Der Amazonas befindet sich im Wandel. Seine Regenwälder sind zunehmend durch die expandierende Landwirtschaft bedroht. Brandrodungen und die meist extensive Weidewirtschaft verantworten großflächige Ökosystemschäden und hohe Treibhausgasemissionen. Erdbeobachtungssysteme wie die Landsat und Sentinel-2 Satelliten ermöglichen eine großflächige Analyse dieser Entwicklungen und sind unerlässlich zur Evaluierung von Maßnahmen zum Schutze des Amazonas. Allerdings sind in den Kerntropen Fernerkundungsanalysen aufgrund des Bewölkungsgrades sehr herausfordernd. Diese Arbeit zielt daher auf eine verbesserte Erkennung landwirtschaftlicher Prozesse, wie sie an Entwaldungsfronten und speziell in der Region Novo Progresso, Pará, Brasilien, typisch sind. Dazu wurde zunächst der EO Time Series Explorer entwickelt, um verschiedene Dimensionen dichter Multisensorzeitserien interaktiv zur Erstellung von Referenzdaten in Wert zu setzen. Mit den Clear Observation Sequences (COS) wurde darauf basierend ein neuer Ansatz zur Erfassung hoch-dynamischer landwirtschaftlicher Prozesse entwickelt, etwa Feuer mit geringer Brandlast oder Bodenbearbeitungsmaßnahmen. Darauf aufbauend wurde schließlich der Landnutzungswandel in der Region Novo Progresso zwischen 2014 und 2020 untersucht. Die Ergebnisse zeigen einen alarmierenden Anstieg der Entwaldung und eine Zunahme landwirtschaftlicher Feuer seit der Präsidentschaft von Jair Bolsonaro. Differenziert nach Landnutzungszonen und Betriebsgrößen wird deutlich, dass Schutzgebiete weniger wirksam sind und insbesondere größere Landwirtschaftsbetriebe die Entwaldung vorantreiben. Diese Arbeit zeigt den hohen Wert einer synergetischen Nutzung unterschiedlicher Satellitenzeitserien für die fernerkundliche Analyse landwirtschaftlicher Prozesse. Eine weitere Verdichtung der Zeitserien mit räumlich und spektral höherauflösenden Sensoren bietet weiteres Verbesserungspotential bei der Beschreibung landwirtschaftlicher Dynamiken. / The Amazon is in transition, and its rainforests are increasingly threatened by agricultural expansion. A slash-and-burn agriculture and mostly extensive cattle grazing are responsible for large-scale ecosystem damage and high levels of greenhouse gas emission. Earth observation systems such as the Landsat and Sentinel-2 satellites enable large-scale analysis of these developments and are essential for evaluating measures to protect the Amazon. However, cloud cover makes remote sensing analysis challenging in the core tropics. The present work aims to improve the detection of agricultural processes typical of deforestation frontiers, focusing specifically on the Novo Progresso region, Pará, Brazil. To that end, the EO Time Series Explorer was developed to interactively visualize the different dimensions of dense multi-sensor time series and to create reference data. Based on this software tool, the Clear Observation Sequences (COS) approach was developed to capture highly dynamic agricultural processes such as low-load fires or tillage operations. Finally, the investigation of land-use changes in the Novo Progresso region between 2014 and 2020 shows an alarming increase in deforestation and agricultural fires since Jair Bolsonaro’s accession to the presidency. Analysis by land-use zone and property size shows that protected areas have become less effective and that larger properties are driving deforestation. This work demonstrates the value of synergistic use of satellite time series for remote sensing analysis of agricultural processes. Further densification of time series using higher spatial and spectral resolution sensors promises to further improve the description of agricultural dynamics.

This diploma thesis presents the topic of urban drainage systems. The first theoretical part contains methods and options for urban drainage systems, stormwater management, blue-green infrastructure (BGI) in urban areas and introduction of numerical modelling of sewerage systems. The application of modelling platforms is further described in the feasibility study in the practical part of this thesis. By using the simulation model, the study assesses the hydraulic reliability of the drainage system in the city of Jedovnice in the Czech Republic, identifies hydraulic issues and their causes on the urban drainage system. The second part of the study assesses extra-urban stormwater inflow from fields around the city caused by heavy rainfalls, which causes local flooding in the urban area. The identification as well as proposed solutions and capital expenditures, their comparison and recommendation of the optimal solution are included in the study.

This thesis is aimed at detection of lines of horizontal road markings from a point cloud, which was obtained using mobile laser mapping. The system works interactively in cooperation with user, which marks the beginning of the traffic line. The program gradually detects the remaining parts of the traffic line and creates its vector representation. Initially, a point cloud is projected into a horizontal plane, crating a 2D image that is segmented by a U-Net convolutional neural network. Segmentation marks one traffic line. Segmentation is converted to a polyline, which can be used in a geo-information system. During testing, the U-Net achieved a segmentation accuracy of 98.8\%, a specificity of 99.5\% and a sensitivity of 72.9\%. The estimated polyline reached an average deviation of 1.8cm.

