National Institute for Space Research
Biblioteca Digital da Memória Científica do INPENot a member yet
6175 research outputs found
Sort by
Classificação de imagens orbitais para a análise de áreas úmidas de altitude: uma ferramenta para a conservação
No full textWetland environment zones have great ecological interest whether permanently or eventually flooded. Its estimated that about 20% of South America territory is covered by wetlands. Although mapping is a preliminary step to the preservation process, its delimitation is still imprecise. The Santa Catharina South Plateau presents wetlands occurring intermingled with native meadows localized at 1000 meters of altitude. The aim of this study was mapping and analyzing the metric of wetlands by means of a Landsat-5 and SPOT-4 images. The research area involves 8174 ha, localized at south of Lages City (between 28°07- 28°10 S and 50°30 50°22 W). After the pre-processing, Minimum Noise Fraction (MNF) transformation was applied to 3, 4 and 5 bands of the Landsat image; the images were classified with Maximum Likelihood method. The vectors generated by the semi-automatized classification were used to analyze the size, form, density and distribution of the wetlands. The permanent preservation areas (APP) were calculated according to State Laws. The SPOT image showed similar results comparing with Landsat image. Wetlands presented random distribution; areas varying between 0,02 65 ha (Landsat-5) and 0,1 35 ha (SPOT-4); concentration of polygons smaller than 0,5 ha; APP occupying about 30% of total area. Therefore, the studied wetlands represented an expressive aspect of the landscape, in terms of the occupied area and also for their frequency and polygons distribution.Pages: 3412-341
Comparação de classificadores para o mapeamento de culturas agrícolas anuais em Campo Novo do Parecis - MT, utilizando NDVI/MODIS
No full textThis study aims to evaluate the performance of four classifiers (MaxVer, Decision Tree, SVM and Neural Networks) in mapping areas of annual crops in Campo Novo do Parecis - Mato Grosso, using NDVI/MODIS time series data for 2010/11 season. Parameters used were selected from the time series of 23 images of the MODIS/NDVI. The classifiers MaxVer, Decision Tree, SVM and Neural Networks showed very similar values of accuracy, 0.927, 0.971, 0.930 and 0.927, and Kappa, 0.824, 0.805, 0.829 and 0.825, respectively. The SVM and MaxVer classifiers presented the characteristics most balanced among the classes of annual crops and other uses. Decision Tree is a classifier more aggressive in areas of annual crops and SVM more conservative for these areas.Pages: 742-74
Correlação entre estimativas de proporção de café em imagens-fração de diferentes resoluções espaciais
No full textCoffee is an important product of national agriculture. Minas Gerais State responds currently to 52% of the Brazilian coffee area and the South and Midwest region of the State are the greatest producers. Given the importance of the coffee crops for the Brazilian economy, we must develop and improve methodologies for their monitoring. Remote sensing is a powerful tool which can be used for many agricultural purposes. In this context, the objective of this study was to verify the correlation between the proportion values of coffee obtained by Linear Spectral Mixture Model applied to low and medium spatial resolution images. The study area was Monte Santo de Minas municipality, located in the South part of Minas Gerais State. Images from LISS III, TM and MODIS sensors of rainy and dry seasons were used. Fraction-images of soil, coffee and shadow/water were obtained through the Linear Spectral Mixture Model. The proportion values of coffee were extracted to 414 Modis pixels selected. We also extracted the mean proportion values of coffee of pixels from LISS III and TM in order to correlate them with the values of each corresponding Modis pixels. The results showed that the correlation coefficient was higher in the dry season than in the rainy season. Thus, we conclude that, despite the low spatial resolution, Modis data can be used to estimate the proportion of coffee contained in a pixel, considering the strong correlation with data from others sensors with better spatial resolution.Pages: 608-61
CompPlexus: programa para avaliação de complexidade de padrões em imagens multiespectrais de sensores remotos
No full textLandscape metrics are widely used in remote sensing to evaluate attributes and patterns of targets of interest. To be accepted and easily adopted, a landscape metric must have a consistent theoretical background and , at the same time, be available on an interface which permits to apply this measure without require very specific knowledge about how to implement it in researchers routines of generation and analysis of data. This paper discusses two landscape metrics based on informational entropy that can be used to evaluate the complexity (in sense of heterogeneity) of multispectral images patterns. Also, it is presented CompPlexus, a software designed to facilitate the use of such metrics. The software was tested in an ASTER multispectral sensor image covering the area of Itirapina Ecologic Station (municipality of Itirapina, state of São Paulo, Brazil) and surroundings, where were choosed for the implementation of the measures of complexity areas with more homogeneous patterns and others with heterogeneous patterns. The results had showed that both measures were well-succeeded in differentiating homogeneous and heterogeneous areas. The two metrics also had allowed to locate the position of each pattern within the order-disorder gradient, enabling the evaluation of its level of organization. CompPlexus software was an accurate and easy-to-use tool for the calculation of the complexity of multispectral images patterns.Pages: 6727-673
