results indicate that NO x -derived dry deposition IS a significant term in the ecosystem N balance, comprising 20% of the total annual N inputs to the CAP study area. The most important determinants of NO x -derived N deposition flux are ambient concentrations of NO x species in the atmosphere and the amount of vegetated surface cover (Figure ). For more accurate simulation of N dry deposition rates it is essential to obtain good representation of the amount and type of vegetated surface when applying 3-D Eulerian air quality models. 4. Implications: Elevated N deposition on non-reactive urban surfaces, can lead to high concentrations of nutrients on these surfaces and in the first-flush runoff during rainfall events. Amounts of inorganic N on asphalt parking lot surfaces across the Phoenix metro area averaged 40.2 mg m -2 for NH 4 -N and 151.1 mg m -2 for NO 3 -N, values that were 13 -91 times higher than measured in surface samples of undeveloped desert soils outside the city (Hope et al. submitted b).

This paper introduces a stratified random sampling strategy and an estimated error matrix of pixel proportions of observations for the entire study area. A classified land cover map was created by applying the Maximum likelihood classification method in IDRISI software to a CBERS-2 false color composite satellite image. The classified land cover map identifies four major land use categories: forest, development/cleared land, scrubland, and pasture/agriculture. The study area is a 240sq km municipality boundary of Rosário da Limeira in the state of Minas Gerais, Brazil. A stratified random sampling strategy was developed to select land cover sample sites. The previously collected ground data is compared to the land cover categories on the map. Finally, an error matrix of the estimated pixel proportions of observations is then utilized to show the correct or incorrect proportion of observations for the entire study area. The stratified random sampling strategy and the error matrix of pixel proportions of observations illustrated in this paper are for the Iracambi GIS lab and other organizations interested in collecting land cover ground data.

Supervised classification techniques are commonly used to assign pixels of multispectral satellite imagery to a predefined set of classes in order to generate or update land use or land cover maps from remote sensed data. These techniques have a limited ability in expressing spatial relationships among pixels. We propose a new contextual approach to address this issue. In particular, we present an integer linear programming formulation which restricts the number of distinct objects in the classified image and we propose a heuristic for the resulting problem. We test it on one generated data set and one real data set.

In fire risk, correct description of topographic and fuel properties is critical to improve fire danger assessment and fire behaviour modelling. Many rural areas are now scanned using LIDAR sensors. In some of these areas the information registered by the sensor includes not only the geometric characteristics of the Earth's surface, given by the coordinates (x,y,z) of the LiDAR point

This study investigates whether the presence of green space can attenuate negative health impacts of stressful life events. Individual-level data on health and socio-demographic characteristics were drawn from a representative two-stage sample of 4529 Dutch respondents to the second Dutch National Survey of General Practice (DNSGP-2), conducted in 2000-2002. Health measures included: (1) the number of health complaints in the last 14 days; (2) perceived mental health (measured by the GHQ-12); and (3) a single item measure of perceived general health ranging from 'excellent' to 'poor'. Percentages of green space in a 1-km and 3-km radius around the home were derived from the 2001 National Land cover Classification database (LGN4). Data were analysed using multilevel regression analysis, with GP practices as the group-level units. All analyses were controlled for age, gender, income, education level, and level of urbanity. The results show that the relationships of stressful life events with number of health complaints and perceived general health were significantly moderated by amount of green space in a 3-km radius. Respondents with a high amount of green space in a 3-km radius were less affected by experiencing a stressful life event than respondents with a low amount of green space in this radius. The same pattern was observed for perceived mental health, although it was marginally significant. The moderating effects of green space were found only for green space within 3 km, and not for green space within 1 km of residents' homes, presumably because the 3-km indicator is more affected by the presence of larger areas of green space, that are supposed to sustain deeper forms of restoration. These results support the notion that green space can provide a buffer against the negative health impact of stressful life events.

