Spelling suggestions: "subject:"atatistical metrics"" "subject:"astatistical metrics""
1 |
Mapeamento de soja e milho com mineração de dados e imagens sintéticas landsat e modis / Mapping of soybean and corn with data mining and synthetic images Landsat and MODISOldoni, Lucas Volochen 05 February 2018 (has links)
Submitted by Rosangela Silva (rosangela.silva3@unioeste.br) on 2018-06-04T17:12:56Z
No. of bitstreams: 2
Lucas Oldoni.pdf: 9472745 bytes, checksum: 1b2c1a8ee59169fa471b43d27a762f6e (MD5)
license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Made available in DSpace on 2018-06-04T17:12:56Z (GMT). No. of bitstreams: 2
Lucas Oldoni.pdf: 9472745 bytes, checksum: 1b2c1a8ee59169fa471b43d27a762f6e (MD5)
license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5)
Previous issue date: 2018-02-05 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Studies related to the monitoring of agricultural production play a decisive and strategic role in the economic planning of the country, due to the importance of agribusiness, as well as food safety. Orbital remote sensing is an effective alternative to perform agricultural crop monitoring due to its low cost, large scale and speed of data collection. However, most of the sensors with high spatial resolution are of low temporal resolution, and the ones with higher temporal resolution have low spatial resolution. Therefore, for the monitoring of agricultural crops with a higher spatial solution, cloud covering can be a limiting factor. Such problems can be circumvented by using a fusion of images of several sensors with different spatial and temporal characteristics, thus creating new images, also called synthetic images. Thus, the objective of the work was the mapping of areas sown with soybean and corn using space-temporal fusion, such as Landsat 8 and MODIS images. In the first part of the research, agricultural crops were separated from other targets. The generated classification served as input to one of the classification algorithms, the Flexta Spatiotemporal Data Fusion (FSDAF), in the second part of the research. In addition to this algorithm, both the Spatial and Temporal Adaptive Reflection Fusion Model (STARFM) and the Advanced and Temporal Spatial Adaptive Reflection Fusion Model (ESTARFM) were employed to generate images for the 2016/2017 summer crops. Then, 5 rating scenarios were created. In the 1st and 2nd scenarios, only the images from the Landsat 8 with no occurrence of clouds were considered. For the 3rd, 4th, and 5th were carried out using images generated by STARFM, ESTARFM and FSDAF. In the third scenario, the metric images of images, Landsat 8 and images of fusion algorithms were used, 4th as a summary of statistical metrics, and in the 5th one as phenological metrics of the temporal profile of the Enhanced Vegetation Index (EVI). The scenario using the EVI phenological metrics from images generated by FSDAF and STARFM yielded better results, with global accuracy of 93.11 and 91.33%, respectively. These results are statistically better than those obtained using only existing Landsat 8 images. Thus, the use of phenological metrics obtained from synthetic images are important alternatives for mapping soybean and corn crops. / Estudos referentes ao acompanhamento da produção agrícola têm um peso determinante e estratégico no planejamento econômico do país, devido à importância do agronegócio, e também para segurança alimentar. O sensoriamento remoto orbital é uma alternativa eficaz para realizar o monitoramento das culturas agrícolas, devido ao baixo custo, grande escala de abrangência e rapidez na coleta de dados. Porém, geralmente os sensores com alta resolução espacial possuem baixa resolução temporal, e os com alta resolução temporal possuem baixa resolução espacial. Assim, para se realizar o acompanhamento de culturas agrícolas com uma resolução espacial mais alta, a cobertura por nuvens pode ser um fator limitante. Estes problemas podem ser contornados com a utilização de fusão de imagens de diversos sensores com características temporais e espaciais diferentes, criando, assim, novas imagens, também chamadas de imagens sintéticas. Deste modo, o objetivo do trabalho foi realizar o mapeamento de áreas semeadas com soja e milho utilizando fusão espaço-temporal de imagens Landsat 8 e MODIS. Na primeira parte do trabalho, foram separadas culturas agrícolas de outros alvos. A classificação gerada serviu de entrada em um dos algoritmos de classificação, o Flexible Spatiotemporal Data Fusion (FSDAF), na segunda parte do trabalho. Nessa parte, além deste algoritmo, também foram utilizados os algoritmos Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM) e Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model (ESTARFM) para gerar imagens nas safras de verão 2016/2017. Então, foram criados 5 cenários de classificação. Nos 1º e 2º foram considerados a utilização apenas das imagens espectrais das imagens Landsat 8 livres de nuvens. As 3º, 4º e 5º foram realizadas com as imagens geradas pelo STARFM, ESTARFM e FSDAF. No 3º cenário foram utilizadas as métricas espectrais das imagens Landsat 8 e as imagens espectrais gerados pelos algoritmos de fusão, no 4º foram considerados as métricas estatísticas e no 5º as métricas fenológicas extraídas do perfil temporal do Enhanced Vegetation Index (EVI). Os cenários que utilizaram métricas fenológicas do EVI a partir de imagens geradas pelo FSDAF e STARFM obtiveram melhores resultados, com exatidão global de 93,11 e 91,33%, respectivamente, resultados estes estatisticamente melhores que os obtidos apenas com as imagens Landsat 8 existentes. Assim, a utilização de métricas fenológicas obtidas de imagens sintéticas são importantes alternativas para o mapeamento de soja e milho.