Análise morfométrica da bacia hidrográfica do córrego Tiradentes no município de Rio Bananal (ES)
No full textThe objective of this study was to analyze the morphology of a watershed based on the calculation of parameters targeted watersheds with support of Geographic Information System (GIS) and remote sensing data. The basin is located in the north-central State of Espirito Santo, Brazil. The analyses were done with the software Spring 5.2 and ArcGIS 10 using the 90 resolution digital elevation model of the study area obtained from the Shuttle Radar Topography Mission (SRTM) The parameters measured were: total number of channels, total number of channels in order, total length of the drainage network (km), total length of channels in order (km), average length of the channel order (km), basin area (km²), basin perimeter (km), length of the main river (km), bifurcation ratio, sinuosity index (%), ratio of the average length of each channel orders (km), basin order, basin form factor, river density, drainage density (km/km²), altimetric amplitude (m), relief ratio (m /km), roughness index, asymmetric factor, coefficient of compactness, and circularity index. The results showed that the Tiradentes stream basin has an area of 67.03 km ², low hydrology (Dh = 0.31 km), with low drainage density (Dd = 1.13 km /km ²). It has fast flow (Hm = 442.22 m; Rr = 28.38 m /km; Ir = 500.75), with steep slopes, the terrain relief is rugged and asymmetric (Af = 21.78).Pages: 5919-592
Avaliação de técnicas de classificação automática de dados multi-polarimétricos na banda-L do sensor R99B-SAR para o mapeamento de áreas inundadas do Lago de Coari, Amazônia Central
No full textStudies in Central Amazonia using remote sensing data can contribute to an understanding on a regional scale of its physiographic characteristics, providing support for the preparation of maps depicting the sensitivity to oil spills of the complex ecosystems existent in the region. The study area herein reported is remote, difficult to access, and permanently cloud-covered. In addition, water level variation in the drainage basin can reach as much as 17 meters between wet and dry seasons. Therefore, it is necessary to map the cover types most sensitive to oil spills based on image datasets suitable to portray such a seasonal change. In this context, the present paper used the algorithm USTC (Unsupervised Semivariogram Textural Classifier), complemented by object-based segmentation and classification techniques, to process digitally calibrated L-band images acquired by the Multipolarimetric R99B-SAR system. These data were obtained in the region of Coari (AM) as part of the mission entitled Multi-Application Purpose SAR (MAPSAR). The Brazilian-German MAPSAR mission is a proposal for a light L-band SAR sensor, based on INPE´s Multi-Mission Platform (500 kg class spacecraft). Application of the USTC algorithm in defining super classes for object-based classification constitutes an innovative approach for digital processing of SAR data. To analyze and compare the accuracy of results of USTC and object-based classification, we used the confusion matrix (error matrix) and Kappa index. Research results enhanced macrophyte stands and flooded forests, which are the cover types most sensitive to oil spills in the fluvial scenario of Central Amazonia.Pages: 8397-840
Uso e cobertura da terra na região de Ouro Preto do Oeste, Rondônia, mediante imagens ASTER
No full textRemote sensing plays a key role in monitoring land use and land cover in the Amazon region, since it allows obtaining current and historical information about a large environment and of difficult access. Studies and projects have been developed to map land use and land cover in the region, using different methods and orbital sensors. The aim of this study was to analyse the temporal dynamics of land use and land cover in the region of Ouro Preto do Oeste, state of Rondônia, using ASTER/Terra images and supervisioned digital classification using Bhattacharya algorithm. For that, two ASTER images of the days 29/07/2002 and 01/08/2003 referring to the product of surface reflectance (AST07XT) were used. The classification was processed in the software SPRING 5.0.6 and involved steps of image segmentation and collection of test and training samples. The results showed intense degradation due to deforestation in the region. It was verified that in the study area primary rainforest and pasture land represent 9% and 70%, respectively. Moreover, during the study period, there was suppression of 500.96 ha (≈ 6%) of the primary rainforest areas. The ASTER images have potential for land use and land cover studies in the Amazon region due especially to its finer minimum spatial resolution (15 m) than those sensors commonly used in the region.Pages: 7273-728