Reliable and current availability of land cover knowledge are essential for many studies regarding planning, management, monitoring and updating activities. The optical satellite sensor data has been utilized for the classification of land use/land cover. In this study, the capability of synthetic aperture radar (SAR) interferometric coherence is practiced for land cover classification in Okara, Pakistan using Sentinel-1A imagery. Two Single Look Complex (SLC) product of months April and May 2016 were used and processed to create backscatter and interferometric coherence layers. From backscatter layers of each month, the mean backscatter and backscatter difference layer were obtained. False color composite (FCC) were developed comprising mean backscatter, backscatter difference and coherence, and performed supervised classification using maximum likelihood method to generate land cover classes i.e. water, barren, vegetation and built-up. Kappa statistics were employed for accuracy assessment of the output map. Results showed the good potential of Sentinel-1C-band for land cover classification having 0.69 Kappa coefficient and 80% overall accuracy. This study investigated the potential of C-band backscatter coefficients and coherence map for land cover discrimination. Coherence proved to be efficient in the examination of vegetative and non-vegetative areas.

The recent drought (1998)(1999)(2000)(2001)(2002)(2003)(2004) in water deficient province of Balochistan (the largest province of Pakistan) affected the availability of seasonal rainwater and vegetation cover. Analysis of the spatial rainfall computed through spatial interpolation of point rainfall depicted a reduction of about 53-85 percent and 66-90 percent in volumes of rainwater during summer and winter seasons of the drought period, respectively. Spatio-temporal analysis of the vegetation cover carried out through land cover classification of satellite images revealed a reduction of up to 31 percent in summer season and 5-7 percent in winter season vegetation cover in Balochistan during the drought period. In addition to the direct impacts of this period of dry weather, significant increase in the number of tubewells in order to overcome the water deficiency during recent drought in the province resulted in over-abstraction of groundwater, particularly in the Pishin-Lora river basin. An average lowering of water table by 1-20 m per year in Pishin-Lora river basin triggered the process of identification of potential recharge sites in this heavily populated river basin. A methodology to delineate the potential groundwater recharge zones by means of Geographical Information Systems and Remotely Sensed datasets was devised and spatial analyses were conducted using surface slope, drainage network, lithology and settlements in Pishin-Lora river basin. The results showed that about 50 percent of the basin had medium to high potential for recharging groundwater and the potential recharge areas were identified as the active zones for implementation of artificial recharge mechanisms for enhancement of groundwater recharge.

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.

The objective of this study is to verify the contribution of the spectral bands from the new WorldView-2 satellite for the extraction of urban targets aiming a detailed mapping from the city of São Luis, at the coastal zone of Maranhão State, Brazil. This satellite system has 3 bands in the visible portion of the spectrum and also the following 4 new bands: Coastal (400-450 nm), Yellow (585-625 nm), Red Edge (705-745 nm), and Near Infrared 2 (860-1040 nm). As for the methodology used, initially a fusion was made among the panchromatic and the multispectral bands, combining the spectral information of the multispectral bands with the geometric information of the panchromatic band. Following the ortho-rectification of the dataset was done, using ground control points (GCPs) obtained during field survey. The classification reached high values of Kappa indices. The use of the new bands Red Edge and Near Infrared 2, allowed the improvement of discriminations at tidal flats, mangrove and other vegetation types. The Yellow band improved the discrimination of bare soils -very important information for urban planning -and ceramic roofs. The Coastal band allowed to map the tidal channels which cross the urban area of São Luis, a typical feature of this coastal area. The functionalities of software GEODMA used, allowed an efficient attribute selection which improved the land cover classification from the test sites. The new WorldView-2 bands permit the identification and extraction of the features mentioned, because these bands are positioned at important parts of the electromagnetic spectrum, such as band Red Edge, which strongly improves the discrimination of vegetation conditions. Combining both higher spatial and spectral resolutions, WorldView-2 data allows an improvement on the discrimination of physical characteristics of the targets of interest, thus permitting a higher precision of land use/land cover maps, contributing to urban planning. The test sites of this study represent the main problem areas involving the city of São Luis and the entire region of the Maranhão Island.