|
2 |
Etude par modélisation numérique de la qualité de l’air en Europe dans les climats actuel et futur / Numeral modeling study of European air quality in current and future climatesLacressonnière, Gwendoline 19 December 2012 (has links)
Cette thèse porte sur l’étude de l’évolution de la qualité de l’air en Europe et en France dans les prochaines décennies à l’aide de simulations numériques. Dans les études des impacts de l’évolution du climat sur la qualité de l’air, les modèles de chimie atmosphérique utilisent des sorties de modèles climatiques globaux ou régionaux qui fournissent les « forçages », c’est-à-dire les conditions météorologiques simulées pour les périodes futures. Contrairement aux analyses météorologiques, qui représentent la variabilité jour à jour du temps, les sorties des modèles de climat doivent nécessairement être interprétées de manière statistique : elles ne représentent la météorologie que dans un sens climatologique. Afin de pouvoir commenter utilement les simulations futures de qualité de l’air, il est nécessaire d’évaluer au préalable et pour le climat présent, la qualité des simulations calculées avec des forçages climatiques par rapport aux références que constituent les simulations calculées avec des forçages analysés et, bien entendu, les observations. Trois simulations pluri-annuelles (6 ans) ont été lancées pour la période actuelle (2000-2010) et ont été comparées ; elles différent par l’utilisation d’analyses météorologiques ou de forçages de modèle de climat (pour les paramètres atmosphériques seuls et par ailleurs, pour les paramètres atmosphériques et le calcul des échanges en surface) en entrée du modèle de chimie-transport tridimensionnel de Météo-France, MOCAGE. Nous avons évalué ces différentes simulations par comparaison aux observations de la base de données européenne AirBase. Nous avons ensuite comparé les performances de ces simulations pour un grand nombre de scores quantitatifs, en analysant d’une part les effets liés aux champs météorologiques (température, vent, humidité, etc.) et d’autre part, ceux liés aux échanges en surface (comme les vitesses de dépôts, les émissions biogéniques) qui dépendent également de la météorologie. Nous avons ainsi évalué comment ces changements affectent les distributions horizontales et verticales des polluants. In fine, nous avons caractérisé la fiabilité des simulations de qualité de l’air reposant sur des forçages issus de modèles climatiques pour le climat présent : des indicateurs (biais moyens, biais moyens normalisés, RMSE, déviations standards) et des index de qualité de l’air (comme le dépassement de seuils) se distinguent et peuvent donc servir de base fiable pour l’interprétation des résultats pour les simulations du futur. Enfin dans une troisième partie, ces indicateurs considérés comme pertinents ont été utilisés pour étudier des simulations de qualité de l’air aux horizons 2030 et 2050 (5 ans). Comme attendu, l’évolution des paramètres météorologiques (température, précipitation, vent) modifie les quantités et la dispersion des polluants dans l’atmosphère, mais l’évolution des émissions en Europe et dans le reste du monde joue aussi un rôle important. Ainsi, face à l’évolution du climat et la hausse des émissions dans certains pays du monde, en Asie notamment, les effets des politiques Européennes pour réduire les émissions anthropiques sont mitigés selon les régions et les polluants, dépendant de l’influence relative des phénomènes locaux et du transport de polluant à longue distance. / This thesis aims at predicting how European and French air quality could evolve over the next decades using numerical modeling. In order to study the impacts of climate change up on regional air quality, atmospheric chemistry models rely on global or regional climate models to produce “forcings”, i.e. meteorological conditions for future periods. Unlike meteorological analyses, which can represent specifically each date and hour thanks to the assimilation of observations, climate model outputs need to be averaged and can only be interpreted in a climatological sense. And so are air quality hindcasts relying on them for their forcings. In order to properly interpret air quality simulations in a future climate, it is a pre-requisite to assess how realistic air quality hindcasts are when driven by forcings from climate models for the current period in comparison to the references, which are simulations with the same set-up but relying on meteorological analyses and observations. Three six-year simulations for the current climate (2000-2010) have been run with the three-dimensional chemistry transport model of Météo-France, MOCAGE. These simulations only differ by the meteorological forcings used, either operational meteorological analyses or outputs from climate simulations (for atmospheric parameters only ; for atmospheric parameters as well as surface exchanges, which depend also on the weather). We compared the three simulations and evaluated them against the European air quality database of the European Environment Agency, AirBase. Further, we investigated how statistical skill indicators compare in the different simulations, assessing the effects of meteorology on atmospheric fields (temperature, wind, humidity,...) and on the dependent emissions and deposition processes (such as deposition velocities, volatile organic compound emissions, ...) that depend upon meteorology. We have in particular studied how these factors affect the horizontal and vertical distributions of species. In the end, we have estimated how reliable are skill indicators for the simulations run with “climate” forcings : some indicators (mean bias, mean normalized bias, RMSE, deviation standards, number of exceedance days) are sufficiently close to the ones obtained with the reference configuration (relying on analysed meteorological forcings) to be considered reliable. They can thus be used to interpret simulations for future periods. We have run simulations of European air quality in the 2030s and 2050s (5 years for each period). They are discussed using the indicators previously indentified. As expected, the changes in meteorological parameters (temperature, precipitation, wind, ...) affect the quantities and distributions of pollutants in the atmosphere, but the future evolutions in European and global emissions also play a significant role. Faced with climate change and increased emissions in some countries in the world, as in Asia, the impacts of European policies for reducing anthropogenic emissions are mitigated, depending on the regions and the pollutants due to the respective influence of local emission and of long-range transport of pollutants.