Geoprocessamento e Biodiversidade: contribuições para a modelagem da distribuição de palmeiras Amazônicas
No full textThis work contributes to the knowledge of the Amazonian biodiversity, specifically regarding the climate and environmental characteristic that have influenced the geographical distribution of the Brazilian Amazoni palms. The main goal was to highlight the importance of both remote sensing (RS) data to characterize the palm ecological niche, and geoprocessing tools to obtain and validate distribution maps from potential species distribution modeling (SDM). We selected 21 palm species typical of the Brazilian Amazon that contained at least 10 occurrence points available to generate its SDMs considering the current climate. For each species, from the environmental variables available for the Amazon we first identified a unique subset of variables, according to the species ecology and physiology, avoiding auto-correlation. Applying binary thresholds over the SDMs, we produced the maps of palm species occurrence. Overlaying historical distribution maps with the resulting modeling maps, only two species did not match with the literature reference. A map of palm species richness was the result of summing individual species occurrence maps, depicting the richness distribution along endemism regions in Amazon. Variables derived from remote sensing were essential to characterize the physical environment, especially the topography and soil moisture, both from the SRTM data. Geoprocessing procedures such as reclassification, georeferencing, overlaying, and algebra over maps were essential to characterize the spatial distribution of biodiversity. Even though we have only tested 21 species from 189 Amazon palms, our results evidenced the dependence of SDM approaches on RS data and geoprocessing procedures to characterize the spatial distribution of species diversity.Pages: 6767-677
Processamento de linhas de bases curtas pelo método relativo utilizando efemérides transmitidas e precisas
No full textIn high precision positioning, its necessary that systematic errors inherent in several spatial positioning system, as the Global Navigation Satellite System, be minimized or even eliminated. Among them, stands out orbital errors, which occurs because of the variations in the reference: while the satellites coordinates are determined in relation to a fixed reference system (inertial), the coordinates of the stations located on the earth surface are subject to variations caused by the rotation of the Earth. Thus, errors present in the satellite coordinates are propagated directly to the users position. This fact can be noticed by the processing and adjustment of Global Positioning System datas, analyzing the variations caused by the choice of using either satellites orbits pos-processed (precise ephemeris) or orbits transmitted in real time (broadcast ephemeris).In order to study these discrepancies, 13 stations of the local GNSS lan, located in Federal University of Uberlândia, Santa Mônica campus, were surveyed with GPS receivers of single frequency (L1), considering MGUB (Uberlândia, MG), station belonging to the Brazilian network for continuos monitoring, as the base, referenced to SIRGAS2000 (epoch 2000,4). Two strategies have been adopted in the datas processing: one applying broadcast ephemeris and other accurate ephemeris of the Internacional GNSS Service. It was verified that the variations between the results obtained with broadcast and accurate ephemeris are in the milimeter order for the vertical component, and in the centimetric order for horizontal components. These results match with the values found in the literature and with the INCRA Technical Standard.Pages: 4692-469
Espectrometria de raios gama em metassedimentos e rochas gnáissicas na região de Cavalcante, Goiás
No full textIn this work, airborne gamma-ray spectrometry and digital elevation modeling were used for geological mapping. The integration of the geophysical and elevation data, using a statistical procedure, the cluster analysis, allowed to produce maps of statistical groups that have a good correlation with the geological units of Cavalcante - Goiás area, in the Middle-West of Brazil. In the study area, paleoproterozoic granite-gneiss basement and two proterozoic sedimentary units, Araí and Paranoá Groups, are founded. Shuttle Radar Topography Mission (SRTM) data were used for elevation modeling. Seven radiometric images were produced. The U, Th and K images were selected because showed better visual correlation with geological map than Total Count, Th/K, U/K and U/Th images. The cluster analysis using four clusters and four variables was performed with U, Th, K and topographic data. The cluster map is the visual representation of a statistical procedure which combined geophysical and topographic data from study area. The most important characteristic of cluster map is the presence of two distinctive groups that can be related with radiometric aspects from local rocks. This product showed that higher terrains present high Th and low U. In these areas sedimentary rocks are founded. Lower terrains, with gneissic rocks present low Th and high U.Pages: 3557-356