The objective of this study is to evaluate the potential of WorldView-2 satellite images and the free-access software InterIMAGE to map land cover in an urban coastal area. For the execution of this study, statistical data, information from public institutions and partner companies at this project, were used. Furthermore the following materials were utilized: 1. WorldView-2 satellite scenes obtained in June 1

Land cover classification for the evaluation of land ranges of sensors. Classification of image pixels by features cover changes over certain areas or time periods is crucial for from different spectral band data has been explored recently geospatial modeling, environmental crisis evaluation and urban [5], [9], [14]. Yet these classifications use data of the same open space planning. Remotely sensed images of various spatial and spectral resolutions make it possible to classify land covers on sialeresoltian similar speresolutios andsdataof the level of pixels. Semantic meanings of large regions consisting different spatial resolutions and different spectral resolutions of hundreds of thousands of pixels cannot be revealed by discrete have not been considered together for classification. and individual pixel classes, but can be derived by integrating Classification on the pixel level cannot reveal semantic various groups of pixels using ontologies. This paper combines concepts at higher levels, and the semantic concepts at high data of different resolutions for pixel classification by support levels can be crucial for security protection, environment vector classifiers, and proposes an efficient algorithm to group v pixels based on classes of neighboring pixels. The algorithm evaluation and urban open space research. For instance, if a is linear in the number of pixels of the target area, and is pixel or a few neighboring pixels are classified as water body, scalable to very large regions. It also re-evaluates imprecise the location can be a pool in a residential area, a pond in an classifications according to neighboring classes for region level urban area, or a lake in a park or open rural area. Similarly, semantic interpretations. Experiments on Advanced Spaceborne a . . . . f Thermal Emission and Reflection Radiometer (ASTER) data of a group of buildigs can be for public service i an urban more than six million pixels show that the proposed approach area, for residential purpose in a residential area, or for highly achieves up to 99.8% cross validation accuracy and 89.25% test confidential military use in a desert. It is of great need to accuracy for pixel classification, and can effectively and efficiently develop high level concepts and distinguish them so that the group pixels to generate high level semantic concepts. semantic meanings of pixel classes become clear and accesses to some confidential concepts become controllable for security

With the recent launch of MERIS, a wide range of new possibilities for the periodic land cover characterization at regional scale is available. This sensor offers a combination of innovative features, such as high spectral and temporal resolutions, wide geographical coverage and improved atmospheric correction. We believe that the exploitation of data obtained by this new sensor fills previous technological gaps, improving automatic land cover classes' discrimination. At the same time, the extra spectral information provided by MERIS can introduce some difficulties on land cover characterization with long-established classification techniques, e.g. k-Nearest Neighbour. In this paper we report the performance of artificial neural networks (ANNs) in the context of high spectral dimensional satellite image classification. The main goal of this research is to assess the potential of the Self-Organizing Maps (SOM) neural network to extract complex land cover type information from medium resolution satellite imagery. The study was carried out with MERIS Full Resolution data from 2004 for the continental Portuguese territory.

The PIMAR Project - Program for Monitoring the Atlantic Rainforest Environment and Urban Growth of Rio de Janeiro through Remote Sensing, aims at the development of an operational methodology for monitoring the land cover dynamics on the borders between Atlantic rainforest remnant areas and urban areas in the city of Rio de Janeiro, Brazil. The project will aid the Government of Rio de Janeiro State in the implementation of actions against aggressions to those forested areas and in the definition of urban development and environmental planning policies. The basic input for the methodology is a sequence of stereo pairs of IKONOS images, from which both the vertical and horizontal growth of urban areas are being measured by visual interpretation on a multitemporal basis. The PIMAR Project is currently evaluating the use of an automatic classification model as a way to accelerate land cover change information production to support decision making. This paper presents the first results ...

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.

Climate models participating in the IPCC Fourth Assessment Report indicate that under a 2xC02 environment, runoff would increase faster than precipitation overland. However, observations over large U.S watersheds indicate otherwise. This inconsistency suggests that there may be important feedbacks between climate and land surface unaccounted for in the present generation of models. We postulate that the increase in precipitation associated with the increase in C 0 2 is also increasing vegetation density, which may already be feeding back onto climate. Including this feedback in a climate model simulation resulted in precipitation and runoff trends consistent with observations and reduced the warming by 0.6OC overland. This unaccounted for missing water may be linked to about 10% of the missing land carbon sink. A recent compilation of outputs from 19 coupled atmosphere-ocean general circulation models used in the IPCC Fourth Assessment Report (AR4) shows projected increases in air temperature, precipitation and river discharge for 24 major rivers in the