|
3 |
Uma an?lise comparativa entre as abordagens lingu?stica e estat?stica para extra??o autom?tica de termos relevantes de corporaSantos, Carlos Alberto dos 27 April 2018 (has links)
Submitted by PPG Ci?ncia da Computa??o (ppgcc@pucrs.br) on 2018-07-26T19:48:07Z
No. of bitstreams: 1
CARLOS ALBERTO DOS SANTOS_DIS.pdf: 1271475 bytes, checksum: 856ae87ad633d3c772b413816caa43d1 (MD5) / Approved for entry into archive by Sheila Dias (sheila.dias@pucrs.br) on 2018-08-01T13:39:36Z (GMT) No. of bitstreams: 1
CARLOS ALBERTO DOS SANTOS_DIS.pdf: 1271475 bytes, checksum: 856ae87ad633d3c772b413816caa43d1 (MD5) / Made available in DSpace on 2018-08-01T14:31:21Z (GMT). No. of bitstreams: 1
CARLOS ALBERTO DOS SANTOS_DIS.pdf: 1271475 bytes, checksum: 856ae87ad633d3c772b413816caa43d1 (MD5)
Previous issue date: 2018-04-27 / It is known that linguistic processing of corpora demands high computational effort because of the complexity of its algorithms, but despite this, the results reached are better than that generated by the statistical processing, where the computational demand is lower. This dissertation describes a comparative analysis between the process linguistic and statistical of term extraction. Experiments were carried out through four corpora in English idiom, built from scientific papers, on which terms extractions were carried out using the approaches. The resulting terms lists were refined with use of relevance metrics and stop list, and then compared with the reference lists of the corpora across the recall technical. These lists, in its turn, were built from the context these corpora, whith help of Internet searches. The results shown that the statistical extraction combined with the stop list and relevance metrics can produce superior results to linguistic process extraction using the same metrics. It?s concluded that statistical approach composed by these metrics can be ideal option to relevance terms extraction, by requiring few computational resources and by to show superior results that found in the linguistic processing. / Sabe-se que o processamento lingu?stico de corpora demanda grande esfor?o computacional devido ? complexidade dos seus algoritmos, mas que, apesar disso, os resultados alcan?ados s?o melhores que aqueles gerados pelo processamento estat?stico, onde a demanda computacional ? menor. Esta disserta??o descreve uma an?lise comparativa entre os processos lingu?stico e estat?stico de extra??o de termos. Foram realizados experimentos atrav?s de quatro corpora em l?ngua inglesa, constru?dos a partir de artigos cient?ficos, sobre os quais foram executadas extra??es de termos utilizando essas abordagens. As listas de termos resultantes foram refinadas com o uso de m?tricas de relev?ncia e stop list, e em seguida comparadas com as listas de refer?ncia dos corpora atrav?s da t?cnica do recall. Essas listas, por sua vez, foram constru?das a partir do contexto desses corpora e com ajuda de pesquisas na Internet. Os resultados mostraram que a extra??o estat?stica combinada com as t?cnicas da stop list e as m?tricas de relev?ncia pode produzir resultados superiores ao processo de extra??o lingu?stico refinado pelas mesmas m?tricas. Concluiu se que a abordagem estat?stica composta por essas t?cnicas pode ser a op??o ideal para extra??o de termos relevantes, por exigir poucos recursos computacionais e por apresentar resultados superiores ?queles encontrados no processamento lingu?stico.
|
Page generated in 0.141 seconds