National Institute for Space Research
Biblioteca Digital da Memória Científica do INPENot a member yet
6175 research outputs found
Sort by
Análise exploratória de imagem COSMO-SkyMED para detecção de corte raso em região amazônica
No full textData from deforestation, over time, have been hampered by atmospheric intemperes as intense Cloud constant presence in some areas of the Amazon. Thus, the images SAR (Synthetic Aperture Radar) may be the way to solve this problem and COSMO-SkyMed is an option in band X for that. This study then sought to assess their ability to discriminate clear-cutting in the Amazon rainforest through remote sensing techniques, the study area as having a cutout of the city of Paragominas in the state of Pará in amplitude images were used, filtered (Lee and Frost) and Gram-Schmidt fusion for classification MaxVer-ICM where it was found better results in fused image TM5 + HH kappa (K = 0.75, where K is the Kappa index) this product has a good potential to identify recent clearcutting and clearcutting with regrowth.Pages: 8344-835
Detecção de mudanças na cobertura vegetal orientada à atualização de mapas de uso e cobertura das terras
No full textThe objective of this study was to update a map of land use using a map of intensity of changes. The periodic generation of land use maps, in many cases, is based on the updated of previous period maps, so the operation can be more economical but the accuracy of some classes of the new map can be low and only acceptable for classes of interest. This form of update is mainly used for classes that experiencing changes will change the thematic legends. The update method used a simple cross between map use and land cover map of 2000 with the intensity changes of 2001/2009 for the use map in 2009. The result showed that the natural forest class can be reliably updated and the areas lost by this class, only present confusion to distinguish between class agriculture and forestry in the initial deployment. The class reforestation can be reliably updated from the loss of forest class in its advanced stages, when it has a higher spectral response than the natural forest. This update maps proved unreliable to update the classes of agricultural use, livestock and natural areas in urban areas, but these results can be improved using techniques that consider the characteristics of the transition regions.Pages: 7460-746
Avaliação do albedo em diferentes tipos de uso e cobertura da terra no sudoeste da Amazônia
No full textThe Amazon region has been focus of attention due to the effects that large-scale deforestation can cause on local, regional and global climate. Several field experiments have been conducted in this region to obtain information related to energy exchange between land surface and atmosphere. However, these experiments are concentrated in a few test sites due to the high cost involved in the data acquisition. The present study aimed to estimate the albedo under different land use and land cover types for two dates (29/07/2002 and 01/08/2003), using data from ASTER/Terra sensor, for an area located in the south-western part of the Brazilian Amazon. ASTER/Terra images, referring to the product of surface reflectance (AST07XT), were used and the albedo was calculated according to the algorithm developed by Liang (2001). The estimates of albedo showed to be consistent with the spatial distribution and are in agreement with values found in other studies using orbital data in similar land use and land cover types. Moreover, albedo values obtained for homogeneous areas of forest and pasture were consistent with meteorological studies performed under these conditions in the Amazon region. The proposed methodology allowed to estimate and to analyse the surface albedo in the Amazon with a fine spatial resolution (15 m), that may work as an alternative methodological approach for studies related to the biosphere-atmosphere interaction in this region.Pages: 8706-871
Dados ASTER/Terra aplicados à estimativa do balanço de radiação na Amazônia
No full textCurrently there is a great concern about deforestation of tropical areas, particularly in the Amazon, and its influence on climate. Field experiments have been conducted involving continuous collection of data about the energy exchange between the land surface and the atmosphere in the Amazon region. However, the measurements obtained by these experiments are generally representative of small areas. In this context, the present study aimed to determine the net radiation in an area located in the central-eastern part of Rondônia state (RO) using ASTER data through the SEBAL model. To implement this model, air temperature data, measured in a micrometeorological tower in the scope of LBA Project located in Jaru Biological Reserve (RO), and ASTER images relating to reflectance and surface temperature products (AST07XT and AST08) were used. Then it was possible to verify how much and how the replacement of tropical forest by pasture can modify the surface energy fluxes, showing the potential of the methodology in assessing the impact, in different climatic variables, caused by land use and land cover changes. The use of ASTER data and SEBAL model in the Amazon, where the cloud cover and the large-area are problematic, allowed the estimation and spatialization of radiation balance in a systematic way in a region where most of the information is generated locally.Pages: 8714-872
Análise comparativa de algoritmos de árvore de decisão do sistema WEKA para classificação do uso e cobertura da terra
No full textIn the last years, the data mining techniques are increasingly used for classification purposes, and between several techniques it is highlighted the decision tree. This tool improves the accuracy of classification, and also allows the integration of different data types in the classification. Thus, this work has as main objective to analyze and compare the best data mining algorithm of decision tree available in WEKA software to use for land use and land cover classification in the Tapajos National Forest region. For this, we used a Landsat-5/TM image, the fraction images obtained by the Linear Spectral Mixture Model, the Normalized Difference Vegetation Index, the Normalized Water Index and the Soil-Adjusted Vegetation Index as input data for the creation of the decision trees. To define the best algorithm, the total size of the decision tree, the number of leaves, the time taken for the creation of the decision tree the number of pixels correctly classified, the number of incorrectly classified pixels and Kappa were considered. The algorithm that presented the best results and that best described the classes of land use and land cover of the study area was SimpleCart algorithm, that is an implementation of the Classification and Regression Tress algorithm. The decision tree technique showed satisfactory results in the classification of the images and the results were generated quickly, showing the computational efficiency of this technique.Pages: 2353-236