Choice of a classification algorithm is generally based upon a number of factors, among which are availability of software, ease of use, and performance, measured here by overall classification accuracy. The maximum likelihood (ML) procedure is, for many users, the algorithm of choice because of its ready availability and the fact that it does not require an extended training process. Artificial neural networks (ANNs) are now widely used by researchers, but their operational applications are hindered by the need for the user to specify the configuration of the network architecture and to provide values for a number of parameters, both of which affect performance. The ANN also requires an extended training phase. In the past few years, the use of decision trees (DTs) to classify remotely sensed data has increased. Proponents of the method claim that it has a number of advantages over the ML and ANN algorithms. The DT is computationally fast, make no statistical assumptions, and can handle data that are represented on different measurement scales. Software to implement DTs is readily available over the Internet. Pruning of DTs can make them smaller and more easily interpretable, while the use of boosting techniques can improve performance. In this study, separate test and training data sets from two different geographical areas and two different sensors-multispectral Landsat ETM+ and hyperspectral DAIS-are used to evaluate the performance of univariate and multivariate DTs for land cover classification. Factors considered are: the effects of variations in training data set size and of the dimensionality of the feature space, together with the impact of boosting, attribute selection measures, and pruning. The level of classification accuracy achieved by the DT is compared to results from back-propagating ANN and the ML classifiers. Our results indicate that the performance of the univariate DT is acceptably good in comparison with that of other classifiers, except with high-dimensional data. Classification accuracy increases linearly with training data set size to a limit of 300 pixels per class in this case. Multivariate DTs do not appear to perform better than univariate DTs. While boosting produces an increase in classification accuracy of between 3% and 6%, the use of attribute selection methods does not appear to be justified in terms of accuracy increases. However, neither the univariate DT nor the multivariate DT performed as well as the ANN or ML classifiers with high-dimensional data.

Land cover classification using multispectral satellite image is a very challenging task with numerous practical applications. We propose a multi-stage classifier that involves fuzzy rule extraction from the training data and then generation of a possibilistic label vector for each pixel using the fuzzy rule base. To exploit the spatial correlation of land cover types we propose four different information aggregation methods which use the possibilistic class label of a pixel and those of its eight spatial neighbors for making the final classification decision. Three of the aggregation methods use Dempster-Shafer theory of evidence while the remaining one is modeled after the fuzzy k-NN rule. The proposed methods are tested with two benchmark seven channel satellite images and the results are found to be quite satisfactory. They are also compared with a Markov random field (MRF) model-based contextual classification method and found to perform consistently better.

Satellite Phased Array L-band Synthetic Aperture Radar-2 has great advantages in extracting natural and industrial forest plantation in tropical areas, but it suffers from presence of speckle that create problem to identify the forest body. Optimal fusion of Landsat-8 operational land imager bands with ALOS PALSAR-2 can provide the ideal complementary information for an accurate forest extraction while suppressing unwanted information. The goal of this study is to analyze the potential ability of Landsat-8 OLI and ALOS PALSAR-2 as complementary data resources in order to extract land cover especially forest types. Comprehensive preprocessing analysis (e.g. geometric correction, filtering enhancement and polarization combination) were conducted on ALOS PALSAR-2 dataset in order to make the imagery ready for processing. Principal component index method as one of the most effective Pan-Sharpening fusion approaches was used to synthesize Landsat and ALOS PALSAR-2 images. Three different...

Abstract: In this paper we aim at describing a method to extract information about urban structure on high resolution Synthetic Aperture Radar (SAR) images. The extraction of useful information from SAR images is a difficult task, due to the presence of the characteristic ...