Submitted by Johnny Rodrigues (johnnyrodrigues@ufcg.edu.br) on 2018-05-10T18:06:15Z No. of bitstreams: 1 WOSLLEY SIDNEY NOGUEIRA DE OLIVEIRA - DISSERTAÇÃO PPGSA ACADÊMICO 2018..pdf: 7059892 bytes, checksum: 1ab51771320e5bbd6c88d3c01b4b7aeb (MD5) / Made available in DSpace on 2018-05-10T18:06:15Z (GMT). No. of bitstreams: 1 WOSLLEY SIDNEY NOGUEIRA DE OLIVEIRA - DISSERTAÇÃO PPGSA ACADÊMICO 2018..pdf: 7059892 bytes, checksum: 1ab51771320e5bbd6c88d3c01b4b7aeb (MD5) Previous issue date: 2018-02-22 / Os perímetros irrigados implantados no Estado da Paraíba são considerados uma alternativa econômica bastante rentável, promove a geração de empregos e aumenta a disponibilidade de alimentos. Devido ao manejo inadequado do solo e da água, isso têm causado perdas na qualidade do solo desses perímetros, degradando-os principalmente por salinização. O sensoriamento remoto é uma alternativa tecnológica de baixo custo, boa frequência temporal e possui a capacidade de mapear áreas em processo de desertificação. Essa pesquisa têm por objetivo identificar possíveis áreas afetadas por sais no Perímetro Irrigado de São Gonçalo (PISG), Sousa- PB, por meio de técnicas de sensoriamento remoto. Para esse estudo foi utilizado imagens do satélite LANDSAT 8/OLI (média resolução espacial), órbita 216 / ponto 65 da data de 23/11/2016; imagem do software Google Earth Pro® da data de 29/02/2016 para servir como imagem auxiliar e registros fotográficos das áreas in loco. Realizou-se a técnica de classificação supervisionada, utilizando o SCP (semi- automatic plugin) no software QGIS (Quantum Gis). A aferição da qualidade da classificação se deu por meio da validação cruzada, utilizando de parâmetros estatísticos como a exatidão do produtor (EP), exatidão do usuário (EU), exatidão global (EG) e índice Kappa. A classe área supostamente salinizada (ASS) apresentou EP e EU de 89.15% e 88.88%, respectivamente. O índice Kappa resultou em um valor de 0.8684, a classe ASS foi classificada como sendo de qualidade excelente. A qualidade geral da classificação é avaliada tanto pela EG que apresentou um valor de 0.9350 como pelo índice Kappa geral com valor de 0.9252, sendo valores que representam uma classificação de qualidade excelente. A classe ASS apresentou os maiores valores mínimos e máximos de fator de refletância em todas as bandas da imagem, destacando a banda 6 de valores 0.47 e 0.67, respectivamente. O valor da área classificada como sendo da classe ASS foi de 1736.75 hectares, 31% da área total do PISG. As imagens analisadas possibilitaram discriminar áreas salinizadas e não salinizadas mediante as diferenças de tonalidade e de refletância. As imagens analisadas com o plugin SCP possibilitaram a realização de um mapa de classificação supervisionada, indicando a variabilidade espacial das áreas propícias ao processo de salinização. No entanto, recomenda- se a análise dos parâmetros físicos e químicos do solo dessas áreas para o aumento da confiabilidade na qualidade desse tipo de mapeamento. / The irrigated perimeters implemented in the State of Paraiba are considered a costeffective alternative quite profitable, promotes the generation of jobs and increases the availability of food. Due to inadequate management of soil and water, that have caused losses in soil quality of these perimeters, degrading them mainly by salinization. Remote sensing is an alternative low-cost technology, good temporal and frequency has the ability to map areas in process of desertification. This research aim to identify potential areas affected by salts in the irrigated perimeter of São Gonçalo (PISG), Sousa-PB, through remote sensing techniques. For this study we used LANDSAT satellite images 8/OLI (average spatial resolution), 216/orbit point 65 of 07/11/2016 date; image of the Google Earth Pro software® from date of 29/02/2016 to serve as auxiliary image and photographic records of the areas on the spot. The supervised classification technique, using the SCP (semi-automatic plugin) in software QGIS (Quantum Gis). The measurement of the quality of the classification took place by means of cross-validation, using statistical parameters such as the accuracy of the producer (EP), accuracy of the user (EU), global (EG) accuracy and Kappa index. The area class supposedly salinated (.ASS) presented EP and I of 89.15% and 88.88%, respectively. The Kappa index resulted in a value of .ASS class 0.8684 was classified as being of excellent quality. The overall quality of the classification is assessed both by EG who presented a 0.9350 value as the Kappa index 0.9252 valued General, being values that represent a rating of excellent quality. The class ASS presented the largest minimum and maximum values of reflectance factor in all the bands in the image, highlighting the band 6 0.47 values and 0.67, respectively. The value of the area classified as being of .ASS class was 1736.75 acres, 31% of the total area of the PISG. The images reviewed discriminate salinated areas and not allowed saline through the variations of shade and reflectance. The images analyzed with the SCP plugin enabled the creation of a map of supervised classification, indicating the spatial variability of the areas prone to salinization process. However, it is recommended that the analysis of the physical and chemical soil parameters of these areas for increased reliability in the quality of this type of mapping.