Proposta metodológica para extração de veículos em Imagens de alta resolução: Estudos iniciais
No full textThis paper aims to identify and eliminate high resolution images concerning vehicles on urban roads. The use of high-resolution images in urban areas has increased significantly in the past years. Due to the large amount of information contained in these images, the classification process may lead to less accurate results, because vehicles are present many similarities to other spectral response patterns. Therefore, we have used an unsupervised classification using fuzzy ARTMAP technique to identify the roads and to remove the vehicles. Subsequently, the original brightness value of the image was recovered and it was calculated the value of the pixel needed to fill the places where there were vehicles taking into account the average of non-zero pixels present in a 3x3 mask that ran from the image. The results showed that the fuzzy ARTMAP classifier was effective for recognizing the vehicle and that the filing of pixels by the proposed methodology was satisfactory. Other studies to be performed, lie in the fact of using automatic segmentation techniques for roads reconstruction of the original image, since it does not always have a file containing road network in vector format, and often the purpose of buying a picture high resolution is to obtain these pathways. Some tests could be performed, obtaining a good way of individualization, however, it is still necessary to integrate these segmented images in rebuilding roads. Further studies are recommended, especially concerning technical replacement of removed pixels.Pages: 8106-811
Análise do novo Código Florestal em relação a Áreas de Preservação Permanentes para a mesorregião Sul/Sudoeste de Minas Gerais
No full textThe present study aimed at mapping some categories of Areas of Permanent Preservation (APP) for the regions of Sul/Sudoeste in Minas Gerais State, based on current Forest Code and new laws to be elaborated in the Brazilian Congress, in the context of dairy chain. From the altimetric information from MDE, it was possible to extract morphological and morphometrical data to estimate the areas of APP of hill tops and along the margins of rivers or drainage in this case. The microregions with the largest area of APP along hydrography were Varginha in both current and new environment laws, with 378 and 181 km², respectively, and in relation to APP of hill top were Pouso Alegre and Passos considering current and new Forest Code, with 1,038 and 235 km², respectively.Pages: 4792-479
Caracterização e dinâmica do uso do solo em área de influência direta do município de Conceição do Araguaia, Pará: o caso da sub-bacia do córrego São Domingos
No full textThe São Domingos stream, a tributary of the Araguaia River, establishes itself as direct area of influence of urban and rural municipality of Conceição do Araguaia, Pará. Thus, this research sought through Geotechnologies as techniques of Remote Sensing and GIS to characterize and verify the dynamics of land use in the sub-basin of the stream São Domingo. Thus, we conducted research library collections and the acquisition of analog and digital products (topographic maps, digital images, cartographic files) that could be processed with the help of ArcGIS 9.3 software. Thus, morphometric variables were obtained from the sub-basin studied, as well as thematic maps of slope and hipsometria, beyond the temporal evaluation (2000-2011) of spatial changes occurring in the same soil.Pages: 5016-502
Influência de padrões morfométricos sobre as ocorrências de inundações na bacia do Córrego Cobiça, Cachoeiro de Itapemirim-ES
No full textWatershed are relief features bounded by topographic divides and drained internally by a main river and its tributaries, this criterion that enables several studies of the physical medium. The geometric characteristics of a basin can influence the runoff and drainage, which justifies its evaluation based on correlation between measurements (morphometry). Rounded Watershed can be fully covered by rainfall and concentrate water in less time, the branching may suffer peaks between flow and concentration and may present elongated profile and balance between flood flows. We carried out the delimitation and morphometric characterization of the Cobiça Stream watershed by applications of Geographic Information System (ArcGIS 9.1\uae), followed by the correlation between the data and parameters identified with the history of flooding in the area. The digital database used consisted of hydrography and contour mapped and orbital images (HRC-CBERS 2B). The geometry and the morphometric indexes (Kc: 1.37; F: 0.55; Ic: 0.534 and Dd: 1.77km/km²) indicate that the basin has high circularity, poor drainage and flow concentration trend. It is estimated that the occurrences of floods in this watershed are influenced by the contrasts of relief and lithology, which intensifies the stream headwaters and reduces the flow of the central portion toward the mouth. The increases in the level of the outflow channel may also influence the flow.Pages: 5769-577
Avaliação da cobertura vegetal por meio dos índices de vegetação SR, NDVI, SAVI e EVI na sub-bacia do vale do rio Itapecerica, alto São Francisco, em Minas Gerais
No full textSince the 1960s, the scientific community is working hard in development of techniques applied to remote sensing products that are able to monitor the vegetation in large scale from the use of images, especially orbitals. Vegetation indices are important digital image processing techniques that aim to reduce the amount of spectral data and enhance the spectral contribution of green vegetation, as well as reducing the contribution of soil, sun angle, senescent vegetation and the atmosphere. Thus, the objective of this study is to evaluate the vegetation cover in the Itapecerica river valley sub basin located in Minas Gerais state, Brazil, using the vegetation indices SR, NDVI, SAVI and EVI applied to TM/Landsat 5 image. The results showed that indices produced reasonable distinction between the defined categories of vegetation. The SR index presented few problems for some areas due to atmospheric interference and soil. The NDVI presented some confusion in the identification of exposed soils and soils with little or no vegetation. Among all the evaluated indices, the best performance can be attributed to SAVI, probably due to soil conditions in the basin, which apparently has a great influence on reflectance values. The EVI presented better separation of classes in densely vegetated categories, showing behavior similar to SAVI, presenting class boundaries very suchlike and equivalent performance in the identification of vegetation cover. The general results showed that the area covered by vegetation can be estimated to an acceptable level, although the limitations of TM/Landsat 5 image properties.Pages: 1472-147