Right from the beginning Selela Forest Management Unit is under the jurisdiction of the Paro Territorial Forest Division. Before the management plan was written the local people had unlimited access to the forest for anything they need. But after the first Management Plan was written, once a remote area was opened to the rest of the Dzongkhag through the construction of forest road and now several farm roads were also built to upgrade the socio-economic status of the people.
In the early nineties although the area is under the control of the CFO Paro, no proper monitoring or patrolling can be done due to lack of staff strength and accessibility to the area. So, the forest is under the mercy of the local people. During the opening of the new FMU and while documenting the first Management Plan in the year 1997-1998 the area is said to be rich in stock density and environmentally intact. This is due to the low population and few households within the FMU and the barrier created by the nature such as Haa Chu and the steep ridges which made accessibility impossible to the outsiders.
Before the creation of Department of Forest in the mid 1950s, people from the neighboring villages of SFMU had unlimited access to the forest for their timber and firewood requirement. Felling of trees for timber and firewood was unregulated. Till 1998 there wasn‟t any written plan that would enable the harvesting of trees in scientific method. Timbers marked for the villagers by the FSD were on an ad-hoc basis.
The first management plan for SFMU was written in 1998 for the duration of ten years (1998 2008). Due to excess allocation of AAC in the previous plan the AAC had been revised in the year 2000 by FRDD. There has been no written management plan prior to 1998 and the last management plan from 1998 to 2008 is the only document for the last decade of harvesting and this is the review of the last ten years. According to the last management plan, the total area of SFMU is 9253 ha.
According to the last management plan, the SFMU was divided into four blocks, which was further divided into compartments and sub-compartments, based on natural features. No proper working circle had been mentioned in the plan but more emphasis is given on erosion control, creation of the ecological reserve and its management, protection of the forest, preservation of the aesthetic beauty as seen from the Thimphu – Haa Highway, and wild life protection.
The silvicultural system that was to be followed was group selection system for Blue Pine, group/single tree selection system (according to slope) for Hemlock and single tree selection for Hardwoods. All the harvesting operation was to be carried out by using cable crane loggings systems. Thinning was also to be carried out in the Blue Pine stand to improve the quality of the stand in the future. The AAC for the last management plan was calculated to be 20,000m3 but it was revised in the year 2000 and the AAC was reduced to 12,600m3 of which 8,568m3 is NRDCLs share and 4,032m3 for rural share.. 23.149 Km of road was constructed during the last plan period by NRDCL.1.4 Forest Condition.
Selela FMU is under commercial harvesting for last ten years and it is obvious that some parts of area are exposed. Presently they are harvesting in the Selela Block. The operated cable lines in the Tegola block have good natural regeneration of hemlock. Some areas devoid of regeneration are mainly due to wood debris left behind after extraction. The sub compartment T2a of Tegola block had been given for rural use in the last plan but now the area had been overharvested with huge gaps created all around. Hence the area had to be protected from now onwards. Allowing the collection of the wood chips had helped a lot for forest sanitation. The natural regeneration of Hemlock and Blue Pine stand in and around the settlements is expanding its territory into the open areas.It was mentioned in the last plan that in areas where natural regeneration fails to establish, it should be supplemented with the artificial plantation. NRDCL has already carried out plantation of 7.33ha in total.

A novel self-organizing neuro-fuzzy multilayered classifier (SONeFMUC) is suggested in this paper, with feature selection capabilities, for the classification of an IKONOS image. The structure of the proposed network is developed in a sequential fashion using the group method of data handling (GMDH) algorithm. The node models, regarded as generic classifiers, are represented by fuzzy rulebased systems, combined with a fusion scheme. A data splitting mechanism is incorporated to discriminate between correctly classified and ambiguous pixels. The classifier was tested on the wetland of international importance of Lake Koronia, Greece, and the surrounding agricultural area. To achieve higher classification accuracy, the image was decomposed into two zones, namely, the wetland and the agricultural zone. Apart from the initial bands, additional input features are considered: textural features, IHS and tasseled cap transformation. To assess the quality of the suggested model, SONeFMUC is compared with a maximum likelihood classifier (MLC). The experimental results show that the SONeFMUC exhibited superior performance than MLC, providing lower confusion of the dominant classes in both zones. Especially in the wetland zone, an overall accuracy of 89.5% was attained.

A novel Self-Organizing Neuro-Fuzzy Multilayered Classifier, the GA-SONeFMUC model, is proposed in this paper for land cover classification of multispectral images. The model is composed of generic fuzzy neuron classifiers (FNCs) arranged in layers, which are implemented by fuzzy rule-based systems. At each layer, parent FNCs are combined to generate a descendant FNC at the next layer with higher classification accuracy. To exploit the information acquired by the parent FNCs, their decision supports are combined using a fusion operator. As a result, a data splitting is devised within each FNC, distinguishing those pixels that are currently correctly classified to a high certainty grade from the ambiguous ones. The former are handled by the fuser, while the ambiguous pixels are further processed to enhance their classification confidence. The GA-SONeFMUC structure is determined in a self-constructing way via a structure-learning algorithm with feature selection capabilities. The parameters of the models obtained after structure learning are optimized using a real-coded genetic algorithm. For effective classification, we formulated three input sets containing spectral and textural feature types. To explore information coming from different feature sources, we apply a classifier fusion approach at the final stage. The outputs of individual classifiers constructed from each input set are combined to provide the final assignments. Our approach is tested on a lake-wetland ecosystem of international importance using an IKONOS image. A high-classification performance of 92.02% and of 75.55% for the wetland zone and the surrounding agricultural zone is achieved, respectively.

An effective methodology for Bohai Sea ice detection based on gray level co-occurrence matrix (GLCM) texture analysis is proposed using MODIS 250 m imagery. The method determines texture measures for sea ice extraction by analyzing the discrepancy of textural features between sea ice and sea water. Sea ice extent and outer edge are recognized accurately by texture segmentation owing to significant differences in texture statistical features between ice and water. The texture analysis method can properly eliminate perturbations on sea ice extraction due to suspended sediment. It effectively solves the problem of spectral confusion and sea ice misassignment in the conventional gray-threshold segmentation and ratio-threshold segmentation methods. The method eliminates the need for threshold range setting for sea ice segmentation. Taking the Bohai Sea as an example, the results of the proposed method are validated using co-temporal HJ1B-CCD 30 m imagery by visual interpretation, and the accuracy of the method are evaluated using confusion matrix. The results show that the proposed method is superior and more reliable for sea ice detection compared to conventional methods, providing an ideal tool for precise sea ice extraction.

A recent article advocated the adoption of a single standard for all land cover classifications. The authors argued that variations in classification were problematic, standards solve problems related to classification heterogeneity and land cover is the fundamental land variable. This letter challenges these arguments: 1) methods exist for integrating disparate data, many based around data semantics; 2) standards are themselves problematic as they are frequently revised (e.g. soils) and because they always lag behind current activities cannot represent the depth of knowledge held within a community such as land cover; 3) scientists working in other disciplines may view land use as the elemental variable driving many other processes and they construct land cover in a very different way This letter argues that as most geographic data and especially land cover is a socially mediated construct (there are no agreed fundamental units), fixing a specific conceptualisation of land cover into the 'aspic' of a formal standard does not represent a scientific advance.

We evaluated the effectiveness of integrating discrete return light detection and ranging (LiDAR) data with high spatial resolution near-infrared digital imagery for object-based classification of land cover types and dominant tree species. In particular we adopted LiDAR ratio features based on pulse attributes that have not been used in past studies. Object-based classifications were performed first on land cover types, and subsequently on dominant tree species within the area classified as trees. In each classification stage, two different data combinations were examined: LiDAR data integrated with digital imagery or digital imagery only. We created basic image objects and calculated a number of spectral, textural, and LiDAR-based features for each image object. Decision tree analysis was performed and important features were investigated in each classification. In the land cover classification, the overall accuracy was improved to 0.975 when using the object-based method and integrating LiDAR data. The mean height value derived from the LiDAR data was effective in separating ''trees'' and ''lawn'' objects having different height. As for the tree species classification, the overall accuracy was also improved by object-based classification with LiDAR data although it remained up to 0.484 because spectral and textural signatures were similar among tree species. We revealed that the LiDAR ratio features associated with laser penetration proportion were important in the object-based classification as they can distinguish tree species having different canopy density. We concluded that integrating LiDAR data was effective in the object-based classifications of land cover and dominant tree species.

Ice can pose a hazard for operations (e.g., transportation, shipping, surveillance, offshore oil and gas exploration) and for infrastructure (e.g., ports, pipelines, offshore structures). There is an increasing need for fine-scale characterization of hazardous ice conditions. This information is of interest to many stakeholders including industry and government agencies. Spaceborne SAR sensors are being used for near-real-time monitoring of the regional ice conditions. Satellite-derived sea ice information products typically rely on the interpretation of ice analysts, sometimes supported by semi-automated tools. However, validation of data products remains a challenge due to limited or no ground truth. Operated from subsurface moorings located below the sea ice canopy, the ULS instrumentation, consisting of the ASL Ice Profiling Sonar (IPS) and the Acoustic Doppler Current Profiler (ADCP), provides accurate measurements of ice draft on a continuous basis and allow detailed character...

